diff options
| author | AlTheKiller <AlTheKiller@svn> | 2009-09-23 01:49:50 +0000 |
|---|---|---|
| committer | AlTheKiller <AlTheKiller@svn> | 2009-09-23 01:49:50 +0000 |
| commit | 45285e8a9300cd754a760560640b75b09f98035e (patch) | |
| tree | ad9f093885ad5c98e9dd4156674e7691c22ed0a2 /system/scrapers | |
step 3/4: Move linuxport to trunk. How'd I get roped into this?
git-svn-id: https://xbmc.svn.sourceforge.net/svnroot/xbmc/trunk@23097 568bbfeb-2a22-0410-94d2-cc84cf5bfa90
Diffstat (limited to 'system/scrapers')
94 files changed, 7014 insertions, 0 deletions
diff --git a/system/scrapers/music/allmusic.gif b/system/scrapers/music/allmusic.gif Binary files differnew file mode 100644 index 0000000000..30c40ef318 --- /dev/null +++ b/system/scrapers/music/allmusic.gif diff --git a/system/scrapers/music/allmusic.xml b/system/scrapers/music/allmusic.xml new file mode 100644 index 0000000000..9a3a9019f6 --- /dev/null +++ b/system/scrapers/music/allmusic.xml @@ -0,0 +1,84 @@ +<?xml version="1.0" encoding="UTF-8"?> +<scraper framework="1.1" date="2009-09-20" name="Allmusic" content="albums" thumb="allmusic.gif" language="en"> + <include>common/htbackdrops.xml</include> + <include>common/allmusic.xml</include> + <GetSettings dest="3"> + <RegExp input="$$5" output="<settings>\1</settings>" dest="3"> + <RegExp input="$$1" output="<setting label="Enable Fanart" type="bool" id="htbfanart" default="true"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting label="Grab thumbs from HTBackdrops" type="bool" id="htbthumbs" default="true"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetSettings> + <NfoUrl dest="3"> + <RegExp input="$$1" output="\1" dest="3"> + <expression>(http://www.allmusic.com/cg/.*)</expression> + </RegExp> + </NfoUrl> + <CreateAlbumSearchUrl dest="3"> + <RegExp input="$$1" output="<url post="yes">http://www.allmusic.com/cg/amg.dll?P=amg&amp;SQL=\1&amp;OPT1=2</url>" dest="3"> + <expression></expression> + </RegExp> + </CreateAlbumSearchUrl> + <GetAlbumSearchResults dest="8"> + <RegExp input="$$5" output="<results>\1</results>" dest="8"> + <RegExp input="$$1" output="<entity><title>\1</title></entity>" dest="5"> + <expression repeat="yes" noclean="1"><TITLE>allmusic[ (]+ ([^>]*) > Overview</expression> + </RegExp> + <RegExp input="$$1" output="<entity><year>\1</year><artist>\2</artist><title>\4</title><url cache="am-\2-\4-album.html">http://www.allmusic.com/cg/amg.dll?p=amg&amp;SQL=\3</url></entity>" dest="5+"> + <expression repeat="yes" noclean="1,2,3">>([0-9]+)</.*?>([a-zA-Z0-9].*?)<.*?sql=([0-9:a-z]*)">(.*?)<.</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetAlbumSearchResults> + <GetAlbumDetails dest="3"> + <RegExp input="$$5" output="<details>\1</details>" dest="3"> + <RegExp input="$$1" output="<url cache="am-$$2-album.html" function="ParseAMGAlbum">http://www.allmusic.com/cg/amg.dll?p=amg&amp;sql=\1</url>" dest="5+"> + <expression noclean="1">sql=([^"]*)">Overview</expression> + </RegExp> + <RegExp input="$$1" output="<url cache="am-$$2-album.html" function="GetAMGReview">http://www.allmusic.com/cg/amg.dll?p=amg&amp;sql=\1</url>" dest="5+"> + <expression noclean="1">sql=([^"]*)">Review</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetAlbumDetails> + <CreateArtistSearchUrl dest="3"> + <RegExp input="$$1" output="<url post="yes">http://www.allmusic.com/cg/amg.dll?P=amg&amp;SQL=\1&amp;OPT1=1</url>" dest="3"> + <expression></expression> + </RegExp> + </CreateArtistSearchUrl> + <GetArtistSearchResults dest="8"> + <RegExp input="$$5" output="<results>\1</results>" dest="8"> + <RegExp input="$$1" output="<entity><title>\1</title></entity>" dest="5"> + <expression repeat="yes" noclean="1"><TITLE>allmusic[ (]+ ([^>]*) > Overview</expression> + </RegExp> + <RegExp input="$$1" output="<entity><title>\2</title><genre>\3</genre><year>\5</year><url cache="am-\2-artist.html">http://www.allmusic.com/cg/amg.dll?p=amg&amp;SQL=\1</url></entity>" dest="5+"> + <expression repeat="yes" noclean="1,2">style="[^=]*<a href="[^"]*sql=([^"]*)">([^<]*)</a>[^-]*[^>]*>([^<]*)</TD>[^>]*>(&nbsp;)?([^<]*)</td></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetArtistSearchResults> + <GetArtistDetails dest="3"> + <RegExp input="$$5" output="<details>\1</details>" dest="3"> + <RegExp input="$$1" output="<url cache="am-$$2-artist.html" function="ParseAMGArtist">http://www.allmusic.com/cg/amg.dll?p=amg&amp;sql=\1</url>" dest="5"> + <expression noclean="1">sql=([^"]*)">Biography</expression> + </RegExp> + <RegExp input="$$2" conditional="htbthumbs" output="<url function="GetHTBThumbs" post="yes">http://www.htbackdrops.com/search.php?search_terms=all&amp;cat_id=5&amp;search_keywords=\1</url>" dest="5+"> + <expression/> + </RegExp> + <RegExp input="$$2" conditional="htbfanart" output="<url function="GetHTBFanart" post="yes">http://www.htbackdrops.com/search.php?search_terms=all&amp;cat_id=1&amp;search_keywords=\1</url>" dest="5+"> + <expression/> + </RegExp> + <RegExp input="$$1" output="<url cache="am-$$2-artist.html" function="GetAMGBiography">http://www.allmusic.com/cg/amg.dll?p=amg&amp;sql=\1</url>" dest="5+"> + <expression noclean="1">sql=([^"]*)">Biography</expression> + </RegExp> + <RegExp input="$$1" output="<url cache="am-$$2-discog.html" function="GetAMGDiscography">http://www.allmusic.com/cg/amg.dll?p=amg&amp;sql=\1</url>" dest="5+"> + <expression noclean="1">sql=([^"]*)">Discography</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetArtistDetails> +</scraper> diff --git a/system/scrapers/music/allmusic_merlin_lastfm.xml b/system/scrapers/music/allmusic_merlin_lastfm.xml new file mode 100644 index 0000000000..a7b30d7b85 --- /dev/null +++ b/system/scrapers/music/allmusic_merlin_lastfm.xml @@ -0,0 +1,149 @@ +<?xml version="1.0" encoding="UTF-8"?> +<scraper framework="1.1" date="2009-09-16" name="allmusic & merlin.pl & lastfm.pl" content="albums" thumb="allmusic.gif"> + <include>common/htbackdrops.xml</include> + <include>common/allmusic.xml</include> + <GetSettings dest="3"> + <RegExp input="$$5" output="<settings>\1</settings>" dest="3"> + <RegExp input="$$1" output="<setting label="Enable Fanart" type="bool" id="htbfanart" default="true"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting label="Grab thumbs from HTBackdrops" type="bool" id="htbthumbs" default="true"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetSettings> + <NfoUrl dest="3"> + <RegExp input="$$1" output="\1" dest="3"> + <expression>(http://www.allmusic.com/cg/.*)</expression> + </RegExp> + </NfoUrl> + <CreateAlbumSearchUrl dest="3"> + <RegExp input="$$1" output="<url post="yes">http://www.allmusic.com/cg/amg.dll?P=amg&amp;SQL=\1&amp;OPT1=2</url>" dest="3"> + <expression></expression> + </RegExp> + </CreateAlbumSearchUrl> + <GetAlbumSearchResults dest="8"> + <RegExp input="$$5" output="<results>\1</results>" dest="8"> + <RegExp input="$$1" output="<entity><title>\1</title></entity>" dest="5"> + <expression repeat="yes" noclean="1"><TITLE>allmusic[ (]+ ([^>]*) > Overview</expression> + </RegExp> + <RegExp input="$$1" output="<entity><year>\1</year><artist>\2</artist><title>\4</title><url cache="allmusic-album.html">http://www.allmusic.com/cg/amg.dll?p=amg&amp;SQL=\3</url></entity>" dest="5+"> + <expression repeat="yes" noclean="1,2,3">>([0-9]+)</.*?>([a-zA-Z0-9].*?)<.*?sql=([0-9:a-z]*)">(.*?)<.</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetAlbumSearchResults> + <GetAlbumDetails dest="3"> + <RegExp input="$$5" output="<details>\1</details>" dest="3"> + <RegExp input="$$1" output="<url cache="allmusic-album.html" function="ParseAMGAlbum">placeholder</url>" dest="5+"> + <expression/> + </RegExp> + <RegExp input="$$1" output="\1" dest="8"> + <expression noclean="1"><TITLE>allmusic[ (]+ ([^>]*) > Overview</expression> + </RegExp> + <RegExp input="$$9" output="\1" dest="6"> + <RegExp input="$$8" output="+\1" dest="9"> + <expression repeat="yes">([^ ]+)</expression> + </RegExp> + <expression>\+(.*)</expression> + </RegExp> + <RegExp input="$$1" output="\1" dest="8"> + <expression noclean="1"><span>Artist</span>[^;]*[^>]*>[^>]*>([^<]*)</expression> + </RegExp> + <RegExp input="$$9" output="\1" dest="7"> + <RegExp input="$$8" output="+\1" dest="9"> + <expression repeat="yes">([^ ]+)</expression> + </RegExp> + <expression>\+(.*)</expression> + </RegExp> + <RegExp input="$$1" output="<url function="MerlinLink">http://merlin.pl/xbmc/browse/search/4,,1.html?phrase=&amp;place=4+simple&amp;carrier=3&amp;offer=O&amp;category=&amp;title=$$6&amp;person=$$7&amp;firm=&amp;date=&amp;isbn=&amp;sort=rank&amp;x=40&amp;y=12</url>" dest="5+"> + <expression noclean="1"></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetAlbumDetails> + <MerlinLink dest="5"> + <RegExp input="$$2" output="<details>\1</details>" dest="5"> + <RegExp input="$$1" output="<url function="GetReview">http://merlin.pl\1</url>" dest="2"> + <expression>Liczba towar[^f]+f="([^"]+)</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </MerlinLink> + <GetReview dest="5"> + <RegExp input="$$2" output="<?xml version="1.0" encoding="iso-8859-2" standalone="yes"?><details><review>\1</review></details>" dest="5"> + <RegExp input="$$1" output="\1" dest="2"> + <expression>class="productDesc">(.*?)</div></expression> + </RegExp> + <RegExp input="$$1" output="\1" dest="2+"> + <expression>piorka_merlina.*?contents">([^<]*)<</expression> + </RegExp> + <RegExp input="$$1" output="\1" dest="2+"> + <expression>contents">([^<]*)<</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetReview> + <CreateArtistSearchUrl dest="3"> + <RegExp input="$$1" output="<url post="yes">http://www.allmusic.com/cg/amg.dll?P=amg&amp;SQL=\1&amp;OPT1=1</url>" dest="3"> + <expression></expression> + </RegExp> + </CreateArtistSearchUrl> + <GetArtistSearchResults dest="8"> + <RegExp input="$$5" output="<results>\1</results>" dest="8"> + <RegExp input="$$1" output="<entity><title>\1</title></entity>" dest="5"> + <expression repeat="yes" noclean="1"><TITLE>allmusic[ (]+ ([^>]*) > Overview</expression> + </RegExp> + <RegExp input="$$1" output="<entity><title>\2</title><genre>\3</genre><year>\5</year><url cache="allmusic-artist.html" >http://www.allmusic.com/cg/amg.dll?p=amg&amp;SQL=\1</url></entity>" dest="5+"> + <expression repeat="yes" noclean="1,2">style="[^=]*<a href="[^"]*sql=([^"]*)">([^<]*)</a>[^-]*[^>]*>([^<]*)</TD>[^>]*>(&nbsp;)?([^<]*)</td></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetArtistSearchResults> + <GetArtistDetails dest="3"> + <RegExp input="$$5" output="<details>\1</details>" dest="3"> + <RegExp input="$$1" output="<url cache="allmusic-artist.html" function="ParseAMGArtist">placeholder</url>" dest="5"> + <expression /> + </RegExp> + <RegExp input="$$2" conditional="htbthumbs" output="<url function="GetHTBThumbs" post="yes">http://www.htbackdrops.com/search.php?search_terms=all&amp;cat_id=5&amp;search_keywords=\1</url>" dest="5+"> + <expression/> + </RegExp> + <RegExp input="$$2" conditional="htbfanart" output="<url function="GetHTBFanart" post="yes">http://www.htbackdrops.com/search.php?search_terms=all&amp;cat_id=1&amp;search_keywords=\1</url>" dest="5+"> + <expression/> + </RegExp> + <RegExp input="$$1" output="\1" dest="8"> + <expression noclean="1"><TITLE>allmusic[ (]+ ([^>]*) > Overview</expression> + </RegExp> + <RegExp input="$$9" output="\1" dest="6"> + <RegExp input="$$8" output="+\1" dest="9"> + <expression repeat="yes">([^ ]+)</expression> + </RegExp> + <expression>\+(.*)</expression> + </RegExp> + <RegExp input="$$1" output="<url function="LastfmLink">http://ws.audioscrobbler.com/2.0/?method=artist.search&amp;artist=$$6&amp;api_key=71e468a84c1f40d4991ddccc46e40f1b</url>" dest="5+"> + <expression noclean="1"></expression> + </RegExp> + <RegExp input="$$1" output="<url function="GetAMGDiscography">http://www.allmusic.com/cg/amg.dll?p=amg&amp;sql=\1</url>" dest="5+"> + <expression noclean="1">sql=([^"]*)">Discography</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetArtistDetails> + <LastfmLink dest="5"> + <RegExp input="$$2" output="<details>\1</details>" dest="5"> + <RegExp input="$$1" output="<url function="GetBiography">http://ws.audioscrobbler.com/2.0/?method=artist.getinfo&amp;mbid=\2&amp;lang=pl&amp;api_key=71e468a84c1f40d4991ddccc46e40f1b</url>" dest="2"> + <expression noclean="1"><artist>.*?<name>([^<]*)</name>.*?<mbid>([^<]*)</mbid></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </LastfmLink> + <GetBiography dest="5"> + <RegExp input="$$2" output="<details>\1</details>" dest="5"> + <RegExp input="$$1" output="<biography>\1</biography>" dest="2"> + <expression clear="yes"><content><!\[CDATA\[(.*?)\]</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetBiography> +</scraper> diff --git a/system/scrapers/music/common/allmusic.xml b/system/scrapers/music/common/allmusic.xml new file mode 100644 index 0000000000..89994e4253 --- /dev/null +++ b/system/scrapers/music/common/allmusic.xml @@ -0,0 +1,141 @@ +<scraperfunctions> + <GetAMGDiscography dest="5"> + <RegExp input="$$2$$3" output="<details>\1</details>" dest="5"> + <RegExp input="$$1" output="<album><year>\1</year><title>\3</title><label>\4</label></album>" dest="2"> + <expression repeat="yes" clear="yes" noclean="1,3,4">sorted-cell">([0-9]+)</td><td[^>]*>(<a href=[^>]*><img [^>]*/></a>|[^<]*)?</td><td[^>]*><a href=[^>]*>([^<]*)</a></td><td[^<]*</td><td[^>]*>([^<]+)</td></expression> + </RegExp> + <RegExp input="$$2" output="\1&amp;\2" dest="3"> + <expression noclean="1,2" repeat="yes">(.*?)&(.+)</expression> + </RegExp> + <RegExp input="$$3" output="" dest="2"> + <expression>(.+)</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetAMGDiscography> + <GetAMGBiography dest="5"> + <RegExp input="$$2" output="<details>\1</details>" dest="5"> + <RegExp input="$$1" output="<biography>\1</biography>" dest="2"> + <expression clear="yes">Biography</td>.*?<p>(.*?)</p></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetAMGBiography> + <GetAMGReview dest="5"> + <RegExp input="$$2" output="<details>\1</details>" dest="5"> + <RegExp input="$$1" output="<review>\1</review>" dest="2"> + <expression>Review</td>.*?<p>(.*?)</p></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetAMGReview> + <ParseAMGArtist dest="3"> + <RegExp input="$$5" output="<details>\1</details>" dest="3"> + <RegExp input="$$1" output="<name>\1</name>" dest="5"> + <expression noclean="1"><TITLE>allmusic[ (]+ ([^>]*) > Overview</expression> + </RegExp> + <RegExp input="$$1" output="<born>\1</born>" dest="5+"> + <expression><span>Born.*?;">(.*?)</td></expression> + </RegExp> + <RegExp input="$$1" output="<died>\1</died>" dest="5+"> + <expression><span>Died.*?;">(.*?)</td></expression> + </RegExp> + <RegExp input="$$1" output="<formed>\1</formed>" dest="5+"> + <expression><span>Formed.*?;">(.*?)</td></expression> + </RegExp> + <RegExp input="$$1" output="<disbanded>\1</disbanded>" dest="5+"> + <expression><span>Disbanded.*?;">(.*?)</td></expression> + </RegExp> + <RegExp input="$$1" output="<yearsactive>\1</yearsactive>" dest="5+"> + <expression><span>Years Active.*?sub">(.*?)</div></expression> + </RegExp> + <RegExp input="$$1" output="\1" dest="6"> + <expression noclean="1">Genre Listing-->([^!]*)<!--Genre Listing</expression> + </RegExp> + <RegExp input="$$6" output="<genre>\1</genre>" dest="5+"> + <expression noclean="1" repeat="yes"><li>[^>]*>([^<]*)</a></li></expression> + </RegExp> + <RegExp input="$$1" output="\1" dest="7"> + <expression noclean="1">Style Listing-->([^!]*)<!--Style Listing</expression> + </RegExp> + <RegExp input="$$7" output="<style>\1</style>" dest="5+"> + <expression noclean="1" repeat="yes"><li>[^>]*>([^<]*)</a></li></expression> + </RegExp> + <RegExp input="$$1" output="\1" dest="8"> + <expression noclean="1">Moods Listing-->([^!]*)<!--Moods Listing</expression> + </RegExp> + <RegExp input="$$8" output="<mood>\1</mood>" dest="5+"> + <expression noclean="1" repeat="yes"><li>[^>]*>([^<]*)</a></li></expression> + </RegExp> + <RegExp input="$$1" output="\1" dest="9"> + <expression noclean="1">Instruments Listing-->([^!]*)<!--Instruments Listing</expression> + </RegExp> + <RegExp input="$$9" output="<instruments>\1</instruments>" dest="5+"> + <expression noclean="1" repeat="yes"><li>[^>]*>([^<]*)</a></li></expression> + </RegExp> + <RegExp input="$$1" output="<thumb>\1</thumb>" dest="5+"> + <expression noclean="1"><img src="([^"]*)" id="Picture"</expression> + </RegExp> + <RegExp input="$$1" output="<thumb>http://image.allmusic.com/00/amg/pic200/dr\1\200/\1\2\3\4/\1\2\3\4\5.jpg</thumb>" dest="5+"> + <expression noclean="1" repeat="yes">"([A-Z^])([0-9^])([0-9^])([0-9^])([^"]*)"</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </ParseAMGArtist> + <ParseAMGAlbum dest="3"> + <RegExp input="$$5" output="<details>\1</details>" dest="3"> + <RegExp input="$$1" output="<title>\1</title>" dest="5"> + <expression noclean="1"><TITLE>allmusic[ (]+ ([^>]*) > Overview</expression> + </RegExp> + <RegExp input="$$1" output="<artist>\1</artist>" dest="5+"> + <expression noclean="1"><span>Artist</span>[^;]*[^>]*>[^>]*>([^<]*)</expression> + </RegExp> + <RegExp input="$$1" output="<rating>\1</rating>" dest="5+"> + <expression noclean="1"><span>Rating</span>[^_]*_r([0-9^]*)</expression> + </RegExp> + <RegExp input="$$1" output="<label>\1</label>" dest="5+"> + <expression noclean="1"><span>Label</span>[^;]*[^>]*>([^<]*)</expression> + </RegExp> + <RegExp input="$$1" output="<type>\1</type>" dest="5+"> + <expression noclean="1"><span>Type</span>[^;]*[^>]*>([^<]*)</expression> + </RegExp> + <RegExp input="$$1" output="<releasedate>\1</releasedate>" dest="5+"> + <expression noclean="1"><span>Release Date</span>[^;]*[^>]*>([^<]*)</expression> + </RegExp> + <RegExp input="$$1" output="<year>\1</year>" dest="5+"> + <expression noclean="1"><span>Release Date.*?([0-9]+)</</expression> + </RegExp> + <RegExp input="$$1" output="\1" dest="6"> + <expression noclean="1">Genre[s]* Listing-->([^!]*)<!--Genre[s]* Listing</expression> + </RegExp> + <RegExp input="$$6" output="<genre>\1</genre>" dest="5+"> + <expression noclean="1" repeat="yes"><li>[^>]*>([^<]*)</a></li></expression> + </RegExp> + <RegExp input="$$1" output="\1" dest="7"> + <expression noclean="1">Style[s]* Listing-->([^!]*)<!--Style[s]* Listing</expression> + </RegExp> + <RegExp input="$$7" output="<style>\1</style>" dest="5+"> + <expression noclean="1" repeat="yes"><li>[^>]*>([^<]*)</a></li></expression> + </RegExp> + <RegExp input="$$1" output="\1" dest="8"> + <expression noclean="1">Mood[s]* Listing-->([^!]*)<!--Mood[s]* Listing</expression> + </RegExp> + <RegExp input="$$8" output="<mood>\1</mood>" dest="5+"> + <expression noclean="1" repeat="yes"><li>[^>]*>([^<]*)</a></li></expression> + </RegExp> + <RegExp input="$$1" output="\1" dest="9"> + <expression noclean="1">Theme[s]* Listing-->([^!]*)<!--Theme[s]* Listing</expression> + </RegExp> + <RegExp input="$$9" output="<theme>\1</theme>" dest="5+"> + <expression noclean="1" repeat="yes"><li>[^>]*>([^<]*)</a></li></expression> + </RegExp> + <RegExp input="$$1" output="<track><position>\1</position><title>\3</title><duration>\4</duration></track>" dest="5+"> + <expression noclean="1,2,3" repeat="yes">"cell">([0-9]+)</.*?sql=([0-9:a-z]*)">(.*?)</a>.*?"right".*?>(.*?)</</expression> + </RegExp> + <RegExp input="$$1" output="<thumb>http://image.allmusic.com/\1</thumb>" dest="5+"> + <expression noclean="1">http://image.allmusic.com/([^"]*)"</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </ParseAMGAlbum> +</scraperfunctions> diff --git a/system/scrapers/music/common/htbackdrops.xml b/system/scrapers/music/common/htbackdrops.xml new file mode 100644 index 0000000000..9872e4717f --- /dev/null +++ b/system/scrapers/music/common/htbackdrops.xml @@ -0,0 +1,18 @@ +<scraperfunctions> + <GetHTBThumbs dest="5"> + <RegExp input="$$13" output="<details>\1</details>" dest="5"> + <RegExp input="$$1" output="<thumb preview="http://www.htbackdrops.com/data/thumbnails/\2">http://www.htbackdrops.com/data/media/\2</thumb>" dest="13"> + <expression noclean="1" repeat="yes">mode=search(&amp;sessionid=[^"]*)?"><img src="./data/thumbnails/([^"]+)"</expression> + </RegExp> + <expression noclean="1"/> + </RegExp> + </GetHTBThumbs> + <GetHTBFanart dest="5"> + <RegExp input="$$13" output="<details><fanart url="http://www.htbackdrops.com/">\1</fanart></details>" dest="5"> + <RegExp input="$$1" output="<thumb preview="data/thumbnails/\2">data/media/\2</thumb>" dest="13"> + <expression noclean="1" repeat="yes">mode=search(&amp;sessionid=[^"]*)?"><img src="./data/thumbnails/([^"]+)"</expression> + </RegExp> + <expression noclean="1"/> + </RegExp> + </GetHTBFanart> +</scraperfunctions> diff --git a/system/scrapers/music/common/lastfm.xml b/system/scrapers/music/common/lastfm.xml new file mode 100644 index 0000000000..307c9a9c45 --- /dev/null +++ b/system/scrapers/music/common/lastfm.xml @@ -0,0 +1,10 @@ +<scraperfunctions> + <GetLastFMDiscography dest="5"> + <RegExp input="$$2" output="<details>\1</details>" dest="5"> + <RegExp input="$$1" output="<album><title>\1</title></album>" dest="2+"> + <expression repeat="yes"><album[^>]*>[^<]*<name>([^<]*?)</name></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetLastFMDiscography> +</scraperfunctions> diff --git a/system/scrapers/music/discogs.gif b/system/scrapers/music/discogs.gif Binary files differnew file mode 100644 index 0000000000..2aba929794 --- /dev/null +++ b/system/scrapers/music/discogs.gif diff --git a/system/scrapers/music/discogs.xml b/system/scrapers/music/discogs.xml new file mode 100644 index 0000000000..d727e9fa93 --- /dev/null +++ b/system/scrapers/music/discogs.xml @@ -0,0 +1,127 @@ +<?xml version="1.0" encoding="UTF-8"?> +<scraper framework="1.1" date="2009-09-16" name="Discogs" content="albums" thumb="discogs.gif" language="en"> + <include>common/htbackdrops.xml</include> + <GetSettings dest="3"> + <RegExp input="$$5" output="<settings>\1</settings>" dest="3"> + <RegExp input="$$1" output="<setting label="Enable Fanart" type="bool" id="htbfanart" default="true"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting label="Grab thumbs from HTBackdrops" type="bool" id="htbthumbs" default="true"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetSettings> + <NfoUrl dest="3"> + <RegExp input="$$1" output="\1" dest="3"> + <expression>(http://www.discogs.com/.*)</expression> + </RegExp> + </NfoUrl> + <CreateAlbumSearchUrl dest="3"> + <RegExp input="$$1" output="http://www.discogs.com/advanced_search?artist=$$2&release_title=\1&btn=Search+Releases" dest="3"> + <expression/> + </RegExp> + </CreateAlbumSearchUrl> + <GetAlbumSearchResults dest="8"> + <RegExp input="$$5" output="<results>\1</results>" dest="8"> + <RegExp input="$$1" output="<entity><artist>\2</artist><title>\3</title><url>http://www.discogs.com\1</url></entity>" dest="5+"> + <expression repeat="yes"><a href="([^"]*/release/[^"]*)".*?>(.*?) - (.*?)</a></expression> + </RegExp> + <expression noclean="1"/> + </RegExp> + </GetAlbumSearchResults> + <GetAlbumDetails dest="3"> + <RegExp input="$$5" output="<details>\1</details>" dest="3"> + <RegExp input="$$1" output="<artist>\1</artist>" dest="5"> + <expression noclean="1"><a href="/artist/[^"]+">(.*?)</a></expression> + </RegExp> + <RegExp input="$$1" output="<title>\1</title>" dest="5+"> + <expression noclean="1">Add .*? - (.*?) to your collection</expression> + </RegExp> + <RegExp input="$$1" output="<rating>\1</rating>" dest="5+"> + <expression noclean="1">Rating:</td>.*?<b>([0-9]+).[0-9]</b>/[0-9]+</expression> + </RegExp> + <RegExp input="$$1" output="<label>\1</label>" dest="5+"> + <expression noclean="1">Label:</td>.*?>([^<]+)</a></expression> + </RegExp> + <RegExp input="$$1" output="<releasedate>\1</releasedate>" dest="5+"> + <expression noclean="1">Released:.*?<td>(.*?)</td></expression> + </RegExp> + <RegExp input="$$1" output="<year>\1</year>" dest="5+"> + <expression noclean="1">Released:.*?<td>.*?([0-9]+)</td></expression> + </RegExp> + <RegExp input="$$1" output="\1" dest="6"> + <expression trim="1">Genre:</td><td>(.*?)</td></expression> + </RegExp> + <RegExp input="$$6" output="<genre>\1</genre>" dest="5+"> + <expression repeat="yes">([^,]+),?</expression> + </RegExp> + <RegExp input="$$1" output="\1" dest="6"> + <expression trim="1">Style:</td><td>(.*?)</td></expression> + </RegExp> + <RegExp input="$$6" output="<style>\1</style>" dest="5+"> + <expression repeat="yes">([^,]+),?</expression> + </RegExp> + <RegExp input="$$1" output="\1" dest="6"> + <expression noclean="1">Tracklisting:(.*?)</table></expression> + </RegExp> + <RegExp input="$$6" output="<track><position>\1</position><title>\2</title><duration>\3</duration></track>" dest="5+"> + <expression repeat="yes">"left">[^<0-9]*([0-9]+)</td>[^:]*<td>([^<]+) \(([:0-9]+)\)</td></expression> + </RegExp> + <RegExp input="$$1" output="<review>\1</review>" dest="5+"> + <expression>User Reviews:</b>[^<]*<blockquote>[^<]*(<div id=.+?)<div id=</expression> + </RegExp> + <RegExp input="$$1" output="<url function="GetImages">http://www.discogs.com\1</url>" dest="5+"> + <expression><a href="(/viewimages\?release=[^"]*)"</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetAlbumDetails> + <GetImages dest="5"> + <RegExp input="$$4" output="<details>\1</details>" dest="5"> + <RegExp input="$$1" output="<thumb>\1</thumb>" dest="4+"> + <expression noclean="1" repeat="yes">img src="(http://www.discogs.com/image/[^"]*)"</expression> + </RegExp> + <expression noclean="1"/> + </RegExp> + </GetImages> + <CreateArtistSearchUrl dest="3"> + <RegExp input="$$2" output="http://www.discogs.com/search?type=artists&q="\1"&btn=Search" dest="3"> + <RegExp input="$$2" output="\1,%20The" dest="2"> + <RegExp input="$$1" output="\1" dest="2"> + <expression noclean="1"/> + </RegExp> + <expression noclean="1" clear="no" repeat="no" trim="1">[Tt]he%20(.+)</expression> + </RegExp> + <expression noclean="1"/> + </RegExp> + </CreateArtistSearchUrl> + <GetArtistSearchResults dest="8"> + <RegExp input="$$5" output="<results>\1</results>" dest="8"> + <RegExp input="$$1" output="<entity><title>\2</title><url>http://www.discogs.com/artist/\1</url></entity>" dest="5+"> + <expression repeat="yes"><a href="/artist/([^"]*)" class="rollover_link">(.+?)</a></expression> + </RegExp> + <expression noclean="1"/> + </RegExp> + </GetArtistSearchResults> + <GetArtistDetails dest="3"> + <RegExp input="$$5" output="<details>\1</details>" dest="3"> + <RegExp input="$$1" output="<name>\1</name>" dest="5"> + <expression noclean="1">Manage your (.*?) collection</expression> + </RegExp> + <RegExp input="$$1" output="<biography>\1</biography>" dest="5+"> + <expression>Profile:(.*?)URLs:</expression> + </RegExp> + <RegExp input="$$2" conditional="htbthumbs" output="<url function="GetHTBThumbs" post="yes">http://www.htbackdrops.com/search.php?search_terms=all&amp;cat_id=5&amp;search_keywords=\1</url>" dest="5+"> + <expression/> + </RegExp> + <RegExp input="$$1" output="<url function="GetImages">http://www.discogs.com\1</url>" dest="5+"> + <expression>a href="(/viewimages\?artist=[^"]*)"</expression> + </RegExp> + <RegExp input="$$2" conditional="htbfanart" output="<url function="GetHTBFanart" post="yes">http://www.htbackdrops.com/search.php?search_terms=all&amp;cat_id=1&amp;search_keywords=\1</url>" dest="5+"> + <expression/> + </RegExp> + <expression noclean="1"/> + </RegExp> + </GetArtistDetails> +</scraper> diff --git a/system/scrapers/music/freebase.png b/system/scrapers/music/freebase.png Binary files differnew file mode 100644 index 0000000000..7b14e32b7b --- /dev/null +++ b/system/scrapers/music/freebase.png diff --git a/system/scrapers/music/freebase.xml b/system/scrapers/music/freebase.xml new file mode 100644 index 0000000000..4399e47a5a --- /dev/null +++ b/system/scrapers/music/freebase.xml @@ -0,0 +1,149 @@ +<?xml version="1.0" encoding="UTF-8"?> +<scraper framework="1.1" date="2009-09-16" name="Freebase" content="albums" thumb="freebase.png"> + <include>common/htbackdrops.xml</include> + <GetSettings dest="3"> + <RegExp input="$$5" output="<settings>\1</settings>" dest="3"> + <RegExp input="$$1" output="<setting label="Enable Fanart" type="bool" id="htbfanart" default="true"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting label="Grab thumbs from HTBackdrops" type="bool" id="htbthumbs" default="true"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetSettings> + <NfoUrl dest="3"> + <RegExp input="$$1" output="<url>http://api.freebase.com/api/service/mqlread?query={%22query%22:{%22type%22:%22/music/artist%22,%22name%22:null,%22genre%22:[],%22album%22:[{%22id%22:%22/\2%22,%22/common/topic/article%22:[{%22guid%22:null,%22optional%22:true,%22limit%22:1}],%22/common/topic/image%22:[{%22guid%22:null,%22optional%22:true,%22limit%22:1}],%22label%22:[],%22engineer%22:[],%22producer%22:[],%22release_type%22:null,%22artist%22:null,%22name%22:null,%22release_date%22:null,%22track%22:[{%22name%22:null,%22length%22:null,%22index%22:null,%22sort%22:%22index%22,%22optional%22:true}],%22sort%22:%22release_date%22}]}}</url>" dest="3+"> + <expression>(http://www.freebase.com/view/)(.*)</expression> + </RegExp> + </NfoUrl> + + <CreateAlbumSearchUrl dest="3"> + <RegExp input="$$1" output="http://api.freebase.com/api/service/mqlread?query={%22query%22:{%22type%22:%22/music/artist%22,%22name~=%22:%22$$2%22,%22album%22:[{%22guid%22:null,%22artist%22:null,%22name%22:null,%22name~=%22:%22\1%22}]}}" dest="3"> + <expression/> + </RegExp> + </CreateAlbumSearchUrl> + + <GetAlbumSearchResults dest="8"> + <RegExp input="$$5" output="<results>\1</results>" dest="8"> + <RegExp input="$$1" output="<entity><artist>\1</artist><title>\3</title><url>http://api.freebase.com/api/service/mqlread?query={%22query%22:{%22type%22:%22/music/artist%22,%22name%22:null,%22genre%22:[],%22album%22:[{%22id%22:%22/guid/\2%22,%22/common/topic/article%22:[{%22guid%22:null,%22optional%22:true,%22limit%22:1}],%22/common/topic/image%22:[{%22guid%22:null,%22optional%22:true,%22limit%22:1}],%22label%22:[],%22engineer%22:[],%22producer%22:[],%22release_type%22:null,%22artist%22:null,%22name%22:null,%22release_date%22:null,%22track%22:[{%22name%22:null,%22length%22:null,%22index%22:null,%22sort%22:%22index%22,%22optional%22:true}],%22sort%22:%22release_date%22}]}}</url></entity>" dest="5+"> + <expression repeat="yes">"artist": "([^"]*)".*?"guid": "#([^"]*)".*?"name": "([^"]*)"</expression> + </RegExp> + <expression noclean="1"/> + </RegExp> + </GetAlbumSearchResults> + + <GetAlbumDetails dest="3"> + <RegExp input="$$5" output="<details>\1</details>" dest="3"> + <RegExp input="$$1" output="<artist>\1</artist>" dest="5"> + <expression noclean="1">"artist": "([^"]*)</expression> + </RegExp> + <RegExp input="$$1" output="<title>\1</title>" dest="5+"> + <expression noclean="1">"name": "([^"]*)</expression> + </RegExp> + <RegExp input="$$1" output="\1" dest="6"> + <expression trim="1">"label": (.*?)\]</expression> + </RegExp> + <RegExp input="$$6" output="<label>\1</label>" dest="5+"> + <expression noclean="1" repeat="yes">"([^"]*)"</expression> + </RegExp> + <RegExp input="$$1" output="<type>\1</type>" dest="5+"> + <expression noclean="1">"release_type": "([^"]*)</expression> + </RegExp> + <RegExp input="$$1" output="<releasedate>\1</releasedate>" dest="5+"> + <expression noclean="1">"release_date": "([^"]*)</expression> + </RegExp> + <RegExp input="$$1" output="<year>\1</year>" dest="5+"> + <expression noclean="1">"release_date": "([0-9]+)</expression> + </RegExp> + <RegExp input="$$1" output="\1" dest="6+"> + <expression trim="1">"genre": (.*?)\]</expression> + </RegExp> + <RegExp input="$$6" output="<genre>\1</genre>" dest="5+"> + <expression noclean="1" repeat="yes">"([^"]*)"</expression> + </RegExp> + <RegExp input="$$1" output="<track><position>\1</position><title>\3</title><duration>\2</duration></track>" dest="5+"> + <expression repeat="yes">"index": ([0-9]+).*?"length": ([0-9]+).*?"name": "([^"]*)"</expression> + </RegExp> + <RegExp input="$$1" output="<thumb>http://www.freebase.com/api/trans/raw/guid/\1</thumb>" dest="5+"> + <expression noclean="1">image":.*?"guid": "#([^"]*)"</expression> + </RegExp> + <RegExp input="$$1" output="<url function="GetAlbumReview">http://www.freebase.com/api/trans/raw/guid/\1</url>" dest="5+"> + <expression noclean="1">article":.*?"guid": "#([^"]*)"</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetAlbumDetails> + + <GetAlbumReview dest="5"> + <RegExp input="$$2" output="<details>\1</details>" dest="5"> + <RegExp input="$$1" output="<review>\1</review>" dest="2"> + <expression noclean="1"><p>([^<]*)</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetAlbumReview> + + <CreateArtistSearchUrl dest="3"> + <RegExp input="$$1" output="http://www.freebase.com/api/service/search?query=\1&type=/music/artist&indent=1" dest="3"> + <expression/> + </RegExp> + </CreateArtistSearchUrl> + + <GetArtistSearchResults dest="8"> + <RegExp input="$$5" output="<results>\1</results>" dest="8"> + <RegExp input="$$1" output="<entity><title>\2</title><url>http://api.freebase.com/api/service/mqlread?query={%22query%22:{%22type%22:%22/music/artist%22,%22id%22:%22/guid/\1%22,%22/common/topic/article%22:[{%22guid%22:null,%22optional%22:true,%22limit%22:1}],%22/common/topic/image%22:[{%22guid%22:null,%22optional%22:true,%22limit%22:1}],%22name%22:null,%22genre%22:[],%22album%22:[],%22active_start%22:null,%22active_end%22:null,%22instruments_played%22:[]}}</url></entity>" dest="5+"> + <expression repeat="yes">"guid": "#([^"]*)".*?"name": "([^"]*)"</expression> + </RegExp> + <expression noclean="1"/> + </RegExp> + </GetArtistSearchResults> + + <GetArtistDetails dest="3"> + <RegExp input="$$5" output="<details>\1</details>" dest="3"> + <RegExp input="$$1" output="<name>\1</name>" dest="5"> + <expression noclean="1">"name": "([^"]*)</expression> + </RegExp> + <RegExp input="$$1" output="<formed>\1</formed>" dest="5+"> + <expression noclean="1">"active_start": "([^"]*)</expression> + </RegExp> + <RegExp input="$$1" output="<disbanded>\1</disbanded>" dest="5+"> + <expression noclean="1">"active_end": "([^"]*)</expression> + </RegExp> + <RegExp input="$$1" output="\1" dest="6"> + <expression trim="1">"genre": (.*?)\]</expression> + </RegExp> + <RegExp input="$$6" output="<genre>\1</genre>" dest="5+"> + <expression repeat="yes">"([^"]*)"</expression> + </RegExp> + <RegExp input="$$1" output="\1" dest="6"> + <expression trim="1">"album": (.*?)\]</expression> + </RegExp> + <RegExp input="$$6" output="<album><title>\1</title></album>" dest="5+"> + <expression noclean="1" repeat="yes">"([^"]*)"</expression> + </RegExp> + <RegExp input="$$1" output="<thumb>http://www.freebase.com/api/trans/raw/guid/\1</thumb>" dest="5+"> + <expression noclean="1">image":.*?"guid": "#([^"]*)"</expression> + </RegExp> + <RegExp input="$$2" conditional="htbthumbs" output="<url function="GetHTBThumbs" post="yes">http://www.htbackdrops.com/search.php?search_terms=all&amp;cat_id=5&amp;search_keywords=\1</url>" dest="5+"> + <expression/> + </RegExp> + <RegExp input="$$2" conditional="htbfanart" output="<url function="GetHTBFanart" post="yes">http://www.htbackdrops.com/search.php?search_terms=all&amp;cat_id=1&amp;search_keywords=\1</url>" dest="5+"> + <expression/> + </RegExp> + <RegExp input="$$1" output="<url function="GetBiography">http://www.freebase.com/api/trans/raw/guid/\1</url>" dest="5+"> + <expression noclean="1">article":.*?"guid": "#([^"]*)"</expression> + </RegExp> + <expression noclean="1"/> + </RegExp> + </GetArtistDetails> + + <GetBiography dest="5"> + <RegExp input="$$2" output="<details>\1</details>" dest="5"> + <RegExp input="$$1" output="<biography>\1</biography>" dest="2"> + <expression noclean="1"><p>([^<]*)</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetBiography> +</scraper> diff --git a/system/scrapers/music/israel-music.png b/system/scrapers/music/israel-music.png Binary files differnew file mode 100644 index 0000000000..932c3abf75 --- /dev/null +++ b/system/scrapers/music/israel-music.png diff --git a/system/scrapers/music/israel-music.xml b/system/scrapers/music/israel-music.xml new file mode 100644 index 0000000000..62010304ee --- /dev/null +++ b/system/scrapers/music/israel-music.xml @@ -0,0 +1,53 @@ +<?xml version="1.0" encoding="UTF-8"?> +<scraper framework="1.0" date="2009-05-22" name="IsraelMusic" content="albums" thumb="israel-music.png"> + <NfoUrl dest="3"> + <RegExp input="$$1" output="\1/" dest="3"> + <expression>(http://he.israel-music.com.*)</expression> + </RegExp> + </NfoUrl> + <CreateAlbumSearchUrl dest="3"> + <RegExp input="$$1" output='http://he.israel-music.com/search/\1/' dest="3"> + <expression></expression> + </RegExp> + <RegExp input="$$2%20$$1" output='http://he.israel-music.com/search/\1/' dest="3"> + <expression></expression> + </RegExp> + </CreateAlbumSearchUrl> + <GetAlbumSearchResults dest="8"> + <RegExp input="$$5" output='<?xml version="1.0" encoding="utf-8" standalone="yes"?><results>\1</results>' dest="8"> + <RegExp input="$$1" output='<entity><url>http://he.israel-music.com\1</url><title>\2</title></entity>' dest="5+"> + <expression repeat="yes" noclean='1,2'><![CDATA[<strong><a dir="rtl" href="(.[^"]*)">(.[^<]*)</a>]]></expression> + </RegExp> + <RegExp input="$$1" output='<entity><url>http://he.israel-music.com\1</url><title>\2</title><artist>\4</artist><year>\5</year></entity>' dest="5+"> + <expression repeat="yes" noclean='1,3,4,5'><![CDATA[<strong><a dir="rtl" href="(.[^"]*)">(.*?)</a> - <a dir="rtl" href="(.[^"]*)">(.[^<]*)</a> <span>([0-9]*)</span></strong>]]></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetAlbumSearchResults> + <GetAlbumDetails dest="3"> + <RegExp input="$$5" output="<details>\1</details>" dest="3"> + <RegExp input="$$1" output="<title>\1</title>" dest="5"> + <expression noclean="1"><![CDATA[<h1>(.[^<]*)</h1>]]></expression> + </RegExp> + <RegExp input="$$1" output="<artist>\1</artist>" dest="5+"> + <expression noclean="1"><![CDATA[</span><a.[^>]*>(.[^<]*)</a>]]></expression> + </RegExp> + <RegExp input="$$1" output="<review>\1</review>" dest="5+"> + <expression><![CDATA[<ul id="details".[^>]*>(.*)</ul>.[^<]*<p style="clear:both">]]></expression> + </RegExp> + <RegExp input="$$1" output="<thumb>\1</thumb>" dest="5+"> + <expression noclean="1"><![CDATA[id="imagef" src="(.[^"]*)"]]></expression> + </RegExp> + <RegExp input="$$1" output="<year>\1</year>" dest="5+"> + <expression noclean="1"><![CDATA[<li><b>Ч©Ч Ч”: </b> ([0-9]+)</li>]]></expression> + </RegExp> + <RegExp input="$$1" output="\1" dest="6"> + <expression noclean="1"><![CDATA[<li><b>Ч–'ЧђЧ ЧЁ:</b>(.*)</li>]]></expression> + </RegExp> + <RegExp input="$$6" output="<genre>\1</genre>" dest="5+"> + <expression noclean="1" repeat="yes"><![CDATA[<a.[^>]*">(.[^<]*)</a>]]></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetAlbumDetails> +</scraper> diff --git a/system/scrapers/music/lastfm.xml b/system/scrapers/music/lastfm.xml new file mode 100644 index 0000000000..494b3a5b58 --- /dev/null +++ b/system/scrapers/music/lastfm.xml @@ -0,0 +1,121 @@ +<?xml version="1.0" encoding="UTF-8"?> +<scraper framework="1.1" date="2009-09-16" name="Last.FM" content="albums" thumb="lastfm.gif"> + <include>common/htbackdrops.xml</include> + <include>common/lastfm.xml</include> + <GetSettings dest="3"> + <RegExp input="$$5" output="<settings>\1</settings>" dest="3"> + <RegExp input="$$1" output="<setting label="Enable Fanart" type="bool" id="htbfanart" default="true"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting label="Grab thumbs from HTBackdrops" type="bool" id="htbthumbs" default="true"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetSettings> + <NfoUrl dest="3"> + <RegExp input="$$1" output="http://ws.audioscrobbler.com/2.0/?method=album.getinfo&album=\2&artist=\1&api_key=71e468a84c1f40d4991ddccc46e40f1b" dest="3"> + <expression>(http://www.last.fm/music/([^/]*)/(.*))</expression> + </RegExp> + <RegExp input="$$1" output="http://ws.audioscrobbler.com/2.0/?method=artist.getinfo&artist=\1&api_key=71e468a84c1f40d4991ddccc46e40f1b" dest="3"> + <expression>(http://www.last.fm/music/([^/]*))$</expression> + </RegExp> + </NfoUrl> + <CreateAlbumSearchUrl dest="3"> + <RegExp input="$$1" output="http://ws.audioscrobbler.com/2.0/?method=album.search&album=\1&artist=$$2&api_key=71e468a84c1f40d4991ddccc46e40f1b" dest="3"> + <expression/> + </RegExp> + </CreateAlbumSearchUrl> + <GetAlbumSearchResults dest="8"> + <RegExp input="$$5" output="<results>\1</results>" dest="8"> + <RegExp input="$$1" output="<entity><artist>\2</artist><title>\1</title><url>http://ws.audioscrobbler.com/2.0/?method=album.getinfo&amp;album=\4&amp;artist=\3&amp;api_key=71e468a84c1f40d4991ddccc46e40f1b</url></entity>" dest="5+"> + <expression repeat="yes"><album>.*?<name>([^<]*)</name>.*?<artist>([^<]*)</artist>.*?<url>http://www.last.fm/music/([^/]+)/(.*?)</url></expression> + </RegExp> + <expression noclean="1"/> + </RegExp> + </GetAlbumSearchResults> + <GetAlbumDetails dest="3"> + <RegExp input="$$5" output="<details>\1</details>" dest="3"> + <RegExp input="$$1" output="<artist>\1</artist>" dest="5"> + <expression noclean="1"><artist>(.*?)</artist></expression> + </RegExp> + <RegExp input="$$1" output="<title>\1</title>" dest="5+"> + <expression noclean="1"><name>(.*?)</name></expression> + </RegExp> + <RegExp input="$$1" output="<releasedate>\1</releasedate>" dest="5+"> + <expression trim="1"><releasedate>:(.*?)</releasedate></expression> + </RegExp> + <RegExp input="$$1" output="<year>\1</year>" dest="5+"> + <expression trim="1"><releasedate>[^,]*([0-9]+),</releasedate></expression> + </RegExp> + <RegExp input="$$1" output="<review>\1</review>" dest="5+"> + <expression><content><!\[CDATA\[(.*)\]\]></content></expression> + </RegExp> + <RegExp input="$$1" output="<thumb>\2</thumb>" dest="5+"> + <expression noclean="1"><image size="(extra)?large">([^<]*)</image></expression> + </RegExp> + <RegExp input="$$1" output="<url function="GetTracks">http://ws.audioscrobbler.com/2.0/?method=playlist.fetch&amp;playlistURL=lastfm://playlist/album/\1&amp;api_key=71e468a84c1f40d4991ddccc46e40f1b</url>" dest="5+"> + <expression noclean="1"><id>(.*?)</id></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetAlbumDetails> + <GetTracks dest="5"> + <RegExp input="$$4" output="<details>\1</details>" dest="5"> + <RegExp input="$$1" output="<track><title>\1</title></track>" dest="4+"> + <expression noclean="1" repeat="yes"><track>.*?<title>(.*?)</title></expression> + </RegExp> + <expression noclean="1"/> + </RegExp> + </GetTracks> + <CreateArtistSearchUrl dest="3"> + <RegExp input="$$1" output="http://ws.audioscrobbler.com/2.0/?method=artist.search&artist=\1&api_key=71e468a84c1f40d4991ddccc46e40f1b" dest="3"> + <expression/> + </RegExp> + </CreateArtistSearchUrl> + <GetArtistSearchResults dest="8"> + <RegExp input="$$5" output="<results>\1</results>" dest="8"> + <RegExp input="$$1" output="<entity><title>\1</title><url>http://ws.audioscrobbler.com/2.0/?method=artist.getinfo&amp;mbid=\2&amp;api_key=71e468a84c1f40d4991ddccc46e40f1b</url></entity>" dest="5+"> + <expression repeat="yes"><artist>.*?<name>([^<]*)</name>.*?<mbid>([^<]*)</mbid></expression> + </RegExp> + <expression noclean="1"/> + </RegExp> + </GetArtistSearchResults> + <GetArtistDetails dest="3"> + <RegExp input="$$5" output="<details>\1</details>" dest="3"> + <RegExp input="$$1" output="<name>\1</name>" dest="5"> + <expression noclean="1"><name>([^<]*)</name></expression> + </RegExp> + <RegExp input="$$1" output="<biography>\1</biography>" dest="5+"> + <expression><content><!\[CDATA\[(.*?)\]</expression> + </RegExp> + <RegExp input="$$1" output="<thumb>\2</thumb>" dest="5+"> + <expression noclean="1"><image size="(extra)?large">([^<]*)</image></expression> + </RegExp> + <RegExp input="$$2" conditional="htbthumbs" output="<url function="GetHTBThumbs" post="yes">http://www.htbackdrops.com/search.php?search_terms=all&amp;cat_id=5&amp;search_keywords=\1</url>" dest="5+"> + <expression/> + </RegExp> + <RegExp input="$$2" conditional="htbfanart" output="<url function="GetHTBFanart" post="yes">http://www.htbackdrops.com/search.php?search_terms=all&amp;cat_id=1&amp;search_keywords=\1</url>" dest="5+"> + <expression/> + </RegExp> + <RegExp input="$$1" output="<url function="GetGenres">http://ws.audioscrobbler.com/2.0/?method=artist.gettoptags&amp;artist=\1&amp;api_key=71e468a84c1f40d4991ddccc46e40f1b</url>" dest="5+"> + <expression noclean="1"><url>http://www.last.fm/music/(.*?)</url></expression> + </RegExp> + <RegExp input="$$1" output="<url function="GetLastFMDiscography">http://ws.audioscrobbler.com/2.0/?method=artist.gettopalbums&amp;artist=\1&amp;api_key=71e468a84c1f40d4991ddccc46e40f1b</url>" dest="5+"> + <expression noclean="1"><url>http://www.last.fm/music/(.*?)</url></expression> + </RegExp> + <expression noclean="1"/> + </RegExp> + </GetArtistDetails> + <GetGenres dest="5"> + <RegExp input="$$4" output="<details>\1</details>" dest="5"> + <RegExp input="$$1" output="<genre>\1</genre>" dest="4"> + <expression noclean="1"><tag>.*?<name>([^/lt;]*)</name></expression> + </RegExp> + <RegExp input="$$1" output="<style>\1</style>" dest="4+"> + <expression repeat="yes" noclean="1"><tag>.*?<name>([^/lt;]*)</name></expression> + </RegExp> + <expression noclean="1"/> + </RegExp> + </GetGenres> +</scraper> diff --git a/system/scrapers/video/AdultFilmdatabase.gif b/system/scrapers/video/AdultFilmdatabase.gif Binary files differnew file mode 100644 index 0000000000..1b9203dca8 --- /dev/null +++ b/system/scrapers/video/AdultFilmdatabase.gif diff --git a/system/scrapers/video/Excalibur.jpg b/system/scrapers/video/Excalibur.jpg Binary files differnew file mode 100644 index 0000000000..95a5e25dc1 --- /dev/null +++ b/system/scrapers/video/Excalibur.jpg diff --git a/system/scrapers/video/Excalibur.xml b/system/scrapers/video/Excalibur.xml new file mode 100644 index 0000000000..da94649863 --- /dev/null +++ b/system/scrapers/video/Excalibur.xml @@ -0,0 +1,65 @@ +<?xml version="1.0" encoding="utf-8" standalone="yes"?> +<scraper framework="1.1" date="2009-08-05" name="Excalibur" content="movies" thumb="excalibur.jpg" language="en"> + <NfoUrl dest="3"> + <RegExp input="$$1" output="<url>http://www.excaliburfilms.com/AdultDVD/\1.htm</url>" dest="3"> + <expression noclean="1">excaliburfilms.com/AdultDVD/([^.]*)</expression> + </RegExp> + </NfoUrl> + + <CreateSearchUrl dest="3"> + <RegExp input="$$1" output="<url>http://search.excaliburfilms.com/excals.htm?searchString=\1&amp;Search=AdultDVDMovies&amp;SearchFor=Title.x</url>" dest="3"> + <expression noclean="1"></expression> + </RegExp> + </CreateSearchUrl> + + <GetSearchResults dest="6"> + <RegExp input="$$5" output="<?xml version="1.0" encoding="iso-8859-1" standalone="yes"?><results>\1</results>" dest="6"> + <RegExp input="$$1" output="<entity><title>\2</title><url>http://www.excaliburfilms.com/AdultDVD/\1</url></entity>" dest="5+"> + <expression trim="1" repeat="yes"><a href="http://www.ExcaliburFilms.com/AdultDVD/([^"]*)".*?<font class="searchTitle13">([^<]*)</font></a></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetSearchResults> + + <GetDetails dest="3"> + <RegExp input="$$5" output="<details>\1</details>" dest="3"> + <RegExp input="$$1" output="<thumb>http://images.excaliburfilms.com/DVD/reviews/imagesBB020609/largemoviepic/\1.jpg</thumb>" dest="5+"> + <expression>imagesBB020609/([^\.]*)</expression> + </RegExp> + <RegExp input="$$1" output="<thumb>http://images.excaliburfilms.com/DVD/reviews/imagesBB020609/largemoviepic/\1-b.jpg</thumb>" dest="5+"> + <expression>imagesBB020609/([^\.]*)</expression> + </RegExp> + <RegExp input="$$1" output="<title>\1</title>" dest="5+"> + <expression>Rent ([^D]*) </expression> + </RegExp> + <RegExp input="$$1" output="<actor><name>\3</name><thumb>http://images.excaliburfilms.com/pornlist/starpicsAA020309/\2.jpg</thumb></actor>" dest="5+"> + <expression repeat="yes"><a href="/pornlist/(starpgs|malepgs)/([^\.]*)[^>]*>([^<]*)</expression> + </RegExp> + <RegExp input="$$1" output="<genre>\1</genre>" dest="5+"> + <expression>href='/DVD/[^>]*>([^<]*)</a><BR></expression> + </RegExp> + <RegExp input="$$1" output="<year>\1</year>" dest="5+"> + <expression>/excal/release[^>]*>[0-9]*/[0-9]*/([^<]*)</expression> + </RegExp> + <RegExp input="$$1" output="<runtime>\1 mins</runtime>" dest="5+"> + <expression>Time: </font><[^>]*>([0-9]*)</expression> + </RegExp> + <RegExp input="$$1" output="<plot>\1</plot>" dest="5+"> + <expression>Description:</font></a>[^>]*>([^<]*)</expression> + </RegExp> + <RegExp input="$$1" output="<studio>\1</studio>" dest="5+"> + <expression>href='/dvd/adultstudios/[^>]*>([^<]*)</expression> + </RegExp> + <RegExp input="$$1" output="<director>\1</director>" dest="5+"> + <expression>>Director: </font>[^>]*>([^<]*)</expression> + </RegExp> + <RegExp input="$$1" output="<rating>\1.\2</rating>" dest="5+"> + <expression>/Stars_([0-9])-([0-9]).gif</expression> + </RegExp> + <RegExp input="$$1" output="<rating>\1</rating>" dest="5+"> + <expression>/Stars_([0-9]).gif</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetDetails> +</scraper> diff --git a/system/scrapers/video/KinoPoisk.gif b/system/scrapers/video/KinoPoisk.gif Binary files differnew file mode 100644 index 0000000000..dbc35ac38e --- /dev/null +++ b/system/scrapers/video/KinoPoisk.gif diff --git a/system/scrapers/video/KinoPoisk.xml b/system/scrapers/video/KinoPoisk.xml new file mode 100644 index 0000000000..7fbef0ce05 --- /dev/null +++ b/system/scrapers/video/KinoPoisk.xml @@ -0,0 +1,342 @@ +<?xml version="1.0" encoding="windows-1251"?> +<scraper framework="1.1" date="2009-08-08" name="KinoPoisk_v2.001.ru" content="movies" thumb="KinoPoisk.gif" ServerContentEncoding="CP1251" language="ru"> + + <GetSettings dest="3"> + <RegExp input="$$5" output="<settings>\1</settings>" dest="3"> + <RegExp input="$$1" output="<setting label="фанарт в виде фото (выбрать одно)" type="bool" id="fanartP" default="false"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting label="фанарт в виде обоев (выбрать одно)" type="bool" id="fanartR" default="false"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting label="фанарт с сайта (www.themoviedb.org) (выбрать одно)" type="bool" id="fanart" default="true"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting label="фанарт с сайта (www.themoviedb.org) +24 (выбрать одно)" type="bool" id="fanartF" default="false"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting label="трейлер с сайта (www.imdb.com)" type="bool" id="trailerI" default="true"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetSettings> + + <CreateSearchUrl dest="3"> + <RegExp input="$$1" output="http://www.kinopoisk.ru/index.php?level=7&from=forma&result=adv&m_act%5Bfrom%5D=forma&m_act%5Bwhat%5D=content&m_act%5Bfind%5D=\1&m_act%5Byear%5D=&m_act%5Bcountry%5D=&m_act%5Bgenre%5D=&m_act%5Bcompany%5D=&m_act%5Bmpaa%5D=&m_act%5Bactor%5D=&m_act%5Bcast%5D=&m_act%5Bcontent_find%5D=" dest="3"> + <expression noclean="1"></expression> + </RegExp> + </CreateSearchUrl> + <GetSearchResults dest="8"> + <RegExp input="$$5" output="<?xml version="1.0" encoding="windows-1251" standalone="yes"?><results>\1</results>" dest="8"> + + <RegExp input="$$7" output="<entity>\1</entity>" dest="5+"> + <RegExp input="$$1" output="$$1" dest="6"> + <expression clear="yes">img src="/images/title002.gif"</expression> + </RegExp> + + <RegExp input="$$6" output="<title>\1 " dest="7"> + <expression trim="1"><title>([^<]+)</title></expression> + </RegExp> + <RegExp input="$$6" output="(\1)</title>" dest="7+"> + <expression><tr><td class="desc-title" height=25><img[^>]+>год</td><td class="desc-data"><a[^>]+>([0-9]+)</a></td></tr></expression> + </RegExp> + <RegExp input="$$6" output="<url>http://www.kinopoisk.ru/level/1/film/\1</url>" dest="7+"> + <expression>href="/level/19/film/([^"]+)" class="all">\.\.\.</a></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + + <RegExp input="$$1" output="<entity><title>\2 (\3)</title><url>http://www.kinopoisk.ru/level\1</url></entity>" dest="5+"> + <expression repeat="yes"><a class="all" href="/level([^"]*)">([^<]*)</a>,&nbsp;<a[^>]*>([^<]*)<</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetSearchResults> + <GetDetails clearbuffers="no" dest="3"> + <RegExp input="$$5" output="<details>\1</details>" dest="3"> + <!-- TITLE --> + <RegExp input="$$1" output="<title>\1</title>" dest="5"> + <expression trim="1"><title>([^<]+)</title></expression> + </RegExp> + <!-- YEAR --> + <RegExp input="$$1" output="<year>\1</year>" dest="5+"> + <expression><tr><td class="desc-title" height=25><img[^>]+>год</td><td class="desc-data"><a[^>]+>([0-9]+)</a></td></tr></expression> + </RegExp> + <!-- TOP250 --> + <RegExp input="$$1" output="<top250>\1</top250>" dest="5+"> + <expression>Top250: <a[^>]+>([0-9]*)</a></expression> + </RegExp> + <!-- MPAA --> + <RegExp input="$$1" output="<mpaa>\1</mpaa>" dest="5+"> + <expression><tr><td class="desc-title" height=25><img[^>]+>рейтинг MPAA</td><td class="desc-data".*?><a href=.+?/rn/(.+?)/</expression> + </RegExp> + + <RegExp input="$$1" output="<certification>\1 \3</certification>" dest="5+"> + <expression repeat="yes"><a href="/List\?certificates=[^"]*">([^<]*)</a>[^<]*(<i>([^<]*)</i>)?</expression> + </RegExp> + <!-- TAGLINE --> + <RegExp input="$$1" output="<tagline>\1</tagline>" dest="5+"> + <expression><tr><td class="desc-title" height=25><img[^>]+>слоган</td><td class="desc-data">&laquo;([^<]*)&raquo;</td></tr></expression> + </RegExp> + <!-- RUNTIME --> + <RegExp input="$$1" output="<runtime>\1 min</runtime>" dest="5+"> + <expression><tr><td class="desc-title" height=25><img[^>]+>время</td><td class="desc-data">([0-9]+) мин</td></tr></expression> + </RegExp> + <!-- RATING --> + <RegExp input="$$1" output="<rating>\1</rating><votes>\2</votes>" dest="5+"> + <expression>IMDB: ([0-9.]+) \(([0-9\s]+)\)</div></expression> + </RegExp> + <!-- GENRE --> + <RegExp input="$$1" output="<genre>\1</genre>" dest="5+"> + <expression repeat="yes">genre[^"]*" class="all">([^<]*)</a></expression> + </RegExp> + <!-- STUDIO --> + <RegExp input="$$1" output="<url function="STT">http://www.kinopoisk.ru\1</url>" dest="5+"> + <expression>href="([^"]+)"><b><font color="#ff6600">с</font><font color="#555555">тудии</font></expression> + </RegExp> + + <!-- DIRECTORS WRITERS ACTORS --> + <RegExp input="$$1" output="<url function="PEOPLE">http://www.kinopoisk.ru/level/19/film/\1</url>" dest="5+"> + <expression>href="/level/19/film/([^"]+)" class="all">\.\.\.</a></expression> + </RegExp> + + <!-- OUTLINE PLOT --> + <RegExp input="$$1" output="\1&#133;&hellip;&laquo;&#151;&mdash;&nbsp;" dest="2"> + <expression noclean="1"><tr><td colspan=3 style="padding:10px;padding-left:20px;" class="news">[\t\n]*(.*?)[\t\n]*</td></tr>\s+<tr><td colspan=3 height=5><spacer type=block height=5></td></tr></expression> + </RegExp> + <RegExp input="$$4" output="<outline>\1</outline><plot>\1</plot>" dest="5+"> + <RegExp input="$$2" output="\1 " dest="4"> + <expression repeat="yes">(.*?)&nbsp;</expression> + </RegExp> + <RegExp input="$$4" output="\1- " dest="4"> + <expression repeat="yes">(.*?)&mdash;</expression> + </RegExp> + <RegExp input="$$4" output="\1- " dest="4"> + <expression repeat="yes">(.*?)&#151;</expression> + </RegExp> + <RegExp input="$$4" output="\1"" dest="4"> + <expression repeat="yes">(.*?)&[lr]aquo;</expression> + </RegExp> + <RegExp input="$$4" output="\1..." dest="4"> + <expression repeat="yes">(.*?)&hellip;</expression> + </RegExp> + <RegExp input="$$4" output="\1..." dest="4"> + <expression repeat="yes">(.*?)&#133;</expression> + </RegExp> + <expression trim="1">(.*)\.\.\.</expression> + </RegExp> + <!-- POSTERS --> + + <RegExp input="$$1" output="<url function="GMP">http://www.kinopoisk.ru/level/17/film/\1</url>" dest="5+"> + <expression>href="/level/17/film/([^"]+)"</expression> + </RegExp> + <RegExp input="$$1" output="<url cache="KinoPoisk_opa.html" function="GetPoster">http://www.kinopoisk.ru/</url>" dest="5+"> + <expression></expression> + </RegExp> + + <RegExp conditional="fanartP" input="$$1" output="<url function="GMF">http://www.kinopoisk.ru/level/13/film/\1</url>" dest="5+"> + <expression>href="/level/13/film/([^"]+)"</expression> + </RegExp> + <RegExp conditional="fanartR" input="$$1" output="<url function="GMFR">http://www.kinopoisk.ru/level/12/film/\1</url>" dest="5+"> + <expression>href="/level/12/film/([^"]+)"</expression> + </RegExp> + + <RegExp conditional="fanart" input="$$8" output="<url function="GMA">http://themoviedb.org/search?search%5Btext%5D=\1</url>" dest="5+"> + <RegExp input="$$1" output="\2" dest="8"> + <expression><span style="font-size:13px;color:#666">(The )?([^<]+)</span></expression> + </RegExp> + <expression/> + </RegExp> + + <RegExp conditional="fanartF" input="$$8" output="<url function="GMAF">http://themoviedb.org/search?search%5Btext%5D=\1</url>" dest="5+"> + <RegExp input="$$1" output="\2" dest="8"> + <expression><span style="font-size:13px;color:#666">(The )?([^<]+)</span></expression> + </RegExp> + <expression/> + </RegExp> + + <RegExp input="$$1" output="<url cache="KinoPoisk_opa.html" function="GetFanart">http://www.kinopoisk.ru/</url>" dest="5+"> + <expression></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + + </GetDetails> + + <GMTR dest="10"> + <RegExp input="$$9" output="<details>\1</details>" dest="10"> + <RegExp input="$$1" output="<url>http://akas.imdb.com/video/imdb/vi\1/player</url>" dest="7"> + <expression noclean="1"><imdb>tt([^"]+)</imdb></expression> + </RegExp> + <RegExp input="$$7" output="<url function="GetTrailer">\1</url>" dest="9"> + <expression noclean="1"><url>([^<]+)</url></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GMTR> + + <GetTrailer clearbuffers="no" dest="5"> + <RegExp conditional="trailer" input="$$2" output="<details><trailer urlencoded="yes">\1</trailer></details>" dest="5"> + <RegExp input="$$1" output="\1" dest="2"> + <expression noclean="1">"file"[^"]*"([^"]*)</expression> + </RegExp> + <RegExp input="$$1" output="$$2/\1" dest="2"> + <expression noclean="1">"id"[^"]*"([^"]*)</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetTrailer> + + <GMA dest="10"> + <RegExp input="$$9" output="<details>\1</details>" dest="10"> + <RegExp input="$$1" output="<url>http://www.themoviedb.org/movie/\1/backdrops</url>" dest="7"> + <expression noclean="1"><a href="/movie/([^"]+)"></expression> + </RegExp> + <RegExp input="$$7" output="<url function="GMFFS">\1</url>" dest="9"> + <expression noclean="1"><url>([^<]+)</url></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GMA> + + <GMAF dest="10"> + <RegExp input="$$9" output="<details>\1</details>" dest="10"> + <RegExp input="$$1" output="<url>http://www.themoviedb.org/movie/\1/backdrops</url>" dest="7"> + <expression repeat="yes" noclean="1"><a href="/movie/([^"]+)"></expression> + </RegExp> + <RegExp input="$$7" output="<url function="GMFFS">\1</url>" dest="9"> + <expression repeat="yes" noclean="1"><url>([^<]+)</url></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GMAF> + + <GMFFS dest="10" clearbuffers="no"> + <RegExp input="$$1" output="<thumb>http://www.themoviedb.org/image/backdrops/\1</thumb>" dest="6+"> + <expression repeat="yes" noclean="1"><a href="/image/backdrops/([^"]+)"</expression> + </RegExp> + </GMFFS> + + <GMFFSS clearbuffers="no" dest="5"> + <RegExp input="$$1" output="<thumb>\1</thumb>" dest="6+"> + <expression repeat="yes" noclean="1"><backdrop size="original">([^<]+)</backdrop></expression> + </RegExp> + </GMFFSS> + + <GMF dest="10"> + <RegExp input="$$9" output="<details>\1</details>" dest="10"> + <RegExp input="$$1" output="<url>http://www.kinopoisk.ru\1</url>" dest="7"> + <expression repeat="yes" noclean="1"><a href="([^"]+)"><img[^>]+alt="Просмотр фото"></expression> + </RegExp> + <RegExp input="$$7" output="<url function="GMFF">\1</url>" dest="9"> + <expression repeat="yes" noclean="1"><url>([^<]+)</url></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GMF> + + <GMFF clearbuffers="no" dest="5"> + <RegExp input="$$1" output="<thumb>http://www.kinopoisk.ru\1</thumb>" dest="6+"> + <expression noclean="1"><img.+?src='([^']*kadr[^']*)'.+?style="border:1px solid #777" onLoad=''></a></td></expression> + </RegExp> + </GMFF> + + <GMFR dest="10"> + <RegExp input="$$9" output="<details>\1</details>" dest="10"> + <RegExp input="$$1" output="<url>http://www.kinopoisk.ru\1</url>" dest="7"> + <expression repeat="yes" noclean="1"><a href="([^"]+)" class=all></expression> + </RegExp> + <RegExp input="$$7" output="<url function="GMFFR">\1</url>" dest="9"> + <expression repeat="yes" noclean="1"><url>([^<]+)</url></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GMFR> + + <GMFFR clearbuffers="no" dest="5"> + <RegExp input="$$1" output="<thumb>http://www.kinopoisk.ru\1</thumb>" dest="6+"> + <expression noclean="1"><img.+?src='([^']*wallpaper[^']*)'.+?style="border:1px solid #777" onLoad=''></a></td></expression> + </RegExp> + </GMFFR> + + + <GMP dest="10"> + <RegExp input="$$9" output="<details>\1</details>" dest="10"> + <RegExp input="$$1" output="<url>http://www.kinopoisk.ru\1</url>" dest="7"> + <expression repeat="yes" noclean="1"><a href="([^"]+)"><img[^>]+alt="Просмотр постера"></expression> + </RegExp> + <RegExp input="$$7" output="<url function="GMPP">\1</url>" dest="9"> + <expression repeat="yes" noclean="1"><url>([^<]+)</url></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GMP> + + <GMPP clearbuffers="no" dest="5"> + <RegExp input="$$1" output="<thumb>http://www.kinopoisk.ru\1</thumb>" dest="6+"> + <expression noclean="1"><img.+?src='([^']*poster[^']*)'.+?style="border:1px solid #777" onLoad=''></a></td></expression> + </RegExp> + </GMPP> + + + <GetPoster dest="5"> + <RegExp input="$$6" output="<details>\1</details>" dest="5"> + <expression noclean="1"></expression> + </RegExp> + </GetPoster> + + <GetFanart dest="5"> + <RegExp input="$$6" output="<details><fanart>\1</fanart></details>" dest="5"> + <expression noclean="1"></expression> + </RegExp> + </GetFanart> + + + <STT dest="10"> + <RegExp input="$$1" output="\2" dest="8"> + <expression noclean="2"><table(.*?)Прокат:(.*?)</table></expression> + </RegExp> + <RegExp input="$$8" output="<details><studio>\1</studio></details>" dest="10"> + <expression trim="1"><a [^>]*>([^<]+)</a></expression> + </RegExp> + </STT> + + <PEOPLE dest="10"> + <RegExp input="$$7" output="<details>\1</details>" dest="10"> + <!-- actors --> + <RegExp input="$$1" output="\1" dest="8"> + <expression noclean="1"><tr><td colspan=3><a name="actor"></td></tr>(.*?)<tr><td colspan=3 style="border-top:1px solid #ccc"><br /></td></tr></expression> + </RegExp> + <RegExp input="$$8" output="<actor><name>\3</name><thumb spoof="http://www.kinopoisk.ru/">http://www.kinopoisk.ru/\1</thumb></actor>" dest="7"> + <expression trim="3" repeat="yes">img src="([^"]+)" width=52 style="border:1px solid #ccc"(.+?)a href=".*?people[^"]*" class="all">([^<]+)</a></expression> + </RegExp> + + <!-- director --> + <RegExp input="$$1" output="\1" dest="8"> + <expression noclean="1"><tr><td colspan=3><a name="director"></td></tr>(.*?)<tr><td colspan=3 style="border-top:1px solid #ccc"><br /></td></tr></expression> + </RegExp> + <RegExp input="$$8" output="<director>\3</director>" dest="7+"> + <expression trim="3" repeat="yes">img src="([^"]+)" width=52 style="border:1px solid #ccc"(.+?)a href=".*?people[^"]*" class="all">([^<]+)</a></expression> + </RegExp> + + <!-- writers --> + <RegExp input="$$1" output="\1" dest="8"> + <expression noclean="1"><tr><td colspan=3><a name="writer"></td></tr>(.*?)<tr><td colspan=3 style="border-top:1px solid #ccc"><br /></td></tr></expression> + </RegExp> + <RegExp input="$$8" output="<credits>\3</credits>" dest="7+"> + <expression trim="3" repeat="yes">img src="([^"]+)" width=52 style="border:1px solid #ccc"(.+?)a href=".*?people[^"]*" class="all">([^<]+)</a></expression> + </RegExp> + + <!-- producer --> + <RegExp input="$$1" output="\1" dest="8"> + <expression noclean="1"><tr><td colspan=3><a name="producer"></td></tr>(.*?)<tr><td colspan=3 style="border-top:1px solid #ccc"><br /></td></tr></expression> + </RegExp> + <RegExp input="$$8" output="<producer>\3</producer>" dest="7+"> + <expression trim="3" repeat="yes">img src="([^"]+)" width=52 style="border:1px solid #ccc"(.+?)a href=".*?people[^"]*" class="all">([^<]+)</a></expression> + </RegExp> + + <expression noclean="1"></expression> + </RegExp> + </PEOPLE> +</scraper> diff --git a/system/scrapers/video/MyMovies.png b/system/scrapers/video/MyMovies.png Binary files differnew file mode 100644 index 0000000000..b822706a9c --- /dev/null +++ b/system/scrapers/video/MyMovies.png diff --git a/system/scrapers/video/MyMovies.xml b/system/scrapers/video/MyMovies.xml new file mode 100644 index 0000000000..1598be8a7e --- /dev/null +++ b/system/scrapers/video/MyMovies.xml @@ -0,0 +1,333 @@ +п»ї<?xml version="1.0" encoding="UTF-8"?> +<scraper framework="1.1" date="2009-08-08" name="MyMovies.it" content="movies" thumb="MyMovies.png" language="it"> + + <GetSettings dest="3"> + <RegExp input="$$5" output="<settings>\1</settings>" dest="3"> + <RegExp input="$$1" output="<setting label="Full Cast Credits" type="bool" id="fullcredits" default="true"></setting>" dest="5"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting label="Full Cast Credits > Get Director(s)" type="bool" id="viewDirector" default="true"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting label="Full Cast Credits > Get Actors" type="bool" id="viewActors" default="true"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting label="Full Cast Credits > Get Filmmakers" type="bool" id="viewFilmmakers" default="true"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting label="Full Cast Credits > Cast Separeted by Fake Actor" type="bool" id="fakeactor" default="false"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting type="sep"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting label="Get All Thumbs" type="bool" id="thumbs" default="true"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting label="Get Trailer" type="bool" id="trailer" default="true"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting label="Get Critic" type="bool" id="critic" default="false"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting label="Get Pubblic" type="bool" id="public" default="false"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting label="Get Forum" type="bool" id="forum" default="false"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting label="Get Frasi Celebri" type="bool" id="frasi_celebri" default="false"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting label="Rating x of 10" type="bool" id="rating_ten" default="false"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting type="sep"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting label="Director, Critics, and Public Rating" type="bool" id="other_rating" default="true"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting label="Director, Critics, and Public Rating > Insert into:" type="labelenum" values="mpaa|outline|studio" id="tag_other_rating" default="mpaa"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetSettings> + + <NfoUrl dest="3"> + <RegExp input="$$1" output="\1" dest="3"> + <expression noclean="1">(http://www\.mymovies\.it/dizionario/recensione\.asp\?id=[0-9]+)</expression> + </RegExp> + </NfoUrl> + + <CreateSearchUrl dest="3"> + <RegExp input="$$1" output="http://www.mymovies.it/database/ricerca/?q=\1" dest="3"> + <expression noclean="1"/> + </RegExp> + </CreateSearchUrl> + + <GetSearchResults dest="8"> + <RegExp input="$$5" output="<?xml version="1.0" encoding="iso-8859-1" standalone="yes"?><results>\1</results>" dest="8"> + <RegExp input="$$1" output="<entity><title>\2 (\3, \4)</title><url>http://www.mymovies.it/dizionario/recensione.asp?id=\1</url><id>\1</id></entity>" dest="5"> + <expression repeat="yes" noclean="1"><h3 style="margin:0px;">[^<]*<a href="http://www\.mymovies\.it/dizionario/recensione\.asp\?id=([0-9]+)" title="[^"]+">([^<]+)</a>[^7]+<div class="linkblu2" style="padding-right:7px; text-align:justify;">[^<]+Un film di <[ab][> ][^A-Z]*([^<]+)</[ab]>[^;]+anno=([^"]+)</expression> + </RegExp> + <expression noclean="1"/> + </RegExp> + </GetSearchResults> + + <GetDetails clearbuffers="no" dest="3"> + <RegExp input="$$5" output="<details>\1</details>" dest="3"> + <RegExp input="$$1" output="<title>\1</title><year>\2</year>" dest="5+"> + <expression noclean="1"><title>(.+).\((.+)\)</title></expression> + </RegExp> + + <RegExp input="$$1" output="<runtime>\1</runtime>" dest="5+"> + <expression noclean="1">durata ([0-9]*) min\.</expression> + </RegExp> + + <RegExp input="$$1" output="<genre>\1</genre>" dest="5+"> + <expression noclean="1"><a title="Film ([^"]*)" href="http://www.mymovies.it/film/([^"]*)">([^<]*)</a></expression> + </RegExp> + + <!-- Director(s) --> + <RegExp input="$$1" output="\1" dest="4"> + <expression noclean="1">Un film di (.+)Con <a</expression> + </RegExp> + + <RegExp input="$$4" output="<director>\1</director>" dest="5+"> + <expression repeat="yes" noclean="1">[^A-Z]*([^\.^,^<]+)[^\.^,]*[\.,]</expression> + </RegExp> + + <!-- Writer(s) --> + <RegExp output="<url cache="$$2-fullcredits.html" function="GetMovieWriters">http://www.mymovies.it/cast/?id=$$2</url>" dest="5+"> + <expression></expression> + </RegExp> + + <!-- Tagline --> + <RegExp input="$$1" output="<tagline>\1</tagline>" dest="5+"> + <expression noclean="1"><strong class="courier" style="font-size:23px; margin-bottom:10px; color:#ff0066; display:block;">([^<]+)</strong></expression> + </RegExp> + + <!-- Rating --> + <!-- rating map array --> + <RegExp conditional="rating_ten" output="<r>0.1-0.2</r><r>0.2-0.4</r><r>0.3-0.6</r><r>0.4-0.8</r><r>0.5-1.0</r><r>0.6-1.2</r><r>0.7-1.4</r><r>0.8-1.6</r><r>0.9-1.8</r><r>1.0-2.0</r><r>1.1-2.2</r><r>1.2-2.4</r><r>1.3-2.6</r><r>1.4-2.8</r><r>1.5-3.0</r><r>1.6-3.2</r><r>1.7-3.4</r><r>1.8-3.6</r><r>1.9-3.8</r><r>2.0-4.0</r><r>2.1-4.2</r><r>2.2-4.4</r><r>2.3-4.6</r><r>2.4-4.8</r><r>2.5-5.0</r><r>2.6-5.2</r><r>2.7-5.4</r><r>2.8-5.6</r><r>2.9-5.8</r><r>3.0-6.0</r><r>3.1-6.2</r><r>3.2-6.4</r><r>3.3-6.6</r><r>3.4-6.8</r><r>3.5-7.0</r><r>3.6-7.2</r><r>3.7-7.4</r><r>3.8-7.6</r><r>3.9-7.8</r><r>4.0-8.0</r><r>4.1-8.2</r><r>4.2-8.4</r><r>4.3-8.6</r><r>4.4-8.8</r><r>4.5-9.0</r><r>4.6-9.2</r><r>4.7-9.4</r><r>4.8-9.6</r><r>4.9-9.8</r><r>5.0-10.0</r><r>n.d- </r>" dest="4"> + <expression noclean="1"></expression> + </RegExp> + + <!-- Average --> + <RegExp input="$$1" output="\1.\2" dest="6"> + <expression noclean="1">Giudizio medio[^,]+([0-9]),?([0-9]?)/5</expression> + </RegExp> + + <RegExp conditional="rating_ten" input="$$4" output="\1" dest="6"> + <expression noclean="1">$$6-([^<]+)</expression> + </RegExp> + <RegExp input="$$6" output="<rating>\1</rating>" dest="5+"> + <expression noclean="1"></expression> + </RegExp> + + <!-- Others Rating --> + <RegExp conditional="other_rating" input="$$6" output="<$INFO[tag_other_rating]>\1</$INFO[tag_other_rating]>" dest="5+"> + <!-- Dizionari --> + <RegExp input="$$1" output="Dizionari: \1.\2" dest="6"> + <expression noclean="1"><b>Dizionari</b> \(([0-9]+),?([0-9]?)/5\)<br /></expression> + </RegExp> + <!-- Critica --> + <RegExp input="$$1" output=" / Critica: \1.\2" dest="6+"> + <expression noclean="1"><b>Critica</b> \(([0-9]+),?([0-9]?)/5\)<br /></expression> + </RegExp> + <!-- Pubblico --> + <RegExp input="$$1" output=" / Pubblico: \1.\2" dest="6+"> + <expression noclean="1"><b>Pubblico</b> \(([0-9]+),?([0-9]?)/5\)<br /></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + + <!-- Cast --> + <RegExp conditional="!fullcredits" input="$$6" output="\1" dest="5+"> + <RegExp input="$$1" output="\1" dest="4"> + <expression noclean="1">Un film di .*Con (.*\.)[^G]+Genere <a</expression> + </RegExp> + <RegExp input="$$4" output="<actor><name>\1</name></actor>" dest="6"> + <expression repeat="yes" noclean="1">[^A-Z]*([^\.^,^<]+)[^\.^,]*[\.,]</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + + <!-- Full Cast --> + <RegExp conditional="fullcredits" output="<url cache="$$2-fullcredits.html" function="GetMovieCast">http://www.mymovies.it/cast/?id=$$2</url>" dest="5+"> + <expression></expression> + </RegExp> + + <!-- Thumb --> + <RegExp conditional="!thumbs" input="$$1" output="<thumb>\1</thumb>" dest="5+"> + <expression noclean="1"><a title="Locandina[^"]+" href="[^"]+"><img style="border:solid 1px \#AEAEAE; padding:3px;" src="([^"]+)" width="150px" alt="Locandina[^"]+" /></a></expression> + </RegExp> + <!-- else All Thumbs --> + <RegExp conditional="thumbs" output="<url function="GetMovieThumbs">http://www.mymovies.it/poster/?id=$$2</url>" dest="5+"> + <expression noclean="1"></expression> + </RegExp> + + <!-- Trailer: not always present --> + <RegExp conditional="trailer" input="$$1" output="<url function="GetMovieTrailer">\1</url>" dest="5+"> + <expression><span class="rec_link_disattivo"><a title="[^"]+" href="([^"]+)">Trailer</a></span></expression> + </RegExp> + + <!-- Plot: It is used for report many information...It will be write on correct buffer (5) at the end --> + <RegExp input="$$1" output="\1" dest="10"> + <expression><p style="text-align:justify;">(.+?)</p></expression> + </RegExp> + + <!-- Plot: all that you want append... --> + <RegExp conditional="critic" output="<url function="GetMovieCritic">http://www.mymovies.it/recensioni/?id=$$2</url>" dest="5+"> + <expression noclean="1"></expression> + </RegExp> + <RegExp conditional="public" output="<url function="GetMoviePublic">http://www.mymovies.it/pubblico/?id=$$2</url>" dest="5+"> + <expression noclean="1"></expression> + </RegExp> + <RegExp conditional="forum" output="<url function="GetMovieForum">http://www.mymovies.it/forum/?id=$$2</url>" dest="5+"> + <expression noclean="1"></expression> + </RegExp> + <RegExp conditional="frasi_celebri" output="<url function="GetMovieFrasiCelebri">http://www.mymovies.it/battute/?id=$$2</url>" dest="5+"> + <expression noclean="1"></expression> + </RegExp> + + <!-- Plot: ...at the end --> + <!-- non serve a un cazzo l'url (giГ cached), ma ГЁ l'unico modo per vedere il buffer 10 istanziato nella GetDetails, e i buffer 11,12,13 istanziati nelle altre fuzioni --> + <RegExp output="<url cache="$$2-fullcredits.html" function="AppendToPlot">http://www.mymovies.it/cast/?id=$$2</url>" dest="5+"> + <expression></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetDetails> + + <GetMovieWriters clearbuffers="no" dest="5"> + <RegExp input="$$2" output="<details>\1</details>" dest="5"> + <RegExp input="$$1" output="<credits>\1</credits>" dest="2"> + <expression repeat="yes" noclean="1">http://www\.mymovies\.it/biografia/\?s=[^>]+>([^<]+)[^\(]+\(Soggetto\)</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetMovieWriters> + + <GetMovieTrailer clearbuffers="no" dest="5"> + <RegExp input="$$1" output="<details><trailer>\1</trailer></details>" dest="5"> + <expression noclean="1">flashvars="file=([^&]+)</expression> + </RegExp> + </GetMovieTrailer> + + <GetMovieThumbs clearbuffers="no" dest="5"> + <RegExp input="$$3" output="<details>\1</details>" dest="5"> + <!-- Mask unnecessary part of source...is necessary when there is one thumb only --> + <RegExp input="$$1" output="\1" dest="2"> + <expression noclean="1"><table cellpadding="0" cellspacing="3" style="width:100%">(.+?)</table></expression> + </RegExp> + <!-- Match All img --> + <RegExp input="$$2" output="<thumb>\1</thumb>" dest="3"> + <expression repeat="yes" noclean="1"><img[^/]+src="([^"]+)" /></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetMovieThumbs> + + <GetMovieCast clearbuffers="no" dest="5"> + <RegExp input="$$2" output="<details>\1</details>" dest="5"> + <!-- Director(s) --> + <RegExp conditional="viewDirector" input="$$3" output="\1" dest="2"> + <!-- Fake Actor for Section delimiter: Regia --> + <RegExp conditional="fakeactor" output="<actor><name>.Regia</name></actor>" dest="3"> + <expression noclean="1"></expression> + </RegExp> + <!-- Mask unnecessary part of source --> + <RegExp input="$$1" output="\1" dest="6"> + <expression noclean="1"><b>Regia</b>[^<]*</div>(.*?)<div class="linkblu" style="padding-left:12px; padding-right:7px; text-align:left; background-color:\#E1E1E1; margin-left:5px; margin-top:20px; font-size:120%;"></expression> + </RegExp> + <RegExp input="$$6" output="<actor><thumb>\1</thumb><name>\2</name><role>Regista</role></actor>" dest="3+"> + <expression repeat="yes" noclean="1"><div class="linkblu" style="padding:3px; padding-left:20px;">[^<]+(?:<a href="[^"]+" title="[^"]+"><img style="float:left; margin-right:3px;" src="([^"]+)" width="80" alt="[^"]+" /></a>)?[^\?]+\?[^>]+>([^<]+)</a></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + + <!-- Actors --> + <RegExp conditional="viewActors" input="$$3" output="\1" dest="2"> + <!-- Fake Actor for Section delimiter: Cast artistico --> + <RegExp conditional="fakeactor" output="<actor><name>.Cast artistico</name></actor>" dest="3+"> + <expression noclean="1"></expression> + </RegExp> + <!-- Mask unnecessary part of source --> + <RegExp input="$$1" output="\1" dest="6"> + <expression noclean="1"><b>Cast artistico</b>[^<]*</div>(.*?)(?(?=<div class="linkblu" style="padding-left:12px; padding-right:7px; text-align:left; background-color:\#E1E1E1; margin-left:5px; margin-top:20px; font-size:120%;">)<div class="linkblu" style="padding-left:12px; padding-right:7px; text-align:left; background-color:\#E1E1E1; margin-left:5px; margin-top:20px; font-size:120%;">|<div style="height:17px; clear:both;"></div>)</expression> + </RegExp> + <!-- \1: thumb (opt) \2: name \3: role (opt) --> + <RegExp input="$$6" output="<actor><thumb>\1</thumb><name>\2</name><role>\3</role></actor>" dest="3+"> + <expression repeat="yes" noclean="1"><div class="linkblu" style="padding:3px; padding-left:20px;">[^<]+(?:<a href="[^"]+" title="[^"]+"><img style="float:left; margin-right:3px;" src="([^"]+)" width="80" alt="[^"]+" /></a>)?[^\?]+\?[^>]+>([^<]+)</a>[^%]+(?:<span style="font-size:120%; padding-left:10px;">interpreta <strong><em>([^<]+)</strong></em></span>)?[^<]+<span style="font-size:120%;"></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + + <!-- Filmmakers --> + <RegExp conditional="viewFilmmakers" input="$$3" output="\1" dest="2"> + <!-- Fake Actor for Section delimiter: Filmmakers --> + <RegExp conditional="fakeactor" output="<actor><name>.Filmmakers</name></actor>" dest="3+"> + <expression noclean="1"></expression> + </RegExp> + <!-- Mask unnecessary part of source --> + <RegExp input="$$1" output="\1" dest="6"> + <expression noclean="1"><b>Filmmakers</b>[^<]*</div>(.*?)<div style="background-color:\#eeeeee; color:\#00336C; font-weight:bold; padding-left:5px;" ></expression> + </RegExp> + <!-- \1: thumb (opt) \2: name \3: role --> + <RegExp input="$$6" output="<actor><thumb>\1</thumb><name>\2</name><role>\3</role></actor>" dest="3+"> + <expression repeat="yes" noclean="1"><div class="linkblu" style="padding:3px; padding-left:20px;">[^<]+(?:<a href="[^"]+" title="[^"]+"><img style="float:left; margin-right:3px;" src="([^"]+)" width="80" alt="[^"]+" /></a>)?[^\?]+\?[^>]+>([^<]+)</a>[^\(]+\(([^\)]+)\)</span></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetMovieCast> + + <!-- Mancano i commenti estesi --> + <GetMovieCritic clearbuffers="no" dest="5"> + <RegExp input="$$2" output="\n\n:: Critica\n\1" dest="11"> + <RegExp input="$$1" output="\n\1 - \2\n\3\n" dest="2"> + <expression repeat="yes"><a href="http://www.mymovies.it/dizionario/critica.asp\?id=[^"]+">([^<]+)</a>[^<]*<strong><em>([^<]+)</em></strong>.+?style="color:\#333333; margin-bottom:10px;">(.+?)</span></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetMovieCritic> + + <!-- Mancano i commenti estesi --> + <GetMoviePublic clearbuffers="no" dest="5"> + <RegExp input="$$2" output="\n\n:: Pubblico\n\1" dest="12"> + <RegExp input="$$1" output="\1\n" dest="2"> + <expression repeat="yes"><div id="parziale[^"]+" class="linkrosa" style="color:\#333333; margin-bottom:10px;">(.+?)</div></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetMoviePublic> + + <!-- Mancano i commenti estesi --> + <GetMovieForum clearbuffers="no" dest="5"> + <RegExp input="$$2" output="\n\n:: Forum\n\1" dest="13"> + <RegExp input="$$1" output="\1\n" dest="2"> + <expression repeat="yes"><div id="parziale[^"]+" class="linkrosa" style="color:\#333333; margin-bottom:10px;">(.+?)</div></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetMovieForum> + + <GetMovieFrasiCelebri clearbuffers="no" dest="5"> + <RegExp input="$$2" output="\n\n:: Frasi Celebri\n\1" dest="14"> + <RegExp input="$$1" output="\1\n" dest="2"> + <expression repeat="yes"><div style="text-align:justify;">(.+?)</div></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetMovieFrasiCelebri> + + <AppendToPlot clearbuffers="no" dest="5"> + <RegExp input="$$10$$11$$12$$13$$14" output="<details><plot>\1</plot></details>" dest="5"> + <expression noclean="1"></expression> + </RegExp> + </AppendToPlot> +</scraper> diff --git a/system/scrapers/video/adultcdmovies.jpg b/system/scrapers/video/adultcdmovies.jpg Binary files differnew file mode 100644 index 0000000000..a6310957d1 --- /dev/null +++ b/system/scrapers/video/adultcdmovies.jpg diff --git a/system/scrapers/video/adultcdmovies.xml b/system/scrapers/video/adultcdmovies.xml new file mode 100644 index 0000000000..0c87de6256 --- /dev/null +++ b/system/scrapers/video/adultcdmovies.xml @@ -0,0 +1,67 @@ +<scraper framework="1.0" date="2009-05-22" name="Adult CD Movies" content="movies" thumb="adultcdmovies.jpg" language="en"> + <NfoUrl dest="3"> + <RegExp input="$$1" output="http://www.adultcdmovies.net/\1.aspx" dest="3"> + <expression noclean="1">adultcdmovies.net/(.*).aspx</expression> + </RegExp> + </NfoUrl> + <CreateSearchUrl dest="2"> + <RegExp input="$$1" output="http://www.adultcdmovies.net/e-pmsearch.aspx?SearchType=0&SearchTerm=\1" dest="2"> + <expression noclean="1"></expression> + </RegExp> + </CreateSearchUrl> + <GetSearchResults dest="8"> + <RegExp input="$$5" output="<?xml version="1.0" encoding="iso-8859-1" standalone="yes"?><results>\1</results>" dest="8"> + <RegExp input="$$1" output="<entity><title>\2</title><url>http://www.adultcdmovies.net/\1.aspx?</url></entity>" dest="5"> + <expression repeat="yes"></td><td><a href="(adult-dvd-[^genre].[^\.]*)\.aspx\?">(.[^/]*)</a></td></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetSearchResults> + <GetDetails dest="3"> + <RegExp input="$$5" output="<details>\1</details>" dest="3"> + <RegExp input="$$1" output="<title>\1</title>" dest="5"> + <expression><title>(.*)</title></expression> + </RegExp> + <RegExp input="$$1" output="<year>\1</year>" dest="5+"> + <expression>Movie Year: ([0-9]+)</expression> + </RegExp> + <RegExp input="$$1" output="<director>\1 \2</director>" dest="5+"> + <expression>Director:<.[^>]*>(.[^/]*)</a></expression> + </RegExp> + <RegExp input="$$1" output="<manufacturer>\1 \2</manufacturer>" dest="5+"> + <expression>Studio: <.[^>]*>(.[^/]*)</a></expression> + </RegExp> + <RegExp input="$$1" output="<genre>Adult</genre>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="\1" dest="8"> + <expression noclean="1">Categories &amp; Genres:(.*)Starring</expression> + </RegExp> + <RegExp input="$$8" output="<genre>\1</genre>" dest="5+"> + <expression repeat="yes"><a href="adult-dvd-genre[^>]*>([^<]*)</a></expression> + </RegExp> + <RegExp input="$$1" output="<plot>\1</plot>" dest="5+"> + <expression></div><hr style="clear:both" />([^<]*)<p align="left"></expression> + </RegExp> + <RegExp input="$$1" output="<thumb spoof="http://www.adultcdmovies.net">http://www.adultcdmovies.net/images/Product/medium/\1.jpg</thumb><thumb spoof="http://www.adultcdmovies.net">http://www.adultcdmovies.net/images/Product/large/\1.jpg</thumb>" dest="5+"> + <expression>images/Product/medium/([0-9]+)\.jpg</expression> + </RegExp> + <RegExp input="$$7" output="<genre>Adult\1</genre>" dest="5+"> + <RegExp input="$$6" output=" / \1" dest="7"> + <RegExp input="$$1" output="\1" dest="6"> + <expression>Categories &amp; Genres:(.*)Director:</expression> + </RegExp> + <expression repeat="yes"><a href=".[^"]*">(.[^/]*) DVD</a></expression> + </RegExp> + <expression></expression> + </RegExp> + <RegExp input="$$6" output="<actor><name>\1</name></actor>" dest="5+"> + <RegExp input="$$1" output="\1" dest="6"> + <expression clear="yes">Starring:(.*)E-mail this movie to a friend</expression> + </RegExp> + <expression repeat="yes">\?">(.[^/]*)</a></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetDetails> +</scraper> diff --git a/system/scrapers/video/adultdvdempire.jpg b/system/scrapers/video/adultdvdempire.jpg Binary files differnew file mode 100644 index 0000000000..9f9e944bd5 --- /dev/null +++ b/system/scrapers/video/adultdvdempire.jpg diff --git a/system/scrapers/video/adultdvdempire.xml b/system/scrapers/video/adultdvdempire.xml new file mode 100644 index 0000000000..36781ba3f1 --- /dev/null +++ b/system/scrapers/video/adultdvdempire.xml @@ -0,0 +1,76 @@ +<?xml version="1.0" encoding="utf-8" standalone="yes"?> +<scraper framework="1.1" date="2009-09-19" name="Adult DVD Empire" content="movies" thumb="adultdvdempire.jpg" language="en"> + <NfoUrl dest="3"> + <RegExp input="$$1" output="<url>http://www.adultdvdempire.com/itempage.aspx?item_id=\1</url>" dest="3"> + <expression noclean="1">adultdvdempire.com/itempage.aspx?item_id=([0-9]*)</expression> + </RegExp> + </NfoUrl> + + <CreateSearchUrl dest="3"> + <RegExp input="$$1" output="<url>http://www.adultdvdempire.com/SearchTitlesPage.aspx?SearchString=\1</url>" dest="3"> + <expression noclean="1"></expression> + </RegExp> + </CreateSearchUrl> + + <GetSearchResults dest="6"> + <RegExp input="$$5" output="<?xml version="1.0" encoding="iso-8859-1" standalone="yes"?><results>\1</results>" dest="6"> + <RegExp input="$$1" output="<entity><title>\2</title><url>http://www.adultdvdempire.com/\1</url></entity>" dest="5"> + <expression repeat="yes">ListItem_ItemTitle"><a href="/([^"]*)">([^<]*)</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetSearchResults> + + <GetDetails dest="3"> + <RegExp input="$$5" output="<details>\1</details>" dest="3"> + <RegExp input="$$1" output="<thumb>http://\1.dvdempire.com/res/movies/\2h.jpg</thumb>" dest="5+"> + <expression>BoxCover_Container">[^>]*><img src="http://(.*?).dvdempire.com/res/movies/([^m]*)</expression> + </RegExp> + + <RegExp input="$$1" output="<thumb>http://\1.dvdempire.com/res/movies/\2bh.jpg</thumb>" dest="5+"> + <expression>BoxCover_Container">[^>]*><img src="http://(.*?).dvdempire.com/res/movies/([^m]*)</expression> + </RegExp> + + <RegExp input="$$1" output="<title>\1</title>" dest="5+"> + <expression>Item_Title">([^<]*)</expression> + </RegExp> + + <RegExp input="$$1" output="<studio>\1</studio>" dest="5+"> + <expression>StudioProductionRating">([^<]*)</expression> + </RegExp> + + <RegExp input="$$1" output="<year>\1</year>" dest="5+"> + <expression>Year: ([0-9]*)</expression> + </RegExp> + + <RegExp input="$$1" output="<tagline>\1</tagline><plot>\2</plot>" dest="5+"> + <expression>InfoTagLine">([^<]*)<[^>]*>[^ ]*([^<]*)<</expression> + </RegExp> + + <RegExp input="$$1" output="<plot>\1</plot>" dest="5+"> + <expression>Item_InfoContainer">([^<]*)<</expression> + </RegExp> + + <RegExp input="$$1" output="<actor><name>\3</name><thumb>http://images.dvdempire.com/pornstar/actors/\1.jpg</thumb></actor>" dest="5+"> + <expression repeat="yes"><a href="/([0-9]*)/([^"]*)-Pornstars.html">([^<]*)</expression> + </RegExp> + + <RegExp input="$$1" output="<genre>\1</genre>" dest="5+"> + <expression repeat="yes">media_id=[^i]*item_id=[^>]*>([^<]*)</expression> + </RegExp> + + <RegExp input="$$1" output="<runtime>\1</runtime>" dest="5+"> + <expression>>Length: ([^<]*)<</expression> + </RegExp> + + <RegExp input="$$1" output="<mpaa>\1</mpaa>" dest="5+"> + <expression>>Rating: ([^<]*)</expression> + </RegExp> + + <RegExp input="$$1" output="<director>\3</director>" dest="5+"> + <expression repeat="yes"><a href="/([0-9]*)/([^"]*)-Directors.html">([^<]*)</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetDetails> +</scraper> diff --git a/system/scrapers/video/adultfilmdatabase.xml b/system/scrapers/video/adultfilmdatabase.xml new file mode 100644 index 0000000000..31a10c5fdd --- /dev/null +++ b/system/scrapers/video/adultfilmdatabase.xml @@ -0,0 +1,61 @@ +<scraper framework="1.0" date="2009-05-22" name="adultfilm database" content="movies" thumb="AdultFilmdatabase.gif" language="en"> + <NfoUrl dest="3"> + <RegExp input="$$1" output="<url>http://www.adultfilmdatabase.com/video.cfm?videoid=\1<url>" dest="3"> + <expression noclean="1">adultfilmdatabase.com/video.cfm?videoid=([0-9]*)</expression> + </RegExp> + </NfoUrl> + <CreateSearchUrl dest="3"> + <RegExp input="$$1" output="<url spoof="http://www.adultfilmdatabase.com" post="true">http://www.adultfilmdatabase.com/lookup.cfm?SearchType=Video&amp;Action=Lookup&amp;Find=\1</url>" dest="3"> + <expression noclean="1"></expression> + </RegExp> + </CreateSearchUrl> + <GetSearchResults dest="6"> + <RegExp input="$$5" output="<?xml version="1.0" encoding="iso-8859-1" standalone="yes"?><results>\1</results>" dest="6"> + <RegExp input="$$1" output="\1" dest="4"> + <expression><h2>([^<]*)</h2></expression> + </RegExp> + <RegExp input="$$1" output="<entity><title>$$4</title><url>http://www.adultfilmdatabase.com/video.cfm?videoid=\1</url><id>\1</id></entity>" dest="5"> + <expression>/Front/([0-9]*)\.jpg</expression> + </RegExp> + <RegExp input="$$1" output="<entity><title>\2</title><url>http://www.adultfilmdatabase.com/video.cfm?videoid=\1</url><id>\1</id></entity>" dest="5+"> + <expression repeat="yes"><a HREF="/video\.cfm\?videoid=([0-9]*)"><U>([^<]*)</U></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetSearchResults> + <GetDetails dest="3"> + <RegExp input="$$5" output="<details>\1</details>" dest="3"> + <RegExp input="$$1" output="<thumb>http://www.adultfilmdatabase.com\1</thumb>" dest="5"> + <expression>(/Graphics/Boxes/[0-9]*/Front/$$2\.jpg)</expression> + </RegExp> + <RegExp input="$$1" output="<title>\1</title>" dest="5+"> + <expression trim="1" noclean="1"><h2>([^<]*)</h2></expression> + </RegExp> + <RegExp input="$$1" output="<actor><name>\1</name></actor>" dest="5+"> + <expression repeat="yes"><a HREF="/actor\.cfm\?actorid=[0-9]*"><U>([^<]*)</U></a><br></expression> + </RegExp> + <RegExp input="$$1" output="<runtime>\1 min</runtime>" dest="5+"> + <expression trim="1">Length:</td><td STYLE="font-size:11px;padding-left: 3px;" WIDTH="100%">([0-9]+)</td></expression> + </RegExp> + <RegExp input="$$1" output="<year>\1</year>" dest="5+"> + <expression>Year:</td><td STYLE="font-size:11px;padding-left: 3px;" WIDTH="100%">([0-9]+)</td></expression> + </RegExp> + <RegExp input="$$1" output="<genre>Adult</genre>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$6" output="<genre>\1</genre>" dest="5+"> + <RegExp input="$$1" output="\1" dest="6"> + <expression trim="1">Genres:</td>[^<]*<td STYLE="font-size:11px;padding-left: 3px;" WIDTH="100%">([^<]*)</td></expression> + </RegExp> + <expression repeat="yes" trim="1">([a-zA-Z ]+)</expression> + </RegExp> + <RegExp input="$$1" output="<plot>\1</plot>" dest="5+"> + <expression><tr><td STYLE="font-size:9pt;" COLSPAN="2"><BR>([^<]*)</td></tr></expression> + </RegExp> + <RegExp input="$$1" output="<director>\1</director>" dest="5+"> + <expression>/director\.cfm\?directorid=281">([^<]*)</a></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetDetails> +</scraper> diff --git a/system/scrapers/video/allocine.jpg b/system/scrapers/video/allocine.jpg Binary files differnew file mode 100644 index 0000000000..eb8aae9c42 --- /dev/null +++ b/system/scrapers/video/allocine.jpg diff --git a/system/scrapers/video/allocine.xml b/system/scrapers/video/allocine.xml new file mode 100644 index 0000000000..fdc6a70336 --- /dev/null +++ b/system/scrapers/video/allocine.xml @@ -0,0 +1,204 @@ +<?xml version="1.0" encoding="utf-8" standalone="yes"?> +<scraper framework="1.1" date="2009-08-08" name="allocine.fr" content="movies" thumb="allocine.jpg" language="fr"> + <GetSettings dest="3"> + <RegExp input="$$5" output="<settings>\1</settings>" dest="3"> + <RegExp input="$$1" output="<setting label="Activer les images HD et photos du film" type="bool" id="GetThumbnail" default="true"></setting>" dest="5+"> + <expression></expression> + </RegExp> + + <RegExp input="$$1" output="<setting label="Activer les Informations du film" type="bool" id="info" default="true"></setting>" dest="5+"> + <expression></expression> + </RegExp> + + <RegExp input="$$1" output="<setting label="Activer les Vignettes d'acteurs" type="bool" id="actor" default="falsetrue"></setting>" dest="5+"> + <expression></expression> + </RegExp> + + <RegExp input="$$1" output="<setting label="Activer les Fanart" type="bool" id="fanart" default="true"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetSettings> + + <NfoUrl dest="3"> + <RegExp input="$$1" output="http://www.allocine.fr/film/fichefilm_gen_cfilm=\1.html" dest="3"> + <expression clear="yes" noclean="1">allocine\.fr/film/fichefilm_gen_cfilm=([0-9]*)\.html</expression> + </RegExp> + </NfoUrl> + + <CreateSearchUrl dest="3"> + <RegExp input="$$1" output="http://www.allocine.fr/recherche/default.html?motcle=\1&rub=1&page=1" dest="3"> + <expression>([^\(]+)</expression> + </RegExp> + </CreateSearchUrl> + + <GetSearchResults dest="2"> + <RegExp input="$$5" output="<?xml version="1.0" encoding="iso-8859-1" standalone="yes"?><results>\1</results>" dest="2"> + <RegExp input="$$1" output="<entity><title>\2</title><url>http://www.allocine.fr/film/fichefilm_gen_cfilm=\1.html</url><id>\1</id></entity>" dest="5+"> + <expression repeat="yes"><h4><a href="/film/fichefilm_gen_cfilm=([0-9]+)\.html" class="link1">([^;#]*)</a</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetSearchResults> + + <GetDetails clearbuffers="no" dest="3"> + <RegExp input="$$5" output="<details>\1</details>" dest="3"> + <!-- TITLE --> + <RegExp input="$$1" output="<title>\1</title>" dest="5"> + <expression trim="1" noclean="1"><title>(.*?) \- AlloCin</expression> + </RegExp> + <!-- STUDIO --> + <RegExp conditional="info" input="$$1" output="<studio>\1</studio>" dest="5+"> + <expression>Distribu. par [^>]*>([^<]*)</expression> + </RegExp> + <!-- GENRE --> + <RegExp conditional="info" input="$$1" output="\1" dest="7"> + <expression noclean="1">Genre :([^:]*):</expression> + </RegExp> + + <RegExp conditional="info" input="$$6" output="<genre>\1</genre>" dest="5+"> + <RegExp input="$$7" output=" / \1" dest="6"> + <expression repeat="yes" noclean="1"><a href[^>]*>([^<]*)</a></expression> + </RegExp> + <expression> / (.*)</expression> + </RegExp> + <!-- YEAR --> + <RegExp conditional="info" input="$$1" output="<year>\1</year>" dest="5+"> + <expression>>Ann.e de production : ([^<]*)<</expression> + </RegExp> + <!-- DIRECTOR --> + <RegExp conditional="info" input="$$1" output="<director>\1</director>" dest="5+"> + <expression>R.alis. par <[^>]*>([^<]*)</expression> + </RegExp> + <!-- RUNTIME --> + <RegExp conditional="info" input="$$1" output="<runtime>\1</runtime>" dest="5+"> + <expression>Dur.e : ([^.]*)</expression> + </RegExp> + <!-- RATING VOTES --> + <RegExp conditional="info" input="$$1" output="\1#\2" dest="7"> + <expression><h4>Note moyenne : <([^>]*)>[^0-9]*([0-9]*)</expression> + </RegExp> + + <RegExp conditional="info" input="$$7" output="<rating>\1.toile(s)</rating><votes>\2</votes>" dest="5+"> + <expression>class="etoile_([0-9]*)"[^#]*#([0-9]*)</expression> + </RegExp> + <!-- MPAA --> + <RegExp conditional="info" input="$$1" output="<mpaa>\1</mpaa>" dest="5+"> + <expression>N. de visa[^>]*>[^>]*>[^>]*>[^>]*>[^>]*>([^<]*)<</expression> + </RegExp> + <!-- GetTagline --> + <RegExp conditional="info" input="$$1" output="<url function="GetTagline">http://www.allocine.fr/film/revuedepresse_gen_cfilm=\1.html</url>" dest="5+"> + <expression noclean="1">revuedepresse_gen_cfilm=([0-9]*)</expression> + </RegExp> + <!-- GetCredits --> + <RegExp conditional="info" input="$$1" output="<url function="GetCredits">http://www.allocine.fr/film/casting_gen_cfilm=\1.html</url>" dest="5+"> + <expression>casting_gen_cfilm=([0-9]*)</expression> + </RegExp> + <!-- GetActor --> + <RegExp conditional="info" input="$$1" output="<url function="GetActor">http://www.allocine.fr/film/casting_gen_cfilm=\1.html</url>" dest="5+"> + <expression>casting_gen_cfilm=([0-9]*)</expression> + </RegExp> + <!-- OUTLINE --> + <RegExp conditional="info" input="$$1" output="<outline>\1</outline><plot>\1</plot>" dest="5+"> + <expression><td valign="top" style="padding:10 0 0 0"><div align="justify"><h4>([^\n]*)</expression> + </RegExp> + <!-- PLOT --> + <RegExp conditional="info" input="$$1" output="<plot>\1</plot>" dest="5+"> + <expression><td valign="top" style="padding:10 0 0 0"><div align="justify"><h4>([^#]*)</h4></div></td>[\n\r\t]*</tr>[\n\r\t]*<tr>[\n\r\t]*<td valign</expression> + </RegExp> + <!-- Fanart --> + <RegExp input="$$1" output="<fanart>\1</fanart>" dest="12+"> + <expression>"([^"]*)" border="0" alt="" class="affichette</expression> + </RegExp> + + <RegExp conditional="fanart" input="$$1" output="<url cache="$$2-fanart.html"function="GetFanart">http://www.allocine.fr/film/galerievignette_gen_cfilm=\1.html</url>" dest="5+"> + <expression>galerievignette_gen_cfilm=([0-9]*)</expression> + </RegExp> + <!-- THUMB --> + <RegExp input="$$3" output="<url function="Getallocine" cache="$$2.html">\1</url>" dest="5+"> + <expression noclean="1"></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetDetails> + + <GetFanart dest="5"> + <RegExp input="$$10$$12" output="<details><fanart>\1</fanart></details>" dest="5"> + <RegExp input="$$1" output="\1/rsz/434/x/x/x/medias" dest="9+"> + <expression>var CRP_PATH = "([^"]*)/crp/80/80/x/x/medias"</expression> + </RegExp> + + <RegExp input="$$1" output="<thumb>$$9\1</thumb>" dest="10+"> + <expression repeat="yes" noclean="1">"fichier":"([^"]*)"</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetFanart> + + <Getallocine dest="5"> + <RegExp input="$$10$$11" output="<details>\1</details>" dest="5"> + <RegExp input="$$1" output="<thumb>http://a69.g.akamai.net/n/69/10688/v1/img5.allocine.fr/acmedia/medias/nmedia/\1</thumb>" dest="11+"> + <expression noclean="1"><img src="http://a69.g.akamai.net/n/69/10688/v1/img5.allocine.fr/acmedia/medias/nmedia/([^"]+)" [^>]*></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </Getallocine> + + <GetCredits clearbuffers="no" dest="5"> + <RegExp input="$$3" output="<details>\1</details>" dest="5"> + <RegExp input="$$1" output="\1" dest="7"> + <expression noclean="1">Sc.nario([^(]*)Equipe technique</expression> + </RegExp> + <RegExp input="$$6" output="<credits>\1</credits>" dest="3+"> + <RegExp input="$$7" output=" / \1" dest="6"> + <expression repeat="yes" noclean="1">>([^<]*)</a</expression> + </RegExp> + <expression noclean="1"> / (.*)</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetCredits> + + <GetActor clearbuffers="no" dest="5"> + <RegExp input="$$3" output="<details>\1</details>" dest="5"> + <RegExp input="$$1" output="\1" dest="6"> + <expression noclean="1"><b>Acteurs</b>(.*)<b>Production</b></expression> + </RegExp> + + <RegExp input="$$6" output="<actor><name>\3</name><role>\1</role></actor>" dest="3+"> + <expression repeat="yes" noclean="1,2,3"><h5>([^<]*)</h5></td>[^<]*<[^>]*><h5><a href="/([^"]*)"[^>]*>([^<]*)<</expression> + </RegExp> + + <RegExp conditional="actor" input="$$6" output="<url function="GetActorThumb">http://www.allocine.fr/personne/fichepersonne_gen_cpersonne=\2.html</url><id>\2</id>\n" dest="3+"> + <expression repeat="yes" noclean="1,2,3" ><h5>([^<]*)</h5></td>[^<]*<[^>]*><h5><a href="/personne/fichepersonne_gen_cpersonne=([0-9]*).html"[^>]*>([^<]*)<</expression> + </RegExp> + + <expression noclean="1"></expression> + </RegExp> + </GetActor> + + <GetActorThumb clearbuffers="no" dest="5"> + <RegExp input="$$2" output="<details>\1</details>" dest="5"> + <RegExp conditional="actor" input="$$1" output="<thumb>\1</thumb>" dest="3"> + <expression>src="([^"]*)" width="120" height="160" border="0"><br /></expression> + </RegExp> + + <RegExp conditional="actor" input="$$1" output="<name>\1</name>" dest="4"> + <expression><title>([^<]*)</expression> + </RegExp> + + <RegExp conditional="actor" input="$$1" output="<actor>$$3$$4<role>test</role></actor>" dest="2+"> + <expression>src="([^"]*)" width="120" height="160" border="0"><br /></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetActorThumb> + + <GetTagline clearbuffers="no" dest="5"> + <RegExp input="$$1" output="<details><tagline>\1</tagline></details>" dest="5+"> + <expression clean="1" trim="1"><div align="justify" style="padding: 5 0 5 0"><h4>([^<]*)</expression> + </RegExp> + </GetTagline> +</scraper> + diff --git a/system/scrapers/video/amazonuk.png b/system/scrapers/video/amazonuk.png Binary files differnew file mode 100644 index 0000000000..e872b16315 --- /dev/null +++ b/system/scrapers/video/amazonuk.png diff --git a/system/scrapers/video/amazonuk.xml b/system/scrapers/video/amazonuk.xml new file mode 100644 index 0000000000..0bf5626b7e --- /dev/null +++ b/system/scrapers/video/amazonuk.xml @@ -0,0 +1,154 @@ +<?xml version="1.0" encoding="UTF-8"?> +<!-- Initial basic version doing Studio and Thumb believed to have been written by C-Quel --> +<!-- Then updated by John Lockwood to scrape Title, Year, MPAA, Runtime, Rating, Votes, Plot, Actors, Directors --> +<!-- This version 1.1 dated 12/01/09 includes fix by C-Quel for processing results from Amazon to match recent change --> +<!-- Version 1.1 also now supports the Writers field and fixes an issue to do with film titles --> +<scraper framework="1.0" date="2009-05-22" content="movies" name="Amazon UK" thumb="amazonuk.png" language="en"> + <CreateSearchUrl dest="3"> + <RegExp input="$$1" output="<url>http://www.amazon.co.uk/s/ref=nb_ss_d_h_?url=search-alias%3Ddvd&amp;field-keywords=\1</url>" dest="3"> + <expression noclean="1"></expression> + </RegExp> + </CreateSearchUrl> + <GetSearchResults dest="8"> + <RegExp input="$$5" output="<?xml version="1.0" encoding="iso-8859-1" standalone="yes"?><results>\1</results>" dest="8"> + <RegExp input="$$1" output="<entity><title>\2</title><url>\1</url></entity>" dest="5"> + <expression repeat="yes" clear="yes" noclean="1">productTitle"><a href="([^"]*)">([^<]*)</a></expression> + </RegExp> + <expression clear="yes" noclean="1"></expression> + </RegExp> + </GetSearchResults> + <GetDetails clearbuffers="no" dest="3"> + <RegExp input="$$5" output="<details>\1</details>" dest="3"> + <RegExp input="$$1" output="<title>\1</title>" dest="5"> + <expression noclean="1" trim="1"><title>[Amazon.co.uk: ]*([^\:\(]*) </expression> + </RegExp> + + <RegExp input="$$1" output="<year>\1</year>" dest="5+"> + <expression trim="1">[ \[\(]([0-9]{4})[ \]\)][^<]*</span></expression> + </RegExp> + + <RegExp input="$$1" output="<top250>\1</top250>" dest="5+"> + <expression>Top 250: #([0-9]*)</a></expression> + </RegExp> + + <RegExp input="$$9" output="<mpaa>Exempt</mpaa>" dest="5+"> + <RegExp input="$$1" output="\1" dest="9"> + <expression><b>Classification:</b>.*(Exempt)</expression> + </RegExp> + <expression>(exempt)</expression> + </RegExp> + <RegExp input="$$9" output="<mpaa>U</mpaa>" dest="5+"> + <RegExp input="$$1" output="\1" dest="9"> + <expression><b>Classification:</b>[^_]*/(u)-rating</expression> + </RegExp> + <expression>(u)</expression> + </RegExp> + <RegExp input="$$9" output="<mpaa>Uc</mpaa>" dest="5+"> + <RegExp input="$$1" output="\1" dest="9"> + <expression><b>Classification:</b>[^_]*/(u[cC])-rating</expression> + </RegExp> + <expression>(uc)</expression> + </RegExp> + <RegExp input="$$9" output="<mpaa>12A</mpaa>" dest="5+"> + <RegExp input="$$1" output="\1" dest="9"> + <expression><b>Classification:</b>[^_]*/(12[aA])-rating</expression> + </RegExp> + <expression>(12a)</expression> + </RegExp> + <RegExp input="$$9" output="<mpaa>12</mpaa>" dest="5+"> + <RegExp input="$$1" output="\1" dest="9"> + <expression><b>Classification:</b>[^_]*/(12)-rating</expression> + </RegExp> + <expression>(12)</expression> + </RegExp> + <RegExp input="$$9" output="<mpaa>15</mpaa>" dest="5+"> + <RegExp input="$$1" output="\1" dest="9"> + <expression><b>Classification:</b>[^_]*/(15)-rating</expression> + </RegExp> + <expression>(15)</expression> + </RegExp> + <RegExp input="$$9" output="<mpaa>PG</mpaa>" dest="5+"> + <RegExp input="$$1" output="\1" dest="9"> + <expression><b>Classification:</b>[^_]*/(pg)-rating</expression> + </RegExp> + <expression>(pg)</expression> + </RegExp> + <RegExp input="$$9" output="<mpaa>18</mpaa>" dest="5+"> + <RegExp input="$$1" output="\1" dest="9"> + <expression><b>Classification:</b>[^_]*/(18)-rating</expression> + </RegExp> + <expression>(18)</expression> + </RegExp> + <RegExp input="$$9" output="<mpaa>R18</mpaa>" dest="5+"> + <RegExp input="$$1" output="\1" dest="9"> + <expression><b>Classification:</b>[^_]*/([rR]18)-rating</expression> + </RegExp> + <expression>(R18)</expression> + </RegExp> + <RegExp input="$$9" output="<mpaa>UNRATED</mpaa>" dest="5+"> + <RegExp input="$$1" output="\1" dest="9"> + <expression><b>Classification:</b>[^_]*/(unrated)-rating</expression> + </RegExp> + <expression>(unrated)</expression> + </RegExp> + + <RegExp input="$$1" output="<mpaa>\1</mpaa>" dest="5+"> + <expression><b>Classification:</b> ([^(]*) \(</expression> + </RegExp> + + <RegExp input="$$1" output="<certification>\1</certification>" dest="5+"> + <expression repeat="yes">Classification:</b>[^>]*alt="([0-9]*)"</expression> + </RegExp> + <RegExp input="$$1" output="<tagline>\1</tagline>" dest="5+"> + <expression><h5>Tagline:</h5>([^<]*)</expression> + </RegExp> + <RegExp input="$$1" output="<runtime>\1</runtime>" dest="5+"> + <expression trim="1">Run Time:</b>[^0-9]*([^<]*)</li></expression> + </RegExp> + <RegExp input="$$1" output="<rating>\1.\2</rating><votes>\3</votes>" dest="5+"> + <expression noclean="1">Average Customer Review</b>[^_]*stars-([0-9])-([0-9])[^)]*>([0-9]*) customer reviews</a>\)</expression> + </RegExp> + <RegExp input="$$1" output="<genre>\1</genre>" dest="5+"> + <expression repeat="yes">"/Sections/Genres/[^/]*/">([^<]*)</a></expression> + </RegExp> + <RegExp input="$$1" output="<studio>\1</studio>" dest="5+"> + <expression>Studio:</b> ([^<]*)</li></expression> + </RegExp> + <RegExp input="$$1" output="<outline>\2</outline><plot>\2</plot>" dest="5+"> + <expression trim="1">Plot (Outline|Summary):</h5>([^<]*)</expression> + </RegExp> + + <RegExp input="$$1" output="<plot>\1</plot>" dest="5+"> + <expression><b>Amazon.co.uk Review</b><br />\n ([^\n]*)</expression> + </RegExp> + <RegExp input="$$1" output="<plot>\1</plot>" dest="5+"> + <expression><b>Synopsis</b><br />\n ([^\n]*)</expression> + </RegExp> + <RegExp input="$$1" output="<thumb>\101.L.jpg</thumb>" dest="5+"> + <expression noclean="1">"original_image", "([^"]*)AA2[0-9]0_\.jpg"</expression> + </RegExp> + + <RegExp input="$$9" output="<credits>\1</credits>" dest="5+"> + <RegExp input="$$1" output="\1" dest="9"> + <expression noclean="1"><b>Writers:</b> ([^\n]*</a>)</expression> + </RegExp> + <expression noclean="1" repeat="yes">[^>]*>([^<]+)</a></expression> + </RegExp> + + <RegExp input="$$9" output="<director>\1</director>" dest="5+"> + <RegExp input="$$1" output="\1" dest="9"> + <expression noclean="1"><b>Directors:</b> ([^\n]*</a>)</expression> + </RegExp> + <expression noclean="1" repeat="yes">[^>]*>([^<]+)</a></expression> + </RegExp> + + <RegExp input="$$9" output="<actor><name>\1</name></actor>" dest="5+"> + <RegExp input="$$1" output="\1" dest="9"> + <expression noclean="1"><b>Actors:</b> ([^\n]*</a>)</expression> + </RegExp> + <expression noclean="1" repeat="yes">[^>]*>([^<]+)</a></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetDetails> +</scraper> diff --git a/system/scrapers/video/amazonus.png b/system/scrapers/video/amazonus.png Binary files differnew file mode 100644 index 0000000000..6a3f929d82 --- /dev/null +++ b/system/scrapers/video/amazonus.png diff --git a/system/scrapers/video/amazonus.xml b/system/scrapers/video/amazonus.xml new file mode 100644 index 0000000000..5a7e3a44eb --- /dev/null +++ b/system/scrapers/video/amazonus.xml @@ -0,0 +1,125 @@ +<?xml version="1.0" encoding="UTF-8"?> +<!-- Initial basic version doing Studio and Thumb believed to have been written by C-Quel --> +<!-- Then updated by John Lockwood to scrape Title, Year, MPAA, Runtime, Rating, Votes, Plot, Actors, Directors --> +<!-- This version 1.1 dated 12/01/09 includes fix by C-Quel for processing results from Amazon to match recent change --> +<!-- Version 1.1 also now supports the Writers field --> +<scraper framework="1.0" date="2009-05-22" content="movies" name="Amazon US" thumb="amazonus.png" language="en"> + <CreateSearchUrl dest="3"> + <RegExp input="$$1" output="<url>http://www.amazon.com/s/ref=nb_ss_d_h_?url=search-alias%3Ddvd&amp;field-keywords=\1</url>" dest="3"> + <expression noclean="1"></expression> + </RegExp> + </CreateSearchUrl> + <GetSearchResults dest="8"> + <RegExp input="$$5" output="<?xml version="1.0" encoding="iso-8859-1" standalone="yes"?><results>\1</results>" dest="8"> + <RegExp input="$$1" output="<entity><title>\2</title><url>\1</url></entity>" dest="5"> + <expression repeat="yes" clear="yes" noclean="1">productTitle"><a href="([^"]*)">([^<]*)</a></expression> + </RegExp> + <expression clear="yes" noclean="1"></expression> + </RegExp> + </GetSearchResults> + <GetDetails clearbuffers="no" dest="3"> + <RegExp input="$$5" output="<details>\1</details>" dest="3"> + <RegExp input="$$1" output="<title>\1</title>" dest="5"> + <expression noclean="1"><title>[Amazon.com: ]*([^:]*)</expression> + </RegExp> + <RegExp input="$$1" output="<year>\1</year>" dest="5+"> + <expression trim="1">[ \[\(]([0-9]{4})[ \]\)][^<]*</span></expression> + </RegExp> + <RegExp input="$$1" output="<top250>\1</top250>" dest="5+"> + <expression>Top 250: #([0-9]*)</a></expression> + </RegExp> + + <RegExp input="$$9" output="<mpaa>G</mpaa>" dest="5+"> + <RegExp input="$$1" output="\1" dest="9"> + <expression><b>Rating: </b>[^_]*/(g)._</expression> + </RegExp> + <expression>(g)</expression> + </RegExp> + <RegExp input="$$9" output="<mpaa>PG</mpaa>" dest="5+"> + <RegExp input="$$1" output="\1" dest="9"> + <expression><b>Rating: </b>[^_]*/(pg)._</expression> + </RegExp> + <expression>(pg)</expression> + </RegExp> + <RegExp input="$$9" output="<mpaa>PG-13</mpaa>" dest="5+"> + <RegExp input="$$1" output="\1" dest="9"> + <expression><b>Rating: </b>[^_]*/(pg-13)._</expression> + </RegExp> + <expression>(pg-13)</expression> + </RegExp> + <RegExp input="$$9" output="<mpaa>R</mpaa>" dest="5+"> + <RegExp input="$$1" output="\1" dest="9"> + <expression><b>Rating: </b>[^_]*/(r)._</expression> + </RegExp> + <expression>(r)</expression> + </RegExp> + <RegExp input="$$9" output="<mpaa>NC-17</mpaa>" dest="5+"> + <RegExp input="$$1" output="\1" dest="9"> + <expression><b>Rating: </b>[^_]*/(nc-17)._</expression> + </RegExp> + <expression>(nc-17)</expression> + </RegExp> + <RegExp input="$$9" output="<mpaa>UNRATED</mpaa>" dest="5+"> + <RegExp input="$$1" output="\1" dest="9"> + <expression><b>Rating: </b>[^_]*/(unrated)._</expression> + </RegExp> + <expression>(unrated)</expression> + </RegExp> + + + <RegExp input="$$1" output="<certification>\1</certification>" dest="5+"> + <expression repeat="yes">Classification:</b>[^>]*alt="([0-9]*)"</expression> + </RegExp> + <RegExp input="$$1" output="<tagline>\1</tagline>" dest="5+"> + <expression><h5>Tagline:</h5>([^<]*)</expression> + </RegExp> + <RegExp input="$$1" output="<runtime>\1</runtime>" dest="5+"> + <expression trim="1">Run Time:</b>[^0-9]*([^<]*)</li></expression> + </RegExp> + <RegExp input="$$1" output="<rating>\1.\2</rating><votes>\3</votes>" dest="5+"> + <expression noclean="1">Average Customer Review</b>[^_]*stars-([0-9])-([0-9])[^)]*>([0-9]*) customer reviews</a>\)</expression> + </RegExp> + <RegExp input="$$1" output="<genre>\1</genre>" dest="5+"> + <expression repeat="yes">"/Sections/Genres/[^/]*/">([^<]*)</a></expression> + </RegExp> + <RegExp input="$$1" output="<studio>\1</studio>" dest="5+"> + <expression>Studio:</b> ([^<]*)</li></expression> + </RegExp> + <RegExp input="$$1" output="<outline>\2</outline><plot>\2</plot>" dest="5+"> + <expression trim="1">Plot (Outline|Summary):</h5>([^<]*)</expression> + </RegExp> + <RegExp input="$$1" output="<plot>\1</plot>" dest="5+"> + <expression trim="1"><b>Product Description</b><br /[^>]*>([^<]+)</expression> + </RegExp> + <RegExp input="$$1" output="<thumb>\101.L.jpg</thumb>" dest="5+"> + <expression noclean="1">"original_image", "([^"]*)AA2[0-9]0_\.jpg"</expression> + </RegExp> + + + <RegExp input="$$9" output="<credits>\1</credits>" dest="5+"> + <RegExp input="$$1" output="\1" dest="9"> + <expression noclean="1"><b>Writers:</b> ([^\n]*</a>)</expression> + </RegExp> + <expression noclean="1" repeat="yes">[^>]*>([^<]+)</a></expression> + </RegExp> + + + <RegExp input="$$9" output="<director>\1</director>" dest="5+"> + <RegExp input="$$1" output="\1" dest="9"> + <expression noclean="1"><b>Directors:</b> ([^\n]*</a>)</expression> + </RegExp> + <expression noclean="1" repeat="yes">[^>]*>([^<]+)</a></expression> + </RegExp> + + + <RegExp input="$$9" output="<actor><name>\1</name></actor>" dest="5+"> + <RegExp input="$$1" output="\1" dest="9"> + <expression noclean="1"><b>Actors:</b> ([^\n]*</a>)</expression> + </RegExp> + <expression noclean="1" repeat="yes">[^>]*>([^<]+)</a></expression> + </RegExp> + + <expression noclean="1"></expression> + </RegExp> + </GetDetails> +</scraper> diff --git a/system/scrapers/video/asiandb.gif b/system/scrapers/video/asiandb.gif Binary files differnew file mode 100644 index 0000000000..8ab7b9f271 --- /dev/null +++ b/system/scrapers/video/asiandb.gif diff --git a/system/scrapers/video/asiandb.xml b/system/scrapers/video/asiandb.xml new file mode 100644 index 0000000000..a717a2e02c --- /dev/null +++ b/system/scrapers/video/asiandb.xml @@ -0,0 +1,103 @@ +<scraper framework="1.0" date="2009-05-22" name="AsianDB" content="movies" thumb="asiandb.gif" language="en"> + <NfoUrl dest="3"> + <RegExp input="$$1" output="<url>http://www.asiandb.com/browse/movie_detail.pfm?code=\1<url>" dest="3"> + <expression noclean="1">asiandb.com/browse/movie_detail.pfm?code=([0-9]*)</expression> + </RegExp> + </NfoUrl> + <CreateSearchUrl dest="3"> + <RegExp input="$$1" output="<url spoof="http://www.asiandb.com/browse/advanced.pfm" post="true">http://www.asiandb.com/browse/advanced.pfm?text=\1&amp;x=0&amp;y=0&amp;part=movie</url>" dest="3"> + <expression noclean="1"></expression> + </RegExp> + </CreateSearchUrl> + +<GetSearchResults dest="3"> + <RegExp input="$$5" output="<?xml version="1.0" encoding="iso-8859-1" standalone="yes"?><results>\1</results>" dest="3"> + <RegExp input="$$1" output="<entity><title>\2</title><url>http://www.asiandb.com/browse/movie_detail.pfm?code=\1</url><id>\1</id></entity>" dest="5"> + <expression repeat="yes"><a href=/browse/movie_detail.pfm\?code=([0-9]+)><font class=e3b>(.[^<]*)</font></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> +</GetSearchResults> + +<GetDetails dest="3"> + <RegExp input="$$5" output="<details>\1</details>" dest="3"> + +<!--Title --> + <RegExp input="$$1" output="<title>\1</title>" dest="5"> + <expression trim="1" noclean="1"><td valign=top><font class=e5b>([^<]*)<</expression> + </RegExp> + +<!--Original Title + <RegExp input="$$1" output="<originaltitle>\1</originaltitle>" dest="5"> + <expression>Titolo originale.[^z]*ze="2">([^)<]*)</expression> + </RegExp> --> + +<!--Year Film--> + <RegExp input="$$1" output="<year>\1</year>" dest="5+"> + <expression><font class=e3>\(([0-9]+)\)</expression> + </RegExp> + +<!--Director--> + <RegExp input="$$7" output="<director>\1</director>" dest="5+"> + <RegExp input="$$1" output="\1" dest="7"> + <expression trim="1" noclean="1"><font class=h0b color=white>Crew</font>(.*)</expression> + </RegExp> + <expression repeat="yes"><b>Director: </b[^\n]*>([A-Za-z][^<]*)<</expression> + </RegExp> + +<!--Runtime Film--> + <RegExp input="$$1" output="<runtime>\1</runtime>" dest="5+"> + <expression><b>Runtime:</b>([^<]*)<</expression> + </RegExp> + +<!--Thumbnail --> + <RegExp input="$$1" output="<thumb>http://www.asiandb.com\1</thumb>" dest="5+"> + <expression noclean="1">(/data/title/mini/[^" ]*)</expression> + </RegExp> + +<!--User rating--> + <RegExp input="$$1" output="<rating>\1</rating>" dest="5+"> + <expression>>Users</td>.*img src=/images/rate/[0-9]+.gif>\(([0-9]+)/10\)</expression> + </RegExp> + +<!--Vote count--> + <RegExp input="$$1" output="<votes>\1</votes>" dest="5+"> + <expression><font class=e2>&nbsp;\[([0-9]+)\]&nbsp;&nbsp;</font></expression> + </RegExp> + +<!--Genre Film--> + <RegExp input="$$7" output="<genre>\1</genre>" dest="5+"> + <RegExp input="$$1" output="\1" dest="7"> + <expression trim="1" noclean="1"><b>Genre:(.*)<b>Production:</expression> + </RegExp> + <expression repeat="yes">/browse/search.pfm\?mode=genre&query=[^>]*>([^<]*)<</expression> + </RegExp> + +<!--Actors --> + <RegExp input="$$7" output="<actor><name>\1</name></actor>" dest="5+"> + <RegExp input="$$1" output="\1" dest="7"> + <expression trim="1" noclean="1"><b>Starring</b>:(.*)<b>Director:</b></expression> + </RegExp> + <expression repeat="yes">">([^<]+)</a></expression> + </RegExp> + + +<!--Writer--> + <RegExp input="$$7" output="<credits>\1</credits>" dest="5+"> + <RegExp input="$$1" output="\1" dest="7"> + <expression trim="1" noclean="1"><font class=h0b color=white>Crew</font>(.*)</expression> + </RegExp> + <expression repeat="yes"><b>Writer: </b[^\n]*>([A-Za-z][^<]*)<</expression> + </RegExp> +<!--Plot --> + <RegExp input="$$1" output="<plot>\1</plot>" dest="5+"> + <expression trim="1">Synopsis</td></table><div[^>]*><table[^>]*><td[^>]*><img[^>]*>(.*)</td></table></div><p></expression> + </RegExp> + <RegExp input="$$1" output="<plot>\1</plot>" dest="5+"> + <expression trim="1">Introduction</td></table><div[^>]*><table[^>]*><td[^>]*><img[^>]*>(.*)</td></table></div><p></expression> + </RegExp> + + <expression noclean="1"></expression> + </RegExp> + </GetDetails> +</scraper> diff --git a/system/scrapers/video/cinefacts.png b/system/scrapers/video/cinefacts.png Binary files differnew file mode 100644 index 0000000000..6831e28f30 --- /dev/null +++ b/system/scrapers/video/cinefacts.png diff --git a/system/scrapers/video/cinefacts.xml b/system/scrapers/video/cinefacts.xml new file mode 100644 index 0000000000..0af556e415 --- /dev/null +++ b/system/scrapers/video/cinefacts.xml @@ -0,0 +1,134 @@ +<?xml version="1.0" encoding="iso-8859-1" standalone="yes"?> +<scraper framework="1.1" date="2009-07-28" name="Cinefacts.de" content="movies" thumb="cinefacts.png" language="de"> + <include>common/tmdb.xml</include> + <GetSettings dest="3"> + <RegExp input="$$5" output="<settings>\1</settings>" dest="3"> + <RegExp input="$$1" output="<setting label="Fanart" type="bool" id="fanart" default="true"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetSettings> + + <CreateSearchUrl dest="3" SearchStringEncoding="iso-8859-1"> + <RegExp input="$$1" output="http://www.cinefacts.de/suche/suche.php?name=\1" dest="3"> + <expression noclean="1"/> + </RegExp> + </CreateSearchUrl> + + <GetSearchResults dest="8"> + <RegExp input="$$5" output="<?xml version="1.0" encoding="iso-8859-1" standalone="yes"?><results>\1</results>" dest="8"> + <RegExp input="$$1" output="<entity><title>\3 (\4)</title><url cache="film.xml" >http://www.cinefacts.de/kino/\1/\2/filmdetails.html</url></entity>" dest="5"> + <expression repeat="yes">><a href="/kino/([0-9]*)/(.[^\/]*)/filmdetails.html">[^<]*<b title="([^"]*)" class="headline">[^<]+</b></a><br>[^<]+<br>+[^0-9]+([^<]*)</expression> + </RegExp> + <expression noclean="1"/> + </RegExp> + </GetSearchResults> + + <GetDetails dest="3"> + <RegExp input="$$5" output="<?xml version="1.0" encoding="iso-8859-1" standalone="yes"?><details>\1</details>" dest="3"> + <!--Title--> + <RegExp input="$$1" output="<title>\1</title>" dest="5+"> + <expression trim="1" noclean="1"><h1>([^<]*)</expression> + </RegExp> + + <!--Original Title--> + <RegExp input="$$1" output="<originaltitle>\1</originaltitle>" dest="5+"> + <expression><dt class="c1">Originaltitel:</dt>[^<]*<dd class="first">(.[^<]*)</dd></expression> + </RegExp> + + <!--Genre--> + <RegExp input="$$1" output="\1" dest="4+"> + <expression noclean="1">Genre:([^:]*)Deutschlandstart:</expression> + </RegExp> + <RegExp input="$$4" output="<genre>\1</genre>" dest="5+"> + <expression repeat="yes" noclean="1" trim="1">>*[ A-Za-z]([^<>]*)</a></expression> + </RegExp> + + <!--Director Film--> + <RegExp input="$$1" output="\1" dest="7+"> + <expression noclean="1">Regie:([^:]*)Buch:</expression> + </RegExp> + <RegExp input="$$7" output="<director>\1</director>" dest="5+"> + <expression repeat="yes" ><a href="[^"]*">([^<]*)</a></expression> + </RegExp> + + <!--Actors--> + <RegExp input="$$1" output="\1" dest="7+"> + <expression noclean="1">Darsteller:</td>(.*)</table</expression> + </RegExp> + <RegExp input="$$7" output="<actor><name>\1</name><role>\2</role></actor>" dest="5+"> + <expression repeat="yes">>([^<>]*)</a></td>+[^<]+<[^>]+> als([ A-Za-z]*)</expression> + </RegExp> + + <!--Studio--> + <RegExp input="$$1" output="<studio>\1</studio>" dest="5+"> + <expression>Studio:([^\.]*)\.</expression> + </RegExp> + + <!--Year--> + <RegExp input="$$1" output="<year>\1</year>" dest="5+"> + <expression></a> ([0-9]*) </dd></expression> + </RegExp> + + <!--MPAA--> + <RegExp input="$$1" output="<mpaa>\1</mpaa>" dest="5+"> + <expression>FSK:</dt>[^>]*>([^<]*)<</expression> + </RegExp> + + <!--Runtime--> + <RegExp input="$$1" output="<runtime>\1</runtime>" dest="5+"> + <expression>L.nge:</dt>[^>]*>([^<]*)<</expression> + </RegExp> + + <!--Plot--> + <RegExp input="$$1" output="<plot>\1</plot>" dest="5+"> + <expression>KURZINHALT</h2></li>[^>]*>*([^<]*)[</li>]</expression> + </RegExp> + + <!--Writers--> + <RegExp input="$$1" output="\1" dest="6+"> + <expression noclean="1">Buch:([^:]*)Musik:</expression> + </RegExp> + <RegExp input="$$6" output="<credits>\1</credits>" dest="5+"> + <expression repeat="yes" ><a href="[^"]*">([^<]*)</a></expression> + </RegExp> + + <!--Poster URL--> + <RegExp input="$$1" output="<url function="GetThumbnailLink">http://www.cinefacts.de/kino/film/\1/\2/plakate.html</url>" dest="5+"> + <expression repeat ="yes"><a href="/kino/film/([0-9]*)/([^\/]*)/plakate.html"></expression> + </RegExp> + + <!--IMDB URL--> + <RegExp conditional="fanart" input="$$8" output="<url function="GetTMDBFanartByIMDBId">http://www.google.com/search?q=site:imdb.com\1</url>" dest="5+"> + <RegExp input="$$1" output=""\2"+\1+||+"\2"" dest="7"> + <expression><h1>[^<]*</h1>[^0-9]*([0-9]*) </li>[^:]*:</dt>[^<]*<dd class="first">(.[^<]*)</dd></expression> + </RegExp> + <RegExp input="$$7" output="+\1" dest="8+"> + <expression repeat="yes">([^ ,]+)</expression> + </RegExp> + <expression></expression> + </RegExp> + <expression noclean="1"/> + </RegExp> + </GetDetails> + + <!--Thumbnail--> + <GetThumbnailLink dest="5"> + <RegExp input="$$2" output="<details>\1</details>" dest="5+"> + <RegExp input="$$1" output="<url function="GetThumbnail">http://www.cinefacts.de/kino/film/\1</url>" dest="2+"> + <expression repeat="yes" noclean="1"><a href="/kino/film/([^"]+)">[^<]*<img</expression> + </RegExp> + <expression noclean="1"/> + </RegExp> + </GetThumbnailLink> + + <GetThumbnail dest="5"> + <RegExp input="$$2" output="<details>\1</details>" dest="5"> + <RegExp input="$$1" output="<thumb>http://www.cinefacts.de/kino/plakat/\1</thumb>" dest="2+"> + <expression>src="/kino/plakat/([^"]*)"</expression> + </RegExp> + <expression noclean="1"/> + </RegExp> + </GetThumbnail> +</scraper> diff --git a/system/scrapers/video/common/dtrailer.xml b/system/scrapers/video/common/dtrailer.xml new file mode 100644 index 0000000000..98ded043dc --- /dev/null +++ b/system/scrapers/video/common/dtrailer.xml @@ -0,0 +1,15 @@ +<scraperfunctions> + <GetDTrailerLink dest="3"> + <RegExp input="$$5" dest="3" output="<details>\1</details>"> + <RegExp input="$$1" dest="5" output="<url function="GetDTrailer">http://en.dtrailer.com/movies/watch/\1</url>"> + <expression><a href="http://en.dtrailer.com/movies/watch/(/?[^\>]+)" class="thumb" title="(/?[^\>]+)"></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetDTrailerLink> + <GetDTrailer dest="5"> + <RegExp input="$$1" output="<details><trailer>http://www.dtrailer.com/dupload/trailerz/\1.flv</trailer></details>" dest="5"> + <expression noclean="1">s1\.addVariable\("file","(/?[^\>]+).flv"\);</expression> + </RegExp> + </GetDTrailer> +</scraperfunctions> diff --git a/system/scrapers/video/common/imdb.xml b/system/scrapers/video/common/imdb.xml new file mode 100644 index 0000000000..235edc2460 --- /dev/null +++ b/system/scrapers/video/common/imdb.xml @@ -0,0 +1,74 @@ +<scraperfunctions> + <GetIMDBThumbs dest="5"> + <RegExp input="$$6" output="<details>\1</details>" dest="5"> + <RegExp input="$$1" output="\1_SX$INFO[imdbscale]_SY$INFO[imdbscale]_\2" dest="4"> + <expression noclean="1,2"><a name="poster".*?src="(.*?)_S.*?(.jpg)".*?</a></expression> + </RegExp> + <RegExp input="$$4" output="<thumb>\1</thumb>" dest="6"> + <expression noclean="1">(.*?_SX[0-9]+_SY[0-9]+_.jpg)</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetIMDBThumbs> + <GetIMDBTrailer dest="5"> + <RegExp input="$$2" output="<details><trailer urlencoded="yes">\1</trailer></details>" dest="5"> + <RegExp input="$$1" output="\1" dest="2"> + <expression noclean="1">"file"[^"]*"([^"]*)</expression> + </RegExp> + <RegExp input="$$1" output="$$2/\1" dest="2"> + <expression noclean="1">"id"[^"]*"([^"]*)</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetIMDBTrailer> + <GetIMDBWriters dest="5"> + <RegExp input="$$2" output="<details>\1</details>" dest="5"> + <RegExp conditional="!fullcredits" input="$$1" output="\1" dest="6"> + <expression noclean="1">>Writer.*?:</h5>(.*?)</div></expression> + </RegExp> + <RegExp conditional="fullcredits" input="$$1" output="\1" dest="6"> + <expression noclean="1">Writing credits(.*?)name="</expression> + </RegExp> + <RegExp input="$$6" output="<credits>\1</credits>" dest="2+"> + <expression repeat="yes"><a href="/name/[^>]*>([^<]*)<</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetIMDBWriters> + <GetIMDBCast dest="5"> + <RegExp input="$$2" output="<details>\1</details>" dest="5"> + <RegExp input="$$1" output="\1" dest="6"> + <expression noclean="1"><table class="cast">(.*?)</table></expression> + </RegExp> + <RegExp input="$$6" output="<actor><thumb>\1_SX$INFO[imdbscale]_SY$INFO[imdbscale]_\2</thumb><name>\3</name><role>\4</role></actor>" dest="7"> + <expression clear="yes" repeat="yes" noclean="1,2" trim="3,4"><img src="(?:([^"]*\.)[^"]*(\.jpg))?[^>]*[^"]*"nm"><a href="[^"]*[^>]*>([^<]*)<[^"]*"ddd"> ... [^"]*"char">(.*?)</td></expression> + </RegExp> + <RegExp input="$$7" output="<actor><thumb>\1</thumb>\2</actor>" dest="2+"> + <expression clear="yes" repeat="yes" noclean="1,2,3"><actor><thumb>(?:(http.*?)|_SX[0-9]+_SY[0-9]+_)</thumb>(.*?)</actor></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetIMDBCast> + <GetIMDBDirectors dest="5"> + <RegExp input="$$2" output="<details>\1</details>" dest="5"> + <RegExp conditional="!fullcredits" input="$$1" output="\1" dest="6"> + <expression clear="yes" noclean="1">>Director.*?</h5>(.*?)</div></expression> + </RegExp> + <RegExp conditional="fullcredits" input="$$1" output="\1" dest="6"> + <expression clear="yes" noclean="1">Directed by(.*?)name="</expression> + </RegExp> + <RegExp input="$$6" output="<director>\1</director>" dest="2+"> + <expression clear="yes" repeat="yes"><a href="/name/[^>]*>([^<]*)<</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetIMDBDirectors> + <GetIMDBPlot dest="5"> + <RegExp input="$$2" output="<details>\1</details>" dest="5"> + <RegExp input="$$1" output="<plot>\1</plot>" dest="2"> + <expression clear="yes"><p class="plotpar">\n([^\n]*)</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetIMDBPlot> +</scraperfunctions> diff --git a/system/scrapers/video/common/impa.xml b/system/scrapers/video/common/impa.xml new file mode 100644 index 0000000000..546bc8bb7e --- /dev/null +++ b/system/scrapers/video/common/impa.xml @@ -0,0 +1,24 @@ +<scraperfunctions> + <GetIMPALink dest="5"> + <RegExp input="$$1" output="<details><url function="GetIMPAThumbs">http://\1impawards.com/\2</url></details>" dest="5"> + <expression noclean="1,2">http://([^"]*)impawards.com/([^"]*)"></expression> + </RegExp> + </GetIMPALink> + <GetIMPAThumbs dest="6"> + <RegExp input="$$4" output="<details>\1</details>" dest="6"> + <RegExp input="$$1" output="<url function="GetIMPAThumbs">http://www.impawards.com/\1</url>" dest="4"> + <expression noclean="1"><meta http-equiv="REFRESH" content="0;URL=[^/]*/([^"]*)"></expression> + </RegExp> + <RegExp input="$$1" output="\1" dest="8"> + <expression noclean="1">value="/([0-9]*)/[^"]*\.html"></expression> + </RegExp> + <RegExp input="$$1" output="<thumb>http://www.impawards.com/$$8/posters/\2</thumb>" dest="4+"> + <expression noclean="1"><img (SRC|src)="posters/([^"]*)"</expression> + </RegExp> + <RegExp input="$$1" output="<thumb>http://www.impawards.com/$$8/posters/\1</thumb>" dest="4+"> + <expression repeat="yes" noclean="1">thumbs/imp_([^>]*ver[^>]*.jpg)></expression> + </RegExp> + <expression noclean="1"/> + </RegExp> + </GetIMPAThumbs> +</scraperfunctions> diff --git a/system/scrapers/video/common/movieposterdb.xml b/system/scrapers/video/common/movieposterdb.xml new file mode 100644 index 0000000000..41386c495b --- /dev/null +++ b/system/scrapers/video/common/movieposterdb.xml @@ -0,0 +1,15 @@ +<scraperfunctions> + <GetMoviePosterDBLink dest="6"> + <RegExp input="$$1" output="<details><url function="GetMoviePosterDBThumbs">\1</url></details>" dest="6"> + <expression>(http://www.movieposterdb.com/movie/[^"]*)"</expression> + </RegExp> + </GetMoviePosterDBLink> + <GetMoviePosterDBThumbs dest="5"> + <RegExp input="$$6" output="<details>\1</details>" dest="5"> + <RegExp input="$$1" output="<thumb>\1l_\2</thumb>" dest="6"> + <expression repeat="yes" noclean="1,2">class="poster".*?src="(.*?)[a-z]_(.*?)"</expression> + </RegExp> + <expression noclean="1"/> + </RegExp> + </GetMoviePosterDBThumbs> +</scraperfunctions> diff --git a/system/scrapers/video/common/tmdb.xml b/system/scrapers/video/common/tmdb.xml new file mode 100644 index 0000000000..147807d25f --- /dev/null +++ b/system/scrapers/video/common/tmdb.xml @@ -0,0 +1,40 @@ +<scraperfunctions> + <GetTMDBFanartByIMDBId dest="4"> + <RegExp input="$$5" output="<details>\1</details>" dest="4"> + <RegExp input="$$1" output="<url function="GetTMDBFanart" cache="tmdb-\1.xml">http://api.themoviedb.org/2.0/Movie.getInfo?id=\1&amp;api_key=57983e31fb435df4df77afb854740ea9</url>" dest="5"> + <expression><id>([0-9]*)</id></expression> + </RegExp> + <RegExp input="$$1" output="<url function="GetTMDBFanartByIMDBId" cache="tmdb-trans-\1.xml" >http://api.themoviedb.org/2.0/Movie.imdbLookup?imdb_id=\1&amp;api_key=57983e31fb435df4df77afb854740ea9</url>" dest="5"> + <expression>/title/([t0-9]*)</expression> + </RegExp> + <expression noclean="1"/> + </RegExp> + </GetTMDBFanartByIMDBId> + <GetTMDBFanart dest="5"> + <RegExp input="$$2" output="<details><fanart url="http://themoviedb.org/image/backdrops">\1</fanart></details>" dest="5"> + <RegExp input="$$1" output="<thumb preview="/\1/\2_poster.\3">/\1/\2.\3</thumb>" dest="2"> + <expression repeat="yes"><backdrop size="original">.*?/([0-9]+)/([^\.]+)\.([^<]*)</backdrop></expression> + </RegExp> + <expression noclean="1">(.+)</expression> + </RegExp> + </GetTMDBFanart> + <GetTMDBThumbsByIMDBId dest="4">> + <RegExp input="$$5" output="<details>\1</details>" dest="4"> + <RegExp input="$$1" output="<url function="GetTMDBThumbs" cache="tmdb-\1.xml">http://api.themoviedb.org/2.0/Movie.getInfo?id=\1&amp;api_key=57983e31fb435df4df77afb854740ea9</url>" dest="5"> + <expression><id>([0-9]*)</id></expression> + </RegExp> + <RegExp input="$$1" output="<url function="GetTMDBThumbsByIMDBId" cache="tmdb-trans-\1.xml">http://api.themoviedb.org/2.0/Movie.imdbLookup?imdb_id=\1&amp;api_key=57983e31fb435df4df77afb854740ea9</url>" dest="5"> + <expression>/title/([t0-9]*)</expression> + </RegExp> + <expression noclean="1"/> + </RegExp> + </GetTMDBThumbsByIMDBId> + <GetTMDBThumbs dest="5"> + <RegExp input="$$4" output="<details>\1</details>" dest="5"> + <RegExp input="$$1" output="<thumb preview="\1_mid.\2">\1.\2</thumb>" dest="4"> + <expression repeat="yes"><poster size="original">([^\.]*)\.([^<]*)</poster></expression> + </RegExp> + <expression noclean="1"/> + </RegExp> + </GetTMDBThumbs> +</scraperfunctions> diff --git a/system/scrapers/video/culturalia.gif b/system/scrapers/video/culturalia.gif Binary files differnew file mode 100644 index 0000000000..6b2030bf23 --- /dev/null +++ b/system/scrapers/video/culturalia.gif diff --git a/system/scrapers/video/culturalia.xml b/system/scrapers/video/culturalia.xml new file mode 100644 index 0000000000..33886f0eb6 --- /dev/null +++ b/system/scrapers/video/culturalia.xml @@ -0,0 +1,115 @@ +п»ї<scraper framework="1.0" date="2009-05-22" name="Culturalia.es" content="movies" thumb="culturalia.gif" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" language="es"> + <NfoUrl dest="3"> + <RegExp input="$$1" output="http://www.culturalianet.com/art/ver.php?art=\1" dest="3"> + <expression noclean="1">art/ver\.php\?art=([0-9]*)</expression> + </RegExp> + </NfoUrl> + <CreateSearchUrl dest="3"> + <RegExp input="$$1" output="http://www.culturalianet.com/bus/resu.php?texto=\1&donde=1" dest="3"> + <expression noclean="1"/> + </RegExp> + </CreateSearchUrl> + <GetSearchResults dest="8"> + <RegExp input="$$5" output="<?xml version="1.0" encoding="iso-8859-1" standalone="yes"?><results>\1</results>" dest="8"> + <RegExp input="$$1" output="<entity><title>'\2' de \3 (\4)</title><url>http://www.culturalianet.com/art/ver.php?art=\1</url></entity>" dest="5"> + <expression repeat="yes"><a href='../art/ver.php\?art=([0-9]*)' target='_top'>(.[^<]*)\.</a>.[^\(]*\. De (.[^\(]*) \(([0-9]*)\)</expression> + </RegExp> + <expression noclean="1"/> + </RegExp> + </GetSearchResults> + <GetDetails dest="3"> + <RegExp input="$$5" output="<details>\1</details>" dest="3"> + <!-- Titulos y aГ±o de producciГіn !--> + <RegExp input="$$1" output="<title>\1 (\3)</title><originaltitle>\3</originaltitle><year>\2</year>" dest="5"> + <expression trim="1" noclean="1">'titulo2'>(.[^\<]*)\. \(([0-9]*)\)</font></u><br><br><i>(.[^<]*)</i></expression> + </RegExp> + <!-- DIRECTOR !--> + <RegExp input="$$1" output="\1" dest="7"> + <!-- Ponemos la cadena de Directores (sin limpiar en $$7) !--> + <expression noclean="1">Director:</font>([^:]*)<br><br><font class = 'titulo3'></expression> + </RegExp> + <!-- Ahora hay que extraer cada director ke pueden estar en 2 formatos + <br><a ref=url>Nombre</a> (Cuando tienen ficha de datos) + <br>Nombre</br> (Cuando no tienen ficha de datos) + La forma que se me ocurre es extraer en 2 pasadas distintas dejando un formato comun + Esto reordenarГЎ los directores en la ficha. !--> + <RegExp input="$$6" output="<director>\1</director>" dest="5+"> + <RegExp input="$$7" output=" /JUR\1/JAR" dest="6"> + <expression repeat="yes"><br><a href=[^0-9]*[0-9]*>([^<]*)</a></expression> + </RegExp> + <RegExp input="$$7" output=" /JUR\1/JAR" dest="6+"> + <expression repeat="yes"><br>([^<]*)</expression> + </RegExp> + <expression repeat="yes">/JUR([^/]*)/JAR</expression> + </RegExp> + <!-- GUION - CREDITS !--> + <RegExp input="$$1" output="\1" dest="7"> + <!-- Ponemos la cadena de Directores (sin limpiar en $$7) !--> + <expression noclean="1">Gui.n:</font>([^:]*)<br><br><font class = 'titulo3'></expression> + </RegExp> + <!-- Ahora hay que extraer cada nombre ke pueden estar en 2 formatos + <br><a ref=url>Nombre</a> (Cuando tienen ficha de datos) + <br>Nombre</br> (Cuando no tienen ficha de datos) + La forma que se me ocurre es extraer en 2 pasadas distintas dejando un formato comun + Esto reordenarГЎ los nombres en la ficha. !--> + <RegExp input="$$6" output="<credits>\1</credits>" dest="5+"> + <RegExp input="$$7" output=" /JUR\1/JAR" dest="6"> + <expression repeat="yes"><br><a href=[^0-9]*[0-9]*>([^<]*)</a></expression> + </RegExp> + <RegExp input="$$7" output=" /JUR\1/JAR" dest="6+"> + <expression repeat="yes"><br>([^<]*)</expression> + </RegExp> + <expression repeat="yes">/JUR([^/]*)/JAR</expression> + </RegExp> + <RegExp input="$$1" output="<plot>\1</plot>" dest="5+"> + <expression>Sinopsis:</b>[ ]*<br>([^=]*)<br></expression> + </RegExp> + <RegExp input="$$1" output="<mpaa>\1</mpaa>" dest="5+"> + <expression>moral:</font> ([^<]*)<br></expression> + </RegExp> + <RegExp input="$$1" output="<runtime>\1</runtime>" dest="5+"> + <expression>Duraci.n:</font> ([^<]*)\.<br></expression> + </RegExp> + <RegExp input="$$1" output="<rating>\2</rating><votes>\1</votes>" dest="5+"> + <expression>tulo3'>([0-9]*)</font> votos. Puntuaci.n: <font class = 'titulo3'>([^<]*)</font> / ([^<]*)\.</a></expression> + </RegExp> + <!-- GENEROS - GENRES !--> + <RegExp input="$$1" output="\1 /" dest="7"> + <expression>nero:</font><br>([^<]*)<br></expression> + </RegExp> + <RegExp input="$$6" output="<genre>\1</genre>" dest="5+"> + <RegExp input="$$7" output=" / \1" dest="6"> + <expression repeat="yes">([^/]*) /</expression> + </RegExp> + <expression> / (.*)</expression> + </RegExp> + <RegExp input="$$1" output="http://www.culturalianet.com/imatges/articulos/\1-1.jpg" dest="7"> + <expression>imatges/articulos/([0-9]*)-</expression> + </RegExp> + <RegExp input="$$7" output="<thumb>\1</thumb>" dest="5+"> + <expression/> + </RegExp> + <!-- ACTORES !--> + <RegExp input="$$1" output="\1" dest="7"> + <!-- Ponemos la cadena de actores (sin limpiar en $$7) !--> + <expression noclean="1">Actores:</font>([^:]*)<br><br><font class = 'titulo3'></expression> + </RegExp> + <!-- Ahora hay que extraer cada actor ke pueden estar en 2 formatos + <br><a ref=url>Nombre Actor</a> (Cuando tienen ficha de datos) + <br>Nombre actor</br> (Cuando no tienen ficha de datos) + La forma que se me ocurre es extraer en 2 pasadas distintas dejando un formato comun + Esto reordenarГЎ los actores en la ficha. Pero no me parece mal puesto que pondra primero + los actores con ficha (los mas importantes, no?) !--> + <RegExp input="$$6" output="<actor><name>\1</name></actor>" dest="5+"> + <RegExp input="$$7" output=" /JUR\1/JAR" dest="6"> + <expression repeat="yes"><br><a href=[^0-9]*[0-9]*>([^<]*)</a></expression> + </RegExp> + <RegExp input="$$7" output=" /JUR\1/JAR" dest="6+"> + <expression repeat="yes"><br>([^<]*)</expression> + </RegExp> + <expression repeat="yes">/JUR([^/]*)/JAR</expression> + </RegExp> + <expression noclean="1"/> + </RegExp> + </GetDetails> +</scraper> diff --git a/system/scrapers/video/daum.png b/system/scrapers/video/daum.png Binary files differnew file mode 100644 index 0000000000..63855ec0c7 --- /dev/null +++ b/system/scrapers/video/daum.png diff --git a/system/scrapers/video/daum.xml b/system/scrapers/video/daum.xml new file mode 100644 index 0000000000..6b63f6c997 --- /dev/null +++ b/system/scrapers/video/daum.xml @@ -0,0 +1,183 @@ +п»ї<?xml version="1.0" encoding="UTF-8"?> +<scraper framework="1.1" date="2009-08-08" name="DAUM (Korean)" content="movies" thumb="daum.png" language="ko"> + <GetSettings dest="3"> + <RegExp input="$$5" output="<settings>\1</settings>" dest="3"> + <RegExp input="$$1" output="<setting label="нЏ¬мЉ¤н„°" type="bool" id="EnableThumbs" default="true"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting label="нЊ¬м•„нЉё(кі н™”м§€)" type="bool" id="EnableFanArt" default="true"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting label="к·№мћ‘к°Ђ" type="bool" id="EnableWriter" default="true"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetSettings> + <NfoUrl dest="3"> + <RegExp input="$$1" output="<url>http://movie.daum.net/moviedetail/moviedetailMain.do?movieId=\1</url>" dest="3"> + <expression>movie\.daum\.net[^\?]*\?movieId=([0-9]*)</expression> + </RegExp> + </NfoUrl> + <CreateSearchUrl dest="3" SearchStringEncoding="UTF-8"> + <RegExp input="$$1" output="http://movie.daum.net/search.do?type=movie&q=\1" dest="3"> + <expression noclean="1"/> + </RegExp> + </CreateSearchUrl> + <GetSearchResults dest="8"> + <RegExp input="$$5" output="<?xml version="1.0" encoding="UTF-8" standalone="yes"?><results>\1</results>" dest="8"> + <RegExp input="$$1" output="<entity><title>\2(\4,\3) </title><url>http://movie.daum.net/moviedetail/moviedetailMain.do?movieId=\1</url></entity>" dest="5"> + <expression repeat="yes">class="fl srch"><a href="http://movie.daum.net/moviedetail/moviedetailMain.do\?movieId=([0-9]*)"[^>]*>(.[^\(]*)\(([0-9]*)\)[^\%]*\%"[^"]*"cb"[^"]*"fs13[^"]*">(.[^\/]*)\/</expression> + </RegExp> + <expression noclean="1"/> + </RegExp> + </GetSearchResults> + <GetDetails dest="3"> + <RegExp input="$$8" output="<details>\1</details>" dest="3"> + <!-- Titles, year !--> + <RegExp input="$$1" output="<title>\1 (\3)</title><year>\2</year>" dest="8"> + <expression noclean="1">class="title_kor"><a[^>]*>(.[^<]*)</a></strong>[^>]*>\(<a[^>]*>([0-9]*)</a>\)[^"]*"title_AKA"[^"]*"eng">(.[^<]*)<</expression> + </RegExp> + <!-- Director's names !--> + <RegExp input="$$7" output="<director>\1</director>" dest="8+"> + <RegExp input="$$1" output="\1" dest="7"> + <expression noclean="1"><strong>к°ђлЏ…</strong>(.*)<strong>л“±кё‰</strong></expression> + </RegExp> + <expression noclean="1" repeat="yes"><a[^>]*>(.[^<]*)</a></expression> + </RegExp> + <!-- Actors !--> + <RegExp input="$$7" output="<actor><name>\1</name><role>\2</role><thumb>\3</thumb></actor>" dest="8+"> + <RegExp input="$$1" output="\1" dest="7"> + <expression noclean="1">"div_actors"(.*)"netizenGrade"</expression> + </RegExp> + <expression noclean="1" repeat="yes">class="actor" title="(.[^\|]*)\|\|(.[^"]*)"><img src="(.[^"]*)"</expression> + </RegExp> + <!-- Tag Line !--> + <RegExp input="$$7" output="<tagline>\1</tagline>" dest="8+"> + <RegExp input="$$1" output="\1" dest="7"> + <expression noclean="1">"synopsis"(.*)id="review"</expression> + </RegExp> + <expression>class="txt">[^>]*>(.[^<]*)</b></expression> + </RegExp> + <!-- Plot !--> + <RegExp input="$$7" output="<plot>\1</plot>" dest="8+"> + <expression>class="txt">[^>]*>[^<]*</b><br>(.[^\t]*)\t</expression> + </RegExp> + <RegExp input="$$7" output="<plot>\1</plot>" dest="8+"> + <expression>class="txt">\n\t\t(.[^\t]*)\t</expression> + </RegExp> + <!-- Single Poster !--> + <RegExp conditional="!EnableThumbs" input="$$1" output="<thumb>\1</thumb>" dest="8+"> + <expression><p class="poster"><a[^<]*<img src="([^"]*)"</expression> + </RegExp> + <!-- Premiered !--> + <RegExp input="$$7" output="<premiered>\1.\2</premiered>" dest="8+"> + <RegExp input="$$1" output="\1" dest="7"> + <expression noclean="1"><strong>мљ”м•Ѕм •ліґ</strong>(.*)<strong>к°ђлЏ…</strong></expression> + </RegExp> + <expression noclean="1">([0-9]*)-([0-9]*)-[0-9]*</expression> + </RegExp> + <!-- Genres !--> + <RegExp input="$$7" output="<genre>\1</genre>" dest="8+"> + <RegExp input="$$1" output="\1" dest="7"> + <expression noclean="1"><strong>мљ”м•Ѕм •ліґ</strong>(.*)<strong>к°ђлЏ…</strong></expression> + </RegExp> + <expression noclean="1" repeat="yes"><a href='[^']*'>(.[^<]*)</a></expression> + </RegExp> + <!-- Runtime !--> + <RegExp input="$$7" output="<runtime>\1 분</runtime>" dest="8+"> + <RegExp input="$$1" output="\1" dest="7"> + <expression noclean="1"><strong>мљ”м•Ѕм •ліґ</strong>(.*)<strong>к°ђлЏ…</strong></expression> + </RegExp> + <expression noclean="1">([0-9]*) 분</expression> + </RegExp> + <!-- User rating !--> + <RegExp input="$$1" output="<rating>\1</rating>" dest="8+"> + <expression>"star_big pink"><[^>]*><[^>]*>[^>]*>[^>]*><em>(.[^<]*)</em></expression> + </RegExp> + <!-- User vote !--> + <RegExp input="$$1" output="<votes>\1</votes>" dest="8+"> + <expression><span class="fs11 fc3">\(([0-9]*)лЄ… м°ём—¬\)</span></expression> + </RegExp> + <!-- MPAA !--> + <RegExp input="$$7" output="<mpaa>\1</mpaa>" dest="8+"> + <RegExp input="$$1" output="\1" dest="7"> + <expression noclean="1"><strong>л“±кё‰</strong></dt>[^>]*>(.[^<]*)</dd></expression> + </RegExp> + <expression noclean="1" trim="1">\b(.[^\t]*)\t</expression> + </RegExp> + <!-- Poster !--> + <RegExp conditional="EnableThumbs" input="$$1" output="<url function="GetDaumPoster">\1</url>" dest="8+"> + <expression><p class="poster"><a href="(.[^"]*)"</expression> + </RegExp> + + <!-- FanArts !--> + <RegExp conditional="EnableFanArt" input="$$5" output="<url function="GetTMDBId">http://api.themoviedb.org/2.0/Movie.search?title=\1&amp;api_key=57983e31fb435df4df77afb854740ea9</url>" dest="8+"> + <RegExp input="$$1" output="\1" dest="7"> + <expression noclean="1" clear="yes">"title_AKA"[^"]*"eng">([^<]*)<</expression> + </RegExp> + <RegExp input="$$7" output="+\1" dest="5"> + <expression repeat="yes" clear="yes">([^ ,]+)</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + <!-- FanArts !--> + <RegExp conditional="EnableFanArt" input="$$1" output="<url function="GetDAUMFanart">\1</url>" dest="8+"> + <expression noclean="1">class="title_kor"><a href="(.[^"]*)"</expression> + </RegExp> + <!-- Writer !--> + <RegExp conditional="EnableWriter" input="$$1" output="<url function="GetWriter">\1</url>" dest="8+"> + <expression noclean="1"><a class="tab3" href="(.[^"]*)"</expression> + </RegExp> + <expression noclean="1"/> + </RegExp> + </GetDetails> + + <GetDaumPoster dest="8"> + <RegExp input="$$9" output="<details>\1</details>" dest="8"> + <RegExp input="$$1" output="\1" dest="7"> + <expression noclean="1">id="photo"(.*)id="steal"</expression> + </RegExp> + <RegExp input="$$7" output="<thumb preview="http://\1/C155x225/\2">http://\1/image/\2</thumb>" dest="9"> + <expression noclean="1" repeat="yes"><img src="http://(.[^/]*)/C155x225/(.[^"]*)"</expression> + </RegExp> + <RegExp input="$$7" output="<thumb preview="http://movie.daum-img.net/movie/movie-photo/\1/\2/\3/155_225_\4">http://movie.daum-img.net/movie/movie-photo/\1/\2/\3/\4</thumb>" dest="9+"> + <expression noclean="1" repeat="yes"><img src="http://movie.daum-img.net/movie/movie-photo/([0-9]*)/([0-9]*)/([0-9]*)/[0-9]*_[0-9]*_(.[^"]*)"</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetDaumPoster> + + <GetTMDBId dest="8"> + <RegExp input="$$1" output="<details><url function="GetTMDBFanart">http://api.themoviedb.org/2.0/Movie.getInfo?id=\1&amp;api_key=57983e31fb435df4df77afb854740ea9</url></details>" dest="8"> + <expression><id>([0-9]*)</id></expression> + </RegExp> + </GetTMDBId> + + <GetTMDBFanart clearbuffers="no" dest="8"> + <RegExp input="$$1" output="<thumb preview="http://themoviedb.org/image/backdrops/\1/\2_poster.\3">http://themoviedb.org/image/backdrops/\1/\2.\3</thumb>" dest="10+"> + <expression clear="yes" noclean="1" repeat="yes"><backdrop size="original">.*?/([0-9]+)/([^\.]+)\.([^<]*)</backdrop></expression> + </RegExp> + </GetTMDBFanart> + + <GetDAUMFanart clearbuffers="no" dest="8"> + <RegExp input="$$10$$11" output="<details><fanart>\1</fanart></details>" dest="8"> + <RegExp input="$$1" output="<thumb preview="http://\1/C93x70/\2">http://\1/image/\2</thumb>" dest="11"> + <expression clear="yes" noclean="1" repeat="yes">imgArray\.push\("http://(.[^/]*)/C93x70/(.[^"]*)"\)</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetDAUMFanart> + + <GetWriter dest="8"> + <RegExp input="$$1" output="\1" dest="7"> + <expression noclean="1"><h5>к°Ѓліё</h5>(.*)<h5>촬мЃ</h5></expression> + </RegExp> + <RegExp input="$$6" output="<details>\1</details>" dest="8"> + <RegExp input="$$7" output="<credits>\1</credits>" dest="6"> + <expression noclean="1" repeat="yes">class="em b">(.[^<]*)<</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetWriter> +</scraper> diff --git a/system/scrapers/video/fdbpl.png b/system/scrapers/video/fdbpl.png Binary files differnew file mode 100644 index 0000000000..54f9766cac --- /dev/null +++ b/system/scrapers/video/fdbpl.png diff --git a/system/scrapers/video/fdbpl.xml b/system/scrapers/video/fdbpl.xml new file mode 100644 index 0000000000..547736237b --- /dev/null +++ b/system/scrapers/video/fdbpl.xml @@ -0,0 +1,170 @@ +п»ї<?xml version="1.0" encoding="UTF-8"?> +<scraper framework="1.1" date="2009-08-08" name="fdb.pl" content="movies" thumb="fdbpl.png" language="pl"> + <include>common/tmdb.xml</include> + <NfoUrl dest="3"> + <RegExp input="$$1" output="<url>http://www.\1/title/tt\2/</url><id>tt\2</id>" dest="3"> + <expression clear="yes" noclean="1">(imdb.com/)Title\?([0-9]*)</expression> + </RegExp> + <RegExp input="$$1" output="<url>http://www.\1\2/</url><id>tt\2</id>" dest="3+"> + <expression clear="yes" noclean="1">(imdb.com/title/tt)([0-9]*)</expression> + </RegExp> + </NfoUrl> + <CreateSearchUrl dest="3"> + <RegExp input="$$1" output="<url>http://akas.imdb.com/find?s=tt;q=\1$$4</url>" dest="3"> + <RegExp input="$$2" output="%20(\1)" dest="4"> + <expression clear="yes">(.+)</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </CreateSearchUrl> + <GetSearchResults dest="8"> + <RegExp input="$$5" output="<?xml version="1.0" encoding="iso-8859-1" standalone="yes"?><results>\1</results>" dest="8"> + <RegExp input="$$1" output="\1" dest="7"> + <expression clear="yes">/title/([t0-9]*)/faq</expression> + </RegExp> + <RegExp input="$$1" output="<entity><title>\1 (\2)</title><url>http://akas.imdb.com/title/$$7/</url><id>$$7</id></entity>" dest="5"> + <expression clear="yes" noclean="1"><meta name="title" content="([^"]*) \(([0-9]*)\)</expression> + </RegExp> + <RegExp input="$$1" output="\1" dest="4"> + <expression noclean="1">(><a href="/title.*)</expression> + </RegExp> + <RegExp input="$$4" output="<entity><title>\2 (\3)</title><url>http://akas.imdb.com/title/\1/</url><id>\1</id></entity>" dest="5+"> + <expression repeat="yes" noclean="1,2">><a href="/title/([t0-9]*)/[^>]*>([^<]*)</a> *\(([0-9]*)</expression> + </RegExp> + <expression clear="yes" noclean="1"></expression> + </RegExp> + </GetSearchResults> + <GetDetails clearbuffers="no" dest="3"> + <RegExp input="$$5" output="<details>\1</details>" dest="3"> + <RegExp input="$$2" output="<id>\1</id>" dest="5"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<rating>\1</rating><votes>\2</votes>" dest="5+"> + <expression><b>([0-9.]+)/10</b>[^<]*<a href="ratings" class="tn15more">([0-9,]+) votes</a></expression> + </RegExp> + <RegExp input="$$1" output="<studio>\1</studio>" dest="5+"> + <expression repeat="yes">"/company/[^/]*/">([^<]*)</a></expression> + </RegExp> + <RegExp input="$$1" output="<runtime>\1</runtime>" dest="5+"> + <expression trim="1"><h5>Runtime:</h5>[^0-9]*([^<]*)</expression> + </RegExp> + <RegExp input="$$1" output="\1" dest="8"> + <expression trim="1" noclean="1"><title>([^<|^(]*)</expression> + </RegExp> + <RegExp input="$$9" output="\1" dest="6"> + <RegExp input="$$8" output="+\1" dest="9"> + <expression repeat="yes">([^ ]+)</expression> + </RegExp> + <expression>\+(.*)</expression> + </RegExp> + <RegExp input="$$1" output="\1" dest="7"> + <expression>a href="/Sections/Years/([0-9]*)</expression> + </RegExp> + <RegExp input="$$2" output="<url function="Getfdblink">http://fdb.pl/katalog?catalog[query]=$$6&amp;catalog[year_from]=$$7&amp;catalog[order]=votes</url>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$2" output="<url function="GetTMDBFanartByIMDBId">http://api.themoviedb.org/2.0/Movie.imdbLookup?imdb_id=$$2&amp;api_key=57983e31fb435df4df77afb854740ea9</url>" dest="5+"> + <expression></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetDetails> + <Getfdblink clearbuffers="no" dest="5"> + <RegExp input="$$2" output="<details>\1</details>" dest="5"> + <RegExp input="$$1" output="\1" dest="4"> + <expression noclean="1">class="results"([^~]*)class="clear"</expression> + </RegExp> + <RegExp input="$$4" output="\1" dest="6"> + <expression clear="yes" noclean="1,2">a href="([^>]+)">([^<]+)[^(]+\(([0-9]+)</expression> + </RegExp> + <RegExp input="$$6" output="<url function="GetDetailsfdb">$$6</url>" dest="2"> + <expression></expression> + </RegExp> + <RegExp input="$$6" output="<url function="GetPlotfdb">$$6/opisy</url>" dest="2+"> + <expression></expression> + </RegExp> + <RegExp input="$$6" output="<url function="GetPosterfdb">$$6/plakaty</url>" dest="2+"> + <expression></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </Getfdblink> + <GetDetailsfdb dest="5"> + <RegExp input="$$2" output="<details>\1</details>" dest="5"> + <RegExp input="$$1" output="<title>\1</title>" dest="2"> + <expression trim="1"><title>([^//|^(]*)</expression> + </RegExp> + <RegExp input="$$1" output="<year>\1</year>" dest="2+"> + <expression>title>.[^<]*\(([0-9]*)</expression> + </RegExp> + <RegExp input="$$1" output="<studio>\1</studio>" dest="2+"> + <expression>Dystrybutor[^=]+[^>]*>([^<]*)</expression> + </RegExp> + <RegExp input="$$1" output="\1" dest="6"> + <expression noclean="1">Gatunek[^=]+([^\n]*)</expression> + </RegExp> + <RegExp input="$$6" output="<genre>\1</genre>" dest="2+"> + <expression repeat="yes">=[^>]*>([^<]*)</expression> + </RegExp> + <RegExp input="$$1" output="\1" dest="6"> + <expression noclean="1">ReЕјyseria:(.*)Scenariusz:</expression> + </RegExp> + <RegExp input="$$6" output="<director>\1</director>" dest="2+"> + <expression repeat="yes"><a href="[^>]*>([^<]*)<</expression> + </RegExp> + <RegExp input="$$1" output="\1" dest="6"> + <expression noclean="1">Scenariusz:(.*)Premiera:</expression> + </RegExp> + <RegExp input="$$6" output="<credits>\1</credits>" dest="2+"> + <expression repeat="yes"><a href="[^>]*>([^<]*)<</expression> + </RegExp> + <RegExp input="$$1" output="<mpaa>od \1 lat</mpaa>" dest="2+"> + <expression>Od lat[^i]+i>([^<]+)</expression> + </RegExp> + <RegExp input="$$1" output="\1" dest="6"> + <expression noclean="1">Recenzja([^~]*)Dodatkowe informacje</expression> + </RegExp> + <RegExp input="$$6" output="<tagline>\1</tagline>" dest="2+"> + <expression></a>([^<]+)</h3></expression> + </RegExp> + <RegExp input="$$6" output="<outline>\1</outline>" dest="2+"> + <expression><p>([^<]+)</expression> + </RegExp> + <RegExp input="$$1" output="\1" dest="7"> + <expression noclean="1">Obsada([^~]*)class="line"</expression> + </RegExp> + <RegExp input="$$7" output="<actor><thumb>\1</thumb><name>\2</name><role>\3</role></actor>" dest="2+"> + <expression repeat="yes" noclean="1" trim="2,3">src="([^"]+)/[^[A-Z]+([^<]+)</a></td>[^[A-Z]+([^<]+)</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetDetailsfdb> + <GetPlotfdb dest="5"> + <RegExp input="$$2" output="<details><plot>\1</plot></details>" dest="5"> + <RegExp input="$$1" output="\1" dest="4"> + <expression noclean="1">id="resize-text">[^>]+>([^~]*)<div class="line"</expression> + </RegExp> + <RegExp input="$$4" output="\1" dest="7"> + <expression>([^<]+)</expression> + </RegExp> + <RegExp input="$$4" output=" \1" dest="7+"> + <expression repeat="yes">>([^<]+)</expression> + </RegExp> + <RegExp input="$$7" output="\1" dest="2"> + <expression noclean="1" repeat="yes">([^\n]+)</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetPlotfdb> + <GetPosterfdb dest="5"> + <RegExp input="$$2" output="<details>\1</details>" dest="5"> + <RegExp input="$$1" output="<thumb>\1</thumb>" dest="2+"> + <expression noclean="1">"gfx-poster" src="([^"]+)/</expression> + </RegExp> + <RegExp input="$$1" output="<thumb>\1</thumb>" dest="2+"> + <expression noclean="1" repeat="yes">"gfx-poster-gallery" src="([^"]+)/</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetPosterfdb> +</scraper> diff --git a/system/scrapers/video/filmaffinity.gif b/system/scrapers/video/filmaffinity.gif Binary files differnew file mode 100644 index 0000000000..3ce1ecad6f --- /dev/null +++ b/system/scrapers/video/filmaffinity.gif diff --git a/system/scrapers/video/filmaffinity.xml b/system/scrapers/video/filmaffinity.xml new file mode 100644 index 0000000000..fcff1db70c --- /dev/null +++ b/system/scrapers/video/filmaffinity.xml @@ -0,0 +1,283 @@ +<?xml version="1.0" encoding="iso-8859-1" standalone="no"?> +<scraper framework="1.1" date="2009-09-14" content="movies" name="FilmAffinity.es (Spanish)" thumb="filmaffinity.gif" language="es"> + <include>common/tmdb.xml</include> + <include>common/movieposterdb.xml</include> + <include>common/impa.xml</include> + <include>common/imdb.xml</include> + <GetSettings dest="3"> + <RegExp input="$$5" output="<settings>\1</settings>" dest="3"> + <RegExp input="$$1" output="<setting label="Buscar fotos de actores en IMDB (muy lento)" type="bool" id="SearchCastThumb" default="true"></setting>" dest="5"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting label="Habilitar trailers" type="bool" id="EnableTrailers" default="true"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting label="Sуlo pуster de FilmAffinity y sin FanArt (rбpido)" type="bool" id="OnlyFilmAffinity" default="false"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting label="Habilitar FanArt" type="bool" id="EnableFanArt" default="true"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting label="Descargar pуsters de IMPAwards" type="bool" id="EnableIMPAwards" default="true"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting label="Descargar pуsters de MoviePosterDB" type="bool" id="movieposterdb" default="true"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting label="Descargar pуster de IMDB" type="bool" id="EnableIMDBPosters" default="true"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting label="Tamaсo del pуster de IMDB" type="labelenum" values="192|256|384|512|1024" id="imdbscale" default="512"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetSettings> + + <NfoUrl dest="3"> + <RegExp dest="3" input="$$1" output="<url>http://www.filmaffinity.com/es/film\1.html/</url>"> + <expression noclean="1">filmaffinity.com/es/film([0-9]*)</expression> + </RegExp> + </NfoUrl> + + <CreateSearchUrl dest="3" SearchStringEncoding="iso-8859-1"> + <RegExp dest="3" input="$$1" output="<url>http://www.filmaffinity.com/es/search.php?stext=\1&amp;stype=none</url>"> + <expression noclean="1"></expression> + </RegExp> + </CreateSearchUrl> + + <GetSearchResults dest="8"> + <RegExp input="$$5" output="<?xml version="1.0" encoding="iso-8859-1" standalone="yes"?><results>\1</results>" dest="8"> + <RegExp input="$$1" output="\1" dest="7"> + <expression><img src="http://www.filmaffinity.com/imgs/movies/full/[0-9]*/([0-9]*).jpg"></expression> + </RegExp> + <RegExp dest="5" input="$$1" output="<entity><title>\1 (\2)</title><url>http://www.filmaffinity.com/es/film$$7.html</url><id>$$7</id></entity>"> + <expression noclean="1"><title>([^<]*)\(([0-9]*)\) - FilmAffinity</expression> + </RegExp> + <RegExp input="$$1" output="\1" dest="4"> + <expression noclean="1">(<b><a href="/es/film.*)</expression> + </RegExp> + <RegExp dest="5+" input="$$1" output="<entity><title>\2 (\3)</title><url>http://www.filmaffinity.com/es/film\1.html</url><id>\1</id></entity>"> + <expression repeat="yes" noclean="1,2"><a href="/es/film([0-9]*).html[^>]*>([^<]*)</a>[^\(]*\(([0-9]*)</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetSearchResults> + + <GetDetails dest="3"> + <RegExp input="$$5" output="<details>\1</details>" dest="3"> + <RegExp dest="5" input="$$1" output="<title>\1</title>"> + <expression trim="1" noclean="1"><img src="http://www.filmaffinity.com/images/movie.gif" border="0"> ([^<]*)</expression> + </RegExp> + <RegExp input="$$1" output="<plot>\1</plot>" dest="5+"> + <expression>/ SINOPSIS: ([^=]*)\(FILMAFFINITY\)</expression> + </RegExp> + <RegExp input="$$1" output="<plot>\1</plot>" dest="5+"> + <expression>/ SINOPSIS: ([^=]*)FILMAFFINITY\)</expression> + </RegExp> + <RegExp input="$$1" output="<plot>\1</plot>" dest="5+"> + <expression>SINOPSIS LARGA: ([^=]*)\(FILMAFFINITY\)</expression> + </RegExp> + <RegExp input="$$1" output="<outline>\1</outline>" dest="5+"> + <expression>SINOPSIS CORTA: ([^=]*)FILMAFFINITY\)([^=]*)FILMAFFINITY\)([^=]*)FILMAFFINITY\)</expression> + </RegExp> + <RegExp input="$$1" output="<outline>\1</outline>" dest="5+"> + <expression>SINOPSIS CORTA: ([^=]*)FILMAFFINITY\)([^=]*)FILMAFFINITY\)</expression> + </RegExp> + <RegExp input="$$1" output="<plot>\1</plot>" dest="5+"> + <expression>/ [^=]* / ([^=]*)FILMAFFINITY\)</expression> + </RegExp> + <RegExp input="$$1" output="<plot>\1</plot>" dest="5+"> + <expression>/ ([^=]*)FILMAFFINITY\)</expression> + </RegExp> + <RegExp input="$$9" output="<year>\1</year>" dest="5+"> + <RegExp input="$$1" output="\1" dest="9"> + <expression noclean="1"><b>A.O</b></td>(.*)<b>DURACI.N</b></expression> + </RegExp> + <expression repeat="yes"><td >([0-9]*)</expression> + </RegExp> + <RegExp input="$$9" output="<studio>\1</studio>" dest="5+"> + <RegExp input="$$1" output="/ \1 /" dest="9"> + <expression noclean="1"><b>PRODUCTORA</b></td>[^>]*>([^<]*)</td></expression> + </RegExp> + <expression repeat="yes">([^/]*)/</expression> + </RegExp> + <RegExp input="$$9" output="<director>\1</director>" dest="5+"> + <RegExp input="$$1" output="\1" dest="9"> + <expression noclean="1"><b>DIRECTOR</b></td>(.*)<b>GUI.N</b></expression> + </RegExp> + <expression repeat="yes">stext=[^>]*>([^<]+)</expression> + </RegExp> + <RegExp input="$$9" output="<tagline>\1 - " dest="5+"> + <RegExp input="$$1" output="\1" dest="9"> + <expression noclean="1"><b>PA.S</b></td>(.*)<b>DIRECTOR</b></expression> + </RegExp> + <expression repeat="yes">title="([^"]+)</expression> + </RegExp> + <RegExp input="$$9" output="\1</tagline>" dest="5+"> + <RegExp input="$$1" output="\1" dest="9"> + <expression noclean="1"><b>T.TULO ORIGINAL</b></td>(.*)<b>A.O</b></expression> + </RegExp> + <expression repeat="yes"><td ><b>([^<]+)</b></td></expression> + </RegExp> + <RegExp input="$$9" output="<credits>\1</credits>" dest="5+"> + <RegExp input="$$1" output="\1" dest="9"> + <expression noclean="1"><b>GUI.N</b></td>(.*)<b>M.SICA</b></expression> + </RegExp> + <expression repeat="yes"><td >([^<]+)</expression> + </RegExp> + <RegExp input="$$1" output="\1." dest="8"> + <expression noclean="1"><td valign="top">[^/]*/([^/]*) / [^=]*<</expression> + </RegExp> + <RegExp input="$$9" output="<genre>\1</genre>" dest="5+"> + <RegExp input="$$8" output=" / \1" dest="9"> + <expression repeat="yes">([^\.]*)\.</expression> + </RegExp> + <expression>/ (.*)</expression> + </RegExp> + <RegExp input="$$1" output="\1." dest="7"> + <expression noclean="1"><td valign="top">([^/]*) / [^=|^/]*<</expression> + </RegExp> + <RegExp input="$$6" output="<genre>\1</genre>" dest="5+"> + <RegExp input="$$7" output=" / \1" dest="6"> + <expression repeat="yes">([^\.]*)\.</expression> + </RegExp> + <expression>/ (.*)</expression> + </RegExp> + <RegExp input="$$1" output="<mpaa>\1</mpaa>" dest="5+"> + <expression><td valign="top">([^/]*)/[^/]* / [^=]*FILMAFFINITY\)</expression> + </RegExp> + <RegExp input="$$1" output="<rating>\1.\2</rating>" dest="5+"> + <expression>font-size:22px; font-weight: bold;">([0-9]+),([0-9]+)</expression> + </RegExp> + <RegExp input="$$1" output="<votes>\1\2</votes>" dest="5+"> + <expression><tr><td align="center">\(([0-9\.]+) votos\)</expression> + </RegExp> + <RegExp input="$$9" output="<runtime>\1 minutos</runtime>" dest="5+"> + <RegExp input="$$1" output="\1" dest="9"> + <expression noclean="1"><b>DURACI.N</b></td>(.*)<b>PA.S</b></expression> + </RegExp> + <expression repeat="yes"><td>([0-9]*) min\.</expression> + </RegExp> + + <RegExp input="$$20" output="\1" dest="5+"> + <RegExp conditional="!SearchCastThumb" input="$$1" output="<actor><name>\1</name></actor>" dest="20+"> + <expression repeat="yes" noclean="1" trim="1"><a href="search\.php.stype=cast[^>]*>([^<]*)</expression> + </RegExp> + <RegExp conditional="SearchCastThumb" input="$$1" output="<url function="SearchCastThumb">http://spanish.imdb.com/find?s=nm&amp;q=\1</url>" dest="20+"> + <expression repeat="yes" noclean="1" trim="1"><a href="search\.php.stype=cast.stext=([^"]*)[^>]*>([^<]*)</expression> + </RegExp> + <RegExp input="$$1" output="" dest="20"> + <expression><a href="search\.php.stype=cast[^>]*>(Animation)</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + + <RegExp conditional="EnableTrailers" input="$$1" output="<url function="SearchTrailerPage">http://www.filmaffinity.com/es/evideos.php\1</url>" dest="5+"> + <expression noclean="1"><a href="/es/evideos.php([^"]+)</expression> + </RegExp> + + <RegExp input="$$1" output="<thumb>http://pics.filmaffinity.com/\1</thumb>" dest="5+"> + <expression noclean="1,2">href="http://pics.filmaffinity.com/([^=]*large.jpg)"</expression> + </RegExp> + + <RegExp conditional="!OnlyFilmAffinity" input="$$1" output="<url function="GetFilmAffinityPosters">http://www.filmaffinity.com/es/filmimages.php?movie_id=\1</url>" dest="5+"> + <expression noclean="1,2">href="/es/filmimages.php.movie_id=([^"]+)</expression> + </RegExp> + + <!--URL to Google and IMDB--> + <RegExp conditional="!OnlyFilmAffinity" input="$$9" output="<url function="GoogleToIMDB">http://www.google.com/search?q=site:imdb.com\1</url>" dest="5+"> + <RegExp input="$$8" output="+\1" dest="9"> + <RegExp input="$$7" output="\1" dest="8"> + <RegExp input="$$1" output="\1" dest="7"> + <expression noclean="1"><b>T.TULO ORIGINAL</b></td>(.*)<b>A.O</b></expression> + </RegExp> + <expression><td ><b>([^<]+)</b></td></expression> + </RegExp> + <expression repeat="yes">([^ ,]+)</expression> + </RegExp> + <RegExp input="$$6" output="+\1" dest="9+"> + <RegExp input="$$1" output="\1" dest="6"> + <expression noclean="1"><b>A.O</b></td>(.*)<b>DURACI.N</b></expression> + </RegExp> + <expression repeat="yes"><td >([0-9]*)</expression> + </RegExp> + <expression></expression> + </RegExp> + + <expression noclean="1"></expression> + </RegExp> + </GetDetails> + + <SearchCastThumb dest="5"> + <RegExp input="$$1" output="<details><actor><name>\1</name></actor></details>" dest="5"> + <expression noclean="1,9" clear="yes">IMDb Name.*?"q" size="28" value="([^"]*)</expression> + </RegExp> + <RegExp input="$$4" output="<details><actor>\1</actor></details>" dest="5+"> + <RegExp input="$$1" output="<name>\1</name>" dest="4"> + <expression trim="1" noclean="1" clear="yes">"tn15title">.<h1>([^<|^(]*)</expression> + </RegExp> + <RegExp input="$$9" output="<thumb>\1</thumb>" dest="4+"> + <RegExp input="$$1" output="\1" dest="9"> + <expression noclean="1"><div class="photo">(.*) alt="</expression> + </RegExp> + <expression repeat="yes">img border="0" src="([^=]*.jpg)</expression> + </RegExp> + <expression noclean="1">(.*)</expression> + </RegExp> + <RegExp input="$$1" output="<details><url function="GetCastThumb">http://spanish.imdb.com\1</url></details>" dest="5+"> + <expression noclean="1">Media of.nbsp;<a href="([^"]*)</expression> + </RegExp> + </SearchCastThumb> + + <GetCastThumb dest="5"> + <RegExp input="$$4" output="<details><actor>\1</actor></details>" dest="5"> + <RegExp input="$$1" output="<name>\1</name>" dest="4"> + <expression noclean="1" clear="yes" trim="1"><title>([^<|^(]*)</expression> + </RegExp> + <RegExp input="$$9" output="<thumb>\1</thumb>" dest="4+"> + <RegExp input="$$1" output="\1" dest="9"> + <expression noclean="1"><div class="photo">(.*) alt="</expression> + </RegExp> + <expression repeat="yes">img border="0" src="([^=]*.jpg)</expression> + </RegExp> + <expression noclean="1" clear="yes">(.*)</expression> + </RegExp> + </GetCastThumb> + + <SearchTrailerPage dest="5"> + <RegExp input="$$1" output="<details><trailer>http://www.yobajo.com/watch?v=\1</trailer></details>" dest="5+"> + <expression noclean="1">http://www.youtube.com/v/([^&|^\\]*)</expression> + </RegExp> + </SearchTrailerPage> + + <GoogleToIMDB dest="3"> + <RegExp input="$$5" output="<details>\1</details>" dest="3"> + <RegExp input="$$1" output="\1" dest="6"> + <expression noclean="1">/title/tt([0-9]*)</expression> + </RegExp> + <RegExp conditional="EnableFanArt" output="<url function="GetTMDBFanartByIMDBId">http://api.themoviedb.org/2.0/Movie.imdbLookup?imdb_id=tt$$6&amp;api_key=57983e31fb435df4df77afb854740ea9</url>" dest="5"> + <expression></expression> + </RegExp> + <RegExp conditional="EnableIMPAwards" output="<url function="GetIMPALink"cache="$$2.html">http://www.imdb.com/title/tt$$6/posters</url>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp conditional="movieposterdb" output="<url function="GetMoviePosterDBLink">http://www.movieposterdb.com/browse/search?type=movies&amp;query=$$6</url>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp conditional="EnableIMDBPosters" output="<url function="GetIMDBPoster"cache="$$2.html">http://www.imdb.com/title/tt$$6/posters</url>" dest="5+"> + <expression></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GoogleToIMDB> + + <GetFilmAffinityPosters dest="5"> + <RegExp input="$$10" output="<details>\1</details>" dest="5"> + <RegExp input="$$1" output="<thumb>http://pics.filmaffinity.com/\4</thumb>" dest="10"> + <expression repeat="yes" noclean="1">, imgs_([0-9]*):([^']*)'([^']*)', url_l: 'http://pics.filmaffinity.com/([^']*)</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetFilmAffinityPosters> +</scraper> diff --git a/system/scrapers/video/filmdelta.png b/system/scrapers/video/filmdelta.png Binary files differnew file mode 100644 index 0000000000..2b5864f351 --- /dev/null +++ b/system/scrapers/video/filmdelta.png diff --git a/system/scrapers/video/filmdelta.xml b/system/scrapers/video/filmdelta.xml new file mode 100644 index 0000000000..fdbcf31ff6 --- /dev/null +++ b/system/scrapers/video/filmdelta.xml @@ -0,0 +1,90 @@ +<?xml version="1.0" encoding="utf-8" standalone="no"?> +<scraper framework="1.1" date="2009-08-11" content="movies" name="Filmdelta.se" thumb="filmdelta.png" language="sv"> + <include>common/tmdb.xml</include> + <GetSettings dest="3"> + <RegExp input="$$5" output="<settings>\1</settings>" dest="3"> + <RegExp input="$$1" output="<setting label="Fanart" type="bool" id="fanart" default="true"></setting>" dest="5+"> + <expression/> + </RegExp> + <expression noclean="1"/> + </RegExp> + </GetSettings> + + <CreateSearchUrl clearbuffers="no" dest="3"> + <RegExp input="$$1" output="http://www.filmdelta.se/search.php?string=\1&type=movie&more=1" dest="3"> + <RegExp input="$$1" output="\1" dest="9"> + <expression>%28(.*?)%29</expression> + </RegExp> + <expression trim="1"/> + </RegExp> + </CreateSearchUrl> + + <GetSearchResults dest="8"> + <RegExp input="$$5" output="<results>\1</results>" dest="8"> + <RegExp input="$$1" output="<entity><title>\2</title><year>\3</year><url>http://www.filmdelta.se/filmer/\1</url></entity>" dest="5"> + <expression repeat="yes" noclean="1"><a href="/filmer/([^"]*)">([^<]*)</a> \(([0-9]*)\)</expression> + </RegExp> + <RegExp input="$$1" output="<entity><title>\2</title><year>$$9</year><url>http://www.filmdelta.se/filmer/\1</url></entity>" dest="5"> + <expression repeat="yes" noclean="1"><a href="/filmer/([^"]*)">([^<]*)</a> \($$9\)</expression> + </RegExp> + <expression noclean="1"/> + </RegExp> + </GetSearchResults> + + <GetDetails dest="3"> + <RegExp input="$$5" output="<details>\1</details>" dest="3"> + <RegExp input="$$1" output="<title>\1</title>" dest="5"> + <expression noclean="1"><title> (.*?) - Filmdelta -</expression> + </RegExp> + <RegExp input="$$1" output="<year>\1</year><runtime>\2</runtime>" dest="5+"> + <expression noclean="1">([0-9]*)</a>, ([0-9]*) min</expression> + </RegExp> + <RegExp input="$$1" output="<plot>\1</plot><outline>\1</outline>" dest="5+"> + <expression trim="1"><div class="text">[^<]*<p>(.*?)</p></expression> + </RegExp> + <RegExp input="$$7" output="<actor><name>\1</name><role>\2</role></actor>" dest="5+"> + <RegExp input="$$1" output="\1" dest="7"> + <expression noclean="1"><div class="info" id="actors">(.*?)</div></expression> + </RegExp> + <expression repeat="yes" trim="1,2">/'>([^<]*)</a> - ([^<]*)</h5></expression> + </RegExp> + <RegExp input="$$6" output="<director>\1</director>" dest="5+"> + <RegExp input="$$1" output="\1" dest="6"> + <expression noclean="1"><h4>Regiss&ouml;r</h4>(.*?)</div></expression> + </RegExp> + <expression repeat="yes" trim="1">/'>([^<]*)</a></h5></expression> + </RegExp> + <RegExp input="$$6" output="<credits>\1</credits>" dest="5+"> + <RegExp input="$$1" output="\1" dest="6+"> + <expression noclean="1"><h4>Manus</h4>(.*?)</div></expression> + </RegExp> + <expression repeat="yes" trim="1">/'>([^<]*)</a></h5></expression> + </RegExp> + <RegExp input="$$8" output="<genre>\1</genre>" dest="5+"> + <RegExp input="$$1" output="\1" dest="8"> + <expression noclean="1"><h4>Kategori</h4>(.*?)</div></expression> + </RegExp> + <expression repeat="yes" trim="1">type=category'>([^<]*)</a></h5></expression> + </RegExp> + <RegExp input="$$9" output="<url cache="filmdelta-trans.xml" function="GetTMDBThumbsByIMDBId">http://www.google.com/search?q=site:imdb.com\1</url>" dest="5+"> + <RegExp input="$$1" output=""\1"+\2" dest="4+"> + <expression><h4>Originaltitel</h4>[^<]*<h5>([^<]*)</h5>.*?/filmarkiv/([0-9]*)/</expression> + </RegExp> + <RegExp input="$$4" output="+\1" dest="9+"> + <expression repeat="yes">([^ ,]+)</expression> + </RegExp> + <expression/> + </RegExp> + <RegExp conditional="fanart" input="$$9" output="<url cache="filmdelta-trans.xml" function="GetTMDBFanartByIMDBId">http://www.google.com/search?q=site:imdb.com\1</url>" dest="5+"> + <RegExp input="$$1" output=""\1"+\2" dest="4+"> + <expression><h4>Originaltitel</h4>[^<]*<h5>([^<]*)</h5>.*?/filmarkiv/([0-9]*)/</expression> + </RegExp> + <RegExp input="$$4" output="+\1" dest="9+"> + <expression repeat="yes">([^ ,]+)</expression> + </RegExp> + <expression/> + </RegExp> + <expression noclean="1"/> + </RegExp> + </GetDetails> +</scraper> diff --git a/system/scrapers/video/filmstarts.jpg b/system/scrapers/video/filmstarts.jpg Binary files differnew file mode 100644 index 0000000000..01c5335e58 --- /dev/null +++ b/system/scrapers/video/filmstarts.jpg diff --git a/system/scrapers/video/filmstarts.xml b/system/scrapers/video/filmstarts.xml new file mode 100644 index 0000000000..676224405c --- /dev/null +++ b/system/scrapers/video/filmstarts.xml @@ -0,0 +1,159 @@ +<?xml version="1.0" encoding="utf-8" standalone="yes"?> +<scraper framework="1.1" date="2009-08-15" name="Filmstarts" content="movies" thumb="filmstarts.jpg" language="de"> + <include>common/tmdb.xml</include> + <GetSettings dest="3"> + <RegExp input="$$5" output="<settings>\1</settings>" dest="3"> + <RegExp input="$$1" output="<setting label="Fanart" type="bool" id="fanart" default="true"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetSettings> + + <NfoUrl dest="3"> + <RegExp input="$$1" output="<url>http://www.filmstarts.de/kritiken/\2.html</url>" dest="3"> + <expression clear="yes" noclean="1">(filmstarts.de/kritiken/)(.*).html</expression> + </RegExp> + <RegExp input="$$1" output="<url function="GetByIMDBId">http://www.imdb.com/title/tt\2/externalreviews</url>" dest="3+"> + <expression>(imdb.com/)Title\?([0-9]+)</expression> + </RegExp> + <RegExp input="$$1" output="<url function="GetByIMDBId">http://www.imdb.com/title/tt\2/externalreviews</url>" dest="3+"> + <expression>(imdb.com/title/tt)([0-9]*)</expression> + </RegExp> + </NfoUrl> + + <GetByIMDBId dest="3"> + <RegExp input="$$1" output="<url>http://www.filmstarts.de/kritiken/\1</url>" dest="3+"> + <expression noclean="1"><a href="http://www.filmstarts.de/kritiken/([^"]*)"</expression> + </RegExp> + </GetByIMDBId> + + <CreateSearchUrl dest="3" SearchStringEncoding="iso-8859-1"> + <RegExp input="$$1" output="<url>http://www.filmstarts.de/finde.html?t=film&amp;anfrage=\1&amp;x=0&amp;y=0&</url>" dest="3"> + <expression noclean="1"></expression> + </RegExp> + </CreateSearchUrl> + + <GetSearchResults dest="3"> + <RegExp input="$$5" output="<?xml version="1.0" encoding="iso-8859-1" standalone="yes"?><results>\1</results>" dest="3"> + <RegExp input="$$1" output="<entity><title>\2</title><url>http://www.filmstarts.de/kritiken/\1</url>;</entity>" dest="5"> + <expression repeat="yes"><li><a href="/kritiken/([^"]*)">.*?<span class="t">([^<]*)</span></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetSearchResults> + + <GetDetails dest="3"> + <RegExp input="$$5" output="<?xml version="1.0" encoding="iso-8859-1" standalone="yes"?><details>\1</details>" dest="3"> + + <!--Title --> + <RegExp input="$$1" output="<title>\1</title>" dest="5+"> + <expression trim="1" noclean="1"><title>([^<|^(]*)</expression> + </RegExp> + + <!--Original Title--> + <RegExp input="$$1" output="<originaltitle>\1</originaltitle>" dest="5+"> + <expression>Originaltitel:(.*?)</div</expression> + </RegExp> + + <!--Year Film--> + <RegExp input="$$1" output="<year>\1</year>" dest="5+"> + <expression trim="1" noclean="1">genre".*weak".*[A-Z]([0-9 ]*) </div</expression> + </RegExp> + + <!--MPAA Film--> + <RegExp input="$$1" output="<mpaa>ab \1</mpaa>" dest="5+"> + <expression>FSK:.*?Ab ([0-9]*)<</expression> + </RegExp> + + <!--Director Film--> + <RegExp input="$$1" output="<director>\1</director>" dest="5+"> + <expression trim="1" noclean="1">Regie</span>.*?class="name">(.*?)</span</expression> + </RegExp> + + <!--Studio Film--> + <RegExp input="$$1" output="<studio>\1</studio>" dest="5+"> + <expression trim="1" noclean="1">Verleih:.*?">(.*?)</td</expression> + </RegExp> + + <!--Runtime Film--> + <RegExp input="$$1" output="<runtime>\1</runtime>" dest="5+"> + <expression trim="1" noclean="1">Laufzeit:</span>.*?">([A-Za-z0-9 ]*)</td</expression> + </RegExp> + + <!--User rating--> + <RegExp input="$$1" output="<rating>\1</rating><votes>\2</votes>" dest="5+"> + <expression trim="1" noclean="1">"Wertung: ([0-9]*) / ([0-9]*)"</expression> + </RegExp> + + <!--Genre Film--> + <RegExp input="$$1" output="<genre>\1</genre>" dest="5+"> + <expression trim="1" noclean="1"><div class="genre"><a.*?">(.*?)</a</expression> + </RegExp> + + <!--Writer--> + <RegExp input="$$1" output="<credits>\1</credits>" dest="5+"> + <expression trim="1" noclean="1">Drehbuch</span>.*?class="name">([A-Za-z ]*)</span</expression> + </RegExp> + + <!--Plot--> + <RegExp input="$$1" output="<plot>\1</plot>" dest="5+"> + <expression trim="1">"text critic topBorder">([^<]*)</div</expression> + </RegExp> + <RegExp input="$$1" output="<plot>\1</plot>" dest="5+"> + <expression trim="1">"text critic topBorder">.*?<br /><br />(.*?)</div></expression> + </RegExp> + + <!--Actors --> + <RegExp input="$$1" output="<url function="GetActors">http://www.filmstarts.de/kritiken/\1/castcrew.html</url>" dest="5+"> + <expression noclean="1">/kritiken/([-.%a-z0-9A-Z]+)/bilder.html</expression> + </RegExp> + + <!--Posters --> + <RegExp input="$$1" output="<url function="GetPosters">http://www.filmstarts.de/kritiken/\1/bilder.html</url>" dest="5+"> + <expression noclean="1">/kritiken/([-.%a-z0-9A-Z]+)/bilder.html</expression> + </RegExp> + + <!--URL to Google and Fanart--> + <RegExp conditional="fanart" input="$$8" output="<url function="GetTMDBFanartByIMDBId">http://www.google.com/search?q=site:imdb.com\1</url>" dest="5+"> + <RegExp input="$$1" output=""\1"+\3+||+"\2"" dest="7"> + <expression><title>([^<|^(]*).*Originaltitel:(.*?)</div>.*weak".*[A-Z]([0-9 ]*) </div</expression> + </RegExp> + <RegExp input="$$7" output="+\1" dest="8+"> + <expression repeat="yes">([^ ,]+)</expression> + </RegExp> + <expression></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetDetails> + + <!--Poster--> + <GetPosters dest="5"> + <RegExp input="$$2" output="<?xml version="1.0" encoding="iso-8859-1" standalone="yes"><details>\1</details>" dest="5+"> + <RegExp input="$$1" output="<thumb preview="http://thumbs.filmstarts.de/image/\1">http://thumbs.filmstarts.de/wallpaper/\1</thumb>" dest="2"> + <expression repeat="yes">src="http://thumbs.filmstarts.de/mini/([^"]*)"</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetPosters> + + <!--Actors--> + <GetActors dest="5"> + <RegExp input="$$2" output="<?xml version="1.0" encoding="iso-8859-1" standalone="yes"><details>\1</details>" dest="5+"> + <RegExp input="$$1" output="\1" dest="7+"> + <expression noclean="1">Schauspieler</li>(.*)</expression> + </RegExp> + <RegExp input="$$7" output="<actor><name>\1</name><thumb>http://thumbs.filmstarts.de/wallpaper/\2</thumb><role>\3</role></actor>" dest="2+"> + <expression repeat="yes" >img alt="([^"]*)".*?src="http://thumbs.filmstarts.de/nano/([^"]*)".*?role weak.*?">als ([A-Za-z ]*)<</expression> + </RegExp> + <RegExp input="$$7" output="<actor><name>\1</name><role>\2</role></actor>" dest="2+"> + <expression repeat="yes" >img alt="([^"]*)".*?src="/designs/default/images/no_avatar_small.gif".*?role weak.*?">als ([A-Za-z ]*)<</expression> + </RegExp> + <RegExp input="$$7" output="<actor><name>\1</name></actor>" dest="2+"> + <expression repeat="yes" >class="name">([A-Za-z ]*)</span</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetActors> +</scraper> diff --git a/system/scrapers/video/filmup.gif b/system/scrapers/video/filmup.gif Binary files differnew file mode 100644 index 0000000000..c67b7750fe --- /dev/null +++ b/system/scrapers/video/filmup.gif diff --git a/system/scrapers/video/filmup.xml b/system/scrapers/video/filmup.xml new file mode 100644 index 0000000000..3bb5fc9725 --- /dev/null +++ b/system/scrapers/video/filmup.xml @@ -0,0 +1,85 @@ +<scraper framework="1.0" date="2009-05-22" name="FilmUP.it" content="movies" thumb="filmup.gif" language="it"> + <NfoUrl dest="3"> + <RegExp input="$$1" output="<url>http://filmup.leonardo.it/sc_\1.htm</url>" dest="3"> + <expression noclean="1">sc_([a-z0-9__]*)\.htm</expression> + </RegExp> + </NfoUrl> + <CreateSearchUrl dest="3"> + <RegExp input="$$1" output="http://filmup.leonardo.it/cgi-bin/search.cgi?ps=10&fmt=long&q=\1&ul=%25%2Fsc_%25&m=all&wf=222210&o=0&ps=50" dest="3"> + <expression noclean="1"></expression> + </RegExp> + </CreateSearchUrl> + <GetSearchResults dest="8"> + <RegExp input="$$5" output="<?xml version="1.0" encoding="iso-8859-1" standalone="yes"?><results>\1</results>" dest="8"> + <RegExp input="$$1" output="<entity><title>\2</title><url>http://filmup.leonardo.it/sc_\1.htm</url><id>\1</id></entity>" dest="5"> + <expression repeat="yes"><a class="filmup" href="http.[^sc_]*sc_([0-9a-z_]*)\.htm"[^>]*>[^:]*: (.[^\n]*[^>]*)></expression> + + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetSearchResults> + + + <GetDetails dest="3"> + <RegExp input="$$5" output="<details>\1</details>" dest="3"> + + <!--Title --> + <RegExp input="$$1" output="<title>\1</title>" dest="5"> + <expression trim="1" noclean="1"><title>FilmUP - Scheda: ([^<]*)<</expression> + </RegExp> + + <!--Original Title + <RegExp input="$$1" output="<originaltitle>\1</originaltitle>" dest="5"> + <expression>Titolo originale.[^z]*ze="2">([^)<]*)</expression> + </RegExp> --> + + <!--Year Film--> + <RegExp input="$$1" output="<year>\1</year>" dest="5+"> + <expression>Anno.[^z]*ze="2">([0-9]*)<</expression> + </RegExp> + + <!--Director--> + <RegExp input="$$1" output="<director>\1</director>" dest="5+"> + <expression>Regia.[^z]*ze="2">([^\n]*)</font></expression> + </RegExp> + + + + <!--Runtime Film--> + <RegExp input="$$1" output="<runtime>\1</runtime>" dest="5+"> + <expression>Durata.[^z]*ze="2">([^)<]*)</expression> + </RegExp> + + <!--Genre Film--> + <RegExp input="$$6" output="<genre>\1</genre>" dest="5+"> + <RegExp input="$$1" output="\1" dest="6"> + <expression>Genere.[^z]*ze="2">([^)<]*)</expression> + </RegExp> + <expression repeat="yes">([a-zA-Z][^,]*)</expression> + </RegExp> + + <!--Plot outline --> + <RegExp input="$$1" output="<outline>\1</outline><plot>\1</plot>" dest="5+"> + <expression>Trama:<br>([^<]*)</expression> + </RegExp> + + <!--Thumbnail --> + <RegExp input="$$1" output="<thumb>http://filmup.leonardo.it/locand/\1</thumb>" dest="5+"> + <expression noclean="1">src="locan[^/]*/(.[^"]*)</expression> + </RegExp> + + <!--Actors --> + <RegExp input="$$7" output="<actor><name>\1</name></actor>" dest="5+"> + <RegExp input="$$1" output="\1" dest="7"> + <expression trim="1">Cast.[^z]*ze="2">([^\n]*)</td></expression> + </RegExp> + <expression repeat="yes">([a-zA-Z][^,]*)</expression> + </RegExp> + + + + <expression noclean="1"></expression> + </RegExp> + </GetDetails> +</scraper> + diff --git a/system/scrapers/video/filmweb.jpg b/system/scrapers/video/filmweb.jpg Binary files differnew file mode 100644 index 0000000000..3618334e08 --- /dev/null +++ b/system/scrapers/video/filmweb.jpg diff --git a/system/scrapers/video/filmweb.xml b/system/scrapers/video/filmweb.xml new file mode 100644 index 0000000000..0e02bbbb17 --- /dev/null +++ b/system/scrapers/video/filmweb.xml @@ -0,0 +1,194 @@ +п»ї<?xml version="1.0" encoding="utf-8" standalone="yes"?> +<scraper framework="1.0" date="2009-08-08" name="filmweb.pl" content="movies" thumb="filmweb.jpg" language="pl"> + <include>common/tmdb.xml</include> + <GetSettings dest="3"> + <RegExp input="$$5" output="<settings>\1</settings>" dest="3"> + <RegExp input="$$1" output="<setting label="Wysoka jakosc okladki" type="bool" id="poster" default="false"></setting>" dest="5"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting label="Pelna obsada" type="bool" id="fullcredits" default="false"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetSettings> + <NfoUrl dest="3"> + <RegExp input="$$1" output="<url>http://www.filmweb.pl/f\1/xbmc,2008</url>" dest="3"> + <expression noclean="1">filmweb.pl/f([0-9]*)/</expression> + </RegExp> + <RegExp input="$$1" output="<url>http://www.filmweb.pl/f\1/xbmc,2008</url>" dest="3+"> + <expression noclean="1">Film.id=([0-9]*)</expression> + </RegExp> + <RegExp input="$$1" output="<url>http://\1.filmweb.pl</url>" dest="3+"> + <expression noclean="1">http://([^\/]+).filmweb.pl</expression> + </RegExp> + </NfoUrl> + <CreateSearchUrl dest="3"> + <RegExp input="$$1" output="http://www.filmweb.pl/szukaj/film?sort=COUNT&q=\1" dest="3"> + <expression noclean="1"></expression> + </RegExp> + </CreateSearchUrl> + <GetSearchResults dest="8"> + <RegExp input="$$5" output="<?xml version="1.0" encoding="utf-8" standalone="yes"?><results>\1</results>" dest="8"> + <RegExp input="$$1" output="\3\6#\2\5#\1\4#" dest="4"> + <expression repeat="yes">src="http://gfx.filmweb.pl/po/[^/]*/[^/]*/([0-9]*)/.[^#]*">(.[^(]+)\(([0-9]+)|searchResultTitle" href=[^=]*=([0-9]*)">(.[^(]*)\(([0-9]*)</expression> + </RegExp> + <RegExp input="$$4" output="\1" dest="7"> + <expression>([^<]+)</expression> + </RegExp> + <RegExp input="$$4" output="\1" dest="7+"> + <expression repeat="yes">[^>]>([^<]+)</expression> + </RegExp> + <RegExp input="$$7" output="<entity><title>(\1) \2</title><url>http://www.filmweb.pl/Film?id=\3</url></entity>" dest="5+"> + <expression repeat="yes">([0-9]+)#([^#]+)#([0-9]+)</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetSearchResults> + <GetDetails clearbuffers="no" dest="3"> + <RegExp input="$$5" output="<?xml version="1.0" encoding="utf-8" standalone="yes"?><details>\1</details>" dest="3"> + <RegExp input="$$1" output="<title>\1</title>" dest="5"> + <expression trim="1">film-title[^=]+[^>]+[^[0-9A-ZЕ»Е№Д†ЕѓД„ЕљЕЃД]+([^<]+[0-9a-zД…Д™ЕјЕєД‡Е›Е„Е‚])</expression> + </RegExp> + <RegExp input="$$1" output="\1" dest="9"> + <expression>Filmweb.globals.filmId = ([0-9]+)</expression> + </RegExp> + <RegExp input="$$1" output="<year>\1</year>" dest="5+"> + <expression>title>.[^<]*\(([0-9]*)</expression> + </RegExp> + <RegExp input="$$1" output="<votes>\1</votes>" dest="5+"> + <expression>gЕ‚osГіw:[^>]+>([0-9]*)</expression> + </RegExp> + <RegExp input="$$1" output="<rating>\1.\2</rating>" dest="5+"> + <expression>Ељrednia ocena:[^,]+>([0-9]*),([0-9]*)</expression> + </RegExp> + <RegExp input="$$1" output="<top250>\1</top250>" dest="5+"> + <expression>wiat: ([0-9]*)</expression> + </RegExp> + <RegExp input="$$1" output="<studio>\1\2</studio>" dest="5+"> + <expression noclean="1">dyst.:[^=]+[^>]+>([^<]+)</a>|dyst.:[^[A-Z]+([^<]+[a-z])</expression> + </RegExp> + <RegExp input="$$1" output="<tagline>\1</tagline>" dest="5+"> + <expression>recezje-header"[^<]+[^"]+>([^<]+)</</expression> + </RegExp> + <RegExp input="$$1" output="<runtime>\1 min</runtime>" dest="5+"> + <expression>czas trwania: ([0-9]*)</expression> + </RegExp> + <RegExp conditional="!poster" input="$$1" output="<thumb>http://gfx.filmweb.pl/po\1</thumb>" dest="5+"> + <expression noclean="1">src="http://gfx.filmweb.pl/po(.[^"]*)</expression> + </RegExp> + <RegExp input="$$1" output="<outline>\2...</outline>" dest="5+"> + <expression>o-filmie-header"[^<]+[^ ]+(>[^[0-9A-Z]+)([^<]+[a-z])</expression> + </RegExp> + <RegExp input="$$1" output="\1" dest="6"> + <expression noclean="1">gatunek[^=]*[^>]*>([^<]*)</a></expression> + </RegExp> + <RegExp input="$$7" output="<genre>\1</genre>" dest="5+"> + <RegExp input="$$6" output=" / \1" dest="7"> + <expression repeat="yes">([^,]+)</expression> + </RegExp> + <expression> / (.*)</expression> + </RegExp> + <RegExp input="$$1" output="\1" dest="6"> + <expression noclean="1">yseria[^-]+([^~]*)o-filmie-header</expression> + </RegExp> + <RegExp input="$$7" output="<credits>\1</credits>" dest="5+"> + <RegExp input="$$6" output=" / \1" dest="7"> + <expression repeat="yes">title="[^>]+>([^<]+)</a></expression> + </RegExp> + <expression> / (.*)</expression> + </RegExp> + <RegExp input="$$1" output="<director>\1</director>" dest="5+"> + <expression noclean="1">yseria[^>]+>([^<]+)</expression> + </RegExp> + <RegExp conditional="!fullcredits" input="$$1" output="\1" dest="6"> + <expression noclean="1">obsada-header([^~]*)zobacz wiД™cej</expression> + </RegExp> + <RegExp conditional="!fullcredits" input="$$6" output="<actor><name>\1\3</name><role>\2\4</role></actor>" dest="5+"> + <expression repeat="yes" trim="1,3,4">src="[^>]+[^[a-zA-Z]+([^\n]+)[^h]+href=[^>]+>([^<]+)</a>|src="[^>]+[^[a-zA-Z]+([^\n]+)[^=]*[^>]+[^[a-zA-Z]+([^\n]+)</expression> + </RegExp> + <RegExp conditional="!fullcredits" input="$$6" output="<actor><thumb>http://gfx.filmweb.pl/p/\1.1.jpg</thumb><name>\2</name></actor>" dest="5+"> + <expression repeat="yes" trim="1,2">src="http://gfx.filmweb.pl/p/([^\.]+)[^>]+[^[a-zA-Z]+([^\n]+)</expression> + </RegExp> + <RegExp input="$$9" output="<url function="FilmwebToWikipedia">http://pl.wikipedia.org/w/index.php?title=Specjalna%3ALinksearch&amp;target=http://www.filmweb.pl/Film?id=$$9</url>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp conditional="fullcredits" input="$$1" output="<url function="Obsada" >http://\1.filmweb.pl/f\2/xbmc,2008/obsada</url>" dest="5+"> + <expression>http://([^\/]+).filmweb.pl/f([0-9]*)/[^>]+">peЕ‚na obsada</a></expression> + </RegExp> + <RegExp input="$$9" output="<url function="Zwiastuny" >http://www.filmweb.pl/f$$9/xbmc,2008/zwiastuny</url>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$9" output="<url function="Opisy">http://www.filmweb.pl/f$$9/xbmc,2008/opisy</url>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp conditional="poster" input="$$9" output="<url function="Plakaty">http://www.filmweb.pl/f$$9/xbmc,2008/plakaty</url>" dest="5+"> + <expression></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetDetails> + <FilmwebToWikipedia clearbuffers="no" dest="5"> + <RegExp input="$$2" output="<?xml version="1.0" encoding="utf-8" standalone="yes"?><details>\1</details>" dest="5"> + <RegExp input="$$1" output="\1" dest="10"> + <expression>id=$$9</a> linkuje z [^>]+>([^<]+)</expression> + </RegExp> + <RegExp input="$$11" output="<url function="GetTMDBFanartByIMDBId">http://pl.wikipedia.org/wiki/\1</url>" dest="2+"> + <RegExp input="$$10" output="_\1" dest="11"> + <expression repeat="yes">([^ ]+)</expression> + </RegExp> + <expression>_(.*)</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </FilmwebToWikipedia> + <Opisy dest="5"> + <RegExp input="$$8" output="<?xml version="1.0" encoding="utf-8" standalone="yes"?><details><plot>\1</plot></details>" dest="5"> + <RegExp input="$$1" output="\1\n" dest="9"> + <expression>distributors.description.0\][^\[]*justify">(.[^=]*)</p></li></expression> + </RegExp> + <RegExp input="$$1" output="\1" dest="9+"> + <expression>opisy-header[^\[]*justify">([^"]*)</p></li></expression> + </RegExp> + <RegExp input="$$9" output="\1" dest="8"> + <expression>([^<]+)</expression> + </RegExp> + <RegExp input="$$9" output=" \1" dest="8+"> + <expression repeat="yes">>([^<]+)</expression> + </RegExp> + <expression trim="1">[^[0-9A-ZЕ»Е№Д†ЕѓД„ЕљЕЃД]+([^~]+[0-9a-zД…Д™ЕјЕєД‡Е›Е„Е‚\]\.])</expression> + </RegExp> + </Opisy> + <Plakaty dest="5"> + <RegExp input="$$8" output="<?xml version="1.0" encoding="utf-8" standalone="yes"?><details>\1</details>" dest="5"> + <RegExp input="$$1" output="<thumb>http://gfx.filmweb.pl/po\1</thumb>" dest="8+"> + <expression noclean="1" repeat="yes">href="http://gfx.filmweb.pl/po(.[^"]*)</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </Plakaty> + <Obsada dest="5"> + <RegExp input="$$8" output="<?xml version="1.0" encoding="utf-8" standalone="yes"?><details>\1</details>" dest="5"> + <RegExp input="$$1" output="\1" dest="9"> + <expression noclean="1">col">aktorzy</th([^~]*)class="film-poster"</expression> + </RegExp> + <RegExp input="$$9" output="<actor><name>\1\3</name><role>\2\4</role></actor>" dest="8+"> + <expression repeat="yes" trim="1,3,4">src="[^>]+[^[a-zA-Z]+([^\n]+)[^h]+href=[^>]+>([^<]+)</a>|src="[^>]+[^[a-zA-Z]+([^\n]+)[^=]*="film-protagonist[^>]+[^[a-zA-Z]+([^\n]+)&nbsp;</expression> + </RegExp> + <RegExp input="$$9" output="<actor><thumb>http://gfx.filmweb.pl/p/\1.1.jpg</thumb><name>\2</name></actor>" dest="8+"> + <expression repeat="yes" trim="1,2">src="http://gfx.filmweb.pl/p/([^\.]+)[^>]+[^[a-zA-Z]+([^\n]+)</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </Obsada> + <Zwiastuny dest="5"> + <RegExp input="$$1" output="<details><url function="GetTrailerLink">http://www.filmweb.pl/Trailer\1</url></details>" dest="5"> + <expression clear="yes" noclean="1">href="/Trailer([^"]+)">[^\n]*[^[0-9]+4</expression> + </RegExp> + </Zwiastuny> + <GetTrailerLink dest="5"> + <RegExp input="$$1" output="<details><trailer>\1</trailer></details>" dest="5"> + <expression clear="yes" noclean="1">param name="src" value="([^"]+)</expression> + </RegExp> + </GetTrailerLink> +</scraper> diff --git a/system/scrapers/video/imdb tv.xml b/system/scrapers/video/imdb tv.xml new file mode 100644 index 0000000000..78f6276cd2 --- /dev/null +++ b/system/scrapers/video/imdb tv.xml @@ -0,0 +1,218 @@ +<?xml version="1.0" encoding="UTF-8"?> +<scraper framework="1.0" date="2009-08-10" name="IMDb TV" content="tvshows" thumb="imdb.png" language="en"> + <include>common/movieposterdb.xml</include> + <include>common/impa.xml</include> + <include>common/imdb.xml</include> + <GetSettings dest="3"> + <RegExp input="$$5" output="<settings>\1</settings>" dest="3"> + <RegExp input="$$1" output="<setting label="Enable Full Cast Credits" type="bool" id="fullcredits" default="false"></setting>" dest="5"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting label="Enable IMPAwards" type="bool" id="impawards" default="true"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting label="Enable MoviePosterDB" type="bool" id="movieposterdb" default="false"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting label="IMDB Poster &amp; Actor Thumb(s) Size" type="labelenum" values="192|256|384|512|1024" id="imdbscale" default="192"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting type="sep"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting label="Input Alternative IMDb Source" type="text" id="url" default="akas.imdb.com"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetSettings> + <NfoUrl dest="3"> + <RegExp input="$$1" output="<url>http://akas.imdb.com/title/tt\1/</url><id>tt\1</id>" dest="3"> + <expression clear="yes" noclean="1">imdb.com/Title\?([0-9]*)</expression> + </RegExp> + <RegExp input="$$1" output="<url>http://akas.imdb.com/title/tt\1/</url><id>tt\1</id>" dest="3+"> + <expression clear="yes" noclean="1">imdb.com/title/tt([0-9]*)</expression> + </RegExp> + </NfoUrl> + <CreateSearchUrl dest="3"> + <RegExp input="$$1" output="<url>http://$INFO[url]/find?s=tt;q=\1</url>" dest="3"> + <expression noclean="1"></expression> + </RegExp> + </CreateSearchUrl> + <GetSearchResults dest="8"> + <RegExp input="$$5" output="<?xml version="1.0" encoding="iso-8859-1" standalone="yes"?><results>\1</results>" dest="8"> + <RegExp input="$$1" output="\1" dest="7"> + <expression clear="yes">/title/([t0-9]*)/faq</expression> + </RegExp> + <RegExp input="$$1" output="<entity><title>\1 (\2)</title><url>http://$INFO[url]/title/$$7/</url><id>$$7</id></entity>" dest="5"> + <expression clear="yes" noclean="1"><meta name="title" content="([^"]*) \(([0-9]*)\)</expression> + </RegExp> + <RegExp input="$$1" output="\1" dest="4"> + <expression noclean="1">(><a href="/title.*)</expression> + </RegExp> + <RegExp input="$$4" output="<entity><title>\2 (\3)</title><url>http://$INFO[url]/title/\1/</url><id>\1</id></entity>" dest="5+"> + <expression repeat="yes" noclean="1,2">><a href="/title/([t0-9]*)/[^>]*>([^<]*)</a> *\(([0-9]*)</expression> + </RegExp> + <expression clear="yes" noclean="1"></expression> + </RegExp> + </GetSearchResults> + <GetDetails clearbuffers="no" dest="3"> + <RegExp input="$$5" output="<details>\1</details>" dest="3"> + <RegExp input="$$1" output="<title>\1</title>" dest="5"> + <expression trim="1" noclean="1"><title>[^;]*;([^&]*)[^;]*;</expression> + </RegExp> + <RegExp input="$$1" output="<year>\1</year>" dest="5+"> + <expression>a href="/Sections/Years/([0-9]*)"></expression> + </RegExp> + <RegExp input="$$1" output="<rating>\1</rating><votes>\2</votes>" dest="5+"> + <expression><b>User Rating:</b>[^<]*<b>([0-9.]+)/10</b>[^<]*<small>\(<a href="ratings">([0-9,]+) votes</a>\)</small></expression> + </RegExp> + <RegExp input="$$1" output="<genre>\1</genre>" dest="5+"> + <expression repeat="yes">"/Sections/Genres/[^/]*/">([^<]*)</a></expression> + </RegExp> + <RegExp input="$$2" output="<url function="GetSeriesPremiered">http://$INFO[url]/title/\1/episodes</url>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$2" output="<url function="GetSeriesPlot">http://$INFO[url]/title/\1/plotsummary</url>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp conditional="!fullcredits" input="$$2" output="<url cache="$$2-credits.html" function="GetSeriesCast">http://$INFO[url]/title/\1/</url>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp conditional="fullcredits" input="$$2" output="<url cache="$$2-credits.html" function="GetSeriesCast">http://$INFO[url]/title/\1/fullcredits#cast</url>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp conditional="impawards" input="$$2" output="<url cache="$$2-posters.html" function="GetIMPALink">http://$INFO[url]/title/\1/posters</url>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp conditional="movieposterdb" input="$$1" output="<url function="GetMoviePosterDBLink">http://www.movieposterdb.com/browse/search?type=movies&amp;query=\1</url>" dest="5+"> + <expression>/title/tt([t0-9]*)/faq</expression> + </RegExp> + <RegExp input="$$2" output="<url cache="$$2-posters.html" function="GetIMDBThumbs">http://$INFO[url]/title/\1/posters</url>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$2" output="<episodeguide><url>http://www.imdb.com/title/\1/episodes</url></episodeguide>" dest="5+"> + <expression></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetDetails> + <GetSeriesPremiered clearbuffers="no" dest="5"> + <RegExp input="$$1" output="<details><premiered>\1</premiered></details>" dest="5"> + <expression>Original Air Date: ([^<]*)<</expression> + </RegExp> + </GetSeriesPremiered> + <GetSeriesPlot clearbuffers="no" dest="5"> + <RegExp input="$$2" output="<details><plot>\1</plot></details>" dest="5"> + <RegExp input="$$1" output="\1" dest="2"> + <expression><p class="plotpar">\n([^\n]*)</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetSeriesPlot> + <GetSeriesCast clearbuffers="no" dest="5"> + <RegExp input="$$2" output="<details>\1</details>" dest="5"> + <RegExp input="$$1" output="\1" dest="6"> + <expression noclean="1"><table class="cast">(.*?)</table></expression> + </RegExp> + <RegExp input="$$6" output="<actor><thumb>\1_SY$INFO[imdbscale]_\2</thumb><name>\3</name><role>\4</role></actor>" dest="7"> + <expression repeat="yes" noclean="1,2" trim="3,4"><img src="(?:([^"]*VM\.)[^"]*(\.jpg))?[^>]*[^"]*"nm"><a href="[^"]*">([^<]*)<[^"]*"ddd"> ... [^"]*"char">(.*?)</td></expression> + </RegExp> + <RegExp input="$$7" output="<actor><thumb>\1</thumb>\2</actor>" dest="2+"> + <expression repeat="yes" noclean="1,2,3"><actor><thumb>(?:(http.*?)|_SY[0-9]+_)</thumb>(.*?)</actor></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetSeriesCast> + <GetEpisodeList dest="9"> + <RegExp input="$$4" output="<episodeguide>\1</episodeguide>" dest="9"> + <RegExp input="$$1" output="<episode><title>\4</title><url>http://$INFO[url]/title/\3/</url><epnum>\2</epnum><season>\1</season><id>\3</id></episode>" dest="4"> + <expression repeat="yes"><a name="year-[0-9]+"> </a><h4>Season ([0-9]*), Episode ([0-9]*): <a href="/title/([^"]*)/">([^<]*)</a></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetEpisodeList> + <GetEpisodeDetails dest="9"> + <RegExp input="$$4" output="<details>\1</details>" dest="9"> + <RegExp input="$$1" output="<title>\1</title>" dest="4"> + <expression trim="1" noclean="1"><title>[^;]*;[^;]*; ([^<|^(]*)</expression> + </RegExp> + <RegExp input="$$1" output="<season>\1</season>" dest="4+"> + <expression>(Season ([0-9]*), Episode [0-9]+)</expression> + </RegExp> + <RegExp input="$$1" output="<episode>\1</episode>" dest="4+"> + <expression>(Season [0-9]+, Episode ([0-9]*))</expression> + </RegExp> + <RegExp input="$$1" output="\1" dest="6"> + <expression noclean="1"><h5>Director(.*)<h5>Writer</expression> + </RegExp> + <RegExp input="$$6" output="<director>\1</director>" dest="4+"> + <expression repeat="yes"><a href="/name/[^>]*>([^<]*)<</expression> + </RegExp> + <RegExp input="$$1" output="<genre>\1</genre>" dest="4+"> + <expression repeat="yes">"/Sections/Genres/[^/]*/">([^<]*)</a></expression> + </RegExp> + <RegExp input="$$1" output="<aired>\1</aired>" dest="4+"> + <expression trim="1"><div class="info">[^<]*<h5>Original Air Date:</h5>([^\(]*)\(</expression> + </RegExp> + <RegExp input="$$1" output="\1" dest="6"> + <expression noclean="1"><h5>Writer[^>]*([^;]*)</expression> + </RegExp> + <RegExp input="$$7" output="<credits>\1</credits>" dest="4+"> + <RegExp input="$$6" output=" / \1" dest="7"> + <expression repeat="yes"><a href="/name/[^>]*>([^<]*)<</expression> + </RegExp> + <expression> / (.*)</expression> + </RegExp> + <RegExp input="$$1" output="<rating>\1</rating><votes>\2</votes>" dest="4+"> + <expression><b>User Rating:</b>[^<]*<b>([0-9.]+)/10</b>[^<]*<small>\(<a href="ratings">([0-9,]+) votes</a>\)</small></expression> + </RegExp> + <RegExp input="$$1" output="\1" dest="7"> + <expression noclean="1">first billed only: </b></td></tr>(.*)</expression> + </RegExp> + <RegExp input="$$2" output="<url function="GetEpisodePlot">http://$INFO[url]/title/\1/plotsummary</url>" dest="4+"> + <expression></expression> + </RegExp> + <RegExp conditional="!fullcredits" input="$$2" output="<url function="GetEpisodeCast">http://$INFO[url]/title/\1/</url>" dest="4+"> + <expression></expression> + </RegExp> + <RegExp conditional="fullcredits" input="$$2" output="<url function="GetEpisodeCast">http://$INFO[url]/title/\1/fullcredits#cast</url>" dest="4+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<url function="GetEpisodeRuntime">http://$INFO[url]/title/\1</url>" dest="4+"> + <expression><h5>TV Series:</h5>[^<]*<a href="/title/([^/]*)/"></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetEpisodeDetails> + <GetEpisodePlot dest="4"> + <RegExp input="$$2" output="<details><plot>\1</plot></details>" dest="4"> + <RegExp input="$$1" output="\1" dest="2"> + <expression><p class="plotpar">\n([^\n]*)</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetEpisodePlot> + <GetEpisodeCast dest="4"> + <RegExp input="$$2" output="<details>\1</details>" dest="5"> + <RegExp input="$$1" output="\1" dest="6"> + <expression noclean="1"><table class="cast">(.*?)</table></expression> + </RegExp> + <RegExp input="$$6" output="<actor><thumb>\1_SY$INFO[imdbscale]_\2</thumb><name>\3</name><role>\4</role></actor>" dest="7"> + <expression repeat="yes" noclean="1,2" trim="3,4"><img src="(?:([^"]*VM\.)[^"]*(\.jpg))?[^>]*[^"]*"nm"><a href="[^"]*">([^<]*)<[^"]*"ddd"> ... [^"]*"char">(.*?)</td></expression> + </RegExp> + <RegExp input="$$7" output="<actor><thumb>\1</thumb>\2</actor>" dest="2+"> + <expression repeat="yes" noclean="1,2,3"><actor><thumb>(?:(http.*?)|_SY[0-9]+_)</thumb>(.*?)</actor></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetEpisodeCast> + <GetEpisodeRuntime dest="4"> + <RegExp input="$$2" output="<details><runtime>\1</runtime></details>" dest="4"> + <RegExp input="$$1" output="\1" dest="2"> + <expression trim="1"><h5>Runtime:</h5>([^<]*)</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetEpisodeRuntime> +</scraper> diff --git a/system/scrapers/video/imdb-cn.gif b/system/scrapers/video/imdb-cn.gif Binary files differnew file mode 100644 index 0000000000..7d12a89f2b --- /dev/null +++ b/system/scrapers/video/imdb-cn.gif diff --git a/system/scrapers/video/imdb-cn.xml b/system/scrapers/video/imdb-cn.xml new file mode 100644 index 0000000000..694186a300 --- /dev/null +++ b/system/scrapers/video/imdb-cn.xml @@ -0,0 +1,137 @@ +<?xml version="1.0" encoding="gb2312"?>
+<scraper framework="1.1" date="2009-08-10" name="IMDb-CN(Chinese)" content="movies" thumb="imdb-cn.gif" language="zh">
+ <include>common/tmdb.xml</include>
+ <include>common/movieposterdb.xml</include>
+ <GetSettings dest="3">
+ <RegExp input="$$5" output="<settings>\1</settings>" dest="3">
+ <RegExp input="$$1" output="<setting label="Enable MoviePosterDB" type="bool" id="movieposterdb" default="false"></setting>" dest="5">
+ <expression></expression>
+ </RegExp>
+ <RegExp input="$$1" output="<setting label="Enable Fanart" type="bool" id="fanart" default="true"></setting>" dest="5+">
+ <expression></expression>
+ </RegExp>
+ <expression noclean="1"></expression>
+ </RegExp>
+ </GetSettings>
+ <NfoUrl dest="3">
+ <RegExp input="$$1" output="<url>http://www.imdb.cn/title/tt\1</url><id>tt\1</id>" dest="3">
+ <expression clear="yes" noclean="1">imdb.cn/title/tt([0-9]*)</expression>
+ </RegExp>
+ <RegExp input="$$1" output="<url>http://www.imdb.cn/title/tt\1</url><id>tt\1</id>" dest="3+">
+ <expression clear="yes" noclean="1">imdb.com/Title\?([0-9]*)</expression>
+ </RegExp>
+ <RegExp input="$$1" output="<url>http://www.imdb.cn/title/tt\1</url><id>tt\1</id>" dest="3+">
+ <expression clear="yes" noclean="1">imdb.com/title/tt([0-9]*)</expression>
+ </RegExp>
+ </NfoUrl>
+ <CreateSearchUrl dest="3" SearchStringEncoding="gb2312">
+ <RegExp input="$$1" output="<url>http://www.imdb.cn/searchh.php?keyword=\1&amp;a=search&amp;Type=movie</url>" dest="3">
+ <expression noclean="1"></expression>
+ </RegExp>
+ </CreateSearchUrl>
+ <GetSearchResults dest="8">
+ <RegExp input="$$5" output="<?xml version="1.0" encoding="gb2312" standalone="yes"?><results>\1</results>" dest="8">
+ <RegExp input="$$1" output="\1" dest="7">
+ <expression clear="yes">/title/([t0-9]*)/ratings</expression>
+ </RegExp>
+ <RegExp input="$$1" output="\1" dest="6">
+ <expression clear="yes"><td colspan=2><span class=mn>([^<]*)</span></expression>
+ </RegExp>
+ <RegExp input="$$1" output="<entity><title>\1 - $$6 (\2)</title><url>http://www.imdb.cn/title/$$7</url><id>\$$7</id></entity>" dest="5">
+ <expression clear="yes" noclean="1"><h1>([^<]*)</h1>[^"]*"http://www.imdb.cn/Sections/Years/([0-9]*)"</expression>
+ </RegExp>
+ <RegExp input="$$1" output="<entity><title>\2 - \3 (\4)</title><url>http://www.imdb.cn/title/\1</url><id>\1</id></entity>" dest="5+">
+ <expression repeat="yes" noclean="1,3"><a href="title/([t0-9]*)" target="_blank">(.[^-]*)-(.[^<]*)</a> - \(([0-9]*)\)</expression>
+ </RegExp>
+ <expression clear="yes" noclean="1"/>
+ </RegExp>
+ </GetSearchResults>
+ <GetDetails dest="3">
+ <RegExp input="$$5" output="<details>\1</details>" dest="3">
+ <RegExp input="$$2" output="<id>\1</id>" dest="5">
+ <expression></expression>
+ </RegExp>
+ <RegExp input="$$1" output="<title>\1</title>" dest="5+">
+ <expression trim="1" noclean="1"><title>([^<|^-]*)</expression>
+ </RegExp>
+ <RegExp input="$$1" output="<originaltitle>\1</originaltitle>" dest="5+">
+ <expression trim="1" noclean="1">Ж¬ ГыЈє</span>.*?class=mn>(.[^<]*)</expression>
+ </RegExp>
+ <RegExp input="$$1" output="<year>\1</year>" dest="5+">
+ <expression><span class=hhh2>\(([0-9]*)\)</expression>
+ </RegExp>
+ <RegExp input="$$1" output="<top250>\1</top250>" dest="5+">
+ <expression>TOP <font color=red>([0-9]*) </font>></expression>
+ </RegExp>
+ <RegExp input="$$4" output="\1" dest="5+">
+ <RegExp input="$$1" output="<mpaa>\1</mpaa>\n" dest="4">
+ <expression>MPAA[^<]*</span><br>(.[^<]*)</expression>
+ </RegExp>
+ <expression repeat="yes" noclean="1">(.*?)\n</expression>
+ </RegExp>
+ <RegExp input="$$1" output="<certification>\1</certification>" dest="5+">
+ <expression repeat="yes"><a href=http://www.imdb.cn/Sections/Certification/[^>]*>([^<]*)</a></expression>
+ </RegExp>
+ <RegExp input="$$4" output="\1" dest="5+">
+ <RegExp input="$$1" output="<runtime>\1</runtime>\n" dest="4">
+ <expression trim="1"><td colspan="2"><font color=red>([^<]*)</expression>
+ </RegExp>
+ <expression repeat="yes" noclean="1">(.*?)\n</expression>
+ </RegExp>
+ <RegExp input="$$1" output="<rating>\1</rating><votes>\2</votes>" dest="5+">
+ <expression>\.gif" ><b>([0-9.]+)/10</b> \( ([0-9]+)</expression>
+ </RegExp>
+ <RegExp input="$$1" output="<genre>\1</genre>" dest="5+">
+ <expression repeat="yes"><a href=http://www.imdb.cn/Sections/Genre/[^>]*>([^<]*)</a></expression>
+ </RegExp>
+ <RegExp input="$$1" output="<outline>\1</outline><plot>\1</plot>" dest="5+">
+ <expression><div id=imdbjqbody><h2>[^<]*</h2><br>(.*?)</div</expression>
+ </RegExp>
+ <RegExp input="$$2" output="<url cache="$$2-fullcredits.html" function="GetFullcredits">$$3/fulldredits</url>" dest="5+">
+ <expression></expression>
+ </RegExp>
+ <RegExp conditional="movieposterdb" input="$$1" output="<url function="GetMoviePosterDBLink">http://www.movieposterdb.com/browse/search?type=movies&amp;query=\1</url>" dest="5+">
+ <expression>/title/tt([0-9]*)/ratings</expression>
+ </RegExp>
+ <RegExp input="$$1" output="<url function="GetIMDBCNPoster">\1</url>" dest="5+">
+ <expression><a href=(http://posters.imdb.cn/poster/[0-9]*)</expression>
+ </RegExp>
+ <RegExp conditional="fanart" input="$$2" output="<url function="GetTMDBFanartByIMDBId">http://api.themoviedb.org/2.0/Movie.imdbLookup?imdb_id=$$2&amp;api_key=57983e31fb435df4df77afb854740ea9</url>" dest="5+">
+ <expression></expression>
+ </RegExp>
+ <expression noclean="1"></expression>
+ </RegExp>
+ </GetDetails>
+
+ <GetFullcredits clearbuffers="no" dest="5">
+ <RegExp input="$$2" output="<details>\1</details>" dest="5">
+ <RegExp input="$$1" output="\1" dest="6">
+ <expression clear="yes" noclean="1">СЭФ± <table(.*?)</table></expression>
+ </RegExp>
+ <RegExp input="$$6" output="<actor><name>\1</name><role>\2</role></actor>" dest="2">
+ <expression repeat="yes" trim="1,2">href='/name/nm[0-9]+'>([^<]*)</a>[^\.]*\.\.\.\.\.\.([^<]*)</td></expression>
+ </RegExp>
+ <RegExp input="$$1" output="\1" dest="7">
+ <expression clear="yes" noclean="1">µјСЭ <table(.*?)</table></expression>
+ </RegExp>
+ <RegExp input="$$7" output="<director>\1</director>" dest="2+">
+ <expression repeat="yes">href='/name/nm[0-9]+'>([^<]*)</a></expression>
+ </RegExp>
+ <RegExp input="$$1" output="\1" dest="8">
+ <expression clear="yes" noclean="1">±аѕз <table(.*?)</table></expression>
+ </RegExp>
+ <RegExp input="$$8" output="<credits>\1</credits>" dest="2+">
+ <expression repeat="yes">href='/name/nm[0-9]+'>([^<]*)</a></expression>
+ </RegExp>
+ <expression noclean="1"></expression>
+ </RegExp>
+ </GetFullcredits>
+ <GetIMDBCNPoster dest="5">
+ <RegExp input="$$6" output="<details>\1</details>" dest="5">
+ <RegExp input="$$1" output="<thumb>\1</thumb>" dest="6">
+ <expression clear="yes" noclean="1"><img src="(.*?/poster-pic/.*?.jpg)"</expression>
+ </RegExp>
+ <expression noclean="1"></expression>
+ </RegExp>
+ </GetIMDBCNPoster>
+</scraper>
diff --git a/system/scrapers/video/imdb.png b/system/scrapers/video/imdb.png Binary files differnew file mode 100644 index 0000000000..b2edac3fe7 --- /dev/null +++ b/system/scrapers/video/imdb.png diff --git a/system/scrapers/video/imdb.xml b/system/scrapers/video/imdb.xml new file mode 100644 index 0000000000..7cbdb81389 --- /dev/null +++ b/system/scrapers/video/imdb.xml @@ -0,0 +1,159 @@ +<?xml version="1.0" encoding="UTF-8"?> +<scraper framework="1.1" date="2009-08-10" name="IMDb" content="movies" thumb="imdb.png" language="en"> + <include>common/imdb.xml</include> + <include>common/tmdb.xml</include> + <include>common/movieposterdb.xml</include> + <include>common/impa.xml</include> + <include>common/dtrailer.xml</include> + <GetSettings dest="3"> + <RegExp input="$$5" output="<settings>\1</settings>" dest="3"> + <RegExp input="$$1" output="<setting label="Enable Full Cast Credits" type="bool" id="fullcredits" default="false"></setting>" dest="5"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting label="Enable themoviedb.org" type="bool" id="tmdbthumbs" default="false"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting label="Enable IMPAwards" type="bool" id="impawards" default="true"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting label="Enable MoviePosterDB" type="bool" id="movieposterdb" default="false"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting label="Enable IMDB Trailer" type="bool" id="trailer" default="true"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting label="Enable DTrailer.com Trailer" type="bool" id="dtrailer" default="false"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting label="Enable Fanart" type="bool" id="fanart" default="true"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting label="IMDB Poster &amp; Actor Thumb(s) Size" type="labelenum" values="192|256|384|512|1024" id="imdbscale" default="512"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting type="sep"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting label="Input Alternative IMDb Source" type="text" id="url" default="akas.imdb.com"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetSettings> + <NfoUrl dest="3"> + <RegExp input="$$1" output="<url>http://www.\1/title/tt\2/</url><id>tt\2</id>" dest="3"> + <expression clear="yes" noclean="1">(imdb.com/)Title\?([0-9]*)</expression> + </RegExp> + <RegExp input="$$1" output="<url>http://www.\1\2/</url><id>tt\2</id>" dest="3+"> + <expression clear="yes" noclean="1">(imdb.com/title/tt)([0-9]*)</expression> + </RegExp> + </NfoUrl> + <CreateSearchUrl dest="3"> + <RegExp input="$$1" output="<url>http://$INFO[url]/find?s=tt;q=\1$$4</url>" dest="3"> + <RegExp input="$$2" output="%20(\1)" dest="4"> + <expression clear="yes">(.+)</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </CreateSearchUrl> + <GetSearchResults dest="8"> + <RegExp input="$$5" output="<?xml version="1.0" encoding="iso-8859-1" standalone="yes"?><results>\1</results>" dest="8"> + <RegExp input="$$1" output="\1" dest="7"> + <expression clear="yes">/title/([t0-9]*)/faq</expression> + </RegExp> + <RegExp input="$$1" output="<entity><title>\1</title><year>\2</year><url>http://$INFO[url]/title/$$7/</url><id>$$7</id></entity>" dest="5"> + <expression clear="yes" noclean="1"><meta name="title" content="([^"]*) \(([0-9]*)\)</expression> + </RegExp> + <RegExp input="$$1" output="\1" dest="4"> + <expression noclean="1">(><a href="/title.*)</expression> + </RegExp> + <RegExp input="$$4" output="<entity><title>\2</title><year>\3</year><url>http://$INFO[url]/title/\1/</url><id>\1</id></entity>" dest="5+"> + <expression repeat="yes" noclean="1,2">><a href="/title/([t0-9]*)/[^>]*>([^<]*)</a> *\(([0-9]*)</expression> + </RegExp> + <expression clear="yes" noclean="1"></expression> + </RegExp> + </GetSearchResults> + <GetDetails dest="3"> + <RegExp input="$$5" output="<details>\1</details>" dest="3"> + <RegExp input="$$2" output="<id>\1</id>" dest="5"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<title>\1</title>" dest="5+"> + <expression trim="1" noclean="1"><title>([^<|^(]*)</expression> + </RegExp> + <RegExp input="$$1" output="<year>\1</year>" dest="5+"> + <expression>a href="/Sections/Years/([0-9]*)</expression> + </RegExp> + <RegExp input="$$1" output="<top250>\1</top250>" dest="5+"> + <expression>Top 250: #([0-9]*)</a></expression> + </RegExp> + <RegExp input="$$1" output="<mpaa>\1</mpaa>" dest="5+"> + <expression>MPAA</a>:</h5>(.[^<]*)</expression> + </RegExp> + <RegExp input="$$1" output="<certification>\1 \3</certification>" dest="5+"> + <expression repeat="yes"><a href="/List\?certificates=[^"]*">([^<]*)</a>[^<]*(<i>([^<]*)</i>)?</expression> + </RegExp> + <RegExp input="$$1" output="<tagline>\1</tagline>" dest="5+"> + <expression><h5>Tagline:</h5>([^<]*)</expression> + </RegExp> + <RegExp input="$$1" output="<runtime>\1</runtime>" dest="5+"> + <expression trim="1"><h5>Runtime:</h5>[^0-9]*([^<]*)</expression> + </RegExp> + <RegExp input="$$1" output="<rating>\1</rating><votes>\2</votes>" dest="5+"> + <expression><b>([0-9.]+)/10</b>[^<]*<a href="ratings" class="tn15more">([0-9,]+) votes</a></expression> + </RegExp> + <RegExp input="$$1" output="<genre>\1</genre>" dest="5+"> + <expression repeat="yes">"/Sections/Genres/[^/]*/">([^<]*)</a></expression> + </RegExp> + <RegExp input="$$1" output="<studio>\1</studio>" dest="5+"> + <expression repeat="yes">"/company/[^/]*/">([^<]*)</a></expression> + </RegExp> + <RegExp input="$$1" output="<outline>\1</outline><plot>\1</plot>" dest="5+"> + <expression>Plot:</h5>(.*?) \| <a class="tn15more</expression> + </RegExp> + <RegExp input="$$2" output="<url function="GetIMDBPlot">$$3plotsummary</url>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp conditional="!fullcredits" input="$$2" output="<url cache="$$2-credits.html" function="GetIMDBCast">$$3</url>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp conditional="!fullcredits" input="$$2" output="<url cache="$$2-credits.html" function="GetIMDBDirectors">$$3</url>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp conditional="!fullcredits" input="$$2" output="<url cache="$$2-credits.html" function="GetIMDBWriters">$$3</url>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp conditional="fullcredits" input="$$2" output="<url cache="$$2-fullcredits.html" function="GetIMDBCast">$$3fullcredits</url>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp conditional="fullcredits" input="$$2" output="<url cache="$$2-fullcredits.html" function="GetIMDBDirectors">$$3fullcredits</url>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp conditional="fullcredits" input="$$2" output="<url cache="$$2-fullcredits.html" function="GetIMDBWriters">$$3fullcredits</url>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp conditional="tmdbthumbs" input="$$2" output="<url function="GetTMDBThumbsByIMDBId" cache="tmdb-trans-\1.xml">http://api.themoviedb.org/2.0/Movie.imdbLookup?imdb_id=\1&amp;api_key=57983e31fb435df4df77afb854740ea9</url>" dest="5+"> + <expression/> + </RegExp> + <RegExp conditional="impawards" input="$$2" output="<url cache="$$2-posters.html" function="GetIMPALink">$$3posters</url>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp conditional="movieposterdb" input="$$1" output="<url function="GetMoviePosterDBLink">http://www.movieposterdb.com/browse/search?type=movies&amp;query=\1</url>" dest="5+"> + <expression>/title/tt([t0-9]*)/faq</expression> + </RegExp> + <RegExp conditional="trailer" input="$$1" output="<url function="GetIMDBTrailer">http://$INFO[url]/video/imdb/vi\1/player</url>" dest="5+"> + <expression>/vi([0-9]*)/"</expression> + </RegExp> + <RegExp conditional="dtrailer" input="$$1" output="<url function="GetDTrailerLink">http://en.dtrailer.com/movies/search/\1</url>" dest="5+"> + <expression trim="1" noclean="1"><title>([^<|^(]*)</expression> + </RegExp> + <RegExp input="$$2" output="<url cache="$$2-posters.html" function="GetIMDBThumbs">$$3posters</url>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp conditional="fanart" input="$$2" output="<url function="GetTMDBFanartByIMDBId">http://api.themoviedb.org/2.0/Movie.imdbLookup?imdb_id=$$2&amp;api_key=57983e31fb435df4df77afb854740ea9</url>" dest="5+"> + <expression></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetDetails> +</scraper> diff --git a/system/scrapers/video/jaded.jpg b/system/scrapers/video/jaded.jpg Binary files differnew file mode 100644 index 0000000000..b5c2fcadc1 --- /dev/null +++ b/system/scrapers/video/jaded.jpg diff --git a/system/scrapers/video/jadedVideo.xml b/system/scrapers/video/jadedVideo.xml new file mode 100644 index 0000000000..e3b6efb7c2 --- /dev/null +++ b/system/scrapers/video/jadedVideo.xml @@ -0,0 +1,47 @@ +<scraper framework="1.0" date="2009-05-22" name="Jaded Video" content="movies" thumb="jaded.jpg" language="en"> + <NfoUrl dest="3"> + <RegExp input="$$1" output="http://jadedvideo.com/\1" dest="3"> + <expression noclean="1">jadedvideo\.com/(.*)</expression> + </RegExp> + </NfoUrl> + <CreateSearchUrl dest="3"> + <RegExp input="$$1" output="<url>http://jadedvideo.com/Search?q=\1</url>" dest="3"> + <expression noclean="1"></expression> + </RegExp> + + </CreateSearchUrl> + + <GetSearchResults dest="6"> + <RegExp input="$$5" output="<?xml version="1.0" encoding="iso-8859-1" standalone="yes"?><results>\1</results>" dest="6"> + <RegExp input="$$1" output="<entity><title>\3</title><url>http://jadedvideo.com/\1</url></entity>" dest="5+"> + <expression repeat="yes">"jadedcatprodtitle"><a href="http://jadedvideo\.com/([^"]*)"><em>([^>]*)</em> ([^<]*)</a></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + + </GetSearchResults> + + <GetDetails dest="3"> + <RegExp input="$$5" output="<details>\1</details>" dest="3"> + <RegExp input="$$1" output="<thumb>http://jadedvideo.com/images/CoverFront/320x480/\1.jpg</thumb>" dest="5"> + <expression>320x480/([0-9]*).jpg"</expression> + </RegExp> + <RegExp input="$$1" output="<title>\1</title>" dest="5+"> + <expression>jadedprodtitle">([^<]*)</expression> + </RegExp> + <RegExp input="$$1" output="<actor><name>\1</name></actor>" dest="5+"> + <expression repeat="yes"><a[^>]*>([^<]*)</a>,</expression> + </RegExp> + <RegExp input="$$1" output="<genre>\1</genre>" dest="5+"> + <expression>Category</strong>: <a[^>]*>([^<]*)</a></expression> + </RegExp> + <RegExp input="$$1" output="<year>\1</year>" dest="5+"> + <expression>Date</strong>[^<]* ([0-9]*)</expression> + </RegExp> + <RegExp input="$$1" output="<studio>\1</studio>" dest="5+"> + <expression>Studio</strong>: <[^>]*>([^<]*)</a></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetDetails> +</scraper> diff --git a/system/scrapers/video/moviemaze.jpg b/system/scrapers/video/moviemaze.jpg Binary files differnew file mode 100644 index 0000000000..b810048168 --- /dev/null +++ b/system/scrapers/video/moviemaze.jpg diff --git a/system/scrapers/video/moviemaze.xml b/system/scrapers/video/moviemaze.xml new file mode 100644 index 0000000000..3c786ba1ae --- /dev/null +++ b/system/scrapers/video/moviemaze.xml @@ -0,0 +1,174 @@ +<?xml version="1.0" encoding="utf-8" standalone="yes"?> +<scraper framework="1.1" date="2009-07-28" name="MovieMaze" content="movies" thumb="moviemaze.jpg" language="de"> + <include>common/tmdb.xml</include> + <GetSettings dest="3"> + <RegExp input="$$5" output="<settings>\1</settings>" dest="3"> + <RegExp input="$$1" output="<setting label="Fanart" type="bool" id="fanart" default="true"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting label="Trailer" type="bool" id="trailer" default="true"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetSettings> + + <NfoUrl dest="3"> + <RegExp input="$$1" output="<url>\1\2/\3.html</url>" dest="3"> + <expression clear="yes" noclean="1">(http://www.moviemaze.de/filme/)([0-9]*)/(.*).html</expression> + </RegExp> + <RegExp input="$$1" output="<url function="GetByIMDBId">http://www.imdb.com/title/tt\2/externalreviews</url>" dest="3+"> + <expression>(imdb.com/)Title\?([0-9]+)</expression> + </RegExp> + <RegExp input="$$1" output="<url function="GetByIMDBId">http://www.imdb.com/title/tt\2/externalreviews</url>" dest="3+"> + <expression>(imdb.com/title/tt)([0-9]*)</expression> + </RegExp> + </NfoUrl> + + <GetByIMDBId dest="3"> + <RegExp input="$$1" output="<url>http://www.moviemaze.de/filme/\1/\2</url><id>\1</id>" dest="3+"> + <expression noclean="1"><a href="http://www.moviemaze.de/filme/([0-9]*)/([^"]*)</expression> + </RegExp> + </GetByIMDBId> + + <CreateSearchUrl dest="3"> + <RegExp input="$$1" output="http://www.moviemaze.de/suche/result.phtml?searchword=\1" dest="3"> + <expression noclean="1"/> + </RegExp> + </CreateSearchUrl> + + <GetSearchResults dest="8"> + <RegExp input="$$5" output="<?xml version="1.0" encoding="iso-8859-1" standalone="yes"?><results>\1</results>" dest="8"> + <RegExp input="$$1" output="<entity><title>\3</title><url>http://www.moviemaze.de/filme/\1/\2</url></entity>" dest="5+"> + <expression repeat="yes"><a href="/filme/([0-9]+)/([^"]*)"><b style="font-size:9pt;font-weight:bold;">([^<]*)</b></expression> + </RegExp> + <expression noclean="1"/> + </RegExp> + </GetSearchResults> + + <GetDetails dest="3"> + <RegExp input="$$5" output="<details>\1</details>" dest="3"> + <!--Title--> + <RegExp input="$$1" output="<title>\1</title>" dest="5+"> + <expression trim="1" noclean="1"><h1>([^<]*)</expression> + </RegExp> + <!--Original Title--> + <RegExp input="$$1" output="<originaltitle>\1</originaltitle>" dest="5+"> + <expression><h2>\(([^,]*)</expression> + </RegExp> + <!--Year Film--> + <RegExp input="$$1" output="<year>\1</year>" dest="5+"> + <expression>([0-9]+)\)</h2</expression> + </RegExp> + <!--MPAA Film--> + <RegExp input="$$2" output="<mpaa>\1</mpaa>" dest="5+"> + <RegExp input="$$1" output="\2" dest="2"> + <expression trim="2">FSK([^"]*)"standard" valign="top"><nobr>([^<]*)</expression> + </RegExp> + <expression>[\t]\b(.*)\b</expression> + </RegExp> + <!--Director Film--> + <RegExp input="$$7" output="<director>\1</director>" dest="5+"> + <RegExp input="$$1" output="\2" dest="2"> + <expression>Regie([^"]*)"standard_justify">([^<]*)</expression> + </RegExp> + <RegExp input="$$2" output="\1" dest="7"> + <expression>[\t]\b(.*)\b</expression> + </RegExp> + <expression repeat="yes">([^,]+)</expression> + </RegExp> + <!--Runtime Film--> + <RegExp input="$$2" output="<runtime>\1</runtime>" dest="5+"> + <RegExp input="$$1" output="\2" dest="2"> + <expression>L.nge([^"]*)"standard" valign="top"><nobr>([^<]*)</expression> + </RegExp> + <expression>[\t]\b(.*)\b</expression> + </RegExp> + <!--URL to Thumbnail--> + <RegExp input="$$1" output="<url function="GetThumbnailLink">http://www.moviemaze.de/media/poster/\1/\2</url>" dest="5+"> + <expression>/media/poster/([0-9]+)/([^"]*)</expression> + </RegExp> + <!--URL to Trailer--> + <RegExp conditional="trailer" input="$$1" output="<url function="GetTrailerLink">http://www.moviemaze.de/media/trailer/\1.html</url>" dest="5+"> + <expression>/media/trailer/([0-9]+,[a-zA-Z0-9\-]+).html. ti</expression> + </RegExp> + <!--URL to Google and Fanart--> + <RegExp conditional="fanart" input="$$8" output="<url function="GetTMDBFanartByIMDBId">http://www.google.com/search?q=site:imdb.com+moviemaze\1</url>" dest="5+"> + <RegExp input="$$1" output=""\1"" dest="7"> + <expression><h2>\(([^,]*)</expression> + </RegExp> + <RegExp input="$$7" output="+\1" dest="8+"> + <expression repeat="yes">([^ ,]+)</expression> + </RegExp> + <expression></expression> + </RegExp> + <!--User rating--> + <RegExp input="$$1" output="<rating>\1</rating><votes>100</votes>" dest="5+"> + <expression trim="1" noclean="1">Wertung von ([0-9]+)%</expression> + </RegExp> + <!--Genre Film--> + <RegExp input="$$7" output="<genre>\1</genre>" dest="5+"> + <RegExp input="$$1" output="\2" dest="2"> + <expression>Genre([^"]*)"standard" valign="top">([^<]*)</expression> + </RegExp> + <RegExp input="$$2" output="\1" dest="7"> + <expression>[\t]\b(.*)\b</expression> + </RegExp> + <expression repeat="yes">([^, ]+)</expression> + </RegExp> + <!--Writer--> + <RegExp input="$$7" output="<credits>\1</credits>" dest="5+"> + <RegExp input="$$1" output="\2" dest="2"> + <expression trim="2">Drehbuch([^%]*)%>([^<]*)</expression> + </RegExp> + <RegExp input="$$2" output="\1" dest="7"> + <expression>[\t]\b(.*)\b</expression> + </RegExp> + <expression repeat="yes">([^,]+)</expression> + </RegExp> + <!--Plot--> + <RegExp input="$$1" output="<plot>\2</plot>" dest="5+"> + <expression trim="1">Inhalt</h3>.([^>]*)>([^<]*)</expression> + </RegExp> + <!--Actors--> + <RegExp input="$$4" output="<actor><name>\2</name><role>\5</role></actor>" dest="5+"> + <RegExp input="$$1" output="\2" dest="2"> + <expression>Darsteller:([^%]*)%>(.*?)</td</expression> + </RegExp> + <RegExp input="$$2" output="\1" dest="4"> + <expression>[\t]\b(.*)\b</expression> + </RegExp> + <expression repeat="yes">(<a href\="[^>]*>)?(.*?)(</a>)?( \((.*?)\))?, </expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetDetails> + + <!--Thumbnail--> + <GetThumbnailLink dest="5"> + <RegExp input="$$2" output="<?xml version="1.0" encoding="iso-8859-1" standalone="yes"><details>\1</details>" dest="5+"> + <RegExp input="$$1" output="<thumb>http://www.moviemaze.de/filme/\1/poster_lg\2.jpg</thumb>" dest="2"> + <expression repeat="yes">/([0-9]+)/poster([0-9]+)</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetThumbnailLink> + <!--Trailer--> + <GetTrailerLink dest="5"> + <RegExp input="$$2" output="<?xml version="1.0" encoding="iso-8859-1" standalone="yes"><details>\1</details>" dest="5+"> + <RegExp input="$$1" output="<trailer urlencoded="yes">http://www.moviemaze.de/media/trailer/delivery/\1.mov</trailer>" dest="2+"> + <expression>delivery/([0-9]+/[a-z0-9]+/[0-9]+_trailer[0-9]+-de_640).mov</expression> + </RegExp> + <RegExp input="$$1" output="<trailer urlencoded="yes">http://www.moviemaze.de/media/trailer/delivery/\1.mov</trailer>" dest="2+"> + <expression>delivery/([0-9]+/[a-z0-9]+/[0-9]+_trailer[0-9]+-de_480).mov</expression> + </RegExp> + <RegExp input="$$1" output="<trailer urlencoded="yes">http://www.moviemaze.de/media/trailer/delivery/\1.mov</trailer>" dest="2+"> + <expression>delivery/([0-9]+/[a-z0-9]+/[0-9]+_trailer[0-9]+-de_384).mov</expression> + </RegExp> + <RegExp input="$$1" output="<trailer urlencoded="yes">http://www.moviemaze.de/media/trailer/delivery/\1.mov</trailer>" dest="2+"> + <expression>delivery/([0-9]+/[a-z0-9]+/[0-9]+_trailer[0-9]+-de_320).mov</expression> + </RegExp> + <expression clear="yes" noclean="1"></expression> + </RegExp> + </GetTrailerLink> +</scraper> diff --git a/system/scrapers/video/mtime.jpg b/system/scrapers/video/mtime.jpg Binary files differnew file mode 100644 index 0000000000..80ebd7c2ec --- /dev/null +++ b/system/scrapers/video/mtime.jpg diff --git a/system/scrapers/video/mtime.xml b/system/scrapers/video/mtime.xml new file mode 100644 index 0000000000..ae6fea8680 --- /dev/null +++ b/system/scrapers/video/mtime.xml @@ -0,0 +1,137 @@ +<?xml version="1.0" encoding="UTF-8"?>
+<scraper framework="1.1" date="2009-08-04" name="Mtime(Chinese)" content="movies" thumb="mtime.jpg" language="zh">
+ <GetSettings dest="3">
+ <RegExp input="$$5" output="<settings>\1</settings>" dest="3">
+ <RegExp input="$$1" output="<setting label="Enable Fanart" type="bool" id="fanart" default="true"></setting>" dest="5+">
+ <expression></expression>
+ </RegExp>
+ <expression noclean="1"></expression>
+ </RegExp>
+ </GetSettings>
+ <NfoUrl dest="3">
+ <RegExp input="$$1" output="<url>\1</url>" dest="3">
+ <expression clear="yes" noclean="1">(http://www.mtime.com/movie/[0-9]*)</expression>
+ </RegExp>
+ </NfoUrl>
+ <CreateSearchUrl dest="3" SearchStringEncoding="gb2312">
+ <RegExp input="$$1" output="<url>http://www.mtime.com/search/movie?\1</url>" dest="3">
+ <expression noclean="1"></expression>
+ </RegExp>
+ </CreateSearchUrl>
+ <GetSearchResults dest="8">
+ <RegExp input="$$5" output="<?xml version="1.0" encoding="iso-8859-1" standalone="yes"?><results>\1</results>" dest="8">
+ <RegExp input="$$1" output="<entity><title>\2</title><url>http://www.mtime.com/movie/\1/</url><id>\1</id></entity>" dest="5+">
+ <expression repeat="yes" noclean="1"><a id="highlight" name="highlight" href="/movie/([0-9]*)/" target="_blank" title="([^"]*)"></expression>
+ </RegExp>
+ <expression clear="yes" noclean="1"/>
+ </RegExp>
+ </GetSearchResults>
+ <GetDetails dest="3">
+ <RegExp input="$$5" output="<details>\1</details>" dest="3">
+ <RegExp input="$$2" output="<id>\1</id>" dest="5">
+ <expression></expression>
+ </RegExp>
+ <RegExp input="$$1" output="<title>\1</title>" dest="5+">
+ <expression trim="1" noclean="1"><a href="/movie/[0-9]*/" class="px28 bold hei c_000">([^<]*)</a></expression>
+ </RegExp>
+ <RegExp input="$$1" output="<year>\1</year>" dest="5+">
+ <expression>href="/movie/section/year/([0-9]*)/"</expression>
+ </RegExp>
+ <RegExp input="$$2" output="<url cache="$$2-details.html" function="GetDetailsPage">$$3details.html</url>" dest="5+">
+ <expression></expression>
+ </RegExp>
+ <RegExp input="$$1" output="<runtime>\1</runtime>" dest="5+">
+ <expression trim="1">片长：</strong>([^<]*)<</expression>
+ </RegExp>
+ <RegExp input="$$2" output="<url cache="$$2-rating.html" function="GetRating">$$3rating/summary.html</url>" dest="5+">
+ <expression></expression>
+ </RegExp>
+ <RegExp input="$$1" output="<genre>\1</genre>" dest="5+">
+ <expression repeat="yes">href="/movie/section/genre/[^>]*>([^<]*)& </expression>
+ </RegExp>
+ <RegExp input="$$1" output="<outline>\1</outline><plot>\1</plot>" dest="5+">
+ <expression><!-- 剧情介绍 -->.*?<p >(.*?)\t</p></expression>
+ </RegExp>
+ <RegExp input="$$2" output="<url cache="$$2-fullcredits.html" function="GetFullcredits">$$3fullcredits.html</url>" dest="5+">
+ <expression></expression>
+ </RegExp>
+ <RegExp input="$$2" output="<url cache="$$2-posters.html" function="GetMtimePoster">$$3posters_and_images/posters.html</url>" dest="5+">
+ <expression></expression>
+ </RegExp>
+ <RegExp conditional="fanart" input="$$2" output="<url cache="$$2-fanart.html" function="GetMtimeWallpapers">$$3posters_and_images/wallpapers.html</url>" dest="5+">
+ <expression></expression>
+ </RegExp>
+ <RegExp conditional="fanart" input="$$2" output="<url cache="$$2-fanart.html" function="GetMtimeFanart">$$3posters_and_images/stills.html</url>" dest="5+">
+ <expression></expression>
+ </RegExp>
+ <expression noclean="1"></expression>
+ </RegExp>
+ </GetDetails>
+
+ <GetDetailsPage clearbuffers="no" dest="5">
+ <RegExp input="$$2" output="<details>\1</details>" dest="5">
+ <RegExp input="$$1" output="<mpaa>\1</mpaa>" dest="2">
+ <expression clear="yes">MPAAиЇ„зє§пјљ.*?<li>([^<]*)</li></expression>
+ </RegExp>
+ <RegExp input="$$1" output="<certification>\1</certification>" dest="2+">
+ <expression repeat="yes">href="/movie/section/certification/[^"]*">([^<]*)</a></expression>
+ </RegExp>
+ <expression noclean="1"></expression>
+ </RegExp>
+ </GetDetailsPage>
+ <GetRating clearbuffers="no" dest="5">
+ <RegExp input="$$2" output="<details>\1</details>" dest="5">
+ <RegExp input="$$1" output="<rating>\1</rating><votes>\2</votes>" dest="2">
+ <expression clear="yes">е…ЁйѓЁдјље‘.*?class="score">([0-9.]+)</span>.*?class="vote">([0-9]+)</span></expression>
+ </RegExp>
+ <expression noclean="1"></expression>
+ </RegExp>
+ </GetRating>
+ <GetFullcredits clearbuffers="no" dest="5">
+ <RegExp input="$$2" output="<details>\1</details>" dest="5">
+ <RegExp input="$$1" output="\1" dest="6">
+ <expression clear="yes" noclean="1">жј”е‘ Actorпјљ(.*?)</ul></expression>
+ </RegExp>
+ <RegExp input="$$6" output="<actor><thumb>\1.jpg</thumb><name>\2</name><role>\3</role></actor>" dest="7">
+ <expression clear="yes" repeat="yes" noclean="1" trim="2,3"><img alt="[^"]*" src="([^_]*)_22X22.jpg"/></a><a href="/person/[0-9]+/">([^<]*)</a>(.*?)</li></expression>
+ </RegExp>
+ <RegExp input="$$7" output="<actor>\1<role>\2</role></actor>" dest="2">
+ <expression clear="yes" repeat="yes" noclean="1,2"><actor>(.*?)<role>[^<]*\.\.\.\.([^<]*)</role></actor></expression>
+ </RegExp>
+ <RegExp input="$$1" output="\1" dest="8">
+ <expression clear="yes" noclean="1">еЇјжј” Directorпјљ(.*?)</ul></expression>
+ </RegExp>
+ <RegExp input="$$8" output="<director>\1</director>" dest="2+">
+ <expression repeat="no"><a href="/person/[0-9]+/">([^<]*)<</expression>
+ </RegExp>
+ <RegExp input="$$1" output="\1" dest="9">
+ <expression clear="yes" noclean="1">编剧 Writer：(.*?)</ul></expression>
+ </RegExp>
+ <RegExp input="$$9" output="<credits>\1</credits>" dest="2+">
+ <expression repeat="no"><a href="/person/[0-9]+/">([^<]*)<</expression>
+ </RegExp>
+ <expression noclean="1"></expression>
+ </RegExp>
+ </GetFullcredits>
+ <GetMtimePoster dest="5">
+ <RegExp input="$$6" output="<details>\1</details>" dest="5">
+ <RegExp input="$$1" output="<thumb>\1.jpg</thumb>" dest="6">
+ <expression clear="yes" repeat="yes" noclean="1">src="(http://img1.mtime.com/[^_]*)_[0-9X]+.jpg"/></a></expression>
+ </RegExp>
+ <expression noclean="1"></expression>
+ </RegExp>
+ </GetMtimePoster>
+ <GetMtimeWallpapers clearbuffers="no" dest="6">
+ <RegExp input="$$1" output="<thumb preview="\1.jpg">\1_o.jpg</thumb>" dest="10">
+ <expression clear="yes" repeat="yes" noclean="1">src="http://img1.mtime.com/([^_]*)(_[0-9X]+).jpg"/></a></expression>
+ </RegExp>
+ </GetMtimeWallpapers>
+ <GetMtimeFanart dest="5">
+ <RegExp input="$$10$$11" output="<details><fanart url="http://img1.mtime.com/">\1</fanart></details>" dest="5">
+ <RegExp input="$$1" output="<thumb preview="\1.jpg">\1_o.jpg</thumb>" dest="11">
+ <expression clear="yes" repeat="yes" noclean="1">src="http://img1.mtime.com/([^_]*)(_[0-9X]+).jpg"/></a></expression>
+ </RegExp>
+ <expression noclean="1"></expression>
+ </RegExp>
+ </GetMtimeFanart>
+</scraper>
diff --git a/system/scrapers/video/mtv.png b/system/scrapers/video/mtv.png Binary files differnew file mode 100644 index 0000000000..dcb14290b7 --- /dev/null +++ b/system/scrapers/video/mtv.png diff --git a/system/scrapers/video/mtv.xml b/system/scrapers/video/mtv.xml new file mode 100644 index 0000000000..5d85b9de5c --- /dev/null +++ b/system/scrapers/video/mtv.xml @@ -0,0 +1,42 @@ +<?xml version="1.0" encoding="UTF-8"?> +<scraper framework="1.0" date="2009-05-22" name="MTV" content="musicvideos" thumb="mtv.png" language="en"> + <CreateSearchUrl dest="3"> + <RegExp input="$$1" output="<url>http://api.mtvnservices.com/1/video/search/?term=\1</url>" dest="3"> + <expression noclean="1"></expression> + </RegExp> + </CreateSearchUrl> + <GetSearchResults dest="8"> + <RegExp input="$$5" output="<?xml version="1.0" encoding="iso-8859-1" standalone="yes"?><results>\1</results>" dest="8"> + <RegExp input="$$1" output="<entity><title>\1 - \2</title><url>\4</url></entity>" dest="5"> + <expression noclean="1" repeat="yes"><entry>.*?<content>(.*?) \| (.*?) \| (.*?)</content>.*?media:player url="([^"]*)"/></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetSearchResults> + <GetDetails dest="3"> + <RegExp input="$$5" output="<details>\1</details>" dest="3"> + <RegExp input="$$1" output="<title>\1</title>" dest="5"> + <expression trim="1" noclean="1"><meta name="mtv_vt" content="([^"]*)"</expression> + </RegExp> + <RegExp input="$$1" output="<artist>\1</artist>" dest="5+"> + <expression trim="1" noclean="1"><meta name="mtv_an" content="([^"]*)"</expression> + </RegExp> + <RegExp input="$$1" output="<album>\1</album>" dest="5+"> + <expression trim="1" noclean="1"><meta name="mtv_at" content="([^"]*)"</expression> + </RegExp> + <RegExp input="$$1" output="<studio>\1</studio>" dest="5+"> + <expression trim="1" noclean="1"><strong>Label</strong> ([^<]*)</li></expression> + </RegExp> + <RegExp input="$$1" output="<year>\1</year>" dest="5+"> + <expression><meta name="content_create_date" content="([0-9]*)-[^"]*"</expression> + </RegExp> + <RegExp input="$$1" output="<director>\1</director>" dest="5+"> + <expression trim="1" noclean="1"><meta name="mtv_vd" content="([^"]*)"</expression> + </RegExp> + <RegExp input="$$1" output="<thumb>http://www.mtv.com\2.jpg</thumb>" dest="5+"> + <expression noclean="1">"thumbnail"([^"]*)content="([^"]*)\.jpg</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetDetails> +</scraper> diff --git a/system/scrapers/video/musicvideos.xml b/system/scrapers/video/musicvideos.xml new file mode 100644 index 0000000000..e778741760 --- /dev/null +++ b/system/scrapers/video/musicvideos.xml @@ -0,0 +1,37 @@ +п»ї<?xml version="1.0" encoding="UTF-8"?> +<scraper framework="1.0" date="2009-05-22" name="NFO Scraper" content="musicvideos" thumb="mvids.gif" language="en"> + <NfoScrape dest="3"> + <RegExp input="$$5" output="<musicvideo>\1</musicvideo>" dest="3"> + <RegExp input="$$1" output="<title>\1</title>" dest="5"> + <expression trim="1">[Tt]itle[ ]*:<? ([0-9a-zA-Z':\. &]+)[^0-9a-zA-Z':\. &]</expression> + </RegExp> + <RegExp input="$$1" output="<artist>\1</artist>" dest="5+"> + <expression trim="1">[Aa]rtist[ ]*:<? ([0-9a-zA-Z':\. &]+)[^0-9a-zA-Z':\. &]</expression> + </RegExp> + <RegExp input="$$1" output="<genre>\1</genre>" dest="5+"> + <expression trim="1">[Gg]enre[ ]*:<? ([0-9a-zA-Z':\. &]+)[^0-9a-zA-Z':\. &]</expression> + </RegExp> + <RegExp input="$$1" output="<year>\1</year>" dest="5+"> + <expression>[Yy]ear[ ]*:<? ([0-9]*)</expression> + </RegExp> + <RegExp input="$$1" output="<year>\1</year>" dest="5+"> + <expression>[Dd]ate[ ]*:<? [0-9][0-9][/\.][0-9][0-9][/\.]([0-9]*)</expression> + </RegExp> + <RegExp input="$$1" output="<runtime>\1</runtime>" dest="5+"> + <expression>[Rr]untime[ ]*:<? ([:0-9]+)</expression> + </RegExp> + <RegExp input="$$1" output="<runtime>\1</runtime>" dest="5+"> + <expression>[Ll]ength[ ]*:<? ([:0-9]+)</expression> + </RegExp> + <expression noclean="1">(.+)</expression> + </RegExp> + </NfoScrape> + <FileNameScrape dest="3"> + <RegExp input="$$1" output="<?xml version="1.0" encoding="UTF-8"?><musicvideo><artist>\1</artist><title>\2</title><year>\3</year></musicvideo>" dest="3"> + <expression>[\\/]([^\\/]+) - ([^\\/]+) - ([0-9]{4})[^\\/]*</expression> + </RegExp> + <RegExp input="$$1" output="<?xml version="1.0" encoding="UTF-8"?><musicvideo><artist>\1</artist><title>\2</title></musicvideo>" dest="3+"> + <expression>[\\/]([^\\/]+) - ([^\\/]+)[^\\/]*</expression> + </RegExp> + </FileNameScrape> +</scraper> diff --git a/system/scrapers/video/mymoviesdk.png b/system/scrapers/video/mymoviesdk.png Binary files differnew file mode 100644 index 0000000000..5771b4596d --- /dev/null +++ b/system/scrapers/video/mymoviesdk.png diff --git a/system/scrapers/video/mymoviesdk.xml b/system/scrapers/video/mymoviesdk.xml new file mode 100644 index 0000000000..d8b0ffda1a --- /dev/null +++ b/system/scrapers/video/mymoviesdk.xml @@ -0,0 +1,188 @@ +п»ї<?xml version="1.0" encoding="utf-8"?> +<scraper framework="1.1" date="2009-08-11" name="MyMovies.dk" content="movies" thumb="mymoviesdk.png" language="en" requiressettings="true"> + <GetSettings dest="3"> + <RegExp input="$$5" output="<settings>\1</settings>" dest="3"> + <!--Username--> + <RegExp input="$$1" output="<setting label="Username" type="text" id="username" default=""></setting>" dest="5+"> + <expression></expression> + </RegExp> + <!--Password--> + <RegExp input="$$1" output="<setting label="Password" type="text" option="hidden" id="password" default=""></setting>" dest="5+"> + <expression></expression> + </RegExp> + <!--- --> + <RegExp input="$$1" output="<setting type="sep"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <!--Fanart--> + <RegExp input="$$1" output="<setting label="Enable Fanart" type="bool" id="fanart" default="true"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <!--Trailer--> + <RegExp input="$$1" output="<setting label="Preferred trailer bitrate" type="labelenum" values="2048|1024|512|256|128" id="trailerbitrate" default="2048"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <!--Preferred language--> + <RegExp input="$$1" output="<setting label="Preferred language" type="labelenum" values="Arabic|Bulgarian|Cantonese|Mandarin|Czech|Croatian|Danish|Dutch|English|Estonian|Faroese|Farsi|Finnish|Flemish|French|German|Greek|Hebrew|Hindi|Hungarian|Icelandic|Indonesian|Italian|Japanese|Korean|Lithuanian|Norwegian|Polish|Portuguese|Romanian|Russian|Serbian|Slovakian|Slovene|Spanish|Swedish|Tagalog|Thai|Turkish|Malay|Latvian|Ukrainian|Tamil|Catalan|Telugu|Malayalam|Kannada|Bengali|Serbo-Croatian|Gujarati|Esperanto|Vietnamese|Kazakh|Ancient Greek|Mandar|Braj|Macedonian|Zulu|Twi|VolapГјk|Albanian|Khmer|Basque" id="language" default="English"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <!--Include adult movies--> + <RegExp input="$$1" output="<setting label="Include adult movies" type="bool" id="includeadult" default="false"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting type="sep"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetSettings> + + <CreateSearchUrl dest="3"> + <RegExp input="$$1" output="<url>https://api.mymovies.dk/Default.aspx?command=SearchMovieByTitle&amp;title=\1&amp;username=$INFO[username]&amp;password=$INFO[password]&amp;PrimaryLanguage=$INFO[language]&amp;includeadult=$INFO[includeadult]&amp;locale=1033</url>" dest="3"> + <RegExp input="$$2" output="%20(\1)" dest="4"> + <expression clear="yes">(.+)</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </CreateSearchUrl> + + <GetSearchResults dest="8"> + <RegExp input="$$3" output="<results sorted="Yes">\1</results>" dest="8"> + <RegExp input="$$1" output="<entity><title>\2</title><year>\3</year><id>\1</id><url>https://api.mymovies.dk/Default.aspx?command=LoadMovieById&amp;titleid=\1&amp;username=$INFO[username]&amp;password=$INFO[password]&amp;PrimaryLanguage=$INFO[language]&amp;maxtrailerbitrate=$INFO[trailerbitrate]&amp;locale=1033&amp;client=xbmcbased</url></entity>" dest="3"> + <expression repeat="yes">id="(.*?)".*? title="(.*?)"[^<]*year="(.*?)" thumb</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + <RegExp input="$$4" output="<error>\1</error>" dest="8"> + <RegExp input="$$1" output="<title>Error returned from MyMovies.dk</title><message>\1</message>" dest="4"> + <expression repeat="no"><response status="failed">(.*?)</response></expression> + </RegExp> + <expression noclean="1">(.+)</expression> + </RegExp> + </GetSearchResults> + + <NfoUrl dest="3"> + <RegExp input="$$1" output="<url>https://api.mymovies.dk/Default.aspx?command=LoadMovieByIMDBId&imdbid=tt\2&username=$INFO[username]&password=$INFO[password]&PrimaryLanguage=$INFO[language]&includeadult=$INFO[includeadult]&maxtrailerbitrate=$INFO[trailerbitrate]&locale=1033</url>" dest="3"> + <expression>(imdb.com/title/tt)([0-9]*)</expression> + </RegExp> + <RegExp input="$$1" output="<url>https://api.mymovies.dk/Default.aspx?command=LoadMovieByIMDBId&imdbid=tt\2&username=$INFO[username]&password=$INFO[password]&PrimaryLanguage=$INFO[language]&includeadult=$INFO[includeadult]&maxtrailerbitrate=$INFO[trailerbitrate]&locale=1033</url>" dest="3+"> + <expression>(imdb.com/)Title\?([0-9]+)</expression> + </RegExp> + </NfoUrl> + + <GetDetails dest="3"> + <RegExp input="$$5" output="<details>\1</details>" dest="3"> + <!--Title--> + <RegExp input="$$1" output="<title>\1</title>" dest="5"> + <expression><LocalTitle>([^<]*)</LocalTitle></expression> + </RegExp> + <!--SortTitle--> + <RegExp input="$$1" output="<sorttitle>\1</sorttitle>" dest="5+"> + <expression><SortTitle>([^<]*)</SortTitle></expression> + </RegExp> + <!--Year--> + <RegExp input="$$1" output="<year>\1</year>" dest="5+"> + <expression><ProductionYear>([0-9]+)</ProductionYear></expression> + </RegExp> + <!--Thumb--> + <RegExp input="$$1" output="<thumb preview="\1">\2</thumb>" dest="5+"> + <expression repeat="yes"><Poster FileThumb="([^"]*)" File="([^"]*)" Hash=</expression> + </RegExp> + <!--IMDB--> + <RegExp input="$$1" output="<id>\1</id>" dest="5+"> + <expression><IMDB>([^<]*)</IMDB></expression> + </RegExp> + <!--Runtime--> + <RegExp input="$$1" output="<runtime>\1 min</runtime>" dest="5+"> + <expression><RunningTime>([^<]*)</RunningTime></expression> + </RegExp> + <!--Rating--> + <RegExp input="$$1" output="<rating>\1</rating>" dest="5+"> + <expression><Rating>([^<]*)</Rating></expression> + </RegExp> + <!--Votes--> + <RegExp input="$$1" output="<votes>\1</votes>" dest="5+"> + <expression><RatingVotes>([^<]*)</RatingVotes></expression> + </RegExp> + <!--Studio--> + <RegExp input="$$1" output="<studio>\1</studio>" dest="5+"> + <expression repeat="yes"><Studio>([^<]*)</Studio></expression> + </RegExp> + <!--Genre--> + <RegExp input="$$1" output="<genre>\1</genre>" dest="5+"> + <expression repeat="yes"><Genre>([^<]*)</Genre></expression> + </RegExp> + <!--Plot--> + <RegExp input="$$1" output="<plot>\1</plot>" dest="5+"> + <expression><Description>[^<]*<!\[CDATA\[([^<]*)\]\][^<]*</Description></expression> + </RegExp> + <!--Plot Outline--> + <RegExp input="$$1" output="<outline>\1</outline>" dest="5+"> + <expression><Overview>([^<]*)</Overview></expression> + </RegExp> + <!--MPAA--> + <RegExp input="$$4" output="<mpaa>\1</mpaa>" dest="5+"> + <RegExp input="$$1" output="None" dest="4"> + <expression><ParentalRating[^.]*?<Value>-1</Value>[^.]*?<Description></expression> + </RegExp> + <RegExp input="$$1" output="Unrated" dest="4"> + <expression><ParentalRating[^.]*?<Value>0</Value>[^.]*?<Description></expression> + </RegExp> + <RegExp input="$$1" output="G" dest="4"> + <expression><ParentalRating[^.]*?<Value>(1|2)</Value>[^.]*?<Description></expression> + </RegExp> + <RegExp input="$$1" output="PG" dest="4"> + <expression><ParentalRating[^.]*?<Value>3</Value>[^.]*?<Description></expression> + </RegExp> + <RegExp input="$$1" output="PG-13" dest="4"> + <expression><ParentalRating[^.]*?<Value>(4|5)</Value>[^.]*?<Description></expression> + </RegExp> + <RegExp input="$$1" output="R" dest="4"> + <expression><ParentalRating[^.]*?<Value>6</Value>[^.]*?<Description></expression> + </RegExp> + <RegExp input="$$1" output="NC-17" dest="4"> + <expression><ParentalRating[^.]*?<Value>7</Value>[^.]*?<Description></expression> + </RegExp> + <RegExp input="$$1" output="NC-17" dest="4"> + <expression><ParentalRating[^.]*?<Value>8</Value>[^.]*?<Description></expression> + </RegExp> + <expression noclean="1">(.+)</expression> + </RegExp> + <!--Tagline--> + <RegExp input="$$1" output="<tagline>\1</tagline>" dest="5+"> + <expression><TagLine>([^<]*)</TagLine></expression> + </RegExp> + <!--Director--> + <RegExp input="$$1" output="<director>\1</director>" dest="5+"> + <expression repeat="yes"><Person Type="2"[^$]*?<Name>([^<]*)</Name></expression> + </RegExp> + <!--Writer--> + <RegExp input="$$1" output="<credits>\1</credits>" dest="5+"> + <expression repeat="yes"><Person Type="3"[^$]*?<Name>([^<]*)</Name></expression> + </RegExp> + <!--Actor--> + <RegExp input="$$1" output="<actor><name>\1</name><role>\2</role></actor>" dest="5+"> + <expression repeat="yes"><Person Type="1"[^.]*?<Photo />[^.]*?<Name>([^<]*)</Name>[^.]*?<Role>([^<]*)</Role></expression> + </RegExp> + <!--Actor with thumb--> + <RegExp input="$$1" output="<actor><thumb>\1</thumb><name>\2</name><role>\3</role></actor>" dest="5+"> + <expression repeat="yes"><Person Type="1"[^.]*?<Photo>([^<]*)</Photo>[^.]*?<Name>([^<]*)</Name>[^.]*?<Role>([^<]*)</Role></expression> + </RegExp> + <!--Fanart--> + <RegExp conditional="fanart" input="$$4" output="<fanart>\1</fanart>" dest="5+"> + <RegExp input="$$1" output="<thumb preview="\1" dim="1920x1080">\2</thumb>" dest="4"> + <expression repeat="yes">FileThumb="([^"]*)" F[^<]*File1080P="([^"]*)" Hash1080p="[^"]+"</expression> + </RegExp> + <RegExp input="$$1" output="<thumb preview="\1" dim="1280x720">\2</thumb>" dest="4+"> + <expression repeat="yes">FileThumb="([^"]*)" File720P="([^"]*)" Hash720p[^<]*File1080P=""</expression> + </RegExp> + <expression noclean="1">(.+)</expression> + </RegExp> + <!--Trailer - filmtrailer.com--> + <RegExp input="$$1" output="<trailer>\1</trailer>" dest="5+"> + <expression><Trailer[^>]*>([^<]*)</Trailer></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetDetails> +</scraper> diff --git a/system/scrapers/video/naver.png b/system/scrapers/video/naver.png Binary files differnew file mode 100644 index 0000000000..1dd4655c2b --- /dev/null +++ b/system/scrapers/video/naver.png diff --git a/system/scrapers/video/naver.xml b/system/scrapers/video/naver.xml new file mode 100644 index 0000000000..f72b5c41b8 --- /dev/null +++ b/system/scrapers/video/naver.xml @@ -0,0 +1,180 @@ +<?xml version="1.0" encoding="UTF-8"?> +<scraper framework="1.1" date="2009-08-08" name="NAVER (Korean)" content="movies" thumb="naver.png" language="ko"> + <include>common/tmdb.xml</include> + <GetSettings dest="3"> + <RegExp input="$$5" output="<settings>\1</settings>" dest="3"> + <RegExp input="$$1" output="<setting label="БЦї¬№иїм/°ўє»/Б¦АЫ»зїЎ ґлЗС »ујјБ¤єё" type="bool" id="EnableDetail" default="true"></setting>" dest="5"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting label="ЖчЅєЕН" type="bool" id="EnableThumbs" default="true"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting label="ЖтБЎ ЕхЗҐАЪјц" type="bool" id="EnableVoteNum" default="true"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting label="ЖТѕЖЖ®(°нИБъ)" type="bool" id="EnableFanArt" default="true"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetSettings> + <NfoUrl dest="3"> + <RegExp input="$$1" output="<url>http://movie.naver.com/movie/bi/mi/basic.nhn?code=\1</url>" dest="3"> + <expression>movie\.naver\.com[^\?]*\?code=([0-9]*)</expression> + </RegExp> + </NfoUrl> + <CreateSearchUrl dest="3" SearchStringEncoding="EUC-KR"> + <RegExp input="$$1" output="http://movie.naver.com/movie/search/result.nhn?section=movie&query=\1&sort=3" dest="3"> + <expression noclean="1"/> + </RegExp> + </CreateSearchUrl> + <GetSearchResults dest="8"> + <RegExp input="$$5" output="<?xml version="1.0" encoding="EUC-KR" standalone="yes"?><results>\1</results>" dest="8"> + <RegExp input="$$1" output="<entity><title>\2 (\3) </title><url>http://movie.naver.com/movie/bi/mi/basic.nhn?code=\1</url></entity>" dest="5"> + <expression repeat="yes"><dt><a href="/movie/bi/mi/basic.nhn\?code=([0-9]*)">(.[^"]*)"point">[^\?]*\?year=([0-9]*)"</expression> + </RegExp> + <expression noclean="1"/> + </RegExp> + </GetSearchResults> + <GetDetails dest="3"> + <RegExp input="$$8" output="<details>\1</details>" dest="3"> + <!-- Titles, year !--> + <RegExp input="$$1" output="<title>\1 (\2, \3)</title><originaltitle>\2</originaltitle><year>\3</year>" dest="8"> + <expression trim="1" noclean="1">"&main_title="\+encodeURIComponent\("(.[^"]*)"\)\+"&sub_title="\+encodeURIComponent\("(.[^\,]*)\, ([0-9]*)"\)</expression> + </RegExp> + <!-- Director's names !--> + <RegExp input="$$7" output="<director>\1</director>" dest="8+"> + <RegExp input="$$1" output="\1" dest="7"> + <expression noclean="1">db_htype_1_text_3\.gif"(.*)db_htype_1_text_4\.gif"</expression> + </RegExp> + <expression noclean="1" repeat="yes"><a[^>]*>(.[^<]*)</a></expression> + </RegExp> + <!-- Actors !--> + <RegExp conditional="!EnableDetail" input="$$7" output="<actor><name>\1</name><role>\2</role></actor>" dest="8+"> + <RegExp input="$$1" output="\1" dest="7"> + <expression noclean="1">db_htype_1_text_4\.gif"(.*)"more_inline"</expression> + </RegExp> + <expression noclean="1" repeat="yes"><a[^>]*>(.[^<]*)</a>\((.[^\)]*)\)</expression> + </RegExp> + <!-- MPAA !--> + <RegExp input="$$1" output="<mpaa>\1</mpaa>" dest="8+"> + <expression noclean="1" repeat="yes"><em>[^<]*</em> <a[^>]*>(.[^<]*)<</expression> + </RegExp> + <!-- Plot !--> + <RegExp input="$$1" output="<plot>\1\2</plot>" dest="8+"> + <expression><p class="c_ff4200">(.[^<]*)</p>[^<]*<p>(.*)</p>[^<]*<ul class="other"</expression> + </RegExp> + <RegExp input="$$1" output="<plot>\1</plot>" dest="8+"> + <expression><p class="c_ff4200"></p>[^<]*<p>(.*)</p>[^<]*<ul class="other"</expression> + </RegExp> + <!-- Thumbnail !--> + <RegExp conditional="!EnableThumbs" input="$$1" output="<thumb><url>\1</url></thumb>" dest="8+"> + <expression><div class="poster">[^\(]*\('([^']*)'</expression> + </RegExp> + <!-- Premiered !--> + <RegExp input="$$1" output="<premiered>\1.\2</premiered>" dest="8+"> + <expression noclean="1">>[^<]*</font> <a[^>]*>([0-9]*)</a>\.<[^>]*>(.[^<]*)<</expression> + </RegExp> + <!-- Genres !--> + <RegExp input="$$7" output="<genre>\1</genre>" dest="8+"> + <RegExp input="$$1" output="\1" dest="7"> + <expression noclean="1">db_htype_1_text_2\.gif"(.*)nhn\?nation</expression> + </RegExp> + <expression noclean="1" repeat="yes">nhn\?genre[^>]*>(.[^<]*)<</expression> + </RegExp> + <!-- Runtime !--> + <RegExp input="$$1" output="<runtime>\1 minute</runtime>" dest="8+"> + <expression noclean="1">class='ls0'>([0-9]*)<</expression> + </RegExp> + <!-- User rating !--> + <RegExp input="$$1" output="<rating>\1</rating>" dest="8+"> + <expression>"&star_point=([0-9]\.+[0-9]*)"\;</expression> + </RegExp> + <!-- Tag Line !--> + <RegExp input="$$1" output="<tagline>\1</tagline>" dest="8+"> + <expression noclean="1"><p class="c_ff4200">(.[^<]*)</p></expression> + </RegExp> + <!-- Detail !--> + <RegExp conditional="EnableDetail" input="$$1" output="<url function="GetNAVERDetail">http://movie.naver.com/movie/bi/mi/detail.nhn?code=\1</url>" dest="8+"> + <expression noclean="1"><a href="/movie/bi/mi/photo.nhn\?code=([0-9]*)"</expression> + </RegExp> + <!-- Thumbs !--> + <RegExp conditional="EnableThumbs" input="$$1" output="<url function="GetNAVERPhoto">http://movie.naver.com/movie/bi/mi/photo.nhn?code=\1</url>" dest="8+"> + <expression noclean="1"><a href="/movie/bi/mi/photo.nhn\?code=([0-9]*)"</expression> + </RegExp> + <!-- Votes !--> + <RegExp conditional="EnableVoteNum" input="$$1" output="<url function="GetNAVERVote">http://movie.naver.com/movie/bi/mi/point.nhn?code=\1</url>" dest="8+"> + <expression noclean="1"><a href="/movie/bi/mi/photo.nhn\?code=([0-9]*)"</expression> + </RegExp> + <!-- FanArts !--> + <RegExp conditional="EnableFanArt" input="$$5" output="<url function="GetTMDBFanartByIMDBId">http://api.themoviedb.org/2.0/Movie.search?title=\1&amp;api_key=57983e31fb435df4df77afb854740ea9</url>" dest="8+"> + <RegExp input="$$1" output="\1" dest="7"> + <expression noclean="1" clear="yes">"&sub_title="\+encodeURIComponent\("([^\,]*)\, ([0-9]*)"\)</expression> + </RegExp> + <RegExp input="$$7" output="+\1" dest="5"> + <expression repeat="yes">([^ ,]+)</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + <!-- FanArts !--> + <RegExp conditional="EnableFanArt" input="$$1" output="<url function="GetNAVERFanart">http://movie.naver.com/movie/bi/mi/photo.nhn?code=\1</url>" dest="8+"> + <expression noclean="1"><a href="/movie/bi/mi/photo.nhn\?code=([0-9]*)"</expression> + </RegExp> + <expression noclean="1"/> + </RegExp> + </GetDetails> + <GetNAVERPhoto dest="8"> + <RegExp input="$$9" output="<details>\1</details>" dest="8"> + <RegExp input="$$1" output="\1" dest="7"> + <expression noclean="1">box_story_1 mt_10">(.*)btn_dbreport\.gif</expression> + </RegExp> + <RegExp input="$$7" output="<thumb preview="http://imgmovie.naver.com/mdi/mit120/\1">http://imgmovie.naver.com/mdi/mit500/\1</thumb>" dest="9"> + <expression noclean="1" repeat="yes">"><img src="http://imgmovie.naver.com/mdi/mit120/(.[^"]*)"</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetNAVERPhoto> + <GetNAVERVote dest="8"> + <RegExp input="$$9" output="<details>\1</details>" dest="8"> + <RegExp input="$$1" output="<votes>\1</votes>" dest="9"> + <expression noclean="1" trim="1"><p class="join">[^>]*>(.[^<]*)<</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetNAVERVote> + <GetNAVERDetail dest="8"> + <RegExp input="$$9" output="<details>;\1</details>" dest="8"> + <!-- studio !--> + <RegExp input="$$1" output="<studio>\1</studio>" dest="9"> + <expression noclean="1" repeat="yes"><a href="/movie/bi/ci/filmo\.nhn\?code=[0-9]*">(.[^<]*)<</expression> + </RegExp> + <!-- Actors !--> + <RegExp input="$$1" output="\1" dest="7"> + <expression noclean="1">id="cast_1"(.*)id="cast_2"</expression> + </RegExp> + <RegExp input="$$7" output="<actor><name>\2</name><role>\3</role><thumb>\1</thumb></actor>" dest="9+"> + <expression noclean="1" repeat="yes">class="thumb"><a[^<]*<img src="(.[^"]*)"[^<]*</a></td>[^>]*><a[^>]*>(.[^<]*)</a> <[^<]*</span><br>(.[^<]*)</th></expression> + </RegExp> + <!-- writer !--> + <RegExp input="$$1" output="\1" dest="6"> + <expression noclean="1"><h6>°ўє»</h6>(.*)id="staff_2"</expression> + </RegExp> + <RegExp input="$$6" output="<credits>\1</credits>" dest="9+"> + <expression noclean="1" repeat="yes"><a href="/movie/bi/pi/basic.nhn\?code=[0-9]*">(.[^<]*)</a></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetNAVERDetail> + <GetNAVERFanart clearbuffers="no" dest="8"> + <RegExp input="$$10$$11" output="<details><fanart>\1</fanart></details>" dest="8"> + <RegExp input="$$1" output="\1" dest="7"> + <expression noclean="1">box_story_1 mt_10">(.*)btn_dbreport\.gif</expression> + </RegExp> + <RegExp input="$$7" output="<thumb preview="http://imgmovie.naver.com/mdi/mit120/\1">http://imgmovie.naver.com/mdi/mi/\1</thumb>" dest="11"> + <expression noclean="1" repeat="yes" clear="yes">"><img src="http://imgmovie.naver.com/mdi/mit120/(.[^"]*)"</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetNAVERFanart> +</scraper> + diff --git a/system/scrapers/video/ofdb.png b/system/scrapers/video/ofdb.png Binary files differnew file mode 100644 index 0000000000..9b7ecbad4c --- /dev/null +++ b/system/scrapers/video/ofdb.png diff --git a/system/scrapers/video/ofdb.xml b/system/scrapers/video/ofdb.xml new file mode 100644 index 0000000000..96da60f805 --- /dev/null +++ b/system/scrapers/video/ofdb.xml @@ -0,0 +1,261 @@ +<?xml version="1.0" encoding="utf-8" standalone="yes"?> +<scraper framework="1.1" date="2009-08-10" name="OFDb" content="movies" thumb="ofdb.png" language="de"> + <include>common/tmdb.xml</include> + <include>common/movieposterdb.xml</include> + <GetSettings dest="3"> + <RegExp input="$$5" output="<settings>\1</settings>" dest="3"> + <RegExp input="$$1" output="<setting label="Fanart" type="bool" id="fanart" default="true"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting label="MoviePosterDB" type="bool" id="movieposterdb" default="false"></setting>" dest="5+"> + <expression /> + </RegExp> + <RegExp input="$$1" output="<setting label="Trailer" type="bool" id="trailer" default="true"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting label="Groesse der Darsteller Bilder" type="labelenum" values="192|256|384|512|1024" id="imdbscale" default="512"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetSettings> + + <NfoUrl dest="3"> + <RegExp input="$$1" output="<url>\1\2,\3</url>" dest="3"> + <expression clear="yes" noclean="1">(http://www.ofdb.de/film/)([0-9]*),(.*)</expression> + </RegExp> + <RegExp input="$$1" output="<url function="GetByIMDBId">http://www.ofdb.de/view.php?SText=\2&amp;Kat=IMDb&amp;page=suchergebnis&amp;sourceid=mozilla-search</url>" dest="3+"> + <expression>(imdb.com/)Title\?([0-9]+)</expression> + </RegExp> + <RegExp input="$$1" output="<url function="GetByIMDBId">http://www.ofdb.de/view.php?SText=\2&amp;Kat=IMDb&amp;page=suchergebnis&amp;sourceid=mozilla-search</url>" dest="3+"> + <expression>(imdb.com/title/tt)([0-9]*)</expression> + </RegExp> + </NfoUrl> + + <GetByIMDBId dest="3"> + <RegExp input="$$1" output="<url>http://www.ofdb.de/film/\1,</url><id>\1</id>" dest="3+"> + <expression repeat="yes" noclean="1">"film/([0-9]*),</expression> + </RegExp> + </GetByIMDBId> + + + <CreateSearchUrl dest="3"> + <RegExp input="$$1" output="http://www.ofdb.de/view.php?SText=\1&Kat=Titel&page=suchergebnis&sourceid=mozilla-search" dest="3"> + <expression noclean="1">([^\(]+)</expression> + </RegExp> + </CreateSearchUrl> + + <GetSearchResults dest="8"> + <RegExp input="$$5" output="<?xml version="1.0" encoding="utf-8" standalone="yes"?><results>\1</results>" dest="8"> + <RegExp input="$$1" output="<entity><title>\2 (\4) | \3</title><year>\4</year><url cache="\1.html">http://www.ofdb.de/film/\1,</url><id>\1</id></entity>" dest="5+"> + <expression repeat="yes" noclean="1"><a href=.[^0-9]+([0-9]+)[^ ]+ onmouseover=.Tip..<[^>]+>[^>]+>([^<]+)<font size=.1.> . ([^<]+)</font> \(([0-9]+)\)</a></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetSearchResults> + + <GetDetails dest="3"> + <RegExp input="$$5" output="<?xml version="1.0" encoding="utf-8" standalone="yes"?><details>\1</details>" dest="3"> + <!--Title --> + <RegExp input="$$6" output="\1" dest="5+"> + <RegExp input="$$1" output="<title>\1</title>" dest="6"> + <expression trim="1"><td width="99."><h2><font face="Arial,Helvetica,sans-serif" size="3"><b>([^<]+)</b></font></h2></td></expression> + </RegExp> + <RegExp input="$$7" output="<title>\2 \1</title>" dest="6"> + <RegExp input="$$1" output="\1" dest="7"> + <expression trim="1"><td width="99."><h2><font face="Arial,Helvetica,sans-serif" size="3"><b>([^<]+)</b></font></h2></td></expression> + </RegExp> + <expression>(.*), (Die|Der|Das|The)</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + + <!--Original Title --> + <RegExp input="$$1" output="\1" dest="8"> + <expression noclean="1">class="Normal">Originaltitel(.*)class="Normal">Herstellungsland</expression> + </RegExp> + + <RegExp input="$$7" output="\1" dest="5+"> + <RegExp input="$$8" output="<originaltitle>\1</originaltitle>" dest="7"> + <expression><b>([^>]+)</b></expression> + </RegExp> + <RegExp input="$$6" output="<originaltitle>\2 \1</originaltitle>" dest="7"> + <RegExp input="$$8" output="\1" dest="6"> + <expression noclean="1"><b>([^>]+)</b></expression> + </RegExp> + <expression>(.*), (Die|Der|Das|The)</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + + <!--Year --> + <RegExp input="$$1" output="<year>\1</year>" dest="5+"> + <expression>view.php\?page=blaettern&Kat=Jahr&Text=([0-9]+)</expression> + </RegExp> + + <!--Director--> + <RegExp input="$$9" output="<director>\1</director>" dest="5+"> + <RegExp input="$$1" output="\1" dest="9"> + <expression noclean="1">class="Normal">Regie(.*)class="Normal">Darsteller</expression> + </RegExp> + <expression repeat="yes">Name=[^>]*>([^<]+)</expression> + </RegExp> + + <!--Plot outline --> + <RegExp input="$$1" output="<outline>\1</outline><plot>\1</plot>" dest="5+"> + <expression><b>Inhalt:</b>([^<]+)</expression> + </RegExp> + + <!--Genre--> + <RegExp input="$$1" output="<genre>\1</genre>" dest="5+"> + <expression repeat="yes">view.php\?page=genre&Genre=[^"]+">([^<]*)<</expression> + </RegExp> + + <!--Rating--> + <RegExp input="$$1" output="<rating>\1</rating>" dest="5+"> + <expression>Note: ([0-9\.]+)</expression> + </RegExp> + + <!--Votes--> + <RegExp input="$$1" output="<votes>\1</votes>" dest="5+"> + <expression>Stimmen: ([0-9]+)</expression> + </RegExp> + + <RegExp input="$$1" output="<mpaa>\1</mpaa>" dest="5+"> + <expression>Freigabe: ([^"]*)"</expression> + </RegExp> + + <RegExp input="$$1" output="<id>\2</id>" dest="5+"> + <expression>(imdb.com/)Title\?([0-9]+)</expression> + </RegExp> + + <!--URL to Plot --> + <RegExp input="$$1" output="<url function="Inhaltsangabe">http://www.ofdb.de/plot/\1,\2,</url>" dest="5+"> + <expression>Inhalt:</b>[^<]*<a href="plot/([0-9]+),([0-9]+),</expression> + </RegExp> + + <!--URL to MPAA / Studio / Runtime--> + <RegExp input="$$1" output="<url function="OFDbInfo">http://www.ofdb.de/view.php?page=fassung&amp;fid=\1&amp;vid=\2</url>" dest="5+"> + <expression><a href=.view.php.page=fassung&fid=([0-9]+)&vid=([0-9]+).>Kino:</expression> + </RegExp> + + <!--URL to IMDB--> + <RegExp input="$$1" output="<url function="IMDB-Details">http://www.imdb.de/title/tt\2/</url>" dest="5+"> + <expression>(imdb.com/)Title\?([0-9]+)</expression> + </RegExp> + + <!--URL to IMDBcredits--> + <RegExp input="$$1" output="<url function="GetMovieCast">http://www.imdb.de/title/tt\2/fullcredits#cast</url>" dest="5+"> + <expression>(imdb.com/)Title\?([0-9]+)</expression> + </RegExp> + + <!--URL to Fanart--> + <RegExp conditional="fanart" input="$$1" output="<url function="GetTMDBFanartByIMDBId">http://api.themoviedb.org/2.0/Movie.imdbLookup?imdb_id=tt\2&amp;api_key=57983e31fb435df4df77afb854740ea9</url>" dest="5+"> + <expression>(imdb.com/)Title\?([0-9]+)</expression> + </RegExp> + + <!--URL to MoviePosterDB --> + <RegExp conditional="movieposterdb" input="$$1" output="<url function="GetMoviePosterDBLink">http://www.movieposterdb.com/browse/search?type=movies&amp;query=\2</url>" dest="5+"> + <expression>(imdb.com/)Title\?([0-9]+)</expression> + </RegExp> + + <!-- Ofdb Poster --> + <RegExp input="$$3" output="<url function="GetOfdbThumb" cache="$$2.html">\1</url>" dest="5+"> + <expression noclean="1"></expression> + </RegExp> + + <!-- Cast --> + <RegExp input="$$2" output="<url cache="$$2-credits.html" function="GetMovieCast">$$3</url>" dest="5+"> + <expression></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetDetails> + + <!--Plot--> + <Inhaltsangabe dest="5"> + <RegExp input="$$7" output="<?xml version="1.0" encoding="utf-8" standalone="yes"?><details><plot>\1</plot></details>" dest="5+"> + <RegExp input="$$1" output="\1" dest="9"> + <expression noclean="1">Eine Inhaltsangabe von(.*)<option value="All"></expression> + </RegExp> + <RegExp input="$$9" output="\1" dest="7"> + <expression noclean="1"><br><br>([^<]+)</expression> + </RegExp> + <RegExp input="$$9" output=" \1" dest="7+"> + <expression noclean="1" repeat="yes"><br />([^<]+)</expression> + </RegExp> + <expression></expression> + </RegExp> + </Inhaltsangabe> + + <!--Studio--> + <OFDbInfo dest="5"> + <RegExp input="$$4" output="<?xml version="1.0" encoding="utf-8" standalone="yes"?><details>\1</details>" dest="5"> + <RegExp input="$$1" output="<studio>\1</studio>" dest="4+"> + <expression>Verleih:</font></td><td></td><td><font[^>]+><b>([^<]+)</b></expression> + </RegExp> + <RegExp input="$$1" output="<runtime>\1</runtime>" dest="4+"> + <expression>Laufzeit: ([^<]*)<</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </OFDbInfo> + + <!--IMDB-Details--> + <IMDB-Details dest="5"> + <RegExp input="$$3" output="<?xml version="1.0" encoding="utf-8" standalone="yes"?><details>\1</details>" dest="5"> + <RegExp input="$$1" output="<tagline>\1</tagline>" dest="3+"> + <expression>Werbezeile:</h5>([^<]*)<</expression> + </RegExp> + <RegExp conditional="trailer" input="$$1" output="<url function="GetTrailer">http://akas.imdb.com/video/imdb/vi\1/player</url>" dest="3+"> + <expression>/vi([0-9]*)/"</expression> + </RegExp> + <RegExp input="$$1" output="\1" dest="6"> + <expression noclean="1">Drehbuchautor(.*)Premierendatum</expression> + </RegExp> + <RegExp input="$$6" output="<credits>\1</credits>" dest="3+"> + <expression><a href="/name/nm[^>]*>([^<]*)<</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </IMDB-Details> + + <!-- MovieCast --> + <GetMovieCast dest="5"> + <RegExp input="$$2" output="<details>\1</details>" dest="5"> + <RegExp input="$$1" output="\1" dest="6"> + <expression noclean="1"><table class="cast">(.*?)</table></expression> + </RegExp> + <RegExp input="$$6" output="<actor><thumb>\1_SX$INFO[imdbscale]_SY$INFO[imdbscale]_\2</thumb><name>\3</name><role>\4</role></actor>" dest="7"> + <expression repeat="yes" noclean="1,2" trim="3,4"><img src="(?:([^"]*\.)[^"]*(\.jpg))?[^>]*[^"]*"nm"><a href="[^"]*[^>]*>([^<]*)<[^"]*"ddd"> ... [^"]*"char">(.*?)</td></expression> + </RegExp> + <RegExp input="$$7" output="<actor><thumb>\1</thumb>\2</actor>" dest="2+"> + <expression repeat="yes" noclean="1,2,3"><actor><thumb>(?:(http.*?)|_SX[0-9]+_SY[0-9]+_)</thumb>(.*?)</actor></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetMovieCast> + + <!-- Trailer --> + <GetTrailer dest="5"> + <RegExp conditional="trailer" input="$$2" output="<details><trailer urlencoded="yes">\1</trailer></details>" dest="5"> + <RegExp input="$$1" output="\1" dest="2"> + <expression noclean="1">"file"[^"]*"([^"]*)</expression> + </RegExp> + <RegExp input="$$1" output="$$2/\1" dest="2"> + <expression noclean="1">"id"[^"]*"([^"]*)</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetTrailer> + + <!-- Ofdb Poster --> + <GetOfdbThumb dest="5"> + <RegExp input="$$6" output="<details>\1</details>" dest="5"> + <RegExp input="$$1" output="<thumb>http://img.ofdb.de/film/\1</thumb>" dest="6+"> + <expression><img src="http://img.ofdb.de/film/([^"]+)" [^>]*></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetOfdbThumb> +</scraper> diff --git a/system/scrapers/video/ofdbxml.xml b/system/scrapers/video/ofdbxml.xml new file mode 100644 index 0000000000..313b28a0b1 --- /dev/null +++ b/system/scrapers/video/ofdbxml.xml @@ -0,0 +1,129 @@ +<?xml version="1.0" encoding="utf-8" standalone="yes"?> +<scraper framework="1.1" date="2009-07-28" name="OFDb (XML)" content="movies" thumb="ofdb.png" language="de"> + <include>common/tmdb.xml</include> + <GetSettings dest="3"> + <RegExp input="$$5" output="<settings>\1</settings>" dest="3"> + <RegExp input="$$1" output="<setting label="Fanart" type="bool" id="fanart" default="true"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetSettings> + + <NfoUrl dest="3"> + <RegExp input="$$1" output="<url>http://xml.n4rf.net/ofdbgw/movie/\2</url>" dest="3"> + <expression clear="yes" noclean="1">(http://www.ofdb.de/film/)([0-9]*)</expression> + </RegExp> + <RegExp input="$$1" output="<url function="GetByIMDBId">http://xml.n4rf.net/ofdbgw/imdb2ofdb/tt\2</url>" dest="3+"> + <expression>(imdb.com/)Title\?([0-9]+)</expression> + </RegExp> + <RegExp input="$$1" output="<url function="GetByIMDBId">http://xml.n4rf.net/ofdbgw/imdb2ofdb/tt\2</url>" dest="3+"> + <expression>(imdb.com/title/tt)([0-9]*)</expression> + </RegExp> + </NfoUrl> + + <GetByIMDBId dest="3"> + <RegExp input="$$1" output="<url>http://xml.n4rf.net/ofdbgw/movie/\1</url>" dest="3+"> + <expression repeat="yes"><ofdbid>([^<]+)</expression> + </RegExp> + </GetByIMDBId> + + <CreateSearchUrl dest="3"> + <RegExp input="$$1" output="http://xml.n4rf.net/ofdbgw/search/\1" dest="3"> + <expression noclean="1"/> + </RegExp> + </CreateSearchUrl> + + <GetSearchResults dest="8"> + <RegExp input="$$5" output="<?xml version="1.0" encoding="utf-8" standalone="yes"?><results>\1</results>" dest="8"> + <RegExp input="$$1" output="<entity><title>\2</title><year>\3</year><url>http://xml.n4rf.net/ofdbgw/movie/\1</url><id>\1</id></entity>" dest="5+"> + <expression repeat="yes"><id>(.*?)</id>\n<titel>([^<]+)</titel>.*?<jahr>(.*?)</jahr></expression> + </RegExp> + <expression noclean="1"/> + </RegExp> + </GetSearchResults> + + <GetDetails dest="3"> + <RegExp input="$$5" output="<?xml version="1.0" encoding="utf-8" standalone="yes"?><details>\1</details>" dest="3"> + <RegExp input="$$6" output="\1" dest="5+"> + <RegExp input="$$1" output="<title>\1</title>" dest="6"> + <expression trim="1"><titel>([^<]+)</titel></expression> + </RegExp> + <RegExp input="$$7" output="<title>\2 \1</title>" dest="6"> + <RegExp input="$$1" output="\1" dest="7"> + <expression trim="1"><titel>([^<]+)</titel></expression> + </RegExp> + <expression>(.*), (Die|Der|Das|The)</expression> + </RegExp> + <expression noclean="1"/> + </RegExp> + <RegExp input="$$1" output="<originaltitle>\1</originaltitle>" dest="5+"> + <expression noclean="1"><alternativ>([^<]+)</expression> + </RegExp> + <RegExp input="$$1" output="<year>\1</year>" dest="5+"> + <expression noclean="1"><jahr>([^<]+)</expression> + </RegExp> + <RegExp input="$$9" output="<director>\1</director>" dest="5+"> + <RegExp input="$$1" output="\1" dest="9"> + <expression noclean="1"><regie>(.*?)</regie></expression> + </RegExp> + <expression repeat="yes"><name>([^<]+)</expression> + </RegExp> + <RegExp input="$$9" output="<genre>\1</genre>" dest="5+"> + <RegExp input="$$1" output="\1" dest="9"> + <expression noclean="1"><genre>(.*?)</genre></expression> + </RegExp> + <expression repeat="yes"><titel>([^<]+)</expression> + </RegExp> + <RegExp input="$$9" output="<credits>\1</credits>" dest="5+"> + <RegExp input="$$1" output="\1" dest="9"> + <expression noclean="1"><drehbuch>(.*?)</drehbuch></expression> + </RegExp> + <expression repeat="yes"><name>([^<]+)</expression> + </RegExp> + <RegExp input="$$9" output="<actor><name>\2</name><role>\3</role></actor>" dest="5+"> + <RegExp input="$$1" output="\1" dest="9"> + <expression noclean="1"><besetzung>(.*?)</besetzung></expression> + </RegExp> + <expression repeat="yes"><id>([^<]+)</id>\n<name>([^<]+)</name>\n<rolle>([^<]+)</expression> + </RegExp> + <RegExp input="$$1" output="<outline>\1</outline>" dest="5+"> + <expression><kurzbeschreibung>([^<]+)</expression> + </RegExp> + <RegExp input="$$1" output="<plot>\1</plot>" dest="5+"> + <expression><beschreibung>([^<]+)</expression> + </RegExp> + <RegExp input="$$1" output="<rating>\1</rating>" dest="5+"> + <expression><note>([^<]+)</expression> + </RegExp> + <RegExp input="$$1" output="<votes>\1</votes>" dest="5+"> + <expression><stimmen>([^<]+)</expression> + </RegExp> + <RegExp input="$$1" output="<url function="GetOFDBData">http://xml.n4rf.net/ofdbgw/fassung/\1</url>" dest="5+"> + <expression><fassungen>.*?<id>([^<]+)</expression> + </RegExp> + <RegExp input="$$1" output="<thumb>\1</thumb>" dest="5+"> + <expression><bild>([^<]+)</expression> + </RegExp> + <RegExp conditional="fanart" input="$$1" output="<url function="GetTMDBFanartByIMDBId">http://api.themoviedb.org/2.0/Movie.imdbLookup?imdb_id=tt\1&amp;api_key=57983e31fb435df4df77afb854740ea9</url>" dest="5+"> + <expression><imdbid>([^<]+)</expression> + </RegExp> + <expression noclean="1"/> + </RegExp> + </GetDetails> + + <GetOFDBData dest="5"> + <RegExp input="$$6" output="<?xml version="1.0" encoding="utf-8" standalone="yes"?><details>\1</details>" dest="5+"> + <RegExp input="$$1" output="<studio>\1</studio>" dest="6+"> + <expression trim="1"><label>([^<|^(]*)</expression> + </RegExp> + <RegExp input="$$1" output="<mpaa>\1</mpaa>" dest="6+"> + <expression trim="1"><freigabe>([^<]+)</expression> + </RegExp> + <RegExp input="$$1" output="<runtime>\1</runtime>" dest="6+"> + <expression trim="1"><laufzeit>([^<]+)</expression> + </RegExp> + <expression noclean="1"/> + </RegExp> + </GetOFDBData> +</scraper> diff --git a/system/scrapers/video/ptgate.jpg b/system/scrapers/video/ptgate.jpg Binary files differnew file mode 100644 index 0000000000..b240e4327f --- /dev/null +++ b/system/scrapers/video/ptgate.jpg diff --git a/system/scrapers/video/ptgate.xml b/system/scrapers/video/ptgate.xml new file mode 100644 index 0000000000..629849fd8a --- /dev/null +++ b/system/scrapers/video/ptgate.xml @@ -0,0 +1,280 @@ +п»ї<?xml version="1.0" encoding="UTF-8"?> +<scraper framework="1.1" date="2009-08-10" name="PTGate and IMDb" content="movies" thumb="ptgate.jpg" language="pt"> + <include>common/tmdb.xml</include> + <include>common/movieposterdb.xml</include> + <include>common/impa.xml</include> + <include>common/imdb.xml</include> + <GetSettings dest="3"> + <RegExp input="$$5" output="<settings>\1</settings>" dest="3"> + <RegExp input="$$1" output="<setting label="Utilizar apenas o Cinema PTGate" type="bool" id="ptgateonly" default="false"></setting>" dest="5"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting type="sep"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting label="Incluir detalhes e fotos dos Actores" type="bool" id="fullcredits" default="false"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting label="Incluir detalhes do IMPAwards" type="bool" id="impawards" default="true"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting label="Incluir detalhes do MoviePosterDB" type="bool" id="movieposterdb" default="false"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting label="Incluir trailer" type="bool" id="trailer" default="true"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting label="Incluir Fanart" type="bool" id="fanart" default="true"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting label="Tamanhos do Poster e das Fotos dos Actores" type="labelenum" values="192|256|384|512|1024" id="imdbscale" default="512"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting type="sep"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting label="Url para acesso ao IMDb" type="text" id="url" default="akas.imdb.com"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetSettings> + + <NfoUrl dest="3"> + <RegExp input="$$1" output="\1" dest="3"> + <expression clear="yes" noclean="1">(http://www.cinema.ptgate.pt/filmes/[0-9]*)</expression> + </RegExp> + </NfoUrl> + + <CreateSearchUrl dest="3"> + <RegExp input="$$1" output="<url>http://www.cinema.ptgate.pt/pesquisa/?q=\1</url>" dest="3"> + <expression noclean="1">([^\(]+)</expression> + </RegExp> + </CreateSearchUrl> + + <GetSearchResults dest="8"> + <RegExp input="$$5" output="<?xml version="1.0" encoding="iso-8859-1" standalone="yes"?><results>\1</results>" dest="8"> + <RegExp dest="5+" input="$$1" output="<entity><title>\2 - \4 (\3)</title><url>http://www.cinema.ptgate.pt/filmes/\1</url><id>\1</id></entity>"> + <expression repeat="yes" trim="2,3"><a href="/filmes/([0-9]*)">(.*?)</a> <small>([^\(]*)\(([0-9]*)\)</expression> + </RegExp> + <RegExp dest="5" input="$$1" output="<entity><title>\2 - \3 (\4)</title><url>http://www.cinema.ptgate.pt/filmes/\1</url><id>\1</id></entity>"> + <expression trim="2,4"><a href="/filmes/([0-9]*)" class="upper">([^<]*)</a> <small>\(([0-9]*)\)<br />([^<]*)</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetSearchResults> + + <GetDetails clearbuffers="no" dest="3"> + <RegExp input="$$5" output="<details>\1</details>" dest="3"> + <!-- Id --> + <RegExp dest="5" input="$$2" output="<id>\1</id>"> + <expression></expression> + </RegExp> + + <!-- Title --> + <RegExp dest="5+" input="$$1" output="<title>\1</title>"> + <expression><h2>([^<]*)</h2></expression> + </RegExp> + + <!-- Plot --> + <RegExp dest="5+" input="$$1" output=";<plot>\1</plot>"> + <expression><b>sinopse[^\n]*\s+([^\n]*)</expression> + </RegExp> + + <!-- Outline --> + <RegExp dest="5+" input="$$1" output="<outline>\1</outline>"> + <expression><b>sinopse[^\n]*\s+(.*?[^A-Z]\.)</expression> + </RegExp> + + <!-- Genre --> + <RegExp dest="5+" input="$$9" output="<genre>\1</genre>"> + <RegExp dest="9" input="$$1" output="\1,"> + <expression><b>g.nero</b>: ([^<]*)</expression> + </RegExp> + <expression repeat="yes" trim="1">([^,]+),</expression> + </RegExp> + + <!-- Year --> + <RegExp dest="11" input="$$1" output="<year>\1</year>"> + <expression><b>ano</b>: ([^<]*)</expression> + </RegExp> + <RegExp dest="5+" input="$$11" output="\1"> + <expression noclean="1"></expression> + </RegExp> + + <!-- Actor --> + <RegExp dest="8" input="$$9" output="<actor><name>\1</name></actor>"> + <RegExp input="$$1" output="\1," dest="9"> + <expression><b>int.rpretes</b><br />\s*(.*?)<br /></expression> + </RegExp> + <expression clear="yes" repeat="yes" trim="1">([^,]*),</expression> + </RegExp> + <RegExp dest="4" input="$$8" output="<actors>\1</actors>"> + <expression noclean="1"></expression> + </RegExp> + <RegExp dest="5+" input="$$8" conditional="ptgateonly" output="\1"> + <expression noclean="1"></expression> + </RegExp> + + <!-- Director --> + <RegExp dest="8" input="$$9" output="<director>\1</director>"> + <RegExp input="$$1" output="\1," dest="9"> + <expression><b>realiza..o</b><br />\s*(.*?)<br /></expression> + </RegExp> + <expression repeat="yes" trim="1">([^,]+),</expression> + </RegExp> + <RegExp dest="4+" input="$$8" output="<directors>\1</directors>"> + <expression noclean="1"></expression> + </RegExp> + <RegExp dest="5+" input="$$8" conditional="ptgateonly" output="\1"> + <expression noclean="1"></expression> + </RegExp> + + <!-- Add PTGate image as last Thumb --> + <RegExp dest="12" input="$$2" output="<thumb>http://www.cinema.ptgate.pt/Movies/\1.jpg</thumb>"> + <expression></expression> + </RegExp> + <RegExp dest="5+" input="$$12" conditional="ptgateonly" output="\1"> + <expression noclean="1"></expression> + </RegExp> + + <!-- Add PT Distributor --> + <RegExp dest="13" input="$$1" output="<studio>Dist: \1</studio>"> + <expression><b>distribuidora</b>: ([^<]*)</expression> + </RegExp> + <RegExp dest="5+" input="$$13" conditional="ptgateonly" output="\1"> + <expression noclean="1"></expression> + </RegExp> + + <!-- IMDB Details --> + <RegExp dest="5+" input="$$10" conditional="!ptgateonly" output="\1"> + <!-- Use Google as last resort --> + <RegExp dest="10" input="$$8" output="<url function="GetIMDBfromGoogle">http://www.google.com/search?q=site:imdb.com\1</url>"> + <!-- search google for the original title --> + <RegExp input="$$1" output="\1" dest="7"> + <expression trim="1">\&#8250; ([^<]+)<br /></expression> + </RegExp> + <!-- add year to search string --> + <RegExp input="$$11" output="+\1" dest="8"> + <expression></expression> + </RegExp> + <!-- remove spaces and , --> + <RegExp input="$$7" output="+\1" dest="8+"> + <expression repeat="yes">([^ ,]+)</expression> + </RegExp> + <expression></expression> + </RegExp> + + <!-- use IMDB link on PTGATE, when available --> + <RegExp dest="10" input="$$9" output="<url function="GetIMDBDetails">http://$INFO[url]/title/\1/</url>"> + <RegExp dest="9" input="$$1" output="\1"> + <expression clear="yes">\.imdb\.com/title/(tt[0-9]*)</expression> + </RegExp> + <expression>(.+)</expression> + </RegExp> + + <expression noclean="1"></expression> + </RegExp> + + <expression noclean="1"></expression> + </RegExp> + </GetDetails> + + + <GetIMDBfromGoogle clearbuffers="no" dest="3"> + <RegExp input="$$5" dest="3" output="<details>\1</details>"> + <RegExp input="$$1" dest="5" output="<url function="GetIMDBDetails">http://$INFO[url]/title/\1/</url>"> + <expression>\.imdb\.com/title/(tt[0-9]*)</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetIMDBfromGoogle> + <GetIMDBDetails clearbuffers="no" dest="3"> + <RegExp input="$$5" dest="3" output="<details>\1</details>"> + <!-- IMDB Id --> + <RegExp input="$$1" dest="2" output="\1"> + <expression>id=(tt[t0-9]*)</expression> + </RegExp> + + <!-- Studio --> + <RegExp input="$$1" output="<studio>\1</studio>" dest="5"> + <expression clear="yes" repeat="yes">"/company/[^/]*/">([^<]*)</a></expression> + </RegExp> + <RegExp input="$$13" output="\1" dest="5+"> + <expression noclean="1"></expression> + </RegExp> + + <!-- Certificates --> + <RegExp input="$$1" output="<certification>\1 \3</certification>" dest="5+"> + <expression repeat="yes"><a href="/List\?certificates=[^"]*">([^<]*)</a>[^<]*(<i>([^<]*)</i>)?</expression> + </RegExp> + + <!-- Runtime --> + <RegExp input="$$1" output="<runtime>\1</runtime>" dest="5+"> + <expression trim="1"><h5>Runtime:</h5>[^0-9]*([^<|]*)</expression> + </RegExp> + + <!-- Rating --> + <RegExp input="$$1" output="<rating>\1</rating><votes>\2</votes>" dest="5+"> + <expression><b>([0-9.]+)/10</b>[^<]*<a href="ratings" class="tn15more">([0-9,]+) votes</a></expression> + </RegExp> + + <!-- Actor --> + <RegExp conditional="!fullcredits" input="$$2" output="<url cache="$$2-credits.html" function="GetIMDBCast">$$3</url>" dest="5+"> + <expression></expression> + </RegExp> + + <!-- Director --> + <RegExp conditional="!fullcredits" input="$$2" output="<url cache="$$2-credits.html" function="GetMovieDirectors">$$3</url>" dest="5+"> + <expression></expression> + </RegExp> + + <!-- Credits --> + <RegExp conditional="!fullcredits" input="$$2" output="<url cache="$$2-credits.html" function="GetIMDBWriters">$$3</url>" dest="5+"> + <expression></expression> + </RegExp> + + <!-- Actor, full --> + <RegExp conditional="fullcredits" input="$$2" output="<url cache="$$2-fullcredits.html" function="GetIMDBCast">$$3fullcredits</url>" dest="5+"> + <expression></expression> + </RegExp> + + <!-- Director, full --> + <RegExp conditional="fullcredits" input="$$2" output="<url cache="$$2-fullcredits.html" function="GetIMDBDirectors">$$3fullcredits</url>" dest="5+"> + <expression></expression> + </RegExp> + + <!-- Credits, full --> + <RegExp conditional="fullcredits" input="$$2" output="<url cache="$$2-fullcredits.html" function="GetIMDBWriters">$$3fullcredits</url>" dest="5+"> + <expression></expression> + </RegExp> + + <!-- Thumb, IMPAwards --> + <RegExp conditional="impawards" input="$$2" output="<url cache="$$2-posters.html" function="GetIMPALink">$$3posters</url>" dest="5+"> + <expression></expression> + </RegExp> + + <!-- Poster, MoviePosterDB --> + <RegExp conditional="movieposterdb" input="$$1" output="<url function="GetMoviePosterDBLink">http://www.movieposterdb.com/browse/search?type=movies&amp;query=\1</url>" dest="5+"> + <expression>/title/tt([t0-9]*)/faq</expression> + </RegExp> + + <!-- Trailer --> + <RegExp conditional="trailer" input="$$1" output="<url function="GetIMDBTrailer">http://$INFO[url]/video/imdb/vi\1/player</url>" dest="5+"> + <expression>/vi([0-9]*)/"</expression> + </RegExp> + + <!-- Thumb, IMDB --> + <RegExp input="$$2" output="<url cache="$$2-posters.html" function="GetIMDBPoster">$$3posters</url>" dest="5+"> + <expression></expression> + </RegExp> + + <!-- Fanart --> + <RegExp conditional="fanart" input="$$2" output="<url function="GetTMDBFanartByIMDBId">http://api.themoviedb.org/2.0/Movie.imdbLookup?imdb_id=$$2&amp;api_key=57983e31fb435df4df77afb854740ea9</url>" dest="5+"> + <expression></expression> + </RegExp> + + <expression noclean="1"></expression> + </RegExp> + </GetIMDBDetails> +</scraper> diff --git a/system/scrapers/video/speedallocinev2.xml b/system/scrapers/video/speedallocinev2.xml new file mode 100644 index 0000000000..912c3f5fdd --- /dev/null +++ b/system/scrapers/video/speedallocinev2.xml @@ -0,0 +1,28 @@ +<?xml version="1.0" encoding="utf-8" standalone="yes"?>
+<scraper framework="1.0" date="2009-05-30" name="speedallocinev2.fr" content="movies" thumb="allocine.jpg" language="fr">
+
+ <CreateSearchUrl dest="3">
+ <RegExp input="$$1" output="http://passion-xbmc.org/scraper/index.php?search=\1" dest="3">
+ <expression></expression>
+ </RegExp>
+ </CreateSearchUrl>
+
+ <GetSearchResults dest="2">
+ <RegExp input="$$5" output="<?xml version="1.0" encoding="iso-8859-1" standalone="yes"?><results>\1</results>" dest="2">
+ <RegExp input="$$1" output="\1" dest="5">
+ <expression trim="1" noclean="1">(.*)</expression>
+ </RegExp>
+ <expression noclean="1"></expression>
+ </RegExp>
+ </GetSearchResults>
+
+ <GetDetails clearbuffers="no" dest="3">
+ <RegExp input="$$5" output="\1" dest="3">
+ <RegExp input="$$1" output="\1" dest="5">
+ <expression trim="1" noclean="1">(.*)</expression>
+ </RegExp>
+ <expression noclean="1"></expression>
+ </RegExp>
+ </GetDetails>
+
+</scraper>
diff --git a/system/scrapers/video/sratim.gif b/system/scrapers/video/sratim.gif Binary files differnew file mode 100644 index 0000000000..d3baf9fcc7 --- /dev/null +++ b/system/scrapers/video/sratim.gif diff --git a/system/scrapers/video/sratim.xml b/system/scrapers/video/sratim.xml new file mode 100644 index 0000000000..5eea36a530 --- /dev/null +++ b/system/scrapers/video/sratim.xml @@ -0,0 +1,115 @@ +п»ї<?xml version="1.0" encoding="UTF-8"?>
+<scraper framework="1.0" date="2009-05-22" name="Sratim" content="movies" thumb="sratim.gif" language="il">
+ <GetSettings dest="3">
+ <RegExp input="$$5" output="<settings>\1</settings>" dest="3">
+ <RegExp input="$$1" output="<setting label="Enable Full Cast Credits" type="bool" id="fullcredits" default="false"></setting>" dest="5">
+ <expression></expression>
+ </RegExp>
+ <RegExp input="$$1" output="<setting label="Enable Trailer" type="bool" id="trailer" default="true"></setting>" dest="5+">
+ <expression></expression>
+ </RegExp>
+ <RegExp input="$$1" output="<setting label="Enable Fanart" type="bool" id="fanart" default="false"></setting>" dest="5+">
+ <expression></expression>
+ </RegExp>
+ <RegExp input="$$1" output="<setting label="Get Rank From IMDB" type="bool" id="imdbrank" default="false"></setting>" dest="5+">
+ <expression></expression>
+ </RegExp>
+ <expression noclean="1"></expression>
+ </RegExp>
+ </GetSettings>
+ <CreateSearchUrl dest="3">
+ <RegExp input="$$1" output="<url>http://www.sratim.co.il/movies/search.aspx?Keyword=\1</url>" dest="3">
+ <expression noclean="1"></expression>
+ </RegExp>
+ </CreateSearchUrl>
+ <GetSearchResults dest="8">
+ <RegExp input="$$5" output="<?xml version="1.0" encoding="iso-8859-1" standalone="yes"?><results>\1</results>" dest="8">
+ <RegExp input="$$1" output="<entity><title>\2</title><url>http://www.sratim.co.il/\1</url></entity>" dest="5">
+ <expression clear="yes" noclean="1" repeat="yes"><table class="MovieViews".*?>.*?<a href="(.[^"]*)">(.*?)<div>.*?</table></expression>
+ </RegExp>
+ <expression clear="yes" noclean="1"></expression>
+ </RegExp>
+ </GetSearchResults>
+ <GetDetails clearbuffers="no" dest="3">
+ <RegExp input="$$5" output="<details>\1</details>" dest="3">
+ <RegExp input="$$1" output="<title>\1</title>" dest="5+">
+ <expression trim="1" noclean="1"><div.*?class="Box_Header"><table.*?><tr><td.*?>(.*?)</td></expression>
+ </RegExp>
+ <RegExp input="$$1" output="<plot>\1</plot>" dest="5+">
+ <expression trim="1"><u>ЧЄЧ§Ч¦Ч™ЧЁ:</u></b><br />(.*?)</div></expression>
+ </RegExp>
+ <RegExp input="$$1" output="<genre>\1</genre>" dest="5+">
+ <expression trim="1" repeat="yes"><a href="/movies/search.aspx\?g=.*?">(.*?)</a></expression>
+ </RegExp>
+ <expression noclean="1"></expression>
+ <RegExp input="$$1" output="\1" dest="7">
+ <expression repeat="no" noclean="1"><b>Ч©Ч—Ч§Ч Ч™Чќ:</b>(.*?)<br /></expression>
+ </RegExp>
+ <RegExp input="$$1" output="<studio>\1</studio>" dest="5+">
+ <expression trim="1" repeat="yes"><a href="/movies/companies/view.aspx\?id=.*?">(.*?)</a></expression>
+ </RegExp>
+ <RegExp input="$$7" output="<actor><name>\1</name></actor>" dest="5+">
+ <expression trim="1" repeat="yes"><a href="/movies/actors/view.aspx\?id=.*?">(.*?)</a></expression>
+ </RegExp>
+ <RegExp conditional="!fullcredits" input="$$7" output="<url function="GetActorDetails">http://www.sratim.co.il/movies/actors/view.aspx?id=\1</url>" dest="5+"><expression repeat="yes"><a href="/movies/actors/view.aspx\?id=(.*?)"></expression></RegExp><RegExp input="$$1" output="\1" dest="7">
+ <expression repeat="no" noclean="1"><b>Ч‘ЧћЧђЧ™:</b>(.*?)<br /></expression>
+ </RegExp>
+ <RegExp input="$$7" output="<director>\1</director>" dest="5+">
+ <expression trim="1" repeat="yes"><a href="/movies/actors/view.aspx\?id=.*?">(.*?)</a></expression>
+ </RegExp>
+ <RegExp input="$$1" output="<year>\1</year>" dest="5+">
+ <expression trim="1"><span id="ctl00_ctl00_Body_Body_Box_ProductionYear">(.*?)</span></expression>
+ </RegExp>
+ <RegExp input="$$1" output="<rating>\1</rating>" dest="5+" conditional="!imdbrank">
+ <expression trim="1" repeat="no"><img alt="(.*?)/10" src="/Images/Star_Big_.*?.gif" /></expression>
+ </RegExp>
+ <RegExp input="$$1" output="<votes>\1</votes>" dest="5+" conditional="!imdbrank">
+ <expression trim="1" repeat="no"><span.*?>([0-9]+?) ЧћЧ“ЧЁЧ’Ч™Чќ</span></expression>
+ </RegExp>
+ <RegExp input="$$1" output="<runtime>\1</runtime>" dest="5+">
+ <expression trim="1" repeat="no"><b>ЧђЧ•ЧЁЧљ:</b> ([0-9]+?) Ч“Ч§Ч•ЧЄ.</expression>
+ </RegExp>
+ <RegExp input="$$1" output="<thumb>http://www.sratim.co.il\1</thumb>" dest="5+">
+ <expression trim="1" repeat="no"><img src="(/movies/images/.*?)" id="ctl00_ctl00_Body_Body_Box_MainPicture"</expression>
+ </RegExp>
+ <RegExp input="$$1" output="<url function="GetIMDBDetails">http://www.imdb.com/title/\1/</url>" dest="5+">
+ <expression trim="1" repeat="no">http://www\.imdb\.com/title/(tt\d+?)'</expression>
+ </RegExp>
+ <RegExp conditional="fanart" input="$$1" output="<url function="GetTMDBFanartByIMDBId">http://api.themoviedb.org/2.0/Movie.imdbLookup?imdb_id=\1&amp;api_key=57983e31fb435df4df77afb854740ea9</url>" dest="5+">
+ <expression trim="1" repeat="no">http://www\.imdb\.com/title/(tt\d+?)'</expression>
+ </RegExp>
+ <RegExp conditional="imdbrank" input="$$1" output="<rating>\1</rating>" dest="5+">
+ <expression trim="1" repeat="no"><img alt="([0-9]+\.[0-9]*)/10" src="/Images/Star_IMDB</expression>
+ </RegExp>
+ <RegExp conditional="imdbrank" input="$$1" output="<votes>\1</votes>" dest="5+">
+ <expression trim="1" repeat="no">Star_IMDB.*?<span.*?>([0-9]+?) ЧћЧ“ЧЁЧ’Ч™Чќ</span></expression>
+ </RegExp>
+ </RegExp>
+ </GetDetails>
+ <GetActorDetails clearbuffers="no" dest="5"><RegExp input="$$2" output="<details>\1</details>" dest="5">
+ <expression noclean="1"></expression>
+ <RegExp input="$$1" output="<actor><name>\1</name><thumb>http://www.sratim.co.il\2</thumb></actor>" dest="2+">
+ <expression clear="yes" repeat="no"><div class="Box_Header">.*?<span.*?>(.*?)</span>.*?<div class="Box_Body">.*?<img.*?src="(/movies/Actors/Images/.*?)" /></expression>
+ </RegExp>
+ </RegExp>
+ </GetActorDetails>
+ <GetIMDBDetails dest="5">
+ <RegExp input="$$3" output="<?xml version="1.0" encoding="utf-8" standalone="yes"?><details>\1</details>" dest="5">
+ <RegExp conditional="trailer" input="$$1" output="<url function="GetTrailer">http://akas.imdb.com/video/imdb/vi\1/player</url>" dest="3+">
+ <expression>/vi([0-9]*)/"</expression>
+ </RegExp>
+ <expression noclean="1"></expression>
+ </RegExp>
+ </GetIMDBDetails>
+ <GetTrailer clearbuffers="no" dest="5">
+ <RegExp input="$$2" output="<details><trailer urlencoded="yes">\1</trailer></details>" dest="5">
+ <RegExp input="$$1" output="\1" dest="2">
+ <expression noclean="1">"file"[^"]*"([^"]*)</expression>
+ </RegExp>
+ <RegExp input="$$1" output="$$2/\1" dest="2">
+ <expression noclean="1">"id"[^"]*"([^"]*)</expression>
+ </RegExp>
+ <expression noclean="1"></expression>
+ </RegExp>
+ </GetTrailer>
+</scraper>
diff --git a/system/scrapers/video/tmdb.png b/system/scrapers/video/tmdb.png Binary files differnew file mode 100644 index 0000000000..377f6af8a5 --- /dev/null +++ b/system/scrapers/video/tmdb.png diff --git a/system/scrapers/video/tmdb.xml b/system/scrapers/video/tmdb.xml new file mode 100644 index 0000000000..92c688cf82 --- /dev/null +++ b/system/scrapers/video/tmdb.xml @@ -0,0 +1,99 @@ +<?xml version="1.0" encoding="UTF-8"?> +<scraper framework="1.1" date="2009-07-28" name="themoviedb.org" content="movies" thumb="tmdb.png" language="en"> + <include>common/tmdb.xml</include> + <include>common/dtrailer.xml</include> + <GetSettings dest="3"> + <RegExp input="$$5" output="<settings>\1</settings>" dest="3"> + <RegExp input="$$1" output="<setting label="Enable Fanart" type="bool" id="fanart" default="true"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting label="Enable Trailers from Dtrailer.com" type="bool" id="trailer" default="false"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetSettings> + <CreateSearchUrl dest="3"> + <RegExp input="$$1" output="<url>http://api.themoviedb.org/2.0/Movie.search?title=\1&amp;api_key=57983e31fb435df4df77afb854740ea9</url>" dest="3"> + <RegExp input="$$2" output="%20(\1)" dest="4"> + <expression clear="yes">(.+)</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </CreateSearchUrl> + <NfoUrl dest="3"> + <RegExp input="$$1" output="<url>http://api.themoviedb.org/2.0/Movie.getInfo?id=\2&amp;api_key=57983e31fb435df4df77afb854740ea9</url><id>\2</id>" dest="3"> + <expression clear="yes" noclean="1">(themoviedb.org/movie/)([0-9]*)</expression> + </RegExp> + <RegExp input="$$1" output="<url function="GetTMDBId">http://api.themoviedb.org/2.0/Movie.imdbLookup?imdb_id=tt\2&amp;api_key=57983e31fb435df4df77afb854740ea9</url>" dest="3+"> + <expression>(imdb.com/title/tt)([0-9]*)</expression> + </RegExp> + <RegExp input="$$1" output="<url function="GetTMDBId">http://api.themoviedb.org/2.0/Movie.imdbLookup?imdb_id=tt\2&amp;api_key=57983e31fb435df4df77afb854740ea9</url>" dest="3+"> + <expression>(imdb.com/)Title\?([0-9]+)</expression> + </RegExp> + </NfoUrl> + <GetTMDBId dest="3"> + <RegExp input="$$1" output="<url cache="tmdb-\1.xml">http://api.themoviedb.org/2.0/Movie.getInfo?id=\1&amp;api_key=57983e31fb435df4df77afb854740ea9</url><id>\1</id>" dest="3+"> + <expression><id>([0-9]*)</id></expression> + </RegExp> + </GetTMDBId> + <GetSearchResults dest="8"> + <RegExp input="$$3" output="<results>\1</result>" dest="8"> + <RegExp input="$$1" output="<entity><title>\1</title><id>\2</id><year>\3</year><url cache="tmdb-\2.xml" >http://api.themoviedb.org/2.0/Movie.getInfo?id=\2&amp;api_key=57983e31fb435df4df77afb854740ea9</url></entity>" dest="3"> + <expression repeat="yes"><movie>.*?<title>([^<]*)</title>.*?<id>([^<]*)</id>.*?<release>([0-9]+).*?</movie></expression> + </RegExp> + <expression noclean="1"/> + </RegExp> + </GetSearchResults> + <GetDetails dest="3"> + <RegExp input="$$5" output="<details>\1</details>" dest="3"> + <RegExp input="$$1" output="<title>\1</title>" dest="5"> + <expression><title>([^<]*)</title></expression> + </RegExp> + <RegExp input="$$1" output="<year>\1</year>" dest="5+"> + <expression><release>([0-9]+)</expression> + </RegExp> + <RegExp input="$$1" output="<runtime>\1</runtime>" dest="5+"> + <expression><runtime>([^<]*)</runtime></expression> + </RegExp> + <RegExp input="$$1" output="<rating>\1</rating>" dest="5+"> + <expression><rating>([^<]*)</rating></expression> + </RegExp> + <RegExp input="$$1" output="<genre>\1</genre>" dest="5+"> + <expression repeat="yes"><category>.*?<name>([^<]*)</name></expression> + </RegExp> + <RegExp input="$$1" output="<plot>\1</plot>" dest="5+"> + <expression><short_overview>([^<]*)</short_overview></expression> + </RegExp> + <RegExp input="$$1" output="<outline>\1</outline>" dest="5+"> + <expression><short_overview>([^<]*)</short_overview></expression> + </RegExp> + <RegExp input="$$1" output="<director>\1</director>" dest="5+"> + <expression repeat="yes"><person job="director".*?<name>([^<]*)</name></expression> + </RegExp> + <RegExp input="$$1" output="<credits>\1</credits>" dest="5+"> + <expression repeat="yes"><person job="author".*?<name>([^<]*)</name></expression> + </RegExp> + <RegExp input="$$1" output="<actor><name>\1</name><role>\2</role></actor>" dest="5+"> + <expression repeat="yes"><person job="actor".*?<name>([^<]*)</name>.*?<role>([^<]*)</role></expression> + </RegExp> + <RegExp input="$$1" output="<url cache="tmdb-$$2.xml" function="GetTMDBThumbs">$$3</url>" dest="5+"> + <expression/> + </RegExp> + + <RegExp conditional="fanart" input="$$1" output="<url cache="tmdb-$$2.xml" function="GetTMDBFanart">$$3</url>" dest="5+"> + <expression/> + </RegExp> + <RegExp conditional="trailer" input="$$6" output="<url function="GetDTrailerLink">http://en.dtrailer.com/movies/search/\1</url>" dest="5+"> + <RegExp input="$$4" output="\1-" dest="6"> + <RegExp input="$$1" output="\1" dest="4"> + <expression><title>([^<]*)</title></expression> + </RegExp> + <expression repeat="yes">([a-zA-Z0-9]+)</expression> + </RegExp> + <expression></expression> + </RegExp> + <expression noclean="1"/> + </RegExp> + </GetDetails> +</scraper> diff --git a/system/scrapers/video/tvcom.png b/system/scrapers/video/tvcom.png Binary files differnew file mode 100644 index 0000000000..1901b1e93f --- /dev/null +++ b/system/scrapers/video/tvcom.png diff --git a/system/scrapers/video/tvcom.xml b/system/scrapers/video/tvcom.xml new file mode 100644 index 0000000000..c4f391ebd3 --- /dev/null +++ b/system/scrapers/video/tvcom.xml @@ -0,0 +1,110 @@ +<?xml version="1.0" encoding="UTF-8"?> +<scraper framework="1.0" date="2009-05-22" name="TV.com" content="tvshows" thumb="tvcom.png" language="en"> + <CreateSearchUrl dest="3"> + <RegExp input="$$1" output="<url>http://www.tv.com/search.php?type=Search&amp;stype=ajax_search&amp;qs=\1&amp;search_type=program&amp;pg_results=0&amp;sort=</url>" dest="3"> + <expression></expression> + </RegExp> + </CreateSearchUrl> + <GetSearchResults dest="3"> + <RegExp input="$$4" output="<results>\1</results>" dest="3"> + <RegExp input="$$1" output="<entity><title>\2</title><url>http://www.tv.com/show/\1/summary.html</url><url>http://www.tv.com/show/\1/cast.html</url><url>http://www.tv.com/show/\1/episode_listings.html?season=All</url><id>\1</id></entity>" dest="4"> + <expression repeat="yes" noclean="1"><a href="http://www\.tv\.com/[^/]*/show/([0-9]+)/summary\.html[^"]*"[^>]*>([^<]+)</a></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetSearchResults> + <GetDetails dest="7"> + <RegExp input="$$5" output="<details>\1</details>" dest="7"> + <RegExp input="$$1" output="<title>\1</title>" dest="5"> + <expression noclean="1"><title>([^<]*) on TV\.com</expression> + </RegExp> + <RegExp input="$$1" output="<genre>\1</genre>" dest="5+"> + <expression repeat="yes" noclean="1">;genre;[^>]*>([^<]*)</a></expression> + </RegExp> +<!-- <RegExp input="$$1" output="<plot>\1</plot>" dest="5+"> + <expression>id="summary_fold" class="mt-10">\W*(.*?) *?</div></expression> + </RegExp> --> + + <RegExp input="$$8" output="<plot>\1</plot>" dest="5+"> + <RegExp input="$$1" output="\1" dest="6"> + <expression><span class="long">(.*)</span>[^<]*<span class="short"></expression> + </RegExp> + <RegExp input="$$6" output="\1" dest="8"> + <expression repeat="yes"></expression> + </RegExp> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<rating>\1</rating>" dest="5+"> + <expression><span>Show Score</span>[^0-9]*([0-9\.]*)</expression> + </RegExp> + <RegExp input="$$1" output="<votes>\1</votes>" dest="5+"> + <expression><span>([0-9,]*)</span>[^<]*Votes</expression> + </RegExp> + <RegExp input="$$2" output="<actor><name>\1</name><role>\2</role></actor>" dest="5+"> + <expression repeat="yes">>([^<]*)</a></h3> <a class="photos_link" href="http://www\.tv\.com/[^/]*/person/[0-9]*/photos\.html\?tag=cast;stars;photos;[0-9]*">\(photos\)</a></div><div class="role">Role: ([^<]*)</div></expression> + </RegExp> + <RegExp input="$$1" output="<thumb>\1</thumb>" dest="5+"> + <expression>(http://image\.com\.com/tv/images/content_headers/program_new/[0-9]*\.jpg)</expression> + </RegExp> + <RegExp input="$$1" output="<status>\1</status>" dest="5+"> + <expression trim="1"><span class="program_status_name">([^<]*)</span></expression> + </RegExp> + <RegExp input="$$1" output="<premiered>\1</premiered>" dest="5+"> + <expression trim="1"><span class="start_date">([^<]*)</span></expression> + </RegExp> + <RegExp input="$$8" output="<episodeguide>\1</episodeguide>" dest="5+"> + <RegExp input="$$3" output="<url>http://www.tv.com/show/$$4/episode_listings.html?season=\1</url>" dest="8"> + <expression repeat="yes">/show/[0-9]+/episode_listings\.html\?season=([0-9]+)</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetDetails> + + <GetEpisodeList dest="3"> + <RegExp input="$$5" output="<episodeguide>\1</episodeguide>" dest="3"> + <RegExp input="$$1" output="\1" dest="6"> + <expression>&nbsp;[^<]*<strong>([0-9]+)</strong></expression> + </RegExp> + <RegExp input="$$1" output="<episode><title>\3</title><id>\2</id><url >http://www.tv.com/episode/\2/summary.html</url><epnum>\1</epnum><season>$$6</season></episode>" dest="5"> + <expression repeat="yes"><div>([0-9]*)</div></td><td class="ep_title"><div><a href="http://www\.tv\.com/[^/]*/[^/]*/episode/([0-9]*)/summary\.html[^>]*>([^<]*)</a></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetEpisodeList> + + <GetEpisodeDetails dest="3"> + <RegExp input="$$5" output="<details>\1</details>" dest="3"> + <RegExp input="$$1" output="<title>\1</title>" dest="5"> + <expression><div class="content_title">[^<]*<h1>[^:]*:([^<]*)</h1></expression> + </RegExp> + <RegExp input="$$1" output="<plot>\1</plot>" dest="5+"> + <expression><p class="deck">([^=]*)<a </expression> + </RegExp> + <RegExp input="$$1" output="<rating>\1</rating>" dest="5+"> + <expression>Episode score[^<]*<span>([0-9\.]*)</span></expression> + </RegExp> + <RegExp input="$$1" output="<aired>\1</aired>" dest="5+"> + <expression><span>First Aired:</span>([^<]*)</li></expression> + </RegExp> + <RegExp input="$$1" output="<actor><name>\1</name><role>\2</role></actor>" dest="5+"> + <expression repeat="yes">">([^<]*)</a> \(([^<]*)\)[^<]*<</expression> + </RegExp> + <RegExp input="$$1" output="<director>\1</director>" dest="5+"> + <expression>Director:</dt><dd><a [^>]*>([^<]*)</a></expression> + </RegExp> + <RegExp input="$$1" output="<credits>\1</credits>" dest="5+"> + <expression>writer;0">([^<]*)</a></expression> + </RegExp> + <RegExp input="$$1" output="<thumb>\1</thumb>" dest="5+"> + <expression>(http://image\.com\.com/tv/images/content_headers/episode_new/[0-9]*\.jpg)</expression> + </RegExp> + <RegExp input="$$1" output="<code>\1</code>" dest="5+"> + <expression><span>Prod Code:</span>([^<]*)</li></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetEpisodeDetails> + +</scraper> diff --git a/system/scrapers/video/tvdb.png b/system/scrapers/video/tvdb.png Binary files differnew file mode 100644 index 0000000000..36c48cdbb6 --- /dev/null +++ b/system/scrapers/video/tvdb.png diff --git a/system/scrapers/video/tvdb.xml b/system/scrapers/video/tvdb.xml new file mode 100644 index 0000000000..559902a12b --- /dev/null +++ b/system/scrapers/video/tvdb.xml @@ -0,0 +1,304 @@ +<?xml version="1.0" encoding="UTF-8"?> +<!-- should be self-explanatory --> +<scraper framework="1.1" date="2009-07-28" name="TheTVDB.com" content="tvshows" thumb="tvdb.png" language="multi"> + <GetSettings dest="3"> + <RegExp input="$$5" output="<settings>\1</settings>" dest="3"> + <RegExp input="$$1" output="<url function="GetLanguages">http://www.thetvdb.com/api/1D62F2F90030C444/languages.xml</url>" dest="5"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting label="Use DVD Order" type="bool" id="dvdorder" default="false" enable="!eq(1,true)"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <!-- Absolute ordering is great for Animes it actually trick XBMC by telling it that the serie has only one season and uses the 'absolute_number' as episodes numbers --> + <RegExp input="$$1" output="<setting label="Use Absolute Ordering (Single Season)" type="bool" id="absolutenumber" default="false" enable="!eq(-1,true)"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting type="sep"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting label="Enable Fanart" type="bool" id="fanart" default="true"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting label="Prefer Posters" type="bool" id="posters" default="false"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting type="sep"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <RegExp input="$$1" output="<setting label="Language Override for Banners" type="bool" id="override" default="false"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetSettings> + + <NfoUrl dest="3"> + <RegExp input="$$1" output="<url cache="\1.xml">http://www.thetvdb.com/api/1D62F2F90030C444/series/\1/all/$INFO[language].zip</url><id>\1</id>" dest="3"> + <expression>http://(?:www\.)?thetvdb.com/(?:index\.php)?\?tab=series&id=([0-9]+)</expression> + </RegExp> + </NfoUrl> + <EpisodeGuideUrl dest="3"> + <RegExp input="$$1" output="\1" dest="3"> + <expression>(.*?http://www.thetvdb.com.*)</expression> + </RegExp> + <RegExp input="$$1" output="\1" dest="3"> + <expression>(.*?http://thetvdb.com.*)</expression> + </RegExp> + </EpisodeGuideUrl> + + <!-- input: $1=query string --> + <!-- returns: the url we should use to do the search --> + <CreateSearchUrl dest="3"> + <RegExp input="$$1" output="http://www.thetvdb.com/api/GetSeries.php?seriesname=\1$$4&language=$INFO[language]" dest="3"> + <RegExp input="$$2" output="%20(\1)" dest="4"> + <expression clear="yes">(.+)</expression> + </RegExp> + <expression></expression> + </RegExp> + </CreateSearchUrl> + + <!-- input: $1=html $2=search query --> + <!-- returns: results in xml format <results><movie><title>*</title><url>*</url>*#urls<extra>*</extra></movie>*</results> --> + <GetSearchResults dest="1"> + <RegExp output="\1" dest="3"> + <expression noclean="1"><Data>(.*)</Data></expression> + </RegExp> + <RegExp input="$$4" output="<?xml version="1.0" encoding="utf-8" standalone="yes"?><results>\1</results>" dest="1"> + <RegExp input="$$3" output="<entity><title>\3</title><language>\2</language><url cache="\1.xml">http://www.thetvdb.com/api/1D62F2F90030C444/series/\1/all/\2.zip</url><id>\1</id></entity>" dest="4"> + <expression repeat="yes"><seriesid>([0-9]*)</seriesid>[^<]*<language>([^<]*)</language>[^<]*<SeriesName>([^<]*)</SeriesName></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetSearchResults> + + <!-- input: $1..#urls=html --> + <!-- input: $(#urls+1): extra !--> + <!-- returns: results in xml format <details><plot>*</plot><director>*</director><premiered>*</premiered><episodeguide>*</episodeguide></details> --> + <GetDetails dest="7"> + <RegExp input="$$4" output="<?xml version="1.0" encoding="utf-8" standalone="yes"?><details>\1</details>" dest="7"> + <RegExp input="$$1" output="\1" dest="5"> + <expression noclean="1"><Series>.*?<id>$$2</id>(.*)</expression> + </RegExp> + <RegExp input="$$5" output="<plot>\1</plot>" dest="4"> + <expression noclean="1"><Overview>([^<]*)</Overview></expression> + </RegExp> + <RegExp input="$$2" output="<id>\1</id>" dest="4+"> + <expression/> + </RegExp> + <RegExp input="$$1" output="<mpaa>\1</mpaa>" dest="4+"> + <expression><ContentRating>([^<]*)</ContentRating></expression> + </RegExp> + <RegExp input="$$5" output="<premiered>\1</premiered>" dest="4+"> + <expression><FirstAired>([^<]*)</FirstAired></expression> + </RegExp> + <RegExp input="$$5" output="<rating>\1</rating>" dest="4+"> + <expression><Rating>([^<]*)</Rating></expression> + </RegExp> + <RegExp input="$$5" output="<studio>\1</studio>" dest="4+"> + <expression><Network>([^<]*)</Network></expression> + </RegExp> + <RegExp input="$$5" output="<title>\1</title>" dest="4+"> + <expression><SeriesName>([^<]*)</SeriesName></expression> + </RegExp> + <RegExp input="$$5" output="\1" dest="10"> + <expression noclean="1"><Genre>([^<]*)</Genre></expression> + </RegExp> + <RegExp input="$$10" output="<genre>\1</genre>" dest="4+"> + <expression repeat="yes">([^\|]*)\|</expression> + </RegExp> + <RegExp input="$$10" output="<genre>\1</genre>" dest="4+"> + <expression repeat="yes">([^,]*),</expression> + </RegExp> + <RegExp input="$$10" output="<genre>\1</genre>" dest="4+"> + <expression>([^\|,]+)$</expression> + </RegExp> +<!-- actors with thumbs --> + <RegExp input="$$5" output="<actor><name>\2</name><role>\3</role><thumb>http://thetvdb.com/banners/\1</thumb></actor>" dest="4+"> + <expression repeat="yes" noclean="1,2,3"><Actor>.*?<Image>([^<]+)</Image>.*?<Name>([^<]*)</Name>.*?<Role>([^<]*)</expression> + </RegExp> +<!-- actors without thumbs --> + <RegExp input="$$5" output="<actor><name>\2</name><role>\3</role></actor>" dest="4+"> + <expression repeat="yes" noclean="1,2,3"><Actor>.*?<Image>([^<]*)</Image>.*?<Name>([^<]*)</Name>.*?<Role>([^<]*)</expression> + </RegExp> + <RegExp conditional="posters" output="<thumb>http://thetvdb.com/banners/\1</thumb>" dest="4+"> + <expression repeat="yes"><BannerPath>([^<]*)</BannerPath>[^<]*<BannerType>poster</BannerType></expression> + </RegExp> + <RegExp conditional="!override" input="$$5" output="<thumb>http://thetvdb.com/banners/\1</thumb>" dest="4+"> + <expression repeat="yes"><BannerPath>([^<]*)</BannerPath>[^<]*<BannerType>series</BannerType>[^<]*<BannerType2>graphical</BannerType2>[^<]*<Language>$INFO[language]</Language></expression> + </RegExp> + <RegExp conditional="override" input="$$5" output="<thumb>http://thetvdb.com/banners/\1</thumb>" dest="4+"> + <expression repeat="yes"><BannerPath>([^<]*)</BannerPath>[^<]*<BannerType>series</BannerType>[^<]*<BannerType2>graphical</BannerType2>[^<]*<Language>[a-z]*</Language></expression> + </RegExp> + <RegExp conditional="!override" input="$$5" output="<thumb>http://thetvdb.com/banners/\1</thumb>" dest="4+"> + <expression repeat="yes"><BannerPath>([^<]*)</BannerPath>[^<]*<BannerType>series</BannerType>[^<]*<BannerType2>text</BannerType2>[^<]*<Language>$INFO[language]</Language></expression> + </RegExp> + <RegExp conditional="override" input="$$5" output="<thumb>http://thetvdb.com/banners/\1</thumb>" dest="4+"> + <expression repeat="yes"><BannerPath>([^<]*)</BannerPath>[^<]*<BannerType>series</BannerType>[^<]*<BannerType2>text</BannerType2>[^<]*<Language>[a-z]*</Language></expression> + </RegExp> + <RegExp input="$$5" output="<thumb>http://thetvdb.com/banners/\1</thumb>" dest="4+"> + <expression repeat="yes"><BannerPath>([^<]*)</BannerPath>[^<]*<BannerType>series</BannerType>[^<]*<BannerType2>blank</BannerType2>[^<]*<Language></Language></expression> + </RegExp> + <RegExp conditional="!override" input="$$5" output="<thumb type="season" season="\2">http://thetvdb.com/banners/\1</thumb>" dest="4+"> + <expression repeat="yes"><BannerPath>([^<]*)</BannerPath>[^<]*<BannerType>season</BannerType>[^<]*<BannerType2>season</BannerType2>[^<]*<Language>$INFO[language]</Language>[^<]*<Season>([0-9]+)</Season></expression> + </RegExp> + <RegExp conditional="override" input="$$5" output="<thumb type="season" season="\2">http://thetvdb.com/banners/\1</thumb>" dest="4+"> + <expression repeat="yes"><BannerPath>([^<]*)</BannerPath>[^<]*<BannerType>season</BannerType>[^<]*<BannerType2>season</BannerType2>[^<]*<Language>[a-z]*</Language>[^<]*<Season>([0-9]+)</Season></expression> + </RegExp> + <RegExp conditional="!override" input="$$5" output="<thumb type="season" season="\2">http://thetvdb.com/banners/\1</thumb>" dest="4+"> + <expression repeat="yes"><BannerPath>([^<]*)</BannerPath>[^<]*<BannerType>season</BannerType>[^<]*<BannerType2>seasonwide</BannerType2>[^<]*<Language>$INFO[language]</Language>[^<]*<Season>([0-9]+)</Season></expression> + </RegExp> + <RegExp conditional="override" input="$$5" output="<thumb type="season" season="\2">http://thetvdb.com/banners/\1</thumb>" dest="4+"> + <expression repeat="yes"><BannerPath>([^<]*)</BannerPath>[^<]*<BannerType>season</BannerType>[^<]*<BannerType2>seasonwide</BannerType2>[^<]*<Language>[a-z]*</Language>[^<]*<Season>([0-9]+)</Season></expression> + </RegExp> + <RegExp input="$$5" conditional="!posters" output="<thumb>http://thetvdb.com/banners/\1</thumb>" dest="4+"> + <expression repeat="yes"><BannerPath>([^<]*)</BannerPath>[^<]*<BannerType>poster</BannerType></expression> + </RegExp> + <RegExp input="$$5" output="<thumb type="season" season="-1">http://thetvdb.com/banners/\1</thumb>" dest="4+"> + <expression repeat="yes"><BannerPath>([^<]*)</BannerPath>[^<]*<BannerType>poster</BannerType></expression> + </RegExp> + <RegExp conditional="fanart" input="$$7" output="<fanart url="http://thetvdb.com/banners/">\1</fanart>" dest="4+"> + <RegExp conditional="!override" input="$$5" output="<thumb dim="\2" colors="\3" preview="_cache/\1">\1</thumb>" dest="7+"> + <expression repeat="yes"><BannerPath>([^<]*)</BannerPath>[^<]*<BannerType>fanart</BannerType>[^<]*<BannerType2>([^<]*)</BannerType2>[^<]*<Colors>([^<]*)</Colors>[^<]*<Language>$INFO[language]</Language></expression> + </RegExp> + <RegExp conditional="override" input="$$5" output="<thumb dim="\2" colors="\3" preview="_cache/\1">\1</thumb>" dest="7+"> + <expression repeat="yes"><BannerPath>([^<]*)</BannerPath>[^<]*<BannerType>fanart</BannerType>[^<]*<BannerType2>([^<]*)</BannerType2>[^<]*<Colors>([^<]*)</Colors>[^<]*<Language>[a-z]*</Language></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + <RegExp input="$$2" output="<episodeguide><url cache="$$2.xml">http://www.thetvdb.com/api/1D62F2F90030C444/series/\1/all/$INFO[language].zip</url></episodeguide>" dest="4+"> + <expression></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetDetails> + + <!-- input: $1=html !--> + <!-- input: $2=series url !--> + <!-- returns: results in xml format <episodeguide><episode><title>*</title><url>*</url><season>*</season><epnum>*</epnum><thumb>*</thumb><id>*</id><aired>*</aired></episode>*</episodeguide> !--> + <GetEpisodeList dest="3"> + <RegExp input="$$4" output="<episodeguide>\1</episodeguide>" dest="3"> + <RegExp input="$$2" output="\1" dest="10"> + <expression>http://www.thetvdb.com/api/1D62F2F90030C444/series/([0-9]*)/all/$INFO[language].zip</expression> + </RegExp> + <!-- If Not Using "dvdorder" then check if the user wants to use "absolute ordering" --> + <RegExp conditional="!dvdorder"> + + <RegExp conditional="absolutenumber"> + <!-- If Using "absolute ordering" then scrape episodes and return them as Season 1 and Absolute_number as Episode number --> + <RegExp conditional="absolutenumber" input="$$1" output="<episode><title>\2</title><url cache="$$10.xml">$$2</url><epnum>\4</epnum><season>1</season><id>\1</id><aired>\3</aired></episode>" dest="4+"> + <expression repeat="yes"><Episode>.*?<id>([0-9]*)</id>.*?<EpisodeName>([^<]*)</EpisodeName>.*?<FirstAired>([^<]*)</FirstAired>.*?<absolute_number>([0-9]+)</absolute_number>.*?</Episode></expression> + </RegExp> + <!-- If Using "absolute ordering" then scrape episodes without absolute_number with normal season/episode numbers to match Specials and other cases --> + <RegExp conditional="absolutenumber" input="$$1" output="<episode><title>\2</title><url cache="$$10.xml">$$2</url><epnum>\3</epnum><season>\5</season><id>\1</id><aired>\4</aired></episode>" dest="4+"> + <expression repeat="yes"><Episode>.*?<id>([0-9]+).*?<EpisodeName>([^<]*).*?<EpisodeNumber>([0-9]+)[^<]*.*?<FirstAired<([^<]*)</FirstAired>.*?<SeasonNumber>(0)</SeasonNumber>.*?<absolute_number></absolute_number>.*?</Episode></expression> + </RegExp> + </RegExp> + + <!-- If not "dvdordering" and not "absolute ordering" ... well use normal regexp ;) --> + <RegExp conditional="!absolutenumber" input="$$1" output="<episode><title>\2</title><url cache="$$10.xml">$$2</url><epnum>\3</epnum><season>\5</season><id>\1</id><aired>\4</aired></episode>" dest="4"> + <expression repeat="yes"><Episode>.*?<id>([0-9]+).*?<EpisodeName>([^<]*).*?<EpisodeNumber>([0-9]+)[^<]*.*?<FirstAired>([^<]*)</FirstAired>.*?<SeasonNumber>([0-9]+)[^<]*.*?</Episode></expression> + </RegExp> + </RegExp> + <!-- ... Or use "dvdordering" --> + <RegExp conditional="dvdorder" input="$$1" output="<episode><title>\4</title><url cache="$$10.xml">$$2</url><epnum>\2</epnum><season>\3</season><id>\1</id><aired>\5</aired></episode>" dest="4"> + <expression repeat="yes"><Episode>.*?<id>([0-9]+).*?<Combined_episodenumber>([0-9]+).*?<Combined_season>([0-9]+).*?<EpisodeName>([^<]*).*?<FirstAired>([^<]*)</FirstAired>.*?</Episode></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetEpisodeList> + + <!-- input: $1=html --> + <!-- returns: results in xml format <details><writer>*</writer><director>*</director><cast>*</cast><rating>*</rating><rank>*</rank><plot>*</plot> --> + <GetEpisodeDetails dest="3"> + <RegExp input="$$4" output="<?xml version="1.0" encoding="utf-8" standalone="yes"?><details>\1</details>" dest="3"> + <RegExp input="$$1" output="\1" dest="8"> + <expression noclean="1"><Episode>.*?<id>$$2</id>(.*?)</Episode></expression> + </RegExp> + <RegExp input="$$8" output="<plot>\1</plot>" dest="4"> + <expression><Overview>([^<]*)</Overview></expression> + </RegExp> + <RegExp input="$$8" output="\1" dest="6"> + <expression noclean="1"><Writer>([^<]*)</Writer></expression> + </RegExp> + <RegExp input="$$6" output="<credits>\1</credits>" dest="4+"> + <expression repeat="yes">([^\|]*)\|</expression> + </RegExp> + <RegExp input="$$6" output="<credits>\1</credits>" dest="4+"> + <expression repeat="yes">([^,]*),</expression> + </RegExp> + <RegExp input="$$6" output="<credits>\1</credits>" dest="4+"> + <expression>([^\|,]+)$</expression> + </RegExp> + <RegExp input="$$8" output="\1" dest="6"> + <expression noclean="1"><Director>([^<]*)</Director></expression> + </RegExp> + <RegExp input="$$6" output="<director>\1</director>" dest="4+"> + <expression repeat="yes">([^\|]*)\|</expression> + </RegExp> + <RegExp input="$$6" output="<director>\1</director>" dest="4+"> + <expression repeat="yes">([^,]*),</expression> + </RegExp> + <RegExp input="$$6" output="<director>\1</director>" dest="4+"> + <expression>([^\|,]+)$</expression> + </RegExp> + <RegExp input="$$8" output="\1" dest="6"> + <expression noclean="1"><GuestStars>([^<]*)</GuestStars></expression> + </RegExp> + <RegExp input="$$6" output="<actor><name>\1</name></actor>" dest="4+"> + <expression repeat="yes">([^\|]*)\|</expression> + </RegExp> + <RegExp input="$$6" output="<actor><name>\1</name></actor>" dest="4+"> + <expression repeat="yes">([^,]*),</expression> + </RegExp> + <RegExp input="$$6" output="<actor><name>\1</name></actor>" dest="4+"> + <expression>([^\|,]+)$</expression> + </RegExp> + <RegExp input="$$8" output="<title>\1</title>" dest="4+"> + <expression><EpisodeName>([^<]*)</EpisodeName></expression> + </RegExp> + <!-- Absolute Numbering --> + <RegExp conditional="absolutenumber" input="$$8" output="<season>1</season>" dest="4+"> + <expression></expression> + </RegExp> + <RegExp conditional="absolutenumber" input="$$8" output="<episode>\1</episode>" dest="4+"> + <expression><absolute_number>([^<]*)</absolute_number></expression> + </RegExp> + <!-- Normal Season/Episodes --> + <RegExp conditional="!absolutenumber" input="$$8" output="<season>\1</season>" dest="4+"> + <expression><SeasonNumber>([^<]*)</SeasonNumber></expression> + </RegExp> + <RegExp conditional="!absolutenumber" input="$$8" output="<episode>\1</episode>" dest="4+"> + <expression><EpisodeNumber>([^<]*)</EpisodeNumber></expression> + </RegExp> + + <RegExp input="$$8" output="<thumb>http://thetvdb.com/banners/\1</thumb>" dest="4+"> + <expression><filename>([^<]+)</filename></expression> + </RegExp> + <RegExp input="$$8" output="<aired>\1</aired>" dest="4+"> + <expression><FirstAired>([^<]+)</FirstAired></expression> + </RegExp> + <RegExp input="$$8" output="<displayseason>\1</displayseason>" dest="4+"> + <expression><airsbefore_season>([^<]+)</airsbefore_season></expression> + </RegExp> + <RegExp input="$$8" output="<displayepisode>\1</displayepisode>" dest="4+"> + <expression><airsbefore_episode>([^<]+)</airsbefore_episode></expression> + </RegExp> + <RegExp input="$$8" output="<displayafterseason>\1</displayafterseason>" dest="4+"> + <expression><airsafter_season>([^<]+)</airsafter_season></expression> + </RegExp> + <RegExp input="$$8" output="<rating>\1</rating>" dest="4+"> + <expression><Rating>([^<]+)</Rating></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetEpisodeDetails> + + <GetLanguages dest="3"> + <RegExp input="$$2" output="<settings><setting label="Language" type="labelenum" id="language" values="$$8" sort="yes" default="en"></setting></settings>" dest="3"> + <RegExp input="$$1" output="\1|" dest="8+"> + <expression repeat="yes"><abbreviation>([^<]*)</abbreviation></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetLanguages> +</scraper> diff --git a/system/scrapers/video/tvrage.jpg b/system/scrapers/video/tvrage.jpg Binary files differnew file mode 100644 index 0000000000..b8b22d6351 --- /dev/null +++ b/system/scrapers/video/tvrage.jpg diff --git a/system/scrapers/video/tvrage.xml b/system/scrapers/video/tvrage.xml new file mode 100644 index 0000000000..c90a929a48 --- /dev/null +++ b/system/scrapers/video/tvrage.xml @@ -0,0 +1,100 @@ +<?xml version="1.0" encoding="UTF-8"?> +<scraper framework="1.0" date="2009-05-22" name="TV Rage" content="tvshows" thumb="tvrage.jpg" language="en"> + <CreateSearchUrl dest="3"> + <RegExp input="$$1" output="<url>http://www.tvrage.com/search.php?search=\1</url>" dest="3"> + <expression></expression> + </RegExp> + </CreateSearchUrl> + <GetSearchResults dest="3"> + <RegExp input="$$4" output="<results>\1</results>" dest="3"> + <RegExp input="$$1" output="<entity><title>\2</title><url>http://www.tvrage.com/\1</url></entity>" dest="4"> + <expression repeat="yes"><td class='b1'><img[^>]*> <a[^h]*href='http://www\.tvrage\.com/([^']*)' >([^<]*)</a></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetSearchResults> + <GetDetails dest="7"> + <RegExp input="$$5" output="<details>\1</details>" dest="7"> + <RegExp input="$$1" output="<title>\1</title>" dest="5"> + <expression></a>"([^"]*)" Summary</h5></expression> + </RegExp> + <RegExp input="$$1" output="<plot>\1</plot>" dest="5+"> + <expression>\n</td></tr></table>(.*)<br>&nbsp;</div></td></tr></expression> + </RegExp> + <RegExp input="$$1" output="<rating>\1</rating><votes>\2</votes>" dest="5+"> + <expression><strong>([0-9.]+)/10 \(([0-9]*) Votes cast\)</td></expression> + </RegExp> + <RegExp input="$$1" output="<actor><name>\1</name><role>\2</role></actor>" dest="5+"> + <expression repeat="yes">a href='/person/[^>]*>([^(<|\(|/)]*)</a></i></span></td><td[^>]*><b>played</b></td><td[^>]*><div[^>]*><i>([^(<|\(|/)]*)</expression> + </RegExp> + <RegExp input="$$1" output="<thumb>\1</thumb>" dest="5+"> + <expression>(http://images\.tvrage\.net/shows/[0-9/]+\.(gif|jpg|png))</expression> + </RegExp> + <RegExp input="$$1" output="<status>\1</status>" dest="5+"> + <expression trim="1">Status: </b></td><td>([^<]*)</td></expression> + </RegExp> + <RegExp input="$$1" output="<premiered>\1</premiered>" dest="5+"> + <expression trim="1">Premiere: </b></td><td>([^<]*)</td></expression> + </RegExp> + <RegExp input="$$1" output="<runtime>\1</runtime>" dest="5+"> + <expression trim="1">Runtime: </b></td><td>([^<]*)</td></expression> + </RegExp> + <RegExp input="$$6" output="<genre>\1</genre>" dest="5+"> + <RegExp input="$$1" output="| \1" dest="6"> + <expression noclean="1">Genre: </b></td><td>([^<]*)</td></expression> + </RegExp> + <expression repeat="yes">\| ([^\|]*)</expression> + </RegExp> + <RegExp input="$$1" output="<episodeguide><url>http://www.tvrage.com\1all</url></episodeguide>" dest="5+"> + <expression trim="1">href='([^']*)'>Episode List</a><br></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetDetails> + <GetEpisodeList dest="3"> + <RegExp input="$$5" output="<episodeguide>\1</episodeguide>" dest="3"> + <RegExp input="$$1" output="<episode><url>http://www.tvrage.com\1</url><epnum>\3</epnum><season>\2</season></episode>" dest="5+"> + <expression repeat="yes">class='b1'><a href='([^']*)'>([0-9]+)+x([0-9]+)</i></a></td></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetEpisodeList> + <GetEpisodeDetails dest="3"> + <RegExp input="$$5" output="<details>\1</details>" dest="3"> + <RegExp input="$$1" output="<title>\1</title>" dest="5"> + <expression>Title: </b></td><td class='b2'>([^<]*)</td></expression> + </RegExp> + <RegExp input="$$1" output="<plot>\1</plot>" dest="5+"> + <expression><td><table width='100%'><tr><td valign='top'><table width='100%'><tr><td>([^<]*)</expression> + </RegExp> + <RegExp input="$$1" output="<rating>\1</rating><votes>\2</votes>" dest="5+"> + <expression><strong>([0-9.]+)/10 \(([0-9]*) Votes cast\)</td></expression> + </RegExp> + <RegExp input="$$1" output="<aired>\1</aired>" dest="5+"> + <expression>Original Airdate: </b></td><td class='b2'>([^<]*)</expression> + </RegExp> + <RegExp input="$$1" output="<thumb>\1</thumb>" dest="5+"> + <expression>(http://images\.tvrage\.net/screencaps/[0-9/]*\.(jpg|gif|png))</expression> + </RegExp> + <RegExp input="$$1" output="<code>\1</code>" dest="5+"> + <expression>Production Number: </b></td><td class='b2'>([^<]*)</td></expression> + </RegExp> + <RegExp input="$$1" output="<actor><name>\1</name><role>\2</role></actor>" dest="5+"> + <expression repeat="yes">a href='/person/[^>]*>([^(<|\(|/)]*)</a></i></td><td[^>]*><b>played</b></td><td><i>([^(<|\(|/)]*)</expression> + </RegExp> + <RegExp input="$$2" output="<director>\1</director>" dest="5+"> + <RegExp input="$$1" output="| \1" dest="2"> + <expression noclean="1">Director: </b></td><td[^>]*><i><a href='/person/[^>]*>([^(<|\(|/)]*)</a></i></expression> + </RegExp> + <expression repeat="yes">\| ([^\|]*)</expression> + </RegExp> + <RegExp input="$$2" output="<credits>\1</credits>" dest="5+"> + <RegExp input="$$1" output="| \1" dest="2"> + <expression noclean="1">Writer: </b></td><td[^>]*><i><a href='/person/[^>]*>([^(<|\(|/)]*)</a></i></expression> + </RegExp> + <expression repeat="yes">\| ([^\|]*)</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetEpisodeDetails> +</scraper> diff --git a/system/scrapers/video/worldart.gif b/system/scrapers/video/worldart.gif Binary files differnew file mode 100644 index 0000000000..f91a81969a --- /dev/null +++ b/system/scrapers/video/worldart.gif diff --git a/system/scrapers/video/worldart.xml b/system/scrapers/video/worldart.xml new file mode 100644 index 0000000000..22a2c87c97 --- /dev/null +++ b/system/scrapers/video/worldart.xml @@ -0,0 +1,256 @@ +<?xml version="1.0" encoding="windows-1251" standalone="yes"?>
+<scraper framework="1.1" date="2009-09-13" content="movies" name="World Art Movies (Russian)" thumb="worldart.gif" language="ru">
+ <include>common/tmdb.xml</include>
+ <include>common/imdb.xml</include>
+ <GetSettings dest="3">
+ <RegExp input="$$5" output="<settings>\1</settings>" dest="3">
+ <RegExp input="$$1" output="<setting label="Постер из TMDB (Иначе стандартный)" type="bool" id="thepos" default="true"></setting>" dest="5+">
+ <expression/>
+ </RegExp>
+ <RegExp input="$$1" output="<setting label="Фанарт из TMDB" type="bool" id="fanart" default="true"></setting>" dest="5+">
+ <expression/>
+ </RegExp>
+ <RegExp input="$$1" output="<setting label="Актеры из IMDB (Рекомендованно)" type="bool" id="imdbcast" default="true"></setting>" dest="5+">
+ <expression/>
+ </RegExp>
+ <RegExp input="$$1" output="<setting label="Директор из IMDB" type="bool" id="imdbdir" default="false"></setting>" dest="5+">
+ <expression/>
+ </RegExp>
+ <RegExp input="$$1" output="<setting label="Сценаристы из IMDB" type="bool" id="imdbcred" default="false"></setting>" dest="5+">
+ <expression/>
+ </RegExp>
+ <expression noclean="1"/>
+ </RegExp>
+ </GetSettings>
+
+ <CreateSearchUrl dest="3" SearchStringEncoding="CP1251">
+ <RegExp input="$$1" output="http://www.world-art.ru/search.php?name=\1&global_sector=all" dest="3">
+ <expression noclean="1"/>
+ </RegExp>
+ </CreateSearchUrl>
+
+ <GetSearchResults dest="8">
+ <RegExp input="$$5" output="<?xml version="1.0" encoding="windows-1251" standalone="yes"?><results>\1</results>" dest="8">
+ <RegExp input="$$1" output="<entity><title>\4</title><url>http://www.world-art.ru/\1/\1.php?id=\3</url><id>\3</id></entity>" dest="5+">
+ <expression repeat="yes">"([^/]+)/([^=]+)=([^"]+)" class='estimation'>([^<]+)<</expression>
+ </RegExp>
+ <expression noclean="1"/>
+ </RegExp>
+ </GetSearchResults>
+
+ <GetDetails dest="3">
+ <RegExp input="$$5" output="<details>\1</details>" dest="3">
+ <!-- ID -->
+ <RegExp input="$$2" output="<id>\1</id>" dest="5">
+ <expression/>
+ </RegExp>
+ <!-- TITLE -->
+ <RegExp input="$$1" output="<title>\1</title>" dest="5+">
+ <expression trim="1">name='description' content='([^']+)'</expression>
+ </RegExp>
+
+ <!-- TYPE -->
+ <RegExp input="$$1" output="\1" dest="7">
+ <expression trim="1">/([^/]+)/review.php</expression>
+ </RegExp>
+
+ <!-- MPAA -->
+ <RegExp input="$$1" output="<url cache="\1-imdb.html" function="MPAA">http://www.imdb.com/title/tt\1</url>" dest="5+">
+ <expression>imdb.com/title/tt([0-9]*)</expression>
+ </RegExp>
+
+ <!-- TOP250 -->
+ <RegExp input="$$1" output="<url cache="\1-imdb.html" function="TOP">http://www.imdb.com/title/tt\1</url>" dest="5+">
+ <expression>imdb.com/title/tt([0-9]*)</expression>
+ </RegExp>
+
+ <!-- YEAR -->
+ <RegExp input="$$1" output="<year>\1</year>" dest="5+">
+ <expression>year=([^"]+)"</expression>
+ </RegExp>
+
+ <!-- RUNTIME -->
+ <RegExp input="$$1" output="<runtime>\1 min</runtime>" dest="5+">
+ <expression noclean="1">полнометражный фильм, ([^м]+) мин</expression>
+ </RegExp>
+ <RegExp input="$$1" output="<url cache="\1-imdb.html" function="RUNTIME">http://www.imdb.com/title/tt\1</url>" dest="5+">
+ <expression noclean="1" clear="no">imdb.com/title/tt([0-9]*)</expression>
+ </RegExp>
+
+ <!-- RATING -->
+ <RegExp input="$$1" output="<url function="GetRating">http://www.world-art.ru/$$7/votes_history.php?id=$$2</url>" dest="5+">
+ <expression/>
+ </RegExp>
+ <RegExp input="$$1" output="<url cache="\1-imdb.html" function="GetRatingAndVotes">http://www.imdb.com/title/tt\1</url>" dest="5+">
+ <expression noclean="1" clear="no">imdb.com/title/tt([0-9]*)</expression>
+ </RegExp>
+
+ <!-- GENRE -->
+ <RegExp input="$$1" output="<genre>\2</genre>" dest="5+">
+ <expression repeat="yes">genre=([^>]+)>([^<]+)<</expression>
+ </RegExp>
+
+ <!-- STUDIO -->
+ <RegExp input="$$1" output="<url function="GetStudios">http://www.world-art.ru/$$7/$$7_full_production.php?id=$$2</url>" dest="5+">
+ <expression/>
+ </RegExp>
+
+ <RegExp input="$$1" output="<url cache="\1-imdb.html" function="GetIMDBStudios">http://www.imdb.com/title/tt\1</url>" dest="5+">
+ <expression noclean="1" clear="no">imdb.com/title/tt([0-9]*)</expression>
+ </RegExp>
+
+ <!-- DIRECTORS, WRITERS, ACTORS -->
+ <RegExp input="$$2" output="<url function="GetPeople">http://www.world-art.ru/$$7/$$7_full_cast.php?id=$$2</url>" dest="5+">
+ <expression/>
+ </RegExp>
+ <RegExp conditional="imdbcast" input="$$1" output="<url cache="\1-fullcredits.html" function="GetIMDBCast">http://www.imdb.com/title/tt\1/fullcredits</url>"dest="5+">
+ <expression noclean="1" clear="no">imdb.com/title/tt([0-9]*)</expression>
+ </RegExp>
+ <RegExp conditional="imdbdir" input="$$1" output="<url cache="\1-fullcredits.html" function="GetIMDBDirectors">http://www.imdb.com/title/tt\1/fullcredits</url>"dest="5+">
+ <expression noclean="1" clear="no">imdb.com/title/tt([0-9]*)</expression>
+ </RegExp>
+ <RegExp conditional="imdbcred" input="$$1" output="<url cache="\1-fullcredits.html" function="GetIMDBCastWriters">http://www.imdb.com/title/tt\1/fullcredits</url>"dest="5+">
+ <expression noclean="1" clear="no">imdb.com/title/tt([0-9]*)</expression>
+ </RegExp>
+
+ <!-- OUTLINE, PLOT -->
+ <RegExp input="$$1" output="\1&#133;&hellip;&laquo;&#151;&mdash;&nbsp;" dest="2">
+ <expression noclean="1">class='review'>[\t\n]*(.*?)[\t\n]*</p></expression>
+ </RegExp>
+ <RegExp input="$$2" output="<outline>\1</outline><plot>\1</plot>" dest="5+">
+ <RegExp input="$$2" output="\1 " dest="2">
+ <expression noclean="1" repeat="yes">(.*?)&nbsp;</expression>
+ </RegExp>
+ <RegExp input="$$2" output="\1-" dest="2">
+ <expression noclean="1" repeat="yes">(.*?)&mdash;</expression>
+ </RegExp>
+ <RegExp input="$$2" output="\1-" dest="2">
+ <expression noclean="1" repeat="yes">(.*?)&#151;</expression>
+ </RegExp>
+ <RegExp input="$$2" output="\1"" dest="2">
+ <expression noclean="1" repeat="yes">(.*?)&[lr]aquo;</expression>
+ </RegExp>
+ <RegExp input="$$2" output="\1..." dest="2">
+ <expression noclean="1" repeat="yes">(.*?)&hellip;</expression>
+ </RegExp>
+ <RegExp input="$$2" output="\1..." dest="2">
+ <expression noclean="1" repeat="yes">(.*?)&#133;</expression>
+ </RegExp>
+ <expression trim="1">(.*)\.{3}?</expression>
+ </RegExp>
+
+ <!-- THUMBS -->
+ <RegExp input="$$1" output="<url function="GetCoverThumbs">http://www.world-art.ru/$$7/$$7_poster.php?id=\1</url>" dest="5+">
+ <expression noclean="1">id=([0-9]*)></expression>
+ </RegExp>
+ <RegExp conditional="thepos" input="$$1" output="<url function="GetTMDBThumbsByIMDBId">http://api.themoviedb.org/2.0/Movie.imdbLookup?imdb_id=tt\1&amp;api_key=57983e31fb435df4df77afb854740ea9</url>" dest="5+">
+ <expression noclean="1" clear="no">imdb.com/title/tt([0-9]*)</expression>
+ </RegExp>
+
+ <!-- FANART -->
+ <RegExp conditional="fanart" input="$$1" output="<url function="GetTMDBFanartByIMDBId">http://api.themoviedb.org/2.0/Movie.imdbLookup?imdb_id=tt\1&amp;api_key=57983e31fb435df4df77afb854740ea9</url>" dest="5+">
+ <expression>imdb.com/title/tt([0-9]*)</expression>
+ </RegExp>
+ <expression noclean="1"/>
+ </RegExp>
+ </GetDetails>
+
+ <!-- MPAA -->
+ <MPAA dest="5">
+ <RegExp input="$$1" output="<details><mpaa>\1</mpaa></details>" dest="5+">
+ <expression trim="1">MPAA</a>:</h5>(.[^<]*)</expression>
+ </RegExp>
+ </MPAA>
+
+ <!-- TOP250 -->
+ <TOP dest="5">
+ <RegExp input="$$1" output="<details><top250>\1</top250></details>" dest="5+">
+ <expression trim="1">Top 250: #([0-9]*)</a></expression>
+ </RegExp>
+ </TOP>
+
+ <!-- RUNTIME -->
+ <RUNTIME dest="5">
+ <RegExp input="$$1" output="<details><runtime>\1</runtime></details>" dest="5+">
+ <expression trim="1"><h5>Runtime:</h5>[^0-9]*([^<]*)</expression>
+ </RegExp>
+ </RUNTIME>
+
+ <!-- RATINGS -->
+ <GetRating dest="5">
+ <RegExp input="$$1" output="<details><rating>\1</rating><votes>10</votes></details>" dest="5+">
+ <expression trim="1">Средний рейтинг:</b> ([^,]+),</expression>
+ </RegExp>
+ </GetRating>
+
+ <GetRatingAndVotes dest="5">
+ <RegExp input="$$1" output="<details><rating>\1</rating><votes>\2</votes></details>" dest="5+">
+ <expression trim="1"><b>([0-9.]+)/10</b>[^<]*<a href="ratings" class="tn15more">([0-9,]+) votes</a></expression>
+ </RegExp>
+ </GetRatingAndVotes>
+
+ <!-- STUDIOS -->
+ <GetStudios dest="5">
+ <RegExp input="$$1" output="<details><studio>\1</studio></details>" dest="5+">
+ <expression repeat="yes">class='estimation'>([^<]+)<</expression>
+ </RegExp>
+ </GetStudios>
+
+ <GetIMDBStudios dest="5">
+ <RegExp input="$$1" output="<details><studio>\1</studio></details>" dest="5+">
+ <expression repeat="yes">"/company/[^/]*/">([^<]*)</a></expression>
+ </RegExp>
+ </GetIMDBStudios>
+
+ <!-- ACTORS -->
+ <GetPeople dest="5">
+ <RegExp input="$$7" output="<details>\1</details>" dest="5+">
+ <RegExp input="$$8" output="<actor><name>\1</name></actor>" dest="7+">
+ <RegExp input="$$1" output="\1" dest="8+">
+ <expression noclean="1">В ролях:(.*?)bgcolor=#5D0E0E</expression>
+ </RegExp>
+ <expression repeat="yes">class='estimation'>([^<]+)<</expression>
+ </RegExp>
+ <expression noclean="1"/>
+ </RegExp>
+ </GetPeople>
+
+ <!-- DIRECTORS -->
+ <GetDirector dest="5">
+ <RegExp input="$$7" output="<details>\1</details>" dest="5+">
+ <RegExp input="$$8" output="<director>\1</director>" dest="7+">
+ <RegExp input="$$1" output="\1" dest="8+">
+ <expression noclean="1"> color=#990000(.*?) color=#990000</expression>
+ </RegExp>
+ <expression repeat="yes">class='estimation'>([^<]+)<</expression>
+ </RegExp>
+ <expression noclean="1"/>
+ </RegExp>
+ </GetDirector>
+
+ <!-- WRITERS -->
+ <GetCredits dest="5">
+ <RegExp input="$$7" output="<details>\1</details>" dest="5+">
+ <RegExp input="$$8" output="<credits>\1</credits>" dest="7+">
+ <RegExp input="$$1" output="\1" dest="8+">
+ <expression noclean="1">Сценарий:(.*?):</b></expression>
+ </RegExp>
+ <expression repeat="yes">class='estimation'>([^<]+)<</expression>
+ </RegExp>
+ <expression noclean="1"/>
+ </RegExp>
+ </GetCredits>
+
+ <!-- POSTERS -->
+ <GetCoverThumbs dest="5">
+ <RegExp input="$$8" output="<details>\1</details>" dest="5+">
+ <RegExp input="$$1" output="\1" dest="7">
+ <expression trim="1">/([^/]+)/review.php</expression>
+ </RegExp>
+ <RegExp input="$$1" output="<thumb preview="http://www.world-art.ru/$$7/img/\1/\3/\4.jpg">http://www.world-art.ru/$$7/img/\1/\3/\4.jpg</thumb>" dest="8+">
+ <expression repeat="yes" noclean="1">img/converted_images_(.*?)/(.*?)/(.*?)-(.*?)-(.*?)</expression>
+ </RegExp>
+ <expression noclean="1"/>
+ </RegExp>
+ </GetCoverThumbs>
+</scraper>
diff --git a/system/scrapers/video/yahoomusic.png b/system/scrapers/video/yahoomusic.png Binary files differnew file mode 100644 index 0000000000..007c422028 --- /dev/null +++ b/system/scrapers/video/yahoomusic.png diff --git a/system/scrapers/video/yahoomusic.xml b/system/scrapers/video/yahoomusic.xml new file mode 100644 index 0000000000..b8c3138bd9 --- /dev/null +++ b/system/scrapers/video/yahoomusic.xml @@ -0,0 +1,50 @@ +<?xml version="1.0" encoding="UTF-8"?> +<scraper framework="1.0" date="2009-05-22" name="Yahoo! Music" content="musicvideos" thumb="yahoomusic.png" language="en"> + <GetSettings dest="3"> + <RegExp input="$$5" output="<settings>\1</settings>" dest="3"> + <RegExp input="$$1" output="<setting label="Thumb(s) Size" type="labelenum" values="192|256|384|512|1024" id="yahooscale" default="384"></setting>" dest="5+"> + <expression></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetSettings> + <CreateSearchUrl dest="3"> + <RegExp input="$$1" output="<url>http://us.music.yahooapis.com/video/v1/list/search/all/\1?appid=r6QWQGXV34H0Sl3Wq4vWJI2v9IO1XnZw5JkeifPbTq2p0S9H9kg0JUc3SVN0Yw--&amp;response=artists,images</url>" dest="3"> + <expression noclean="1"></expression> + </RegExp> + </CreateSearchUrl> + <GetSearchResults dest="8"> + <RegExp input="$$5" output="<?xml version="1.0" encoding="iso-8859-1" standalone="yes"?><results>\1</results>" dest="8"> + <RegExp input="$$1" output="<entity><title>\4 - \1 (\2)</title><url>http://us.music.yahooapis.com/video/v1/item/\3?format=xml&amp;appid=r6QWQGXV34H0Sl3Wq4vWJI2v9IO1XnZw5JkeifPbTq2p0S9H9kg0JUc3SVN0Yw--</url></entity>" dest="5"> + <expression repeat="yes"><Video.*?title="([^"]*)".*?copyrightYear="([^"]*)".*?\/([0-9]*)\?size.*?name="([^"]*)".*?</Video></expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetSearchResults> + <GetDetails dest="3"> + <RegExp input="$$5" output="<details>\1</details>" dest="3"> + <RegExp input="$$1" output="<title>\1</title>" dest="5"> + <expression><Video.*?title="([^"]*)"</expression> + </RegExp> + <RegExp input="$$1" output="<artist>\1</artist>" dest="5+"> + <expression repeat="yes"><Artist.*?name="([^"]*)"</expression> + </RegExp> + <RegExp input="$$1" output="<album>\1</album>" dest="5+"> + <expression><Release.*?title="([^"]*)"</expression> + </RegExp> + <RegExp input="$$1" output="<studio>\1</studio>" dest="5+"> + <expression repeat="yes">label="([^"]*)"</expression> + </RegExp> + <RegExp input="$$1" output="<year>\1</year>" dest="5+"> + <expression>copyrightYear="([0-9]*)"</expression> + </RegExp> + <RegExp input="$$1" output="<genre>\1</genre>" dest="5+"> + <expression repeat="yes"><Category.*?name="([^"]*)" rating="([^"]*)" type="Genre"></expression> + </RegExp> + <RegExp input="$$1" output="<thumb>http://d.yimg.com/img.music.yahoo.com/image/v1/video/\1?size=$INFO[yahooscale]</thumb>" dest="5+"> + <expression>\/([0-9]*)\?size</expression> + </RegExp> + <expression noclean="1"></expression> + </RegExp> + </GetDetails> +</scraper> |
