diff options
| author | nadasb <nadasb@svn> | 2010-10-11 19:46:49 +0000 |
|---|---|---|
| committer | nadasb <nadasb@svn> | 2010-10-11 19:46:49 +0000 |
| commit | 86ac17e549023b61417519b62d872b550da54de0 (patch) | |
| tree | 97e8db7ea45b371031d597ffd99e370c4493ec43 /addons/metadata.common.imdb.com | |
| parent | c5155ccc852779b9c26ee1bd1ef6fa182d69719f (diff) | |
sync with scrapers git
git-svn-id: https://xbmc.svn.sourceforge.net/svnroot/xbmc/trunk@34690 568bbfeb-2a22-0410-94d2-cc84cf5bfa90
Diffstat (limited to 'addons/metadata.common.imdb.com')
| -rw-r--r-- | addons/metadata.common.imdb.com/addon.xml | 29 | ||||
| -rw-r--r-- | addons/metadata.common.imdb.com/changelog.txt | 20 | ||||
| -rw-r--r-- | addons/metadata.common.imdb.com/imdb.xml | 205 |
3 files changed, 254 insertions, 0 deletions
diff --git a/addons/metadata.common.imdb.com/addon.xml b/addons/metadata.common.imdb.com/addon.xml new file mode 100644 index 0000000000..4b08fd17a4 --- /dev/null +++ b/addons/metadata.common.imdb.com/addon.xml @@ -0,0 +1,29 @@ +<?xml version="1.0" encoding="UTF-8" standalone="yes"?> +<addon id="metadata.common.imdb.com" + name="IMDB common scraper functions" + version="2.0.5" + provider-name="Team XBMC"> + <requires> + <import addon="xbmc.metadata" version="1.0"/> + </requires> + <extension point="xbmc.metadata.scraper.library" + library="imdb.xml"/> + <extension point="xbmc.addon.metadata"> + <platform>all</platform> + <minversion>20000</minversion> + <summary lang="en">IMDB Scraper Library</summary> + <summary lang="hu">IMDB leolvasó-könyvtár</summary> + <summary lang="kr">IMDB 스크래퍼 라이브러리</summary> + <summary lang="nl">IMDB Scraper Bibliotheek</summary> + <summary lang="pl">Scraper IMDb</summary> + <summary lang="pt">Scraper de filmes IMDb</summary> + <summary lang="se">IMDB Skrapbibliotek</summary> + <description lang="en">Download Movie information from www.imdb.com</description> + <description lang="hu">Film információk letöltése a www.imdb.com webhelyről</description> + <description lang="kr">www.imdb.com 에서 영화 정보 다운로드</description> + <description lang="nl">Download film informatie van www.imdb.com</description> + <description lang="pl">Pobieraj informacje o filmach z www.imdb.com</description> + <description lang="pt">Descarregar informação de filmes de www.imdb.com</description> + <description lang="se">Ladda ner filminformation från www.imdb.com</description> + </extension> +</addon> diff --git a/addons/metadata.common.imdb.com/changelog.txt b/addons/metadata.common.imdb.com/changelog.txt new file mode 100644 index 0000000000..cd4bd77c6d --- /dev/null +++ b/addons/metadata.common.imdb.com/changelog.txt @@ -0,0 +1,20 @@ +[B]2.0.5[/B] +- changed: move scraping of IMDb rating out to IMDb scraper library + +[B]2.0.4[/B] +- reverted: purge fixchars from all functions, since it cause convert fails on Beta2 and Linux + +[B]2.0.3[/B] +- reverted: fixchars removed from cast parsing, leads to Dharma Beta2 hanging, will be added back once next Beta is out + +[B]2.0.2[/B] +- fixed: some more cast won't get scraped under certain circumtances + +[B]2.0.1[/B] +- fixed: cast won't get scraped under certain circumtances + +[B]2.0.0[/B] +- changed: adapted new IMDb layout + +[B]1.0.5[/B] +- fixed: tack /combined onto the end of the url to work around the new layout of IMDb
\ No newline at end of file diff --git a/addons/metadata.common.imdb.com/imdb.xml b/addons/metadata.common.imdb.com/imdb.xml new file mode 100644 index 0000000000..24a3a4f422 --- /dev/null +++ b/addons/metadata.common.imdb.com/imdb.xml @@ -0,0 +1,205 @@ +<scraperfunctions> + <GetIMDBRatingById dest="5"> + <RegExp input="$$1" output="<details><url cache="$$1-main.html" function="ParseIMDBRating">http://akas.imdb.com/title/$$1/</url></details>" dest="5"> + <expression noclean="1" /> + </RegExp> + </GetIMDBRatingById> + <GetIMDBPlotById dest="5"> + <RegExp input="$$1" output="<details><url cache="$$1-main.html" function="ParseIMDBPlot">http://akas.imdb.com/title/$$1/</url></details>" dest="5"> + <expression noclean="1" /> + </RegExp> + </GetIMDBPlotById> + <GetIMDBCastById dest="5"> + <RegExp input="$$1" output="<details><url cache="$$1-main.html" function="ParseIMDBCast">http://akas.imdb.com/title/$$1/</url></details>" dest="5"> + <expression noclean="1" /> + </RegExp> + </GetIMDBCastById> + <GetIMDBScaledCastById dest="5"> + <RegExp input="$$1" output="<details><url cache="$$1-main.html" function="ParseIMDBScaledCast">http://akas.imdb.com/title/$$1/</url></details>" dest="5"> + <expression noclean="1" /> + </RegExp> + </GetIMDBScaledCastById> + <GetIMDBDirectorsById dest="5"> + <RegExp input="$$1" output="<details><url cache="$$1-main.html" function="ParseIMDBDirectors">http://akas.imdb.com/title/$$1/</url></details>" dest="5"> + <expression noclean="1" /> + </RegExp> + </GetIMDBDirectorsById> + <GetIMDBWritersById dest="5"> + <RegExp input="$$1" output="<details><url cache="$$1-main.html" function="ParseIMDBWriters">http://akas.imdb.com/title/$$1/</url></details>" dest="5"> + <expression noclean="1" /> + </RegExp> + </GetIMDBWritersById> + <GetIMDBFullCastById dest="5"> + <RegExp input="$$1" output="<details><url cache="$$1-combined.html" function="ParseIMDBFullCast">http://akas.imdb.com/title/$$1/combined</url></details>" dest="5"> + <expression noclean="1" /> + </RegExp> + </GetIMDBFullCastById> + <GetIMDBScaledFullCastById dest="5"> + <RegExp input="$$1" output="<details><url cache="$$1-combined.html" function="ParseIMDBScaledFullCast">http://akas.imdb.com/title/$$1/combined</url></details>" dest="5"> + <expression noclean="1" /> + </RegExp> + </GetIMDBScaledFullCastById> + <GetIMDBFullDirectorsById dest="5"> + <RegExp input="$$1" output="<details><url cache="$$1-combined.html" function="ParseIMDBFullDirectors">http://akas.imdb.com/title/$$1/combined</url></details>" dest="5"> + <expression noclean="1" /> + </RegExp> + </GetIMDBFullDirectorsById> + <GetIMDBFullWritersById dest="5"> + <RegExp input="$$1" output="<details><url cache="$$1-combined.html" function="ParseIMDBFullWriters">http://akas.imdb.com/title/$$1/combined</url></details>" dest="5"> + <expression noclean="1" /> + </RegExp> + </GetIMDBFullWritersById> + <GetIMDBThumbsById dest="5"> + <RegExp input="$$1" output="<details><url cache="$$1-posters.html" function="ParseIMDBThumbs">http://akas.imdb.com/title/$$1/posters</url></details>" dest="5"> + <expression noclean="1" /> + </RegExp> + </GetIMDBThumbsById> + <GetIMDBTrailerById dest="5"> + <RegExp input="$$1" output="<details><url cache="$$1-main.html" function="GetIMDBTrailerPage">http://akas.imdb.com/title/$$1/</url></details>" dest="5"> + <expression noclean="1" /> + </RegExp> + </GetIMDBTrailerById> + <ParseIMDBRating dest="5"> + <RegExp input="$$2" output="<details>\1</details>" dest="5"> + <RegExp input="$$1" output="<rating>\1</rating><votes>\2</votes>" dest="2"> + <expression><span class="rating-rating">([0-9.]+).*?>([0-9,]+) votes</a></expression> + </RegExp> + <expression noclean="1" /> + </RegExp> + </ParseIMDBRating> + <ParseIMDBPlot dest="5"> + <RegExp input="$$2" output="<details>\1</details>" dest="5"> + <RegExp input="$$1" output="<plot>\1</plot>" dest="2"> + <expression trim="1"><h2>Storyline</h2>\n+<p>([^<]+)</expression> + </RegExp> + <expression noclean="1" /> + </RegExp> + </ParseIMDBPlot> + <ParseIMDBCast dest="5"> + <RegExp input="$$2" output="<details>\1</details>" dest="5"> + <RegExp input="$$1" output="\1" dest="6"> + <expression noclean="1"><table class="cast_list">(.*?)</table></expression> + </RegExp> + <RegExp input="$$6" output="<actor><thumb>\1_SX512_SY512_\2</thumb><name>\3</name><role>\4</role></actor>" dest="7"> + <expression repeat="yes" clear="yes" trim="3,4" noclean="1,2"><img\n[^\n]*\n[^\n]*\n\s+src="(?:([^"]*\.)[^"]*(\.jpg))?[^>]*[^"]*"[^"]*">[^>]*>([^<]*).*?<td class="character">[^>]*>[^>]*>([^<]*)?</expression> + </RegExp> + <RegExp input="$$7" output="<actor><thumb>\1</thumb>\2</actor>" dest="2+"> + <expression repeat="yes" clear="yes" noclean="1,2,3"><actor><thumb>(?:(http.*?)|_SX[0-9]+_SY[0-9]+_)</thumb>(.*?)</actor></expression> + </RegExp> + <expression noclean="1" /> + </RegExp> + </ParseIMDBCast> + <ParseIMDBScaledCast dest="5"> + <RegExp input="$$2" output="<details>\1</details>" dest="5"> + <RegExp input="$$1" output="\1" dest="6"> + <expression noclean="1"><table class="cast_list">(.*?)</table></expression> + </RegExp> + <RegExp input="$$6" output="<actor><thumb>\1_SX$INFO[imdbscale]_SY$INFO[imdbscale]_\2</thumb><name>\3</name><role>\4</role></actor>" dest="7"> + <expression repeat="yes" clear="yes" trim="3,4" noclean="1,2"><img\n[^\n]*\n[^\n]*\n\s+src="(?:([^"]*\.)[^"]*(\.jpg))?[^>]*[^"]*"[^"]*">[^>]*>([^<]*).*?<td class="character">[^>]*>[^>]*>([^<]*)?</expression> + </RegExp> + <RegExp input="$$7" output="<actor><thumb>\1</thumb>\2</actor>" dest="2+"> + <expression repeat="yes" clear="yes" noclean="1,2,3"><actor><thumb>(?:(http.*?)|_SX[0-9]+_SY[0-9]+_)</thumb>(.*?)</actor></expression> + </RegExp> + <expression noclean="1" /> + </RegExp> + </ParseIMDBScaledCast> + <ParseIMDBDirectors dest="5"> + <RegExp input="$$2" output="<details>\1</details>" dest="5"> + <RegExp input="$$1" output="\1" dest="6"> + <expression clear="yes" noclean="1">Director[s]?.*?</h4>(.*?)<div class</expression> + </RegExp> + <RegExp input="$$6" output="<director>\1</director>" dest="2+"> + <expression repeat="yes" clear="yes"><a href="/name[^>]*>([^<]*)<</expression> + </RegExp> + <expression noclean="1" /> + </RegExp> + </ParseIMDBDirectors> + <ParseIMDBWriters dest="5"> + <RegExp input="$$2" output="<details>\1</details>" dest="5"> + <RegExp input="$$1" output="\1" dest="6"> + <expression noclean="1">Writer[s]?.*?</h4>(.*?)<div class</expression> + </RegExp> + <RegExp input="$$6" output="<credits>\1</credits>" dest="2+"> + <expression repeat="yes"><a href="/name[^>]*>([^<]*)<</expression> + </RegExp> + <expression noclean="1" /> + </RegExp> + </ParseIMDBWriters> + <ParseIMDBFullCast 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="$$6" output="<actor><thumb>\1_SX512_SY512_\2</thumb><name>\3</name><role>\5</role></actor>" dest="7"> + <expression repeat="yes" clear="yes" trim="3,5" noclean="1,2"><img src="(?:([^"]*\.)[^"]*(\.jpg))?[^>]*[^"]*"nm"><a href="[^"]*[^>]*>([^<]*)<[^"]*"ddd">([^<]<)?[^"]*"char">(.*?)</td></expression> + </RegExp> + <RegExp input="$$7" output="<actor><thumb>\1</thumb>\2</actor>" dest="2+"> + <expression repeat="yes" clear="yes" noclean="1,2,3"><actor><thumb>(?:(http.*?)|_SX[0-9]+_SY[0-9]+_)</thumb>(.*?)</actor></expression> + </RegExp> + <expression noclean="1" /> + </RegExp> + </ParseIMDBFullCast> + <ParseIMDBScaledFullCast 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="$$6" output="<actor><thumb>\1_SX$INFO[imdbscale]_SY$INFO[imdbscale]_\2</thumb><name>\3</name><role>\5</role></actor>" dest="7"> + <expression repeat="yes" clear="yes" trim="3,5" noclean="1,2"><img src="(?:([^"]*\.)[^"]*(\.jpg))?[^>]*[^"]*"nm"><a href="[^"]*[^>]*>([^<]*)<[^"]*"ddd">([^<]<)?[^"]*"char">(.*?)</td></expression> + </RegExp> + <RegExp input="$$7" output="<actor><thumb>\1</thumb>\2</actor>" dest="2+"> + <expression repeat="yes" clear="yes" noclean="1,2,3"><actor><thumb>(?:(http.*?)|_SX[0-9]+_SY[0-9]+_)</thumb>(.*?)</actor></expression> + </RegExp> + <expression noclean="1" /> + </RegExp> + </ParseIMDBScaledFullCast> + <ParseIMDBFullDirectors 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[s]?:</h5>[^>]*>(.*?)</div</expression> + </RegExp> + <RegExp input="$$6" output="<director>\1</director>" dest="2+"> + <expression repeat="yes" clear="yes"><a href="/name/[^>]*>([^<]*)<</expression> + </RegExp> + <expression noclean="1" /> + </RegExp> + </ParseIMDBFullDirectors> + <ParseIMDBFullWriters dest="5"> + <RegExp input="$$2" output="<details>\1</details>" dest="5"> + <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" /> + </RegExp> + </ParseIMDBFullWriters> + <ParseIMDBThumbs 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="$$4" output="<thumb>\1</thumb>" dest="6"> + <expression noclean="1">(.*?_SX[0-9]+_SY[0-9]+_.jpg)</expression> + </RegExp> + <expression noclean="1" /> + </RegExp> + </ParseIMDBThumbs> + <GetIMDBTrailerPage dest="5"> + <RegExp input="$$1" output="<details><url function="ParseIMDBTrailer">http://akas.imdb.com/video/imdb/vi\1/player</url></details>" dest="5"> + <expression>/vi([0-9]*)/</expression> + </RegExp> + </GetIMDBTrailerPage> + <ParseIMDBTrailer 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" /> + </RegExp> + </ParseIMDBTrailer> +</scraperfunctions> |