diff options
-rw-r--r-- | test/test_utils.py | 31 | ||||
-rwxr-xr-x | youtube_dl/YoutubeDL.py | 20 | ||||
-rw-r--r-- | youtube_dl/extractor/__init__.py | 1 | ||||
-rw-r--r-- | youtube_dl/extractor/daum.py | 17 | ||||
-rw-r--r-- | youtube_dl/extractor/turbo.py | 67 | ||||
-rw-r--r-- | youtube_dl/utils.py | 18 | ||||
-rw-r--r-- | youtube_dl/version.py | 2 |
7 files changed, 149 insertions, 7 deletions
diff --git a/test/test_utils.py b/test/test_utils.py index 70050d2b8..3efbed29d 100644 --- a/test/test_utils.py +++ b/test/test_utils.py @@ -41,6 +41,8 @@ from youtube_dl.utils import ( strip_jsonp, uppercase_escape, limit_length, + escape_rfc3986, + escape_url, ) @@ -294,5 +296,34 @@ class TestUtil(unittest.TestCase): limit_length('foo bar baz asd', 12).startswith('foo bar')) self.assertTrue('...' in limit_length('foo bar baz asd', 12)) + def test_escape_rfc3986(self): + reserved = "!*'();:@&=+$,/?#[]" + unreserved = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-_.~' + self.assertEqual(escape_rfc3986(reserved), reserved) + self.assertEqual(escape_rfc3986(unreserved), unreserved) + self.assertEqual(escape_rfc3986('тест'), '%D1%82%D0%B5%D1%81%D1%82') + self.assertEqual(escape_rfc3986('%D1%82%D0%B5%D1%81%D1%82'), '%D1%82%D0%B5%D1%81%D1%82') + self.assertEqual(escape_rfc3986('foo bar'), 'foo%20bar') + self.assertEqual(escape_rfc3986('foo%20bar'), 'foo%20bar') + + def test_escape_url(self): + self.assertEqual( + escape_url('http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavré_FD.mp4'), + 'http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavre%CC%81_FD.mp4' + ) + self.assertEqual( + escape_url('http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erklärt/Das-Erste/Video?documentId=22673108&bcastId=5290'), + 'http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erkl%C3%A4rt/Das-Erste/Video?documentId=22673108&bcastId=5290' + ) + self.assertEqual( + escape_url('http://тест.рф/фрагмент'), + 'http://тест.рф/%D1%84%D1%80%D0%B0%D0%B3%D0%BC%D0%B5%D0%BD%D1%82' + ) + self.assertEqual( + escape_url('http://тест.рф/абв?абв=абв#абв'), + 'http://тест.рф/%D0%B0%D0%B1%D0%B2?%D0%B0%D0%B1%D0%B2=%D0%B0%D0%B1%D0%B2#%D0%B0%D0%B1%D0%B2' + ) + self.assertEqual(escape_url('http://vimeo.com/56015672#at=0'), 'http://vimeo.com/56015672#at=0') + if __name__ == '__main__': unittest.main() diff --git a/youtube_dl/YoutubeDL.py b/youtube_dl/YoutubeDL.py index 553bf559b..9519594c9 100755 --- a/youtube_dl/YoutubeDL.py +++ b/youtube_dl/YoutubeDL.py @@ -28,6 +28,7 @@ from .utils import ( compat_str, compat_urllib_error, compat_urllib_request, + escape_url, ContentTooShortError, date_from_str, DateRange, @@ -1241,6 +1242,25 @@ class YoutubeDL(object): def urlopen(self, req): """ Start an HTTP download """ + + # According to RFC 3986, URLs can not contain non-ASCII characters, however this is not + # always respected by websites, some tend to give out URLs with non percent-encoded + # non-ASCII characters (see telemb.py, ard.py [#3412]) + # urllib chokes on URLs with non-ASCII characters (see http://bugs.python.org/issue3991) + # To work around aforementioned issue we will replace request's original URL with + # percent-encoded one + url = req if isinstance(req, compat_str) else req.get_full_url() + url_escaped = escape_url(url) + + # Substitute URL if any change after escaping + if url != url_escaped: + if isinstance(req, compat_str): + req = url_escaped + else: + req = compat_urllib_request.Request( + url_escaped, data=req.data, headers=req.headers, + origin_req_host=req.origin_req_host, unverifiable=req.unverifiable) + return self._opener.open(req, timeout=self._socket_timeout) def print_debug_header(self): diff --git a/youtube_dl/extractor/__init__.py b/youtube_dl/extractor/__init__.py index 48683ebcc..7af0cfd8f 100644 --- a/youtube_dl/extractor/__init__.py +++ b/youtube_dl/extractor/__init__.py @@ -365,6 +365,7 @@ from .trutube import TruTubeIE from .tube8 import Tube8IE from .tudou import TudouIE from .tumblr import TumblrIE +from .turbo import TurboIE from .tutv import TutvIE from .tvigle import TvigleIE from .tvp import TvpIE diff --git a/youtube_dl/extractor/daum.py b/youtube_dl/extractor/daum.py index 6033cd94a..45d66e2e6 100644 --- a/youtube_dl/extractor/daum.py +++ b/youtube_dl/extractor/daum.py @@ -11,10 +11,10 @@ from ..utils import ( class DaumIE(InfoExtractor): - _VALID_URL = r'https?://(?:m\.)?tvpot\.daum\.net/.*?clipid=(?P<id>\d+)' + _VALID_URL = r'https?://(?:m\.)?tvpot\.daum\.net/(?:v/|.*?clipid=)(?P<id>[^?#&]+)' IE_NAME = 'daum.net' - _TEST = { + _TESTS = [{ 'url': 'http://tvpot.daum.net/clip/ClipView.do?clipid=52554690', 'info_dict': { 'id': '52554690', @@ -24,11 +24,17 @@ class DaumIE(InfoExtractor): 'upload_date': '20130831', 'duration': 3868, }, - } + }, { + 'url': 'http://tvpot.daum.net/v/vab4dyeDBysyBssyukBUjBz', + 'only_matching': True, + }, { + 'url': 'http://tvpot.daum.net/v/07dXWRka62Y%24', + 'only_matching': True, + }] def _real_extract(self, url): mobj = re.match(self._VALID_URL, url) - video_id = mobj.group(1) + video_id = mobj.group('id') canonical_url = 'http://tvpot.daum.net/v/%s' % video_id webpage = self._download_webpage(canonical_url, video_id) full_id = self._search_regex( @@ -42,7 +48,6 @@ class DaumIE(InfoExtractor): 'http://videofarm.daum.net/controller/api/open/v1_2/MovieData.apixml?' + query, video_id, 'Downloading video formats info') - self.to_screen(u'%s: Getting video urls' % video_id) formats = [] for format_el in urls.findall('result/output_list/output_list'): profile = format_el.attrib['profile'] @@ -52,7 +57,7 @@ class DaumIE(InfoExtractor): }) url_doc = self._download_xml( 'http://videofarm.daum.net/controller/api/open/v1_2/MovieLocation.apixml?' + format_query, - video_id, note=False) + video_id, note='Downloading video data for %s format' % profile) format_url = url_doc.find('result/url').text formats.append({ 'url': format_url, diff --git a/youtube_dl/extractor/turbo.py b/youtube_dl/extractor/turbo.py new file mode 100644 index 000000000..29703a8a9 --- /dev/null +++ b/youtube_dl/extractor/turbo.py @@ -0,0 +1,67 @@ +# coding: utf-8 +from __future__ import unicode_literals + +import re + +from .common import InfoExtractor +from ..utils import ( + ExtractorError, + int_or_none, + qualities, + xpath_text, +) + + +class TurboIE(InfoExtractor): + _VALID_URL = r'https?://(?:www\.)?turbo\.fr/videos-voiture/(?P<id>[0-9]+)-' + _API_URL = 'http://www.turbo.fr/api/tv/xml.php?player_generique=player_generique&id={0:}' + _TEST = { + 'url': 'http://www.turbo.fr/videos-voiture/454443-turbo-du-07-09-2014-renault-twingo-3-bentley-continental-gt-speed-ces-guide-achat-dacia.html', + 'md5': '33f4b91099b36b5d5a91f84b5bcba600', + 'info_dict': { + 'id': '454443', + 'ext': 'mp4', + 'duration': 3715, + 'title': 'Turbo du 07/09/2014 : Renault Twingo 3, Bentley Continental GT Speed, CES, Guide Achat Dacia... ', + 'description': 'Retrouvez dans cette rubrique toutes les vidéos de l\'Turbo du 07/09/2014 : Renault Twingo 3, Bentley Continental GT Speed, CES, Guide Achat Dacia... ', + 'thumbnail': 're:^https?://.*\.jpg$', + } + } + + def _real_extract(self, url): + mobj = re.match(self._VALID_URL, url) + video_id = mobj.group('id') + + webpage = self._download_webpage(url, video_id) + + playlist = self._download_xml(self._API_URL.format(video_id), video_id) + item = playlist.find('./channel/item') + if item is None: + raise ExtractorError('Playlist item was not found', expected=True) + + title = xpath_text(item, './title', 'title') + duration = int_or_none(xpath_text(item, './durate', 'duration')) + thumbnail = xpath_text(item, './visuel_clip', 'thumbnail') + description = self._og_search_description(webpage) + + formats = [] + get_quality = qualities(['3g', 'sd', 'hq']) + for child in item: + m = re.search(r'url_video_(?P<quality>.+)', child.tag) + if m: + quality = m.group('quality') + formats.append({ + 'format_id': quality, + 'url': child.text, + 'quality': get_quality(quality), + }) + self._sort_formats(formats) + + return { + 'id': video_id, + 'title': title, + 'duration': duration, + 'thumbnail': thumbnail, + 'description': description, + 'formats': formats, + } diff --git a/youtube_dl/utils.py b/youtube_dl/utils.py index 3ac0f1f54..b644f4e92 100644 --- a/youtube_dl/utils.py +++ b/youtube_dl/utils.py @@ -1437,6 +1437,24 @@ def uppercase_escape(s): lambda m: unicode_escape(m.group(0))[0], s) + +def escape_rfc3986(s): + """Escape non-ASCII characters as suggested by RFC 3986""" + if sys.version_info < (3, 0) and isinstance(s, unicode): + s = s.encode('utf-8') + return compat_urllib_parse.quote(s, "%/;:@&=+$,!~*'()?#[]") + + +def escape_url(url): + """Escape URL as suggested by RFC 3986""" + url_parsed = compat_urllib_parse_urlparse(url) + return url_parsed._replace( + path=escape_rfc3986(url_parsed.path), + params=escape_rfc3986(url_parsed.params), + query=escape_rfc3986(url_parsed.query), + fragment=escape_rfc3986(url_parsed.fragment) + ).geturl() + try: struct.pack(u'!I', 0) except TypeError: diff --git a/youtube_dl/version.py b/youtube_dl/version.py index 97a5dcacf..cf0d862da 100644 --- a/youtube_dl/version.py +++ b/youtube_dl/version.py @@ -1,2 +1,2 @@ -__version__ = '2014.09.15' +__version__ = '2014.09.15.1' |