X-Git-Url: https://git.rapsys.eu/youtubedl/blobdiff_plain/9f2b33881274af98a9145c533a1d295fad71521a..4cd837fe5fe8d00dd8edcc06fef3a4c9b63c32bc:/youtube_dl/extractor/pbs.py diff --git a/youtube_dl/extractor/pbs.py b/youtube_dl/extractor/pbs.py index f43e3a1..80340f5 100644 --- a/youtube_dl/extractor/pbs.py +++ b/youtube_dl/extractor/pbs.py @@ -4,14 +4,18 @@ from __future__ import unicode_literals import re from .common import InfoExtractor -from ..compat import compat_HTTPError +from ..compat import compat_str from ..utils import ( ExtractorError, determine_ext, int_or_none, + float_or_none, js_to_json, + orderedSet, strip_jsonp, + strip_or_none, unified_strdate, + url_or_none, US_RATINGS, ) @@ -185,42 +189,38 @@ class PBSIE(InfoExtractor): _VALID_URL = r'''(?x)https?:// (?: # Direct video URL - (?:%s)/(?:viralplayer|video)/(?P[0-9]+)/? | + (?:%s)/(?:(?:vir|port)alplayer|video)/(?P[0-9]+)(?:[?/]|$) | # Article with embedded player (or direct video) - (?:www\.)?pbs\.org/(?:[^/]+/){2,5}(?P[^/]+?)(?:\.html)?/?(?:$|[?\#]) | + (?:www\.)?pbs\.org/(?:[^/]+/){1,5}(?P[^/]+?)(?:\.html)?/?(?:$|[?\#]) | # Player (?:video|player)\.pbs\.org/(?:widget/)?partnerplayer/(?P[^/]+)/ ) ''' % '|'.join(list(zip(*_STATIONS))[0]) + _GEO_COUNTRIES = ['US'] + _TESTS = [ { 'url': 'http://www.pbs.org/tpt/constitution-usa-peter-sagal/watch/a-more-perfect-union/', - 'md5': 'ce1888486f0908d555a8093cac9a7362', + 'md5': '173dc391afd361fa72eab5d3d918968d', 'info_dict': { 'id': '2365006249', 'ext': 'mp4', 'title': 'Constitution USA with Peter Sagal - A More Perfect Union', - 'description': 'md5:36f341ae62e251b8f5bd2b754b95a071', + 'description': 'md5:31b664af3c65fd07fa460d306b837d00', 'duration': 3190, }, - 'params': { - 'skip_download': True, # requires ffmpeg - }, }, { 'url': 'http://www.pbs.org/wgbh/pages/frontline/losing-iraq/', - 'md5': '143c98aa54a346738a3d78f54c925321', + 'md5': '6f722cb3c3982186d34b0f13374499c7', 'info_dict': { 'id': '2365297690', 'ext': 'mp4', 'title': 'FRONTLINE - Losing Iraq', - 'description': 'md5:4d3eaa01f94e61b3e73704735f1196d9', + 'description': 'md5:5979a4d069b157f622d02bff62fbe654', 'duration': 5050, }, - 'params': { - 'skip_download': True, # requires ffmpeg - } }, { 'url': 'http://www.pbs.org/newshour/bb/education-jan-june12-cyberschools_02-23/', @@ -229,7 +229,7 @@ class PBSIE(InfoExtractor): 'id': '2201174722', 'ext': 'mp4', 'title': 'PBS NewsHour - Cyber Schools Gain Popularity, but Quality Questions Persist', - 'description': 'md5:95a19f568689d09a166dff9edada3301', + 'description': 'md5:86ab9a3d04458b876147b355788b8781', 'duration': 801, }, }, @@ -242,10 +242,7 @@ class PBSIE(InfoExtractor): 'title': 'Great Performances - Dudamel Conducts Verdi Requiem at the Hollywood Bowl - Full', 'description': 'md5:657897370e09e2bc6bf0f8d2cd313c6b', 'duration': 6559, - 'thumbnail': 're:^https?://.*\.jpg$', - }, - 'params': { - 'skip_download': True, # requires ffmpeg + 'thumbnail': r're:^https?://.*\.jpg$', }, }, { @@ -258,13 +255,10 @@ class PBSIE(InfoExtractor): 'description': 'md5:c741d14e979fc53228c575894094f157', 'title': 'NOVA - Killer Typhoon', 'duration': 3172, - 'thumbnail': 're:^https?://.*\.jpg$', + 'thumbnail': r're:^https?://.*\.jpg$', 'upload_date': '20140122', 'age_limit': 10, }, - 'params': { - 'skip_download': True, # requires ffmpeg - }, }, { 'url': 'http://www.pbs.org/wgbh/pages/frontline/united-states-of-secrets/', @@ -273,6 +267,13 @@ class PBSIE(InfoExtractor): }, 'playlist_count': 2, }, + { + 'url': 'http://www.pbs.org/wgbh/americanexperience/films/great-war/', + 'info_dict': { + 'id': 'great-war', + }, + 'playlist_count': 3, + }, { 'url': 'http://www.pbs.org/wgbh/americanexperience/films/death/player/', 'info_dict': { @@ -280,9 +281,9 @@ class PBSIE(InfoExtractor): 'display_id': 'player', 'ext': 'mp4', 'title': 'American Experience - Death and the Civil War, Chapter 1', - 'description': 'md5:1b80a74e0380ed2a4fb335026de1600d', + 'description': 'md5:67fa89a9402e2ee7d08f53b920674c18', 'duration': 682, - 'thumbnail': 're:^https?://.*\.jpg$', + 'thumbnail': r're:^https?://.*\.jpg$', }, 'params': { 'skip_download': True, # requires ffmpeg @@ -290,6 +291,7 @@ class PBSIE(InfoExtractor): }, { 'url': 'http://www.pbs.org/video/2365245528/', + 'md5': '115223d41bd55cda8ae5cd5ed4e11497', 'info_dict': { 'id': '2365245528', 'display_id': '2365245528', @@ -297,10 +299,7 @@ class PBSIE(InfoExtractor): 'title': 'FRONTLINE - United States of Secrets (Part One)', 'description': 'md5:55756bd5c551519cc4b7703e373e217e', 'duration': 6851, - 'thumbnail': 're:^https?://.*\.jpg$', - }, - 'params': { - 'skip_download': True, # requires ffmpeg + 'thumbnail': r're:^https?://.*\.jpg$', }, }, { @@ -308,17 +307,15 @@ class PBSIE(InfoExtractor): # "', webpage): url = self._search_regex( @@ -426,7 +501,11 @@ class PBSIE(InfoExtractor): if url: break - mobj = re.match(self._VALID_URL, url) + if not url: + url = self._og_search_url(webpage) + + mobj = re.match( + self._VALID_URL, self._proto_relative_url(url.strip())) player_id = mobj.group('player_id') if not display_id: @@ -436,15 +515,29 @@ class PBSIE(InfoExtractor): url, display_id, note='Downloading player page', errnote='Could not download player page') video_id = self._search_regex( - r'