X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;f=youtube_dl%2Fextractor%2Frte.py;h=a6fac6c35d00327c2858f9aead301845c4af572a;hb=ec85ded83cbfa652ba94cb080aab52d8b270212a;hp=47c8331fed6221ccba38874513cba90ed7d609db;hpb=9938a17f92ad8242d8841528a89df647ff759183;p=youtube-dl diff --git a/youtube_dl/extractor/rte.py b/youtube_dl/extractor/rte.py index 47c8331fe..a6fac6c35 100644 --- a/youtube_dl/extractor/rte.py +++ b/youtube_dl/extractor/rte.py @@ -1,94 +1,34 @@ # coding: utf-8 from __future__ import unicode_literals +import re + from .common import InfoExtractor +from ..compat import compat_HTTPError from ..utils import ( float_or_none, parse_iso8601, unescapeHTML, + ExtractorError, ) -class RteIE(InfoExtractor): - IE_NAME = 'rte' - IE_DESC = 'Raidió Teilifís Éireann TV' - _VALID_URL = r'https?://(?:www\.)?rte\.ie/player/[^/]{2,3}/show/[^/]+/(?P[0-9]+)' - _TEST = { - 'url': 'http://www.rte.ie/player/ie/show/iwitness-862/10478715/', - 'info_dict': { - 'id': '10478715', - 'ext': 'flv', - 'title': 'Watch iWitness online', - 'thumbnail': 're:^https?://.*\.jpg$', - 'description': 'iWitness : The spirit of Ireland, one voice and one minute at a time.', - 'duration': 60.046, - }, - 'params': { - 'skip_download': 'f4m fails with --test atm' - } - } - - def _real_extract(self, url): - video_id = self._match_id(url) - webpage = self._download_webpage(url, video_id) - - title = self._og_search_title(webpage) - description = self._html_search_meta('description', webpage, 'description') - duration = float_or_none(self._html_search_meta( - 'duration', webpage, 'duration', fatal=False), 1000) - - thumbnail_id = self._search_regex( - r'', webpage, 'thumbnail') - thumbnail = 'http://img.rasset.ie/' + thumbnail_id + '.jpg' - - feeds_url = self._html_search_meta("feeds-prefix", webpage, 'feeds url') + video_id - json_string = self._download_json(feeds_url, video_id) - - # f4m_url = server + relative_url - f4m_url = json_string['shows'][0]['media:group'][0]['rte:server'] + json_string['shows'][0]['media:group'][0]['url'] - f4m_formats = self._extract_f4m_formats(f4m_url, video_id) - - return { - 'id': video_id, - 'title': title, - 'formats': f4m_formats, - 'description': description, - 'thumbnail': thumbnail, - 'duration': duration, - } - - -class RteRadioIE(InfoExtractor): - IE_NAME = 'rte:radio' - IE_DESC = 'Raidió Teilifís Éireann radio' - # Radioplayer URLs have the specifier #!rii=:::: - # where the IDs are int/empty, the date is DD-MM-YYYY, and the specifier may be truncated. - # An uniquely defines an individual recording, and is the only part we require. - _VALID_URL = r'https?://(?:www\.)?rte\.ie/radio/utils/radioplayer/rteradioweb\.html#!rii=(?:[0-9]*)(?:%3A|:)(?P[0-9]+)' - - _TEST = { - 'url': 'http://www.rte.ie/radio/utils/radioplayer/rteradioweb.html#!rii=16:10507902:2414:27-12-2015:', - 'info_dict': { - 'id': '10507902', - 'ext': 'mp4', - 'title': 'Gloria', - 'thumbnail': 're:^https?://.*\.jpg$', - 'description': 'md5:9ce124a7fb41559ec68f06387cabddf0', - 'timestamp': 1451203200, - 'upload_date': '20151227', - 'duration': 7230.0, - }, - 'params': { - 'skip_download': 'f4m fails with --test atm' - } - } - +class RteBaseIE(InfoExtractor): def _real_extract(self, url): item_id = self._match_id(url) - json_string = self._download_json( - 'http://www.rte.ie/rteavgen/getplaylist/?type=web&format=json&id=' + item_id, - item_id) + try: + json_string = self._download_json( + 'http://www.rte.ie/rteavgen/getplaylist/?type=web&format=json&id=' + item_id, + item_id) + except ExtractorError as ee: + if isinstance(ee.cause, compat_HTTPError) and ee.cause.code == 404: + error_info = self._parse_json(ee.cause.read().decode(), item_id, fatal=False) + if error_info: + raise ExtractorError( + '%s said: %s' % (self.IE_NAME, error_info['message']), + expected=True) + raise # NB the string values in the JSON are stored using XML escaping(!) show = json_string['shows'][0] @@ -102,8 +42,18 @@ class RteRadioIE(InfoExtractor): formats = [] - if mg.get('url') and not mg['url'].startswith('rtmpe:'): - formats.append({'url': mg['url']}) + if mg.get('url'): + m = re.match(r'(?Prtmpe?://[^/]+)/(?P.+)/(?Pmp4:.*)', mg['url']) + if m: + m = m.groupdict() + formats.append({ + 'url': m['url'] + '/' + m['app'], + 'app': m['app'], + 'play_path': m['playpath'], + 'player_url': url, + 'ext': 'flv', + 'format_id': 'rtmp', + }) if mg.get('hls_server') and mg.get('hls_url'): formats.extend(self._extract_m3u8_formats( @@ -126,3 +76,67 @@ class RteRadioIE(InfoExtractor): 'duration': duration, 'formats': formats, } + + +class RteIE(RteBaseIE): + IE_NAME = 'rte' + IE_DESC = 'Raidió Teilifís Éireann TV' + _VALID_URL = r'https?://(?:www\.)?rte\.ie/player/[^/]{2,3}/show/[^/]+/(?P[0-9]+)' + _TEST = { + 'url': 'http://www.rte.ie/player/ie/show/iwitness-862/10478715/', + 'md5': '4a76eb3396d98f697e6e8110563d2604', + 'info_dict': { + 'id': '10478715', + 'ext': 'mp4', + 'title': 'iWitness', + 'thumbnail': r're:^https?://.*\.jpg$', + 'description': 'The spirit of Ireland, one voice and one minute at a time.', + 'duration': 60.046, + 'upload_date': '20151012', + 'timestamp': 1444694160, + }, + } + + +class RteRadioIE(RteBaseIE): + IE_NAME = 'rte:radio' + IE_DESC = 'Raidió Teilifís Éireann radio' + # Radioplayer URLs have two distinct specifier formats, + # the old format #!rii=:::: + # the new format #!rii=b____ + # where the IDs are int/empty, the date is DD-MM-YYYY, and the specifier may be truncated. + # An uniquely defines an individual recording, and is the only part we require. + _VALID_URL = r'https?://(?:www\.)?rte\.ie/radio/utils/radioplayer/rteradioweb\.html#!rii=(?:b?[0-9]*)(?:%3A|:|%5F|_)(?P[0-9]+)' + + _TESTS = [{ + # Old-style player URL; HLS and RTMPE formats + 'url': 'http://www.rte.ie/radio/utils/radioplayer/rteradioweb.html#!rii=16:10507902:2414:27-12-2015:', + 'md5': 'c79ccb2c195998440065456b69760411', + 'info_dict': { + 'id': '10507902', + 'ext': 'mp4', + 'title': 'Gloria', + 'thumbnail': r're:^https?://.*\.jpg$', + 'description': 'md5:9ce124a7fb41559ec68f06387cabddf0', + 'timestamp': 1451203200, + 'upload_date': '20151227', + 'duration': 7230.0, + }, + }, { + # New-style player URL; RTMPE formats only + 'url': 'http://rte.ie/radio/utils/radioplayer/rteradioweb.html#!rii=b16_3250678_8861_06-04-2012_', + 'info_dict': { + 'id': '3250678', + 'ext': 'flv', + 'title': 'The Lyric Concert with Paul Herriott', + 'thumbnail': r're:^https?://.*\.jpg$', + 'description': '', + 'timestamp': 1333742400, + 'upload_date': '20120406', + 'duration': 7199.016, + }, + 'params': { + # rtmp download + 'skip_download': True, + }, + }]