X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;f=youtube_dl%2Fextractor%2Fdramafever.py;h=60ed438f8cd9b254fa1ae8d25cd76ed866cab25d;hb=9f0ee2a3883ec6f6fdccba90085cb925aaa2f617;hp=c4b7c0b68afb3e85c98fca16790a774efe81bd3f;hpb=463b2e5542a85d5cd41b255a71833fec7b4f51e0;p=youtube-dl diff --git a/youtube_dl/extractor/dramafever.py b/youtube_dl/extractor/dramafever.py index c4b7c0b68..60ed438f8 100644 --- a/youtube_dl/extractor/dramafever.py +++ b/youtube_dl/extractor/dramafever.py @@ -3,23 +3,69 @@ from __future__ import unicode_literals import itertools -from .common import InfoExtractor +from .amp import AMPIE from ..compat import ( compat_HTTPError, + compat_urllib_parse, compat_urlparse, ) from ..utils import ( ExtractorError, clean_html, - determine_ext, - int_or_none, - parse_iso8601, + sanitized_Request, ) -class DramaFeverIE(InfoExtractor): +class DramaFeverBaseIE(AMPIE): + _LOGIN_URL = 'https://www.dramafever.com/accounts/login/' + _NETRC_MACHINE = 'dramafever' + + _CONSUMER_SECRET = 'DA59dtVXYLxajktV' + + _consumer_secret = None + + def _get_consumer_secret(self): + mainjs = self._download_webpage( + 'http://www.dramafever.com/static/51afe95/df2014/scripts/main.js', + None, 'Downloading main.js', fatal=False) + if not mainjs: + return self._CONSUMER_SECRET + return self._search_regex( + r"var\s+cs\s*=\s*'([^']+)'", mainjs, + 'consumer secret', default=self._CONSUMER_SECRET) + + def _real_initialize(self): + self._login() + self._consumer_secret = self._get_consumer_secret() + + def _login(self): + (username, password) = self._get_login_info() + if username is None: + return + + login_form = { + 'username': username, + 'password': password, + } + + request = sanitized_Request( + self._LOGIN_URL, compat_urllib_parse.urlencode(login_form).encode('utf-8')) + response = self._download_webpage( + request, None, 'Logging in as %s' % username) + + if all(logout_pattern not in response + for logout_pattern in ['href="/accounts/logout/"', '>Log out<']): + error = self._html_search_regex( + r'(?s)class="hidden-xs prompt"[^>]*>(.+?)<', + response, 'error message', default=None) + if error: + raise ExtractorError('Unable to login: %s' % error, expected=True) + raise ExtractorError('Unable to log in') + + +class DramaFeverIE(DramaFeverBaseIE): IE_NAME = 'dramafever' - _VALID_URL = r'https?://(?:www\.)?dramafever\.com/drama/(?P[0-9]+/[0-9]+)' + _VALID_URL = r'https?://(?:www\.)?dramafever\.com/drama/(?P[0-9]+/[0-9]+)(?:/|$)' _TEST = { 'url': 'http://www.dramafever.com/drama/4512/1/Cooking_with_Shin/', 'info_dict': { @@ -31,73 +77,46 @@ class DramaFeverIE(InfoExtractor): 'timestamp': 1404336058, 'upload_date': '20140702', 'duration': 343, - } + }, + 'params': { + # m3u8 download + 'skip_download': True, + }, } def _real_extract(self, url): video_id = self._match_id(url).replace('/', '.') try: - feed = self._download_json( - 'http://www.dramafever.com/amp/episode/feed.json?guid=%s' % video_id, - video_id, 'Downloading episode JSON')['channel']['item'] + info = self._extract_feed_info( + 'http://www.dramafever.com/amp/episode/feed.json?guid=%s' % video_id) except ExtractorError as e: if isinstance(e.cause, compat_HTTPError): raise ExtractorError( 'Currently unavailable in your country.', expected=True) raise - media_group = feed.get('media-group', {}) - - formats = [] - for media_content in media_group['media-content']: - src = media_content.get('@attributes', {}).get('url') - if not src: - continue - ext = determine_ext(src) - if ext == 'f4m': - formats.extend(self._extract_f4m_formats( - src, video_id, f4m_id='hds')) - elif ext == 'm3u8': - formats.extend(self._extract_m3u8_formats( - src, video_id, 'mp4', m3u8_id='hls')) - else: - formats.append({ - 'url': src, - }) - self._sort_formats(formats) - - title = media_group.get('media-title') - description = media_group.get('media-description') - duration = int_or_none(media_group['media-content'][0].get('@attributes', {}).get('duration')) - thumbnail = self._proto_relative_url( - media_group.get('media-thumbnail', {}).get('@attributes', {}).get('url')) - timestamp = parse_iso8601(feed.get('pubDate'), ' ') - - subtitles = {} - for media_subtitle in media_group.get('media-subTitle', []): - lang = media_subtitle.get('@attributes', {}).get('lang') - href = media_subtitle.get('@attributes', {}).get('href') - if not lang or not href: - continue - subtitles[lang] = [{ - 'ext': 'ttml', - 'url': href, - }] - - return { - 'id': video_id, - 'title': title, - 'description': description, - 'thumbnail': thumbnail, - 'timestamp': timestamp, - 'duration': duration, - 'formats': formats, - 'subtitles': subtitles, - } - - -class DramaFeverSeriesIE(InfoExtractor): + series_id, episode_number = video_id.split('.') + episode_info = self._download_json( + # We only need a single episode info, so restricting page size to one episode + # and dealing with page number as with episode number + r'http://www.dramafever.com/api/4/episode/series/?cs=%s&series_id=%s&page_number=%s&page_size=1' + % (self._consumer_secret, series_id, episode_number), + video_id, 'Downloading episode info JSON', fatal=False) + if episode_info: + value = episode_info.get('value') + if value: + subfile = value[0].get('subfile') or value[0].get('new_subfile') + if subfile and subfile != 'http://www.dramafever.com/st/': + info['subtitiles'].setdefault('English', []).append({ + 'ext': 'srt', + 'url': subfile, + }) + + return info + + +class DramaFeverSeriesIE(DramaFeverBaseIE): IE_NAME = 'dramafever:series' _VALID_URL = r'https?://(?:www\.)?dramafever\.com/drama/(?P[0-9]+)(?:/(?:(?!\d+(?:/|$)).+)?)?$' _TESTS = [{ @@ -118,27 +137,14 @@ class DramaFeverSeriesIE(InfoExtractor): 'playlist_count': 20, }] - _CONSUMER_SECRET = 'DA59dtVXYLxajktV' _PAGE_SIZE = 60 # max is 60 (see http://api.drama9.com/#get--api-4-episode-series-) - def _get_consumer_secret(self, video_id): - mainjs = self._download_webpage( - 'http://www.dramafever.com/static/51afe95/df2014/scripts/main.js', - video_id, 'Downloading main.js', fatal=False) - if not mainjs: - return self._CONSUMER_SECRET - return self._search_regex( - r"var\s+cs\s*=\s*'([^']+)'", mainjs, - 'consumer secret', default=self._CONSUMER_SECRET) - def _real_extract(self, url): series_id = self._match_id(url) - consumer_secret = self._get_consumer_secret(series_id) - series = self._download_json( 'http://www.dramafever.com/api/4/series/query/?cs=%s&series_id=%s' - % (consumer_secret, series_id), + % (self._consumer_secret, series_id), series_id, 'Downloading series JSON')['series'][series_id] title = clean_html(series['name']) @@ -148,11 +154,14 @@ class DramaFeverSeriesIE(InfoExtractor): for page_num in itertools.count(1): episodes = self._download_json( 'http://www.dramafever.com/api/4/episode/series/?cs=%s&series_id=%s&page_size=%d&page_number=%d' - % (consumer_secret, series_id, self._PAGE_SIZE, page_num), + % (self._consumer_secret, series_id, self._PAGE_SIZE, page_num), series_id, 'Downloading episodes JSON page #%d' % page_num) for episode in episodes.get('value', []): + episode_url = episode.get('episode_url') + if not episode_url: + continue entries.append(self.url_result( - compat_urlparse.urljoin(url, episode['episode_url']), + compat_urlparse.urljoin(url, episode_url), 'DramaFever', episode.get('guid'))) if page_num == episodes['num_pages']: break