import re from .common import InfoExtractor from ..utils import ( int_or_none, join_nonempty, parse_duration, urljoin, xpath_element, xpath_text, ) class HBOBaseIE(InfoExtractor): _FORMATS_INFO = { 'pro7': { 'width': 1280, 'height': 720, }, '1920': { 'width': 1280, 'height': 720, }, 'pro6': { 'width': 768, 'height': 432, }, '640': { 'width': 768, 'height': 432, }, 'pro5': { 'width': 640, 'height': 360, }, 'highwifi': { 'width': 640, 'height': 360, }, 'high3g': { 'width': 640, 'height': 360, }, 'medwifi': { 'width': 400, 'height': 224, }, 'med3g': { 'width': 400, 'height': 224, }, } def _extract_info(self, url, display_id): video_data = self._download_xml(url, display_id) video_id = xpath_text(video_data, 'id', fatal=True) episode_title = title = xpath_text(video_data, 'title', fatal=True) series = xpath_text(video_data, 'program') if series: title = f'{series} - {title}' formats = [] for source in xpath_element(video_data, 'videos', 'sources', True): if source.tag == 'size': path = xpath_text(source, './/path') if not path: continue width = source.attrib.get('width') format_info = self._FORMATS_INFO.get(width, {}) height = format_info.get('height') fmt = { 'url': path, 'format_id': join_nonempty('http'. height and f'{height}p'), 'width': format_info.get('width'), 'height': height, } rtmp = re.search(r'^(?Prtmpe?://[^/]+/(?P.+))/(?Pmp4:.+)$', path) if rtmp: fmt.update({ 'url': rtmp.group('url'), 'play_path': rtmp.group('playpath'), 'app': rtmp.group('app'), 'ext': 'flv', 'format_id': fmt['format_id'].replace('http', 'rtmp'), }) formats.append(fmt) else: video_url = source.text if not video_url: continue if source.tag == 'tarball': formats.extend(self._extract_m3u8_formats( video_url.replace('.tar', '/base_index_w8.m3u8'), video_id, 'mp4', 'm3u8_native', m3u8_id='hls', fatal=False)) elif source.tag == 'hls': m3u8_formats = self._extract_m3u8_formats( video_url.replace('.tar', '/base_index.m3u8'), video_id, 'mp4', 'm3u8_native', m3u8_id='hls', fatal=False) for f in m3u8_formats: if f.get('vcodec') == 'none' and not f.get('tbr'): f['tbr'] = int_or_none(self._search_regex( r'-(\d+)k/', f['url'], 'tbr', default=None)) formats.extend(m3u8_formats) elif source.tag == 'dash': formats.extend(self._extract_mpd_formats( video_url.replace('.tar', '/manifest.mpd'), video_id, mpd_id='dash', fatal=False)) else: format_info = self._FORMATS_INFO.get(source.tag, {}) formats.append({ 'format_id': f'http-{source.tag}', 'url': video_url, 'width': format_info.get('width'), 'height': format_info.get('height'), }) thumbnails = [] card_sizes = xpath_element(video_data, 'titleCardSizes') if card_sizes is not None: for size in card_sizes: path = xpath_text(size, 'path') if not path: continue width = int_or_none(size.get('width')) thumbnails.append({ 'id': width, 'url': path, 'width': width, }) subtitles = None caption_url = xpath_text(video_data, 'captionUrl') if caption_url: subtitles = { 'en': [{ 'url': caption_url, 'ext': 'ttml', }], } return { 'id': video_id, 'title': title, 'duration': parse_duration(xpath_text(video_data, 'duration/tv14')), 'series': series, 'episode': episode_title, 'formats': formats, 'thumbnails': thumbnails, 'subtitles': subtitles, } class HBOIE(HBOBaseIE): IE_NAME = 'hbo' _VALID_URL = r'https?://(?:www\.)?hbo\.com/(?:video|embed)(?:/[^/]+)*/(?P[^/?#]+)' _TEST = { 'url': 'https://www.hbo.com/video/game-of-thrones/seasons/season-8/videos/trailer', 'md5': '8126210656f433c452a21367f9ad85b3', 'info_dict': { 'id': '22113301', 'ext': 'mp4', 'title': 'Game of Thrones - Trailer', }, 'expected_warnings': ['Unknown MIME type application/mp4 in DASH manifest'], } def _real_extract(self, url): display_id = self._match_id(url) webpage = self._download_webpage(url, display_id) location_path = self._parse_json(self._html_search_regex( r'data-state="({.+?})"', webpage, 'state'), display_id)['video']['locationUrl'] return self._extract_info(urljoin(url, location_path), display_id)