From 266a1b5d52d4a48a966d0a0b6286ca2740482409 Mon Sep 17 00:00:00 2001 From: Ashish Gupta <39122144+Ashish0804@users.noreply.github.com> Date: Thu, 18 Nov 2021 21:28:51 +0530 Subject: [PATCH] [ESPNCricInfo] Add extractor (#1652) Closes: #1635 Authored by: Ashish0804 --- yt_dlp/extractor/espn.py | 43 ++++++++++++++++++++++++++++++++++ yt_dlp/extractor/extractors.py | 1 + 2 files changed, 44 insertions(+) diff --git a/yt_dlp/extractor/espn.py b/yt_dlp/extractor/espn.py index d4a66c29f..dc50f3b8b 100644 --- a/yt_dlp/extractor/espn.py +++ b/yt_dlp/extractor/espn.py @@ -7,7 +7,9 @@ from .once import OnceIE from ..compat import compat_str from ..utils import ( determine_ext, + dict_get, int_or_none, + unified_strdate, unified_timestamp, ) @@ -236,3 +238,44 @@ class FiveThirtyEightIE(InfoExtractor): webpage, 'embed url') return self.url_result(embed_url, 'AbcNewsVideo') + + +class ESPNCricInfoIE(InfoExtractor): + _VALID_URL = r'https?://(?:www\.)?espncricinfo\.com/video/[^#$&?/]+-(?P\d+)' + _TESTS = [{ + 'url': 'https://www.espncricinfo.com/video/finch-chasing-comes-with-risks-despite-world-cup-trend-1289135', + 'info_dict': { + 'id': '1289135', + 'ext': 'mp4', + 'title': 'Finch: Chasing comes with \'risks\' despite World Cup trend', + 'description': 'md5:ea32373303e25efbb146efdfc8a37829', + 'upload_date': '20211113', + 'duration': 96, + }, + 'params': {'skip_download': True} + }] + + def _real_extract(self, url): + id = self._match_id(url) + data_json = self._download_json(f'https://hs-consumer-api.espncricinfo.com/v1/pages/video/video-details?videoId={id}', id)['video'] + formats, subtitles = [], {} + for item in data_json.get('playbacks') or []: + if item.get('type') == 'HLS' and item.get('url'): + m3u8_frmts, m3u8_subs = self._extract_m3u8_formats_and_subtitles(item['url'], id) + formats.extend(m3u8_frmts) + subtitles = self._merge_subtitles(subtitles, m3u8_subs) + elif item.get('type') == 'AUDIO' and item.get('url'): + formats.append({ + 'url': item['url'], + 'vcodec': 'none', + }) + self._sort_formats(formats) + return { + 'id': id, + 'title': data_json.get('title'), + 'description': data_json.get('summary'), + 'upload_date': unified_strdate(dict_get(data_json, ('publishedAt', 'recordedAt'))), + 'duration': data_json.get('duration'), + 'formats': formats, + 'subtitles': subtitles, + } diff --git a/yt_dlp/extractor/extractors.py b/yt_dlp/extractor/extractors.py index a60e27186..a3674d836 100644 --- a/yt_dlp/extractor/extractors.py +++ b/yt_dlp/extractor/extractors.py @@ -418,6 +418,7 @@ from .espn import ( ESPNIE, ESPNArticleIE, FiveThirtyEightIE, + ESPNCricInfoIE, ) from .esri import EsriVideoIE from .europa import EuropaIE