aboutsummaryrefslogtreecommitdiff
path: root/youtube_dl/extractor/mlb.py
diff options
context:
space:
mode:
Diffstat (limited to 'youtube_dl/extractor/mlb.py')
-rw-r--r--youtube_dl/extractor/mlb.py250
1 files changed, 172 insertions, 78 deletions
diff --git a/youtube_dl/extractor/mlb.py b/youtube_dl/extractor/mlb.py
index e242b897f..b69301d97 100644
--- a/youtube_dl/extractor/mlb.py
+++ b/youtube_dl/extractor/mlb.py
@@ -4,95 +4,156 @@ import re
from .common import InfoExtractor
from ..utils import (
+ determine_ext,
+ int_or_none,
parse_duration,
parse_iso8601,
+ try_get,
)
-class MLBIE(InfoExtractor):
+class MLBBaseIE(InfoExtractor):
+ def _real_extract(self, url):
+ display_id = self._match_id(url)
+ video = self._download_video_data(display_id)
+ video_id = video['id']
+ title = video['title']
+ feed = self._get_feed(video)
+
+ formats = []
+ for playback in (feed.get('playbacks') or []):
+ playback_url = playback.get('url')
+ if not playback_url:
+ continue
+ name = playback.get('name')
+ ext = determine_ext(playback_url)
+ if ext == 'm3u8':
+ formats.extend(self._extract_m3u8_formats(
+ playback_url, video_id, 'mp4',
+ 'm3u8_native', m3u8_id=name, fatal=False))
+ else:
+ f = {
+ 'format_id': name,
+ 'url': playback_url,
+ }
+ mobj = re.search(r'_(\d+)K_(\d+)X(\d+)', name)
+ if mobj:
+ f.update({
+ 'height': int(mobj.group(3)),
+ 'tbr': int(mobj.group(1)),
+ 'width': int(mobj.group(2)),
+ })
+ mobj = re.search(r'_(\d+)x(\d+)_(\d+)_(\d+)K\.mp4', playback_url)
+ if mobj:
+ f.update({
+ 'fps': int(mobj.group(3)),
+ 'height': int(mobj.group(2)),
+ 'tbr': int(mobj.group(4)),
+ 'width': int(mobj.group(1)),
+ })
+ formats.append(f)
+ self._sort_formats(formats)
+
+ thumbnails = []
+ for cut in (try_get(feed, lambda x: x['image']['cuts'], list) or []):
+ src = cut.get('src')
+ if not src:
+ continue
+ thumbnails.append({
+ 'height': int_or_none(cut.get('height')),
+ 'url': src,
+ 'width': int_or_none(cut.get('width')),
+ })
+
+ language = (video.get('language') or 'EN').lower()
+
+ return {
+ 'id': video_id,
+ 'title': title,
+ 'formats': formats,
+ 'description': video.get('description'),
+ 'duration': parse_duration(feed.get('duration')),
+ 'thumbnails': thumbnails,
+ 'timestamp': parse_iso8601(video.get(self._TIMESTAMP_KEY)),
+ 'subtitles': self._extract_mlb_subtitles(feed, language),
+ }
+
+
+class MLBIE(MLBBaseIE):
_VALID_URL = r'''(?x)
https?://
(?:[\da-z_-]+\.)*mlb\.com/
(?:
(?:
- (?:.*?/)?video/(?:topic/[\da-z_-]+/)?v|
+ (?:[^/]+/)*video/[^/]+/c-|
(?:
shared/video/embed/(?:embed|m-internal-embed)\.html|
(?:[^/]+/)+(?:play|index)\.jsp|
)\?.*?\bcontent_id=
)
- (?P<id>n?\d+)|
- (?:[^/]+/)*(?P<path>[^/]+)
+ (?P<id>\d+)
)
'''
_TESTS = [
{
- 'url': 'http://m.mlb.com/sea/video/topic/51231442/v34698933/nymsea-ackley-robs-a-home-run-with-an-amazing-catch/?c_id=sea',
- 'md5': 'ff56a598c2cf411a9a38a69709e97079',
+ 'url': 'https://www.mlb.com/mariners/video/ackleys-spectacular-catch/c-34698933',
+ 'md5': '632358dacfceec06bad823b83d21df2d',
'info_dict': {
'id': '34698933',
'ext': 'mp4',
'title': "Ackley's spectacular catch",
'description': 'md5:7f5a981eb4f3cbc8daf2aeffa2215bf0',
'duration': 66,
- 'timestamp': 1405980600,
- 'upload_date': '20140721',
- 'thumbnail': 're:^https?://.*\.jpg$',
+ 'timestamp': 1405995000,
+ 'upload_date': '20140722',
+ 'thumbnail': r're:^https?://.*\.jpg$',
},
},
{
- 'url': 'http://m.mlb.com/video/topic/81536970/v34496663/mianym-stanton-practices-for-the-home-run-derby',
- 'md5': 'd9c022c10d21f849f49c05ae12a8a7e9',
+ 'url': 'https://www.mlb.com/video/stanton-prepares-for-derby/c-34496663',
+ 'md5': 'bf2619bf9cacc0a564fc35e6aeb9219f',
'info_dict': {
'id': '34496663',
'ext': 'mp4',
'title': 'Stanton prepares for Derby',
'description': 'md5:d00ce1e5fd9c9069e9c13ab4faedfa57',
'duration': 46,
- 'timestamp': 1405105800,
+ 'timestamp': 1405120200,
'upload_date': '20140711',
- 'thumbnail': 're:^https?://.*\.jpg$',
+ 'thumbnail': r're:^https?://.*\.jpg$',
},
},
{
- 'url': 'http://m.mlb.com/video/topic/vtp_hrd_sponsor/v34578115/hrd-cespedes-wins-2014-gillette-home-run-derby',
- 'md5': '0e6e73d509321e142409b695eadd541f',
+ 'url': 'https://www.mlb.com/video/cespedes-repeats-as-derby-champ/c-34578115',
+ 'md5': '99bb9176531adc600b90880fb8be9328',
'info_dict': {
'id': '34578115',
'ext': 'mp4',
'title': 'Cespedes repeats as Derby champ',
'description': 'md5:08df253ce265d4cf6fb09f581fafad07',
'duration': 488,
- 'timestamp': 1405399936,
+ 'timestamp': 1405414336,
'upload_date': '20140715',
- 'thumbnail': 're:^https?://.*\.jpg$',
+ 'thumbnail': r're:^https?://.*\.jpg$',
},
},
{
- 'url': 'http://m.mlb.com/video/v34577915/bautista-on-derby-captaining-duties-his-performance',
- 'md5': 'b8fd237347b844365d74ea61d4245967',
+ 'url': 'https://www.mlb.com/video/bautista-on-home-run-derby/c-34577915',
+ 'md5': 'da8b57a12b060e7663ee1eebd6f330ec',
'info_dict': {
'id': '34577915',
'ext': 'mp4',
'title': 'Bautista on Home Run Derby',
'description': 'md5:b80b34031143d0986dddc64a8839f0fb',
'duration': 52,
- 'timestamp': 1405390722,
+ 'timestamp': 1405405122,
'upload_date': '20140715',
- 'thumbnail': 're:^https?://.*\.jpg$',
+ 'thumbnail': r're:^https?://.*\.jpg$',
},
},
{
- 'url': 'http://m.mlb.com/news/article/118550098/blue-jays-kevin-pillar-goes-spidey-up-the-wall-to-rob-tim-beckham-of-a-homer',
- 'md5': 'b190e70141fb9a1552a85426b4da1b5d',
- 'info_dict': {
- 'id': '75609783',
- 'ext': 'mp4',
- 'title': 'Must C: Pillar climbs for catch',
- 'description': '4/15/15: Blue Jays outfielder Kevin Pillar continues his defensive dominance by climbing the wall in left to rob Tim Beckham of a home run',
- 'timestamp': 1429124820,
- 'upload_date': '20150415',
- }
+ 'url': 'https://www.mlb.com/video/hargrove-homers-off-caldwell/c-1352023483?tid=67793694',
+ 'only_matching': True,
},
{
'url': 'http://m.mlb.com/shared/video/embed/embed.html?content_id=35692085&topic_id=6479266&width=400&height=224&property=mlb',
@@ -107,7 +168,7 @@ class MLBIE(InfoExtractor):
'only_matching': True,
},
{
- 'url': 'http://m.cardinals.mlb.com/stl/video/v51175783/atlstl-piscotty-makes-great-sliding-catch-on-line/?partnerId=as_mlb_20150321_42500876&adbid=579409712979910656&adbpl=tw&adbpr=52847728',
+ 'url': 'https://www.mlb.com/cardinals/video/piscottys-great-sliding-catch/c-51175783',
'only_matching': True,
},
{
@@ -115,59 +176,92 @@ class MLBIE(InfoExtractor):
'url': 'http://mlb.mlb.com/shared/video/embed/m-internal-embed.html?content_id=75609783&property=mlb&autoplay=true&hashmode=false&siteSection=mlb/multimedia/article_118550098/article_embed&club=mlb',
'only_matching': True,
},
- {
- 'url': 'http://washington.nationals.mlb.com/mlb/gameday/index.jsp?c_id=was&gid=2015_05_09_atlmlb_wasmlb_1&lang=en&content_id=108309983&mode=video#',
- 'only_matching': True,
- }
]
+ _TIMESTAMP_KEY = 'date'
- def _real_extract(self, url):
- mobj = re.match(self._VALID_URL, url)
- video_id = mobj.group('id')
+ @staticmethod
+ def _get_feed(video):
+ return video
- if not video_id:
- video_path = mobj.group('path')
- webpage = self._download_webpage(url, video_path)
- video_id = self._search_regex(
- [r'data-video-?id="(\d+)"', r'content_id=(\d+)'], webpage, 'video id')
+ @staticmethod
+ def _extract_mlb_subtitles(feed, language):
+ subtitles = {}
+ for keyword in (feed.get('keywordsAll') or []):
+ keyword_type = keyword.get('type')
+ if keyword_type and keyword_type.startswith('closed_captions_location_'):
+ cc_location = keyword.get('value')
+ if cc_location:
+ subtitles.setdefault(language, []).append({
+ 'url': cc_location,
+ })
+ return subtitles
- detail = self._download_xml(
- 'http://m.mlb.com/gen/multimedia/detail/%s/%s/%s/%s.xml'
- % (video_id[-3], video_id[-2], video_id[-1], video_id), video_id)
+ def _download_video_data(self, display_id):
+ return self._download_json(
+ 'http://content.mlb.com/mlb/item/id/v1/%s/details/web-v1.json' % display_id,
+ display_id)
- title = detail.find('./headline').text
- description = detail.find('./big-blurb').text
- duration = parse_duration(detail.find('./duration').text)
- timestamp = parse_iso8601(detail.attrib['date'][:-5])
- thumbnails = [{
- 'url': thumbnail.text,
- } for thumbnail in detail.findall('./thumbnailScenarios/thumbnailScenario')]
+class MLBVideoIE(MLBBaseIE):
+ _VALID_URL = r'https?://(?:www\.)?mlb\.com/(?:[^/]+/)*video/(?P<id>[^/?&#]+)'
+ _TEST = {
+ 'url': 'https://www.mlb.com/mariners/video/ackley-s-spectacular-catch-c34698933',
+ 'md5': '632358dacfceec06bad823b83d21df2d',
+ 'info_dict': {
+ 'id': 'c04a8863-f569-42e6-9f87-992393657614',
+ 'ext': 'mp4',
+ 'title': "Ackley's spectacular catch",
+ 'description': 'md5:7f5a981eb4f3cbc8daf2aeffa2215bf0',
+ 'duration': 66,
+ 'timestamp': 1405995000,
+ 'upload_date': '20140722',
+ 'thumbnail': r're:^https?://.+',
+ },
+ }
+ _TIMESTAMP_KEY = 'timestamp'
- formats = []
- for media_url in detail.findall('./url'):
- playback_scenario = media_url.attrib['playback_scenario']
- fmt = {
- 'url': media_url.text,
- 'format_id': playback_scenario,
- }
- m = re.search(r'(?P<vbr>\d+)K_(?P<width>\d+)X(?P<height>\d+)', playback_scenario)
- if m:
- fmt.update({
- 'vbr': int(m.group('vbr')) * 1000,
- 'width': int(m.group('width')),
- 'height': int(m.group('height')),
- })
- formats.append(fmt)
+ @classmethod
+ def suitable(cls, url):
+ return False if MLBIE.suitable(url) else super(MLBVideoIE, cls).suitable(url)
- self._sort_formats(formats)
+ @staticmethod
+ def _get_feed(video):
+ return video['feeds'][0]
- return {
- 'id': video_id,
- 'title': title,
- 'description': description,
- 'duration': duration,
- 'timestamp': timestamp,
- 'formats': formats,
- 'thumbnails': thumbnails,
+ @staticmethod
+ def _extract_mlb_subtitles(feed, language):
+ subtitles = {}
+ for cc_location in (feed.get('closedCaptions') or []):
+ subtitles.setdefault(language, []).append({
+ 'url': cc_location,
+ })
+
+ def _download_video_data(self, display_id):
+ # https://www.mlb.com/data-service/en/videos/[SLUG]
+ return self._download_json(
+ 'https://fastball-gateway.mlb.com/graphql',
+ display_id, query={
+ 'query': '''{
+ mediaPlayback(ids: "%s") {
+ description
+ feeds(types: CMS) {
+ closedCaptions
+ duration
+ image {
+ cuts {
+ width
+ height
+ src
}
+ }
+ playbacks {
+ name
+ url
+ }
+ }
+ id
+ timestamp
+ title
+ }
+}''' % display_id,
+ })['data']['mediaPlayback'][0]
generated by cgit v1.2.3 (git 2.26.2) at 2025-11-02 07:46:34 +0000