]> Raphaƫl G. Git Repositories - youtubedl/blobdiff - youtube_dl/extractor/laola1tv.py
debian/README.source: Change 'markup'.
[youtubedl] / youtube_dl / extractor / laola1tv.py
index 3190b187c9dfb8fa9204e9761b47ded0c17f5f2d..fa217365a37853f471440d8138d0e20bc4115d52 100644 (file)
@@ -1,6 +1,9 @@
 # coding: utf-8
 from __future__ import unicode_literals
 
+import json
+import re
+
 from .common import InfoExtractor
 from ..utils import (
     ExtractorError,
@@ -8,15 +11,15 @@ from ..utils import (
     urlencode_postdata,
     xpath_element,
     xpath_text,
-    urljoin,
     update_url_query,
+    js_to_json,
 )
 
 
 class Laola1TvEmbedIE(InfoExtractor):
     IE_NAME = 'laola1tv:embed'
     _VALID_URL = r'https?://(?:www\.)?laola1\.tv/titanplayer\.php\?.*?\bvideoid=(?P<id>\d+)'
-    _TEST = {
+    _TESTS = [{
         # flashvars.premium = "false";
         'url': 'https://www.laola1.tv/titanplayer.php?videoid=708065&type=V&lang=en&portal=int&customer=1024',
         'info_dict': {
@@ -26,7 +29,31 @@ class Laola1TvEmbedIE(InfoExtractor):
             'uploader': 'ITTF - International Table Tennis Federation',
             'upload_date': '20161211',
         },
-    }
+    }]
+
+    def _extract_token_url(self, stream_access_url, video_id, data):
+        return self._download_json(
+            self._proto_relative_url(stream_access_url, 'https:'), video_id,
+            headers={
+                'Content-Type': 'application/json',
+            }, data=json.dumps(data).encode())['data']['stream-access'][0]
+
+    def _extract_formats(self, token_url, video_id):
+        token_doc = self._download_xml(
+            token_url, video_id, 'Downloading token',
+            headers=self.geo_verification_headers())
+
+        token_attrib = xpath_element(token_doc, './/token').attrib
+
+        if token_attrib['status'] != '0':
+            raise ExtractorError(
+                'Token error: %s' % token_attrib['comment'], expected=True)
+
+        formats = self._extract_akamai_formats(
+            '%s?hdnea=%s' % (token_attrib['url'], token_attrib['auth']),
+            video_id)
+        self._sort_formats(formats)
+        return formats
 
     def _real_extract(self, url):
         video_id = self._match_id(url)
@@ -68,29 +95,16 @@ class Laola1TvEmbedIE(InfoExtractor):
         else:
             data_abo = urlencode_postdata(
                 dict((i, v) for i, v in enumerate(_v('req_liga_abos').split(','))))
-            token_url = self._download_json(
-                'https://club.laola1.tv/sp/laola1/api/v3/user/session/premium/player/stream-access',
-                video_id, query={
+            stream_access_url = update_url_query(
+                'https://club.laola1.tv/sp/laola1/api/v3/user/session/premium/player/stream-access', {
                     'videoId': _v('id'),
                     'target': self._search_regex(r'vs_target = (\d+);', webpage, 'vs target'),
                     'label': _v('label'),
                     'area': _v('area'),
-                }, data=data_abo)['data']['stream-access'][0]
-
-        token_doc = self._download_xml(
-            token_url, video_id, 'Downloading token',
-            headers=self.geo_verification_headers())
-
-        token_attrib = xpath_element(token_doc, './/token').attrib
-
-        if token_attrib['status'] != '0':
-            raise ExtractorError(
-                'Token error: %s' % token_attrib['comment'], expected=True)
+                })
+            token_url = self._extract_token_url(stream_access_url, video_id, data_abo)
 
-        formats = self._extract_akamai_formats(
-            '%s?hdnea=%s' % (token_attrib['url'], token_attrib['auth']),
-            video_id)
-        self._sort_formats(formats)
+        formats = self._extract_formats(token_url, video_id)
 
         categories_str = _v('meta_sports')
         categories = categories_str.split(',') if categories_str else []
@@ -107,9 +121,59 @@ class Laola1TvEmbedIE(InfoExtractor):
         }
 
 
-class Laola1TvIE(InfoExtractor):
+class Laola1TvBaseIE(Laola1TvEmbedIE):
+    def _extract_video(self, url):
+        display_id = self._match_id(url)
+        webpage = self._download_webpage(url, display_id)
+
+        if 'Dieser Livestream ist bereits beendet.' in webpage:
+            raise ExtractorError('This live stream has already finished.', expected=True)
+
+        conf = self._parse_json(self._search_regex(
+            r'(?s)conf\s*=\s*({.+?});', webpage, 'conf'),
+            display_id,
+            transform_source=lambda s: js_to_json(re.sub(r'shareurl:.+,', '', s)))
+        video_id = conf['videoid']
+
+        config = self._download_json(conf['configUrl'], video_id, query={
+            'videoid': video_id,
+            'partnerid': conf['partnerid'],
+            'language': conf.get('language', ''),
+            'portal': conf.get('portalid', ''),
+        })
+        error = config.get('error')
+        if error:
+            raise ExtractorError('%s said: %s' % (self.IE_NAME, error), expected=True)
+
+        video_data = config['video']
+        title = video_data['title']
+        is_live = video_data.get('isLivestream') and video_data.get('isLive')
+        meta = video_data.get('metaInformation')
+        sports = meta.get('sports')
+        categories = sports.split(',') if sports else []
+
+        token_url = self._extract_token_url(
+            video_data['streamAccess'], video_id,
+            video_data['abo']['required'])
+
+        formats = self._extract_formats(token_url, video_id)
+
+        return {
+            'id': video_id,
+            'display_id': display_id,
+            'title': self._live_title(title) if is_live else title,
+            'description': video_data.get('description'),
+            'thumbnail': video_data.get('image'),
+            'categories': categories,
+            'formats': formats,
+            'is_live': is_live,
+        }
+
+
+class Laola1TvIE(Laola1TvBaseIE):
     IE_NAME = 'laola1tv'
     _VALID_URL = r'https?://(?:www\.)?laola1\.tv/[a-z]+-[a-z]+/[^/]+/(?P<id>[^/?#&]+)'
+
     _TESTS = [{
         'url': 'http://www.laola1.tv/de-de/video/straubing-tigers-koelner-haie/227883.html',
         'info_dict': {
@@ -157,20 +221,45 @@ class Laola1TvIE(InfoExtractor):
     }]
 
     def _real_extract(self, url):
-        display_id = self._match_id(url)
+        return self._extract_video(url)
 
-        webpage = self._download_webpage(url, display_id)
 
-        if 'Dieser Livestream ist bereits beendet.' in webpage:
-            raise ExtractorError('This live stream has already finished.', expected=True)
+class EHFTVIE(Laola1TvBaseIE):
+    IE_NAME = 'ehftv'
+    _VALID_URL = r'https?://(?:www\.)?ehftv\.com/[a-z]+(?:-[a-z]+)?/[^/]+/(?P<id>[^/?#&]+)'
+
+    _TESTS = [{
+        'url': 'https://www.ehftv.com/int/video/paris-saint-germain-handball-pge-vive-kielce/1166761',
+        'info_dict': {
+            'id': '1166761',
+            'display_id': 'paris-saint-germain-handball-pge-vive-kielce',
+            'ext': 'mp4',
+            'title': 'Paris Saint-Germain Handball - PGE Vive Kielce',
+            'is_live': False,
+            'categories': ['Handball'],
+        },
+        'params': {
+            'skip_download': True,
+        },
+    }]
 
-        iframe_url = urljoin(url, self._search_regex(
-            r'<iframe[^>]*?id="videoplayer"[^>]*?src="([^"]+)"',
-            webpage, 'iframe url'))
+    def _real_extract(self, url):
+        return self._extract_video(url)
 
-        return {
-            '_type': 'url',
-            'display_id': display_id,
-            'url': iframe_url,
-            'ie_key': 'Laola1TvEmbed',
-        }
+
+class ITTFIE(InfoExtractor):
+    _VALID_URL = r'https?://tv\.ittf\.com/video/[^/]+/(?P<id>\d+)'
+    _TEST = {
+        'url': 'https://tv.ittf.com/video/peng-wang-wei-matsudaira-kenta/951802',
+        'only_matching': True,
+    }
+
+    def _real_extract(self, url):
+        return self.url_result(
+            update_url_query('https://www.laola1.tv/titanplayer.php', {
+                'videoid': self._match_id(url),
+                'type': 'V',
+                'lang': 'en',
+                'portal': 'int',
+                'customer': 1024,
+            }), Laola1TvEmbedIE.ie_key())