[zattoo] Adopt to website redesign (closes #22258)

youtube_dl/extractor/extractors.py

@@ -1512,7 +1512,6 @@ from .zattoo import (
     VTXTVIE,
     WalyTVIE,
     ZattooIE,
-    ZattooLiveIE,
 )
 from .zdf import ZDFIE, ZDFChannelIE
 from .zingmp3 import ZingMp3IE

youtube_dl/extractor/zattoo.py

@@ -58,7 +58,8 @@ class ZattooPlatformBaseIE(InfoExtractor):
             r'appToken\s*=\s*(["\'])(?P<token>(?:(?!\1).)+?)\1',
             webpage, 'app token', group='token')
         app_version = self._html_search_regex(
-            r'<!--\w+-(.+?)-', webpage, 'app version', default='2.8.2')
+            r'<!--\w+-(.+?)-', webpage,
+            'app version', default='2.8.2').lstrip('v')
 
         # Will setup appropriate cookies
         self._request_webpage(
@@ -73,6 +74,20 @@ class ZattooPlatformBaseIE(InfoExtractor):
 
         self._login()
 
+    def _extract_video_id_from_recording(self, recid):
+        playlist = self._download_json(
+            '%s/zapi/v2/playlist' % self._host_url(),
+            recid, 'Downloading playlist')
+        try:
+            recordings = playlist['recordings']
+            return next(
+                compat_str(item['program_id']) for item in recordings
+                if item.get('program_id') and compat_str(
+                    item.get('id')) == recid
+            )
+        except (StopIteration, KeyError):
+            raise ExtractorError('Could not extract video id from recording')
+
     def _extract_cid(self, video_id, channel_name):
         channel_groups = self._download_json(
             '%s/zapi/v2/cached/channels/%s' % (self._host_url(),
@@ -121,7 +136,24 @@ class ZattooPlatformBaseIE(InfoExtractor):
 
         return cid, info_dict
 
-    def _extract_formats(self, cid, video_id, record_id=None, is_live=False):
+    def _extract_ondemand_info(self, ondemand_id):
+        data = self._download_json(
+            '%s/zapi/avod/videos/%s' % (self._host_url(), ondemand_id),
+            ondemand_id, 'Downloading ondemand information'
+        )
+        info_dict = {
+            'id': ondemand_id,
+            'title': data['title'],
+            'description': data.get('description'),
+            'duration': int_or_none(data.get('duration')),
+            'release_year': int_or_none(data.get('year')),
+            'episode_number': int_or_none(data.get('episode_number')),
+            'season_number': int_or_none(data.get('season_number')),
+            'categories': try_get(data, lambda x: x['categories'], list),
+        }
+        return info_dict
+
+    def _extract_formats(self, cid, video_id, record_id=None, ondemand_id=None, is_live=False):
         postdata_common = {
             'https_watch_urls': True,
         }
@@ -131,6 +163,8 @@ class ZattooPlatformBaseIE(InfoExtractor):
             url = '%s/zapi/watch/live/%s' % (self._host_url(), cid)
         elif record_id:
             url = '%s/zapi/watch/recording/%s' % (self._host_url(), record_id)
+        elif ondemand_id:
+            url = '%s/zapi/avod/videos/%s/watch' % (self._host_url(), ondemand_id)
         else:
             url = '%s/zapi/watch/recall/%s/%s' % (self._host_url(), cid, video_id)
 
@@ -187,18 +221,34 @@ class ZattooPlatformBaseIE(InfoExtractor):
         self._sort_formats(formats)
         return formats
 
-    def _extract_video(self, channel_name, video_id, record_id=None, is_live=False):
+    def _extract_video(self, channel_name, video_id, record_id=None):
-        if is_live:
+        cid, info_dict = self._extract_cid_and_video_info(video_id)
-            cid = self._extract_cid(video_id, channel_name)
+        formats = self._extract_formats(cid, video_id, record_id=record_id)
+        info_dict['formats'] = formats
+        return info_dict
+
+    def _extract_live(self, channel_name):
+        cid = self._extract_cid(channel_name, channel_name)
         info_dict = {
             'id': channel_name,
             'title': self._live_title(channel_name),
             'is_live': True,
         }
-        else:
+        formats = self._extract_formats(cid, cid, is_live=True)
+        info_dict['formats'] = formats
+        return info_dict
+
+    def _extract_record(self, record_id):
+        video_id = self._extract_video_id_from_recording(record_id)
         cid, info_dict = self._extract_cid_and_video_info(video_id)
+        formats = self._extract_formats(cid, video_id, record_id=record_id)
+        info_dict['formats'] = formats
+        return info_dict
+
+    def _extract_ondemand(self, ondemand_id):
+        info_dict = self._extract_ondemand_info(ondemand_id)
         formats = self._extract_formats(
-            cid, video_id, record_id=record_id, is_live=is_live)
+            None, ondemand_id, ondemand_id=ondemand_id)
         info_dict['formats'] = formats
         return info_dict
 
@@ -234,8 +284,8 @@ class QuicklineLiveIE(QuicklineBaseIE):
         return False if QuicklineIE.suitable(url) else super(QuicklineLiveIE, cls).suitable(url)
 
     def _real_extract(self, url):
-        channel_name = video_id = self._match_id(url)
+        channel_name = self._match_id(url)
-        return self._extract_video(channel_name, video_id, is_live=True)
+        return self._extract_live(channel_name)
 
 
 class ZattooBaseIE(ZattooPlatformBaseIE):
@@ -248,6 +298,50 @@ def _make_valid_url(tmpl, host):
 
 
 class ZattooIE(ZattooBaseIE):
+    _VALID_URL_TEMPLATE = r'''(?x)
+                            https?://(?:www\.)?%s/
+                            (?:
+                                .+\?recording=(?P<recid>[0-9]+)|
+                                .+\?video=(?P<ondemandid>[A-Za-z0-9]+)|
+                                .+\?channel=(?P<channelid>[^&]+)(?:
+                                    &program=(?P<programid>\d+)
+                                )?
+                            )'''
+    _VALID_URL = _make_valid_url(_VALID_URL_TEMPLATE, ZattooBaseIE._HOST)
+    _TESTS = [{
+        'url': 'https://zattoo.com/channels/german?channel=srf_zwei',
+        'only_matching': True,
+    }, {
+        'url': 'https://zattoo.com/guide/german?channel=srf1&program=169860555',
+        'only_matching': True,
+    }, {
+        'url': 'https://zattoo.com/channels/german?channel=3plus&program=169860178',
+        'only_matching': True,
+    }, {
+        'url': 'https://zattoo.com/recordings?recording=193615508',
+        'only_matching': True,
+    }, {
+        'url': 'https://zattoo.com/tc/ptc_recordings_all_recordings?recording=193615420',
+        'only_matching': True,
+    }, {
+        'url': 'https://zattoo.com/ondemand?video=dvSHj79rsRqHuAHFqsYZHdox',
+        'only_matching': True,
+    }]
+
+    def _real_extract(self, url):
+        record_id, ondemand_id, channel_id, program_id = re.match(
+            self._VALID_URL, url).groups()
+        if record_id:
+            return self._extract_record(record_id)
+        elif ondemand_id:
+            return self._extract_ondemand(ondemand_id)
+        elif channel_id:
+            if program_id:
+                return self._extract_video(channel_id, program_id)
+            return self._extract_live(channel_id)
+
+
+class ZattooOldIE(ZattooBaseIE):
     _VALID_URL_TEMPLATE = r'https?://(?:www\.)?%s/watch/(?P<channel>[^/]+?)/(?P<id>[0-9]+)[^/]+(?:/(?P<recid>[0-9]+))?'
     _VALID_URL = _make_valid_url(_VALID_URL_TEMPLATE, ZattooBaseIE._HOST)
 
@@ -262,11 +356,13 @@ class ZattooIE(ZattooBaseIE):
     }]
 
     def _real_extract(self, url):
-        channel_name, video_id, record_id = re.match(self._VALID_URL, url).groups()
+        channel_name, video_id, record_id = re.match(
-        return self._extract_video(channel_name, video_id, record_id)
+            self._VALID_URL, url).groups()
+        return self._extract_video(
+            channel_name, video_id, record_id=record_id)
 
 
-class ZattooLiveIE(ZattooBaseIE):
+class ZattooOldLiveIE(ZattooBaseIE):
     _VALID_URL = r'https?://(?:www\.)?zattoo\.com/watch/(?P<id>[^/]+)'
 
     _TEST = {
@@ -276,18 +372,18 @@ class ZattooLiveIE(ZattooBaseIE):
 
     @classmethod
     def suitable(cls, url):
-        return False if ZattooIE.suitable(url) else super(ZattooLiveIE, cls).suitable(url)
+        return False if ZattooOldIE.suitable(url) else super(ZattooOldLiveIE, cls).suitable(url)
 
     def _real_extract(self, url):
-        channel_name = video_id = self._match_id(url)
+        channel_name = self._match_id(url)
-        return self._extract_video(channel_name, video_id, is_live=True)
+        return self._extract_live(channel_name)
 
 
-class NetPlusIE(ZattooIE):
+class NetPlusIE(ZattooOldIE):
     _NETRC_MACHINE = 'netplus'
     _HOST = 'netplus.tv'
     _API_HOST = 'www.%s' % _HOST
-    _VALID_URL = _make_valid_url(ZattooIE._VALID_URL_TEMPLATE, _HOST)
+    _VALID_URL = _make_valid_url(ZattooOldIE._VALID_URL_TEMPLATE, _HOST)
 
     _TESTS = [{
         'url': 'https://www.netplus.tv/watch/abc/123-abc',
@@ -295,10 +391,10 @@ class NetPlusIE(ZattooIE):
     }]
 
 
-class MNetTVIE(ZattooIE):
+class MNetTVIE(ZattooOldIE):
     _NETRC_MACHINE = 'mnettv'
     _HOST = 'tvplus.m-net.de'
-    _VALID_URL = _make_valid_url(ZattooIE._VALID_URL_TEMPLATE, _HOST)
+    _VALID_URL = _make_valid_url(ZattooOldIE._VALID_URL_TEMPLATE, _HOST)
 
     _TESTS = [{
         'url': 'https://tvplus.m-net.de/watch/abc/123-abc',
@@ -306,10 +402,10 @@ class MNetTVIE(ZattooIE):
     }]
 
 
-class WalyTVIE(ZattooIE):
+class WalyTVIE(ZattooOldIE):
     _NETRC_MACHINE = 'walytv'
     _HOST = 'player.waly.tv'
-    _VALID_URL = _make_valid_url(ZattooIE._VALID_URL_TEMPLATE, _HOST)
+    _VALID_URL = _make_valid_url(ZattooOldIE._VALID_URL_TEMPLATE, _HOST)
 
     _TESTS = [{
         'url': 'https://player.waly.tv/watch/abc/123-abc',
@@ -317,11 +413,11 @@ class WalyTVIE(ZattooIE):
     }]
 
 
-class BBVTVIE(ZattooIE):
+class BBVTVIE(ZattooOldIE):
     _NETRC_MACHINE = 'bbvtv'
     _HOST = 'bbv-tv.net'
     _API_HOST = 'www.%s' % _HOST
-    _VALID_URL = _make_valid_url(ZattooIE._VALID_URL_TEMPLATE, _HOST)
+    _VALID_URL = _make_valid_url(ZattooOldIE._VALID_URL_TEMPLATE, _HOST)
 
     _TESTS = [{
         'url': 'https://www.bbv-tv.net/watch/abc/123-abc',
@@ -329,11 +425,11 @@ class BBVTVIE(ZattooIE):
     }]
 
 
-class VTXTVIE(ZattooIE):
+class VTXTVIE(ZattooOldIE):
     _NETRC_MACHINE = 'vtxtv'
     _HOST = 'vtxtv.ch'
     _API_HOST = 'www.%s' % _HOST
-    _VALID_URL = _make_valid_url(ZattooIE._VALID_URL_TEMPLATE, _HOST)
+    _VALID_URL = _make_valid_url(ZattooOldIE._VALID_URL_TEMPLATE, _HOST)
 
     _TESTS = [{
         'url': 'https://www.vtxtv.ch/watch/abc/123-abc',
@@ -341,11 +437,11 @@ class VTXTVIE(ZattooIE):
     }]
 
 
-class MyVisionTVIE(ZattooIE):
+class MyVisionTVIE(ZattooOldIE):
     _NETRC_MACHINE = 'myvisiontv'
     _HOST = 'myvisiontv.ch'
     _API_HOST = 'www.%s' % _HOST
-    _VALID_URL = _make_valid_url(ZattooIE._VALID_URL_TEMPLATE, _HOST)
+    _VALID_URL = _make_valid_url(ZattooOldIE._VALID_URL_TEMPLATE, _HOST)
 
     _TESTS = [{
         'url': 'https://www.myvisiontv.ch/watch/abc/123-abc',
@@ -353,10 +449,10 @@ class MyVisionTVIE(ZattooIE):
     }]
 
 
-class GlattvisionTVIE(ZattooIE):
+class GlattvisionTVIE(ZattooOldIE):
     _NETRC_MACHINE = 'glattvisiontv'
     _HOST = 'iptv.glattvision.ch'
-    _VALID_URL = _make_valid_url(ZattooIE._VALID_URL_TEMPLATE, _HOST)
+    _VALID_URL = _make_valid_url(ZattooOldIE._VALID_URL_TEMPLATE, _HOST)
 
     _TESTS = [{
         'url': 'https://iptv.glattvision.ch/watch/abc/123-abc',
@@ -364,11 +460,11 @@ class GlattvisionTVIE(ZattooIE):
     }]
 
 
-class SAKTVIE(ZattooIE):
+class SAKTVIE(ZattooOldIE):
     _NETRC_MACHINE = 'saktv'
     _HOST = 'saktv.ch'
     _API_HOST = 'www.%s' % _HOST
-    _VALID_URL = _make_valid_url(ZattooIE._VALID_URL_TEMPLATE, _HOST)
+    _VALID_URL = _make_valid_url(ZattooOldIE._VALID_URL_TEMPLATE, _HOST)
 
     _TESTS = [{
         'url': 'https://www.saktv.ch/watch/abc/123-abc',
@@ -376,10 +472,10 @@ class SAKTVIE(ZattooIE):
     }]
 
 
-class EWETVIE(ZattooIE):
+class EWETVIE(ZattooOldIE):
     _NETRC_MACHINE = 'ewetv'
     _HOST = 'tvonline.ewe.de'
-    _VALID_URL = _make_valid_url(ZattooIE._VALID_URL_TEMPLATE, _HOST)
+    _VALID_URL = _make_valid_url(ZattooOldIE._VALID_URL_TEMPLATE, _HOST)
 
     _TESTS = [{
         'url': 'https://tvonline.ewe.de/watch/abc/123-abc',
@@ -387,11 +483,11 @@ class EWETVIE(ZattooIE):
     }]
 
 
-class QuantumTVIE(ZattooIE):
+class QuantumTVIE(ZattooOldIE):
     _NETRC_MACHINE = 'quantumtv'
     _HOST = 'quantum-tv.com'
     _API_HOST = 'www.%s' % _HOST
-    _VALID_URL = _make_valid_url(ZattooIE._VALID_URL_TEMPLATE, _HOST)
+    _VALID_URL = _make_valid_url(ZattooOldIE._VALID_URL_TEMPLATE, _HOST)
 
     _TESTS = [{
         'url': 'https://www.quantum-tv.com/watch/abc/123-abc',
@@ -399,10 +495,10 @@ class QuantumTVIE(ZattooIE):
     }]
 
 
-class OsnatelTVIE(ZattooIE):
+class OsnatelTVIE(ZattooOldIE):
     _NETRC_MACHINE = 'osnateltv'
     _HOST = 'tvonline.osnatel.de'
-    _VALID_URL = _make_valid_url(ZattooIE._VALID_URL_TEMPLATE, _HOST)
+    _VALID_URL = _make_valid_url(ZattooOldIE._VALID_URL_TEMPLATE, _HOST)
 
     _TESTS = [{
         'url': 'https://tvonline.osnatel.de/watch/abc/123-abc',
@@ -410,11 +506,11 @@ class OsnatelTVIE(ZattooIE):
     }]
 
 
-class EinsUndEinsTVIE(ZattooIE):
+class EinsUndEinsTVIE(ZattooOldIE):
     _NETRC_MACHINE = '1und1tv'
     _HOST = '1und1.tv'
     _API_HOST = 'www.%s' % _HOST
-    _VALID_URL = _make_valid_url(ZattooIE._VALID_URL_TEMPLATE, _HOST)
+    _VALID_URL = _make_valid_url(ZattooOldIE._VALID_URL_TEMPLATE, _HOST)
 
     _TESTS = [{
         'url': 'https://www.1und1.tv/watch/abc/123-abc',
@@ -422,10 +518,10 @@ class EinsUndEinsTVIE(ZattooIE):
     }]
 
 
-class SaltTVIE(ZattooIE):
+class SaltTVIE(ZattooOldIE):
     _NETRC_MACHINE = 'salttv'
     _HOST = 'tv.salt.ch'
-    _VALID_URL = _make_valid_url(ZattooIE._VALID_URL_TEMPLATE, _HOST)
+    _VALID_URL = _make_valid_url(ZattooOldIE._VALID_URL_TEMPLATE, _HOST)
 
     _TESTS = [{
         'url': 'https://tv.salt.ch/watch/abc/123-abc',