[youtube] Fix extraction.

[youtube-dl] / youtube_dl / extractor / iprima.py

diff --git a/youtube_dl/extractor/iprima.py b/youtube_dl/extractor/iprima.py
index da2cdc656ac90f15a575eceabf33309b084c8f28..648ae6741f11f27ae297aed06f41b0584972b395 100644 (file)
--- a/youtube_dl/extractor/iprima.py
+++ b/youtube_dl/extractor/iprima.py
@@ -8,20 +8,30 @@ from .common import InfoExtractor
 from ..utils import (
     determine_ext,
     js_to_json,
-    sanitized_Request,
 )
 
 
 class IPrimaIE(InfoExtractor):
-    _VALID_URL = r'https?://play\.iprima\.cz/(?:.+/)?(?P<id>[^?#]+)'
+    _VALID_URL = r'https?://(?:[^/]+)\.iprima\.cz/(?:[^/]+/)*(?P<id>[^/?#&]+)'
+    _GEO_BYPASS = False
 
     _TESTS = [{
-        'url': 'http://play.iprima.cz/gondici-s-r-o-33',
+        'url': 'https://prima.iprima.cz/particka/92-epizoda',
         'info_dict': {
-            'id': 'p136534',
+            'id': 'p51388',
             'ext': 'mp4',
-            'title': 'Gondíci s. r. o. (34)',
-            'description': 'md5:16577c629d006aa91f59ca8d8e7f99bd',
+            'title': 'Partička (92)',
+            'description': 'md5:859d53beae4609e6dd7796413f1b6cac',
+        },
+        'params': {
+            'skip_download': True,  # m3u8 download
+        },
+    }, {
+        'url': 'https://cnn.iprima.cz/videa/70-epizoda',
+        'info_dict': {
+            'id': 'p681554',
+            'ext': 'mp4',
+            'title': 'HLAVNÍ ZPRÁVY 3.5.2020',
         },
         'params': {
             'skip_download': True,  # m3u8 download
@@ -29,20 +39,64 @@ class IPrimaIE(InfoExtractor):
     }, {
         'url': 'http://play.iprima.cz/particka/particka-92',
         'only_matching': True,
+    }, {
+        # geo restricted
+        'url': 'http://play.iprima.cz/closer-nove-pripady/closer-nove-pripady-iv-1',
+        'only_matching': True,
+    }, {
+        # iframe api.play-backend.iprima.cz
+        'url': 'https://prima.iprima.cz/my-little-pony/mapa-znameni-2-2',
+        'only_matching': True,
+    }, {
+        # iframe prima.iprima.cz
+        'url': 'https://prima.iprima.cz/porady/jak-se-stavi-sen/rodina-rathousova-praha',
+        'only_matching': True,
+    }, {
+        'url': 'http://www.iprima.cz/filmy/desne-rande',
+        'only_matching': True,
+    }, {
+        'url': 'https://zoom.iprima.cz/10-nejvetsich-tajemstvi-zahad/posvatna-mista-a-stavby',
+        'only_matching': True,
+    }, {
+        'url': 'https://krimi.iprima.cz/mraz-0/sebevrazdy',
+        'only_matching': True,
+    }, {
+        'url': 'https://cool.iprima.cz/derava-silnice-nevadi',
+        'only_matching': True,
+    }, {
+        'url': 'https://love.iprima.cz/laska-az-za-hrob/slib-dany-bratrovi',
+        'only_matching': True,
+    }, {
+        'url': 'https://autosalon.iprima.cz/motorsport/7-epizoda-1',
+        'only_matching': True,
     }]
 
     def _real_extract(self, url):
         video_id = self._match_id(url)
 
-        webpage = self._download_webpage(url, video_id)
+        self._set_cookie('play.iprima.cz', 'ott_adult_confirmed', '1')
 
-        video_id = self._search_regex(r'data-product="([^"]+)">', webpage, 'real id')
+        webpage = self._download_webpage(url, video_id)
 
-        req = sanitized_Request(
-            'http://play.iprima.cz/prehravac/init?_infuse=1'
-            '&_ts=%s&productId=%s' % (round(time.time()), video_id))
-        req.add_header('Referer', url)
-        playerpage = self._download_webpage(req, video_id, note='Downloading player')
+        title = self._og_search_title(
+            webpage, default=None) or self._search_regex(
+            r'<h1>([^<]+)', webpage, 'title')
+
+        video_id = self._search_regex(
+            (r'<iframe[^>]+\bsrc=["\'](?:https?:)?//(?:api\.play-backend\.iprima\.cz/prehravac/embedded|prima\.iprima\.cz/[^/]+/[^/]+)\?.*?\bid=(p\d+)',
+             r'data-product="([^"]+)">',
+             r'id=["\']player-(p\d+)"',
+             r'playerId\s*:\s*["\']player-(p\d+)',
+             r'\bvideos\s*=\s*["\'](p\d+)'),
+            webpage, 'real id')
+
+        playerpage = self._download_webpage(
+            'http://play.iprima.cz/prehravac/init',
+            video_id, note='Downloading player', query={
+                '_infuse': 1,
+                '_ts': round(time.time()),
+                'productId': video_id,
+            }, headers={'Referer': url})
 
         formats = []
 
@@ -65,7 +119,7 @@ class IPrimaIE(InfoExtractor):
 
         options = self._parse_json(
             self._search_regex(
-                r'(?s)var\s+playerOptions\s*=\s*({.+?});',
+                r'(?s)(?:TDIPlayerOptions|playerOptions)\s*=\s*({.+?});\s*\]\]',
                 playerpage, 'player options', default='{}'),
             video_id, transform_source=js_to_json, fatal=False)
         if options:
@@ -82,14 +136,14 @@ class IPrimaIE(InfoExtractor):
                 extract_formats(src)
 
         if not formats and '>GEO_IP_NOT_ALLOWED<' in playerpage:
-            self.raise_geo_restricted()
+            self.raise_geo_restricted(countries=['CZ'])
 
         self._sort_formats(formats)
 
         return {
             'id': video_id,
-            'title': self._og_search_title(webpage),
-            'thumbnail': self._og_search_thumbnail(webpage),
+            'title': title,
+            'thumbnail': self._og_search_thumbnail(webpage, default=None),
             'formats': formats,
-            'description': self._og_search_description(webpage),
+            'description': self._og_search_description(webpage, default=None),
         }