Update upstream source from tag 'upstream/2020.05.08'

[youtubedl] / youtube_dl / extractor / iprima.py

diff --git a/youtube_dl/extractor/iprima.py b/youtube_dl/extractor/iprima.py
index 1d58d6e850724f226d66f5822777010f8a8b8d38..53a550c11e4407813deb12f646a0c714436862b5 100644 (file)
--- a/youtube_dl/extractor/iprima.py
+++ b/youtube_dl/extractor/iprima.py
@@ -12,16 +12,26 @@ from ..utils import (
 
 
 class IPrimaIE(InfoExtractor):
 
 
 class IPrimaIE(InfoExtractor):

-    _VALID_URL = r'https?://(?:play|prima)\.iprima\.cz/(?:.+/)?(?P<id>[^?#]+)'
+    _VALID_URL = r'https?://(?:[^/]+)\.iprima\.cz/(?:[^/]+/)*(?P<id>[^/?#&]+)'

     _GEO_BYPASS = False
 
     _TESTS = [{
     _GEO_BYPASS = False
 
     _TESTS = [{

-        'url': 'http://play.iprima.cz/gondici-s-r-o-33',
+        'url': 'https://prima.iprima.cz/particka/92-epizoda',

         'info_dict': {
         'info_dict': {

-            'id': 'p136534',
+            'id': 'p51388',

             'ext': 'mp4',
             '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
         },
         'params': {
             'skip_download': True,  # m3u8 download


@@ -41,6 +51,24 @@ class IPrimaIE(InfoExtractor):
         # iframe prima.iprima.cz
         'url': 'https://prima.iprima.cz/porady/jak-se-stavi-sen/rodina-rathousova-praha',
         '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):
     }]
 
     def _real_extract(self, url):


@@ -50,9 +78,15 @@ class IPrimaIE(InfoExtractor):
 
         webpage = self._download_webpage(url, video_id)
 
 
         webpage = self._download_webpage(url, video_id)
 

+        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+)',
         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'data-product="([^"]+)">',
+             r'id=["\']player-(p\d+)"',
+             r'playerId\s*:\s*["\']player-(p\d+)'),

             webpage, 'real id')
 
         playerpage = self._download_webpage(
             webpage, 'real id')
 
         playerpage = self._download_webpage(


@@ -107,8 +141,8 @@ class IPrimaIE(InfoExtractor):
 
         return {
             'id': video_id,
 
         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,
             'formats': formats,

-            'description': self._og_search_description(webpage),
+            'description': self._og_search_description(webpage, default=None),

         }
         }