[youtube] Fix extraction.

[youtube-dl] / youtube_dl / extractor / varzesh3.py

diff --git a/youtube_dl/extractor/varzesh3.py b/youtube_dl/extractor/varzesh3.py
index 85dcd8734c35fce16b3439cbadb5473f289d1601..f474ed73f861910d9c593510a4aff6be8244e903 100644 (file)
--- a/youtube_dl/extractor/varzesh3.py
+++ b/youtube_dl/extractor/varzesh3.py
@@ -1,63 +1,79 @@
 # coding: utf-8
 from __future__ import unicode_literals
+
 from .common import InfoExtractor
+from ..compat import (
+    compat_urllib_parse_urlparse,
+    compat_parse_qs,
+)
 from ..utils import (
-    ExtractorError,
+    clean_html,
+    remove_start,
 )
-import re
 
 
 class Varzesh3IE(InfoExtractor):
-    _VALID_URL = r'(?P<url>(https?://(?:www\.)?video\.varzesh3\.com)/(?P<id>.+))'
-    _TESTS = [
-    {
-        'url': 'http://video.varzesh3.com/video/%D8%AF%D8%A7%D9%86%D9%85%D8%A7%D8%B1%DA%A9-3-2-%D8%A2%D9%85%D8%B1%DB%8C%DA%A9%D8%A7/',
-        'md5': 'c4b850780df9374b078463c5cf6b3318',
-        'info_dict': {
-            'url': 'http://dl1.video.varzesh3.com/video/clip94/1/international/friendly/usa_2_3_danmark.mp4',
-            'id': '76464',
-            'ext': 'mp4',
-            'title': u'دانمارک  ۳-۲   آمریکا (گلهای بازی)',
-            'thumbnail': 'http://video.varzesh3.com/wp-content/uploads/250315_usa_danmark_site.jpg',
-            'description': u'بازی دوستانه ۲۰۱۵',
-
-        }
-    },
-    {
+    _VALID_URL = r'https?://(?:www\.)?video\.varzesh3\.com/(?:[^/]+/)+(?P<id>[^/]+)/?'
+    _TESTS = [{
         'url': 'http://video.varzesh3.com/germany/bundesliga/5-%D9%88%D8%A7%DA%A9%D9%86%D8%B4-%D8%A8%D8%B1%D8%AA%D8%B1-%D8%AF%D8%B1%D9%88%D8%A7%D8%B2%D9%87%E2%80%8C%D8%A8%D8%A7%D9%86%D8%A7%D9%86%D8%9B%D9%87%D9%81%D8%AA%D9%87-26-%D8%A8%D9%88%D9%86%D8%AF%D8%B3/',
         'md5': '2a933874cb7dce4366075281eb49e855',
         'info_dict': {
-            'url': 'http://dl1.video.varzesh3.com/video/clip94/1/video/namayeshi/saves_week26.mp4',
             'id': '76337',
             'ext': 'mp4',
-            'title': u'۵ واکنش برتر دروازه‌بانان؛هفته ۲۶ بوندسلیگا',
-            'thumbnail': 'http://video.varzesh3.com/wp-content/uploads/230315_saves_week26.jpg',
-            'description': u'فصل ۲۰۱۵-۲۰۱۴',
-        }
-    },
-    ]
+            'title': '۵ واکنش برتر دروازه‌بانان؛هفته ۲۶ بوندسلیگا',
+            'description': 'فصل ۲۰۱۵-۲۰۱۴',
+            'thumbnail': r're:^https?://.*\.jpg$',
+        },
+        'skip': 'HTTP 404 Error',
+    }, {
+        'url': 'http://video.varzesh3.com/video/112785/%D8%AF%D9%84%D9%87-%D8%B9%D9%84%DB%8C%D8%9B-%D8%B3%D8%AA%D8%A7%D8%B1%D9%87-%D9%86%D9%88%D8%B8%D9%87%D9%88%D8%B1-%D9%84%DB%8C%DA%AF-%D8%A8%D8%B1%D8%AA%D8%B1-%D8%AC%D8%B2%DB%8C%D8%B1%D9%87',
+        'md5': '841b7cd3afbc76e61708d94e53a4a4e7',
+        'info_dict': {
+            'id': '112785',
+            'ext': 'mp4',
+            'title': 'دله علی؛ ستاره نوظهور لیگ برتر جزیره',
+            'description': 'فوتبال 120',
+        },
+        'expected_warnings': ['description'],
+    }]
 
     def _real_extract(self, url):
-        video_id = self._match_id(url)
-        webpage = self._download_webpage(url, video_id)
+        display_id = self._match_id(url)
+
+        webpage = self._download_webpage(url, display_id)
+
+        video_url = self._search_regex(
+            r'<source[^>]+src="([^"]+)"', webpage, 'video url')
+
+        title = remove_start(self._html_search_regex(
+            r'<title>([^<]+)</title>', webpage, 'title'), 'ویدیو ورزش 3 | ')
+
+        description = self._html_search_regex(
+            r'(?s)<div class="matn">(.+?)</div>',
+            webpage, 'description', default=None)
+        if description is None:
+            description = clean_html(self._html_search_meta('description', webpage))
 
-        if not 'shortlink' in webpage:
-            raise ExtractorError('URL has no videos or there is a problem.')
+        thumbnail = self._og_search_thumbnail(webpage, default=None)
+        if thumbnail is None:
+            fb_sharer_url = self._search_regex(
+                r'<a[^>]+href="(https?://www\.facebook\.com/sharer/sharer\.php?[^"]+)"',
+                webpage, 'facebook sharer URL', fatal=False)
+            sharer_params = compat_parse_qs(compat_urllib_parse_urlparse(fb_sharer_url).query)
+            thumbnail = sharer_params.get('p[images][0]', [None])[0]
 
-        title = self._html_search_regex(r'meta[^>]+property="og:title"[^>]+content="([^"]+)"', webpage, 'title')
-        video_link = self._html_search_regex(r'source[^>]+src="([^"]+)"', webpage, 'video_link')
-        vid_id = self._html_search_regex(r"link[^>]+rel='canonical'[^>]+href='\/\?p=([^']+)'\/>", webpage, 'vid_id')
-        try:
-            description = self._html_search_regex(r'<div class="matn">(.*?)</div>', webpage, 'description', flags=re.DOTALL)
-        except:
-            description = title
-        thumbnail = self._html_search_regex(r'link[^>]+rel="image_src"[^>]+href="([^"]+)"', webpage, 'thumbnail')
+        video_id = self._search_regex(
+            r"<link[^>]+rel='(?:canonical|shortlink)'[^>]+href='/\?p=([^']+)'",
+            webpage, display_id, default=None)
+        if video_id is None:
+            video_id = self._search_regex(
+                r'var\s+VideoId\s*=\s*(\d+);', webpage, 'video id',
+                default=display_id)
 
         return {
-            'url': video_link,
-            'id': vid_id,
+            'url': video_url,
+            'id': video_id,
             'title': title,
-            'ext': video_link.split(".")[-1],
             'description': description,
             'thumbnail': thumbnail,
         }