[youtube] Fix extraction.

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

diff --git a/youtube_dl/extractor/playtvak.py b/youtube_dl/extractor/playtvak.py
index 4e5034dc61529a5a379bf7120d174175443f2676..4c5f57919b4e786b95570ce523790afd62b4e3ed 100644 (file)
--- a/youtube_dl/extractor/playtvak.py
+++ b/youtube_dl/extractor/playtvak.py
@@ -4,7 +4,7 @@ from __future__ import unicode_literals
 from .common import InfoExtractor
 from ..compat import (
     compat_urlparse,
-    compat_urllib_parse,
+    compat_urllib_parse_urlencode,
 )
 from ..utils import (
     ExtractorError,
@@ -24,8 +24,8 @@ class PlaytvakIE(InfoExtractor):
             'id': 'A150730_150323_hodinovy-manzel_kuko',
             'ext': 'mp4',
             'title': 'Vyžeňte vosy a sršně ze zahrady',
-            'description': 'md5:f93d398691044d303bc4a3de62f3e976',
-            'thumbnail': 're:(?i)^https?://.*\.(?:jpg|png)$',
+            'description': 'md5:4436e61b7df227a093778efb7e373571',
+            'thumbnail': r're:(?i)^https?://.*\.(?:jpg|png)$',
             'duration': 279,
             'timestamp': 1438732860,
             'upload_date': '20150805',
@@ -36,9 +36,19 @@ class PlaytvakIE(InfoExtractor):
         'info_dict': {
             'id': 'A150624_164934_planespotting_cat',
             'ext': 'flv',
-            'title': 're:^Přímý přenos iDNES.cz [0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2}$',
+            'title': 're:^Planespotting [0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2}$',
             'description': 'Sledujte provoz na ranveji Letiště Václava Havla v Praze',
-            'thumbnail': 're:(?i)^https?://.*\.(?:jpg|png)$',
+            'is_live': True,
+        },
+        'params': {
+            'skip_download': True,  # requires rtmpdump
+        },
+    }, {  # another live stream, this one without Misc.videoFLV
+        'url': 'https://slowtv.playtvak.cz/zive-sledujte-vlaky-v-primem-prenosu-dwi-/hlavni-nadrazi.aspx?c=A151218_145728_hlavni-nadrazi_plap',
+        'info_dict': {
+            'id': 'A151218_145728_hlavni-nadrazi_plap',
+            'ext': 'flv',
+            'title': 're:^Hlavní nádraží [0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2}$',
             'is_live': True,
         },
         'params': {
@@ -52,7 +62,7 @@ class PlaytvakIE(InfoExtractor):
             'ext': 'mp4',
             'title': 'Zavřeli jsme mraženou pizzu do auta. Upekla se',
             'description': 'md5:01e73f02329e2e5760bd5eed4d42e3c2',
-            'thumbnail': 're:(?i)^https?://.*\.(?:jpg|png)$',
+            'thumbnail': r're:(?i)^https?://.*\.(?:jpg|png)$',
             'duration': 39,
             'timestamp': 1438969140,
             'upload_date': '20150807',
@@ -66,11 +76,24 @@ class PlaytvakIE(InfoExtractor):
             'ext': 'mp4',
             'title': 'Táhni! Demonstrace proti imigrantům budila emoce',
             'description': 'md5:97c81d589a9491fbfa323c9fa3cca72c',
-            'thumbnail': 're:(?i)^https?://.*\.(?:jpg|png)$',
+            'thumbnail': r're:(?i)^https?://.*\.(?:jpg|png)$',
             'timestamp': 1439052180,
             'upload_date': '20150808',
             'is_live': False,
         }
+    }, {  # metro.cz
+        'url': 'http://www.metro.cz/video-pod-billboardem-se-na-vltavske-roztocil-kolotoc-deti-vozil-jen-par-hodin-1hx-/metro-extra.aspx?c=A141111_173251_metro-extra_row',
+        'md5': '84fc1deedcac37b7d4a6ccae7c716668',
+        'info_dict': {
+            'id': 'A141111_173251_metro-extra_row',
+            'ext': 'mp4',
+            'title': 'Recesisté udělali z billboardu kolotoč',
+            'description': 'md5:7369926049588c3989a66c9c1a043c4c',
+            'thumbnail': r're:(?i)^https?://.*\.(?:jpg|png)$',
+            'timestamp': 1415725500,
+            'upload_date': '20141111',
+            'is_live': False,
+        }
     }, {
         'url': 'http://www.playtvak.cz/embed.aspx?idvideo=V150729_141549_play-porad_kuko',
         'only_matching': True,
@@ -82,7 +105,7 @@ class PlaytvakIE(InfoExtractor):
         webpage = self._download_webpage(url, video_id)
 
         info_url = self._html_search_regex(
-            r'Misc\.videoFLV\(\s*{\s*data\s*:\s*"([^"]+)"', webpage, 'info url')
+            r'Misc\.video(?:FLV)?\(\s*{\s*data\s*:\s*"([^"]+)"', webpage, 'info url')
 
         parsed_url = compat_urlparse.urlparse(info_url)
 
@@ -93,7 +116,7 @@ class PlaytvakIE(InfoExtractor):
         })
 
         info_url = compat_urlparse.urlunparse(
-            parsed_url._replace(query = compat_urllib_parse.urlencode(qs, True)))
+            parsed_url._replace(query=compat_urllib_parse_urlencode(qs, True)))
 
         json_info = self._download_json(
             info_url, video_id,
@@ -107,7 +130,7 @@ class PlaytvakIE(InfoExtractor):
         if not item:
             raise ExtractorError('No suitable stream found')
 
-        quality = qualities(['low', 'middle', 'high'])
+        quality = qualities(('low', 'middle', 'high'))
 
         formats = []
         for fmt in item['video']:
@@ -119,7 +142,7 @@ class PlaytvakIE(InfoExtractor):
             format_id = '%s_%s' % (format_, fmt['quality'])
             preference = None
 
-            if format_ in ['mp4', 'webm']:
+            if format_ in ('mp4', 'webm'):
                 ext = format_
             elif format_ == 'rtmp':
                 ext = 'flv'
@@ -146,6 +169,8 @@ class PlaytvakIE(InfoExtractor):
         is_live = item['type'] == 'stream'
         if is_live:
             title = self._live_title(title)
+        description = self._og_search_description(webpage, default=None) or self._html_search_meta(
+            'description', webpage, 'description', default=None)
         timestamp = None
         duration = None
         if not is_live:
@@ -157,7 +182,7 @@ class PlaytvakIE(InfoExtractor):
         return {
             'id': video_id,
             'title': title,
-            'description': self._og_search_description(webpage),
+            'description': description,
             'thumbnail': item.get('image'),
             'duration': duration,
             'timestamp': timestamp,