[ctsnews] Detect youtube embedde videos

youtube_dl/extractor/ctsnews.py

@@ -1,13 +1,27 @@
-# coding: utf-8
+# -*- coding: utf-8 -*-
 from __future__ import unicode_literals
 
 from .common import InfoExtractor
+from ..compat import compat_urllib_request
 from ..utils import parse_iso8601, ExtractorError
 
 
 class CtsNewsIE(InfoExtractor):
+    # https connection failed (Connection reset)
     _VALID_URL = r'http://news\.cts\.com\.tw/[a-z]+/[a-z]+/\d+/(?P<id>\d+)\.html'
     _TESTS = [{
+        'url': 'http://news.cts.com.tw/cts/international/201501/201501291578109.html',
+        'md5': 'a9875cb790252b08431186d741beaabe',
+        'info_dict': {
+            'id': '201501291578109',
+            'ext': 'mp4',
+            'title': '以色列.真主黨交火 3人死亡',
+            'description': 'md5:95e9b295c898b7ff294f09d450178d7d',
+            'timestamp': 1422528540,
+            'upload_date': '20150129',
+        }
+    }, {
+        # News count not appear on page but still available in database
         'url': 'http://news.cts.com.tw/cts/international/201309/201309031304098.html',
         'md5': '3aee7e0df7cdff94e43581f54c22619e',
         'info_dict': {
@@ -19,33 +33,70 @@ class CtsNewsIE(InfoExtractor):
             'timestamp': 1378205880,
             'upload_date': '20130903',
         }
+    }, {
+        # With Youtube embedded video
+        'url': 'http://news.cts.com.tw/cts/money/201501/201501291578003.html',
+        'md5': '1d842c771dc94c8c3bca5af2cc1db9c5',
+        'add_ie': ['Youtube'],
+        'info_dict': {
+            'id': 'OVbfO7d0_hQ',
+            'ext': 'mp4',
+            'title': 'iPhone6熱銷 蘋果財報亮眼',
+            'description': 'md5:f395d4f485487bb0f992ed2c4b07aa7d',
+            'thumbnail': 're:^https?://.*\.jpg$',
+            'upload_date': '20150128',
+            'uploader_id': 'TBSCTS',
+            'uploader': '中華電視公司',
+        }
     }]
 
     def _real_extract(self, url):
         news_id = self._match_id(url)
         page = self._download_webpage(url, news_id)
 
-        if not self._search_regex(r'(CTSPlayer2)', page, 'CTSPlayer2 identifier', fatal=False):
+        if self._search_regex(r'(CTSPlayer2)', page, 'CTSPlayer2 identifier', default=None):
-            raise ExtractorError('The news includes no videos!')
+            feed_url = self._html_search_regex(
+                r'(http://news\.cts\.com\.tw/action/mp4feed\.php\?news_id=\d+)',
+                page, 'feed url')
+            video_url = self._download_webpage(feed_url, news_id)
+        else:
+            self.to_screen('Not CTSPlayer video, trying Youtube...')
+            youtube_url = self._search_regex(
+                r'src="(//www\.youtube\.com/embed/[^"]+)"', page, 'youtube url',
+                default=None)
+            if not youtube_url:
+                raise ExtractorError('The news includes no videos!', expected=True)
 
-        feed_pattern = r'(http://news.cts.com.tw/action/mp4feed.php\?news_id=\d+)'
+            return {
-        feed_url = self._html_search_regex(feed_pattern, page, 'feed url')
+                '_type': 'url',
-        feed_page = self._download_webpage(feed_url, news_id)
+                'url': youtube_url,
+                'ie_key': 'Youtube',
+            }
 
         description = self._html_search_meta('description', page)
         title = self._html_search_meta('title', page)
         thumbnail = self._html_search_meta('image', page)
 
-        datetime_pattern = r'(\d{4}/\d{2}/\d{2} \d{2}:\d{2})'
+        datetime_str = self._html_search_regex(
-        datetime_str = self._html_search_regex(datetime_pattern, page, 'date and time')
+            r'(\d{4}/\d{2}/\d{2} \d{2}:\d{2})', page, 'date and time')
-        time = (datetime_str + ':00+08:00').replace('/', '-')
+        # Transform into ISO 8601 format with timezone info
-        timestamp = parse_iso8601(time, delimiter=' ')
+        datetime_str = datetime_str.replace('/', '-') + ':00+0800'
+        timestamp = parse_iso8601(datetime_str, delimiter=' ')
+
+        # Note: the news count may decrease as time goes by
+        # It should be a bug in CTS website
+        req = compat_urllib_request.Request(
+            'http://news.cts.com.tw/action/news_count.php?callback=cb&news_id=' + news_id)
+        req.add_header('Referer', url)
+        newscount_page = self._download_webpage(req, news_id)
+        news_count = self._search_regex(r'cb\((\d+)\)', newscount_page, 'news count')
 
         return {
             'id': news_id,
+            'url': video_url,
             'title': title,
             'description': description,
-            'url': feed_page,
             'thumbnail': thumbnail,
-            'timestamp': timestamp
+            'timestamp': timestamp,
+            'view_count': news_count,
         }