]> git.bitcoin.ninja Git - youtube-dl/commitdiff
[echomsk] Add extractor (Closes #4600)
authorSergey M․ <dstftw@gmail.com>
Wed, 31 Dec 2014 12:03:51 +0000 (18:03 +0600)
committerSergey M․ <dstftw@gmail.com>
Wed, 31 Dec 2014 12:03:51 +0000 (18:03 +0600)
youtube_dl/extractor/__init__.py
youtube_dl/extractor/echomsk.py [new file with mode: 0644]

index c15786ad7f6823adc3ce54647d95dabc63899b53..45b50792e5b6154a690f9e13d4222b5896b01512 100644 (file)
@@ -99,6 +99,7 @@ from .discovery import DiscoveryIE
 from .divxstage import DivxStageIE
 from .dropbox import DropboxIE
 from .ebaumsworld import EbaumsWorldIE
+from .echomsk import EchoMskIE
 from .ehow import EHowIE
 from .eighttracks import EightTracksIE
 from .einthusan import EinthusanIE
diff --git a/youtube_dl/extractor/echomsk.py b/youtube_dl/extractor/echomsk.py
new file mode 100644 (file)
index 0000000..d2d9404
--- /dev/null
@@ -0,0 +1,46 @@
+# coding: utf-8
+from __future__ import unicode_literals
+
+import re
+
+from .common import InfoExtractor
+
+
+class EchoMskIE(InfoExtractor):
+    _VALID_URL = r'http://(?:www\.)?echo\.msk\.ru/sounds/(?P<id>\d+)'
+    _TEST = {
+        'url': 'http://www.echo.msk.ru/sounds/1464134.html',
+        'md5': '2e44b3b78daff5b458e4dbc37f191f7c',
+        'info_dict': {
+            'id': '1464134',
+            'ext': 'mp3',
+            'title': 'Особое мнение - 29 декабря 2014, 19:08',
+        },
+    }
+
+    def _real_extract(self, url):
+        video_id = self._match_id(url)
+
+        webpage = self._download_webpage(url, video_id)
+
+        audio_url = self._search_regex(
+            r'<a rel="mp3" href="([^"]+)">', webpage, 'audio URL')
+
+        title = self._html_search_regex(
+            r'<a href="/programs/[^"]+" target="_blank">([^<]+)</a>',
+            webpage, 'title')
+
+        air_date = self._html_search_regex(
+            r'(?s)<div class="date">(.+?)</div>',
+            webpage, 'date', fatal=False, default=None)
+
+        if air_date:
+            air_date = re.sub(r'(\s)\1+', r'\1', air_date)
+            if air_date:
+                title = '%s - %s' % (title, air_date)
+
+        return {
+            'id': video_id,
+            'url': audio_url,
+            'title': title,
+        }