Move YahooSearchIE to youtube_dl.extractor.yahoo
[youtube-dl] / youtube_dl / extractor / yahoo.py
index 4f3af17d719443d9d19e32193e95cbb7f802acfb..4b3aec9d1f5a8ce91f676d585e5f98dd3ba126aa 100644 (file)
@@ -1,9 +1,12 @@
 import datetime
+import itertools
 import json
 import re
 
-from .common import InfoExtractor
+from .common import InfoExtractor, SearchInfoExtractor
 from ..utils import (
+    compat_urllib_parse,
+
     ExtractorError,
 )
 
@@ -74,3 +77,37 @@ class YahooIE(InfoExtractor):
                      'ext': 'flv',
                      }
         return info_dict
+
+class YahooSearchIE(SearchInfoExtractor):
+    """Information Extractor for Yahoo! Video search queries."""
+
+    _MAX_RESULTS = 1000
+    IE_NAME = u'screen.yahoo:search'
+    _SEARCH_KEY = 'yvsearch'
+
+    def _get_n_results(self, query, n):
+        """Get a specified number of results for a query"""
+
+        res = {
+            '_type': 'playlist',
+            'id': query,
+            'entries': []
+        }
+        for pagenum in itertools.count(0): 
+            result_url = u'http://video.search.yahoo.com/search/?p=%s&fr=screen&o=js&gs=0&b=%d' % (compat_urllib_parse.quote_plus(query), pagenum * 30)
+            webpage = self._download_webpage(result_url, query,
+                                             note='Downloading results page '+str(pagenum+1))
+            info = json.loads(webpage)
+            m = info[u'm']
+            results = info[u'results']
+
+            for (i, r) in enumerate(results):
+                if (pagenum * 30) +i >= n:
+                    break
+                mobj = re.search(r'(?P<url>screen\.yahoo\.com/.*?-\d*?\.html)"', r)
+                e = self.url_result('http://' + mobj.group('url'), 'Yahoo')
+                res['entries'].append(e)
+            if (pagenum * 30 +i >= n) or (m[u'last'] >= (m[u'total'] -1 )):
+                break
+
+        return res