Browse Source

[fix] broken google images parsing

Adam Tauber 9 years ago
parent
commit
9331fc28a8
2 changed files with 6 additions and 8 deletions
  1. 3 5
      searx/engines/google_images.py
  2. 3 3
      tests/unit/engines/test_google_images.py

+ 3 - 5
searx/engines/google_images.py

@@ -49,8 +49,6 @@ def response(resp):
 
 
     # parse results
     # parse results
     for result in dom.xpath('//div[@data-ved]'):
     for result in dom.xpath('//div[@data-ved]'):
-        data_url = result.xpath('./a/@href')[0]
-        data_query = {k: v[0] for k, v in parse_qs(data_url.split('?', 1)[1]).iteritems()}
 
 
         metadata = loads(result.xpath('./div[@class="rg_meta"]/text()')[0])
         metadata = loads(result.xpath('./div[@class="rg_meta"]/text()')[0])
 
 
@@ -60,11 +58,11 @@ def response(resp):
         thumbnail_src = thumbnail_src.replace("http://", "https://")
         thumbnail_src = thumbnail_src.replace("http://", "https://")
 
 
         # append result
         # append result
-        results.append({'url': data_query['imgrefurl'],
+        results.append({'url': metadata['ru'],
                         'title': metadata['pt'],
                         'title': metadata['pt'],
                         'content': metadata['s'],
                         'content': metadata['s'],
-                        'thumbnail_src': metadata['tu'],
-                        'img_src': data_query['imgurl'],
+                        'thumbnail_src': thumbnail_src,
+                        'img_src': metadata['ou'],
                         'template': 'images.html'})
                         'template': 'images.html'})
 
 
     # return results
     # return results

+ 3 - 3
tests/unit/engines/test_google_images.py

@@ -41,7 +41,7 @@ class TestGoogleImagesEngine(SearxTestCase):
         </div>
         </div>
       </a>
       </a>
       <div class="rg_meta">
       <div class="rg_meta">
-        {"id":"bQWQ9wz9loJmjM:","isu":"clker.com","ity":"png","md":"/search?tbs\u003dsbi:AMhZZit7u1mHyop9pQisu-5idR-8W_1Itvwc3afChmsjQYPx_1yYMzBvUZgtkcGoojqekKZ-6n_1rjX9ySH0OWA_1eO5OijFY6BBDw_1GApr6xxb1bXJcBcj-DiguMoXWW7cZSG7MRQbwnI5SoDZNXcv_1xGszy886I7NVb_1oRKSliTHtzqbXAxhvYreM","msu":"/search?q\u003dsouth\u0026biw\u003d1364\u0026bih\u003d235\u0026tbm\u003disch\u0026tbs\u003dsimg:CAQSEgltBZD3DP2WgiG-U42R4G0RFw","oh":598,"os":"13KB","ow":504,"pt":"South Arrow Clip Art at Clker.com - vector clip art online ...","rid":"vlONkeBtERfDuM","s":"Download this image as:","sc":1,"si":"/search?q\u003dsouth\u0026biw\u003d1364\u0026bih\u003d235\u0026tbm\u003disch\u0026tbs\u003dsimg:CAESEgltBZD3DP2WgiG-U42R4G0RFw","th":245,"tu":"https://thumbnail.url/","tw":206}
+        {"id":"bQWQ9wz9loJmjM:","isu":"clker.com","ity":"png","md":"/search?tbs\u003dsbi:AMhZZit7u1mHyop9pQisu-5idR-8W_1Itvwc3afChmsjQYPx_1yYMzBvUZgtkcGoojqekKZ-6n_1rjX9ySH0OWA_1eO5OijFY6BBDw_1GApr6xxb1bXJcBcj-DiguMoXWW7cZSG7MRQbwnI5SoDZNXcv_1xGszy886I7NVb_1oRKSliTHtzqbXAxhvYreM","msu":"/search?q\u003dsouth\u0026biw\u003d1364\u0026bih\u003d235\u0026tbm\u003disch\u0026tbs\u003dsimg:CAQSEgltBZD3DP2WgiG-U42R4G0RFw","oh":598,"os":"13KB","ow":504,"pt":"South Arrow Clip Art at Clker.com - vector clip art online ...","rid":"vlONkeBtERfDuM","s":"Download this image as:","sc":1,"si":"/search?q\u003dsouth\u0026biw\u003d1364\u0026bih\u003d235\u0026tbm\u003disch\u0026tbs\u003dsimg:CAESEgltBZD3DP2WgiG-U42R4G0RFw","th":245,"tu":"https://thumbnail.url/","tw":206,"ru":"a","ou":"b"}
       </div>
       </div>
     </div><!--n--><!--m-->
     </div><!--n--><!--m-->
   </div>
   </div>
@@ -52,7 +52,7 @@ class TestGoogleImagesEngine(SearxTestCase):
         self.assertEqual(type(results), list)
         self.assertEqual(type(results), list)
         self.assertEqual(len(results), 1)
         self.assertEqual(len(results), 1)
         self.assertEqual(results[0]['title'], u'South Arrow Clip Art at Clker.com - vector clip art online ...')
         self.assertEqual(results[0]['title'], u'South Arrow Clip Art at Clker.com - vector clip art online ...')
-        self.assertEqual(results[0]['url'], 'http://www.clker.com/clipart-south-arrow.html')
+        self.assertEqual(results[0]['url'], 'a')
         self.assertEqual(results[0]['thumbnail_src'], 'https://thumbnail.url/')
         self.assertEqual(results[0]['thumbnail_src'], 'https://thumbnail.url/')
-        self.assertEqual(results[0]['img_src'], 'http://www.clker.com/cliparts/H/X/l/b/0/0/south-arrow-hi.png')
+        self.assertEqual(results[0]['img_src'], 'b')
         self.assertEqual(results[0]['content'], 'Download this image as:')
         self.assertEqual(results[0]['content'], 'Download this image as:')