on master help sign in
[fix] broken google images parsing
9331fc2Adam Tauber
searx/engines/google_images.py+3−5
@@ -49,8 +49,6 @@ def response(resp):
49
49
50 # parse results
50 # parse results
51 for result in dom.xpath('//div[@data-ved]'):
51 for result in dom.xpath('//div[@data-ved]'):
52− data_url = result.xpath('./a/@href')[0]
53− data_query = {k: v[0] for k, v in parse_qs(data_url.split('?', 1)[1]).iteritems()}
54
52
55 metadata = loads(result.xpath('./div[@class="rg_meta"]/text()')[0])
53 metadata = loads(result.xpath('./div[@class="rg_meta"]/text()')[0])
56
54
@@ -60,11 +58,11 @@ def response(resp):
60 thumbnail_src = thumbnail_src.replace("http://", "https://")
58 thumbnail_src = thumbnail_src.replace("http://", "https://")
61
59
62 # append result
60 # append result
63− results.append({'url': data_query['imgrefurl'],
61+ results.append({'url': metadata['ru'],
64 'title': metadata['pt'],
62 'title': metadata['pt'],
65 'content': metadata['s'],
63 'content': metadata['s'],
66− 'thumbnail_src': metadata['tu'],
64+ 'thumbnail_src': thumbnail_src,
67− 'img_src': data_query['imgurl'],
65+ 'img_src': metadata['ou'],
68 'template': 'images.html'})
66 'template': 'images.html'})
69
67
70 # return results
68 # return results
tests/unit/engines/test_google_images.py+3−3
@@ -41,7 +41,7 @@ class TestGoogleImagesEngine(SearxTestCase):
41 </div>
41 </div>
42 </a>
42 </a>
43 <div class="rg_meta">
43 <div class="rg_meta">
44− {"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}
44+ {"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"}
45 </div>
45 </div>
46 </div><!--n--><!--m-->
46 </div><!--n--><!--m-->
47 </div>
47 </div>
@@ -52,7 +52,7 @@ class TestGoogleImagesEngine(SearxTestCase):
52 self.assertEqual(type(results), list)
52 self.assertEqual(type(results), list)
53 self.assertEqual(len(results), 1)
53 self.assertEqual(len(results), 1)
54 self.assertEqual(results[0]['title'], u'South Arrow Clip Art at Clker.com - vector clip art online ...')
54 self.assertEqual(results[0]['title'], u'South Arrow Clip Art at Clker.com - vector clip art online ...')
55− self.assertEqual(results[0]['url'], 'http://www.clker.com/clipart-south-arrow.html')
55+ self.assertEqual(results[0]['url'], 'a')
56 self.assertEqual(results[0]['thumbnail_src'], 'https://thumbnail.url/')
56 self.assertEqual(results[0]['thumbnail_src'], 'https://thumbnail.url/')
57− self.assertEqual(results[0]['img_src'], 'http://www.clker.com/cliparts/H/X/l/b/0/0/south-arrow-hi.png')
57+ self.assertEqual(results[0]['img_src'], 'b')
58 self.assertEqual(results[0]['content'], 'Download this image as:')
58 self.assertEqual(results[0]['content'], 'Download this image as:')