Browse Source

Merge pull request #1186 from kvch/fix-bing-videos

Fix Bing videos engine
Adam Tauber 7 years ago
parent
commit
360f8fab97
2 changed files with 38 additions and 83 deletions
  1. 5 17
      searx/engines/bing_videos.py
  2. 33 66
      tests/unit/engines/test_bing_videos.py

+ 5 - 17
searx/engines/bing_videos.py

@@ -69,22 +69,11 @@ def response(resp):
     dom = html.fromstring(resp.text)
 
     for result in dom.xpath('//div[@class="dg_u"]'):
-
-        # try to extract the url
-        url_container = result.xpath('.//div[@class="sa_wrapper"]/@data-eventpayload')
-        if len(url_container) > 0:
-            url = loads(url_container[0])['purl']
-        else:
-            url = result.xpath('./a/@href')[0]
-
-            # discard results that do not return an external url
-            # very recent results sometimes don't return the video's url
-            if url.startswith('/videos/search?'):
-                continue
-
-        title = extract_text(result.xpath('./a//div[@class="tl"]'))
-        content = extract_text(result.xpath('.//div[@class="pubInfo"]'))
-        thumbnail = result.xpath('.//div[@class="vthumb"]/img/@src')[0]
+        url = result.xpath('./div[@class="mc_vtvc"]/a/@href')[0]
+        url = 'https://bing.com' + url
+        title = extract_text(result.xpath('./div/a/div/div[@class="mc_vtvc_title"]/@title'))
+        content = extract_text(result.xpath('./div/a/div/div/div/div/text()'))
+        thumbnail = result.xpath('./div/a/div/div/img/@src')[0]
 
         results.append({'url': url,
                         'title': title,
@@ -92,7 +81,6 @@ def response(resp):
                         'thumbnail': thumbnail,
                         'template': 'videos.html'})
 
-        # first page ignores requested number of results
         if len(results) >= number_of_results:
             break
 

+ 33 - 66
tests/unit/engines/test_bing_videos.py

@@ -47,87 +47,54 @@ class TestBingVideosEngine(SearxTestCase):
         self.assertEqual(bing_videos.response(response), [])
 
         html = """
-        <div>
-            <div class="dg_u">
-                <a class="dv_i" href="/videos/search?abcde">
-                    <div class="vthblock">
-                        <div class="vthumb">
+        <div class="dg_u">
+            <div id="mc_vtvc_1" class="mc_vtvc">
+                <a class="mc_vtvc_link" href="/video">
+                    <div class="mc_vtvc_th">
+                        <div class="cico">
                             <img src="thumb_1.jpg" />
                         </div>
-                        <div>
-                            <div class="tl">
-                                Title 1
+                        <div class="mc_vtvc_ban_lo">
+                            <div class="vtbc">
+                                <div class="mc_bc_w b_smText">
+                                    <div class="mc_bc pivot bpi_2">
+                                        <span title="">
+                                             <span class="mv_vtvc_play cipg "></span>
+                                        </span>
+                                    </div>
+                                    <div class="mc_bc items">10:06</div>
+                                </div>
                             </div>
                         </div>
-                    </div>
-                    <div class="videoInfoPanel">
-                        <div class="pubInfo">
-                            <div>Content 1</div>
                         </div>
-                    </div>
-                </a>
-                <div class="sa_wrapper"
-                    data-eventpayload="{&quot;purl&quot;: &quot;https://url.com/1&quot;}">
-                </div>
-            </div>
-        </div>
-        """
-        response = mock.Mock(text=html)
-        results = bing_videos.response(response)
-        self.assertEqual(type(results), list)
-        self.assertEqual(len(results), 1)
-        self.assertEqual(results[0]['title'], 'Title 1')
-        self.assertEqual(results[0]['url'], 'https://url.com/1')
-        self.assertEqual(results[0]['content'], 'Content 1')
-        self.assertEqual(results[0]['thumbnail'], 'thumb_1.jpg')
-
-        html = """
-        <div>
-            <div class="dg_u">
-                <a class="dv_i" href="https://url.com/1">
-                    <div class="vthblock">
-                        <div class="vthumb">
-                            <img src="thumb_1.jpg" />
-                        </div>
-                        <div>
-                            <div class="tl">
-                                Title 1
-                            </div>
-                        </div>
-                    </div>
-                    <div class="videoInfoPanel">
-                        <div class="pubInfo">
-                            <div>Content 1</div>
-                        </div>
-                    </div>
-                </a>
-            </div>
-            <div class="dg_u">
-                <a class="dv_i" href="/videos/search?abcde">
-                    <div class="vthblock">
-                        <div class="vthumb">
-                            <img src="thumb_2.jpg" />
-                        </div>
-                        <div>
-                            <div class="tl">
-                                Title 2
+                        <div class="mc_vtvc_meta">
+                        <div class="mc_vtvc_title" title="Title 1"></div>
+                        <div class="mc_vtvc_meta_block_area">
+                        <div class="mc_vtvc_meta_block">
+                            <div class="mc_vtvc_meta_row">
+                                <span>65,696,000+ views</span>
+                                <span>1 year ago</span>
                             </div>
+                            <div class="mc_vtvc_meta_row mc_vtvc_meta_row_channel">Content 1</div>
+                            <div class="mc_vtvc_meta_row"><span>
+                                <div class="cico mc_vtvc_src_ico">
+                                    <div></div>
+                                </div>
+                                <span>YouTube</span>
+                            </span></div>
                         </div>
-                    </div>
-                    <div class="videoInfoPanel">
-                        <div class="pubInfo">
-                            <div>Content 2</div>
                         </div>
                     </div>
-                </a>
+                    <div class="vrhdata"></div>
+                    </a>
+                </div>
             </div>
-        </div>
         """
         response = mock.Mock(text=html)
         results = bing_videos.response(response)
         self.assertEqual(type(results), list)
         self.assertEqual(len(results), 1)
         self.assertEqual(results[0]['title'], 'Title 1')
-        self.assertEqual(results[0]['url'], 'https://url.com/1')
+        self.assertEqual(results[0]['url'], 'https://bing.com/video')
         self.assertEqual(results[0]['content'], 'Content 1')
         self.assertEqual(results[0]['thumbnail'], 'thumb_1.jpg')