Find ketoconnect image when opengraph isn't set

mtlynch · Oct 1, 2017 · 9f82bcb · 9f82bcb
1 parent 22dc239
commit 9f82bcb
Show file tree

Hide file tree

Showing 2 changed files with 20 additions and 1 deletion.
diff --git a/ketohub/spiders.py b/ketohub/spiders.py
@@ -44,6 +44,12 @@ def find_ketoconnect_image_url(response):
     opengraph_url = _find_opengraph_image(response)
     if opengraph_url:
         return opengraph_url
+
+    for image_url in response.xpath(
+            '//div[@id="tve_editor"]/img/@src').extract():
+        if image_url.endswith('.jpg'):
+            return image_url
+
     raise NoImageFound('Could not find image in source HTML: %s' % response.url)
 
 

diff --git a/tests/test_spiders.py b/tests/test_spiders.py
@@ -7,7 +7,7 @@
 
 class FindKetoConnectImageUrlTest(unittest.TestCase):
 
-    def test_finds_correct_image_on_simple_page(self):
+    def test_finds_image_from_opengraph(self):
         self.assertEqual(
             spiders.find_ketoconnect_image_url(
                 http.TextResponse(
@@ -18,6 +18,19 @@ def test_finds_correct_image_on_simple_page(self):
   <meta property="og:image" content="https://mock.com/recipe-image.jpg" />
 </html>""")), 'https://mock.com/recipe-image.jpg')
 
+    def test_finds_image_when_opengraph_image_is_not_set(self):
+        self.assertEqual(
+            spiders.find_ketoconnect_image_url(
+                http.TextResponse(
+                    url='https://www.foo.com',
+                    request=http.Request('https://www.foo.com'),
+                    body="""
+<html>
+<div id="tve_editor">
+<img class="tve_image" alt="" style="width: 400px;" src="https://mock.com/recipe-image.jpg" width="400" height="600" data-attachment-id="9282">
+</div>
+</html>""")), 'https://mock.com/recipe-image.jpg')
+
     def test_raises_error_when_image_not_found(self):
         with self.assertRaises(spiders.NoImageFound):
             spiders.find_ketoconnect_image_url(