[enh] implement image support for swisscows engine

searx/engines/swisscows.py

@@ -1,5 +1,5 @@
 """
- Swisscows (Web)
+ Swisscows (Web, Images)
 
  @website     https://swisscows.ch
  @provide-api no
@@ -15,7 +15,7 @@ from urllib import urlencode, unquote
 import re
 
 # engine dependent config
-categories = ['general']
+categories = ['general', 'images']
 paging = True
 language_support = True
 
@@ -24,7 +24,7 @@ base_url = 'https://swisscows.ch/'
 search_string = '?{query}&page={page}'
 
 # regex
-regex_json = re.compile('initialData: {"Request":(.|\n)*}\]},\s*environment')
+regex_json = re.compile('initialData: {"Request":(.|\n)*},\s*environment')
 regex_json_remove_start = re.compile('^initialData:\s*')
 regex_json_remove_end = re.compile(',\s*environment$')
 regex_img_url_remove_start = re.compile('^https?://i\.swisscows\.ch/\?link=')
@@ -45,6 +45,10 @@ def request(query, params):
                          'region': region}),
         page=params['pageno'])
 
+    # image search query is something like 'image?{query}&page={page}'
+    if params['category'] == 'images':
+        search_path = 'image' + search_path
+
     params['url'] = base_url + search_path
 
     return params
@@ -63,12 +67,30 @@ def response(resp):
     json_raw = regex_json_remove_end.sub('', regex_json_remove_start.sub('', json_regex.group()))
     json = loads(json_raw)
 
-    # parse normal results
+    # parse results
     for result in json['Results'].get('items', []):
-        # append result
-        results.append({'url': result['Url'].replace(u'\uE000', '').replace(u'\uE001', ''),
-                        'title': result['Title'].replace(u'\uE000', '').replace(u'\uE001', ''),
-                        'content': result['Description'].replace(u'\uE000', '').replace(u'\uE001', '')})
+        result_title = result['Title'].replace(u'\uE000', '').replace(u'\uE001', '')
+
+        # parse image results
+        if result.get('ContentType', '').startswith('image'):
+            img_url = unquote(regex_img_url_remove_start.sub('', result['Url']))
+
+            # append result
+            results.append({'url': result['SourceUrl'],
+                            'title': result['Title'],
+                            'content': '',
+                            'img_src': img_url,
+                            'template': 'images.html'})
+
+        # parse general results
+        else:
+            result_url = result['Url'].replace(u'\uE000', '').replace(u'\uE001', '')
+            result_content = result['Description'].replace(u'\uE000', '').replace(u'\uE001', '')
+
+            # append result
+            results.append({'url': result_url,
+                            'title': result_title,
+                            'content': result_content})
 
     # parse images
     for result in json.get('Images', []):

searx/tests/engines/test_swisscows.py

@@ -51,7 +51,23 @@ class TestSwisscowsEngine(SearxTestCase):
                             "Description":"\uE000This should\uE001 be the content.",
                             "Url":"http://this.should.be.the.link/",
                             "DisplayUrl":"www.\uE000this.should.be.the\uE001.link",
-                            "Id":"782ef287-e439-451c-b380-6ebc14ba033d"}
+                            "Id":"782ef287-e439-451c-b380-6ebc14ba033d"},
+                            {"Title":"Datei:This should1.svg",
+                            "Url":"https://i.swisscows.ch/?link=http%3a%2f%2fts2.mm.This/should1.png",
+                            "SourceUrl":"http://de.wikipedia.org/wiki/Datei:This should1.svg",
+                            "DisplayUrl":"de.wikipedia.org/wiki/Datei:This should1.svg",
+                            "Width":950,
+                            "Height":534,
+                            "FileSize":92100,
+                            "ContentType":"image/jpeg",
+                            "Thumbnail":{
+                                "Url":"https://i.swisscows.ch/?link=http%3a%2f%2fts2.mm.This/should1.png",
+                                "ContentType":"image/jpeg",
+                                "Width":300,
+                                "Height":168,
+                                "FileSize":9134},
+                                "Id":"6a97a542-8f65-425f-b7f6-1178c3aba7be"
+                            }
                         ],"TotalCount":55300,
                         "Query":"This should "
                     },
@@ -94,11 +110,15 @@ class TestSwisscowsEngine(SearxTestCase):
         response = mock.Mock(content=html)
         results = swisscows.response(response)
         self.assertEqual(type(results), list)
-        self.assertEqual(len(results), 2)
+        self.assertEqual(len(results), 3)
         self.assertEqual(results[0]['title'], 'This should be the title')
         self.assertEqual(results[0]['url'], 'http://this.should.be.the.link/')
         self.assertEqual(results[0]['content'], 'This should be the content.')
-        self.assertEqual(results[1]['title'], 'Datei:This should.svg')
-        self.assertEqual(results[1]['url'], 'http://de.wikipedia.org/wiki/Datei:This should.svg')
-        self.assertEqual(results[1]['img_src'], 'http://ts2.mm.This/should.png')
+        self.assertEqual(results[1]['title'], 'Datei:This should1.svg')
+        self.assertEqual(results[1]['url'], 'http://de.wikipedia.org/wiki/Datei:This should1.svg')
+        self.assertEqual(results[1]['img_src'], 'http://ts2.mm.This/should1.png')
         self.assertEqual(results[1]['template'], 'images.html')
+        self.assertEqual(results[2]['title'], 'Datei:This should.svg')
+        self.assertEqual(results[2]['url'], 'http://de.wikipedia.org/wiki/Datei:This should.svg')
+        self.assertEqual(results[2]['img_src'], 'http://ts2.mm.This/should.png')
+        self.assertEqual(results[2]['template'], 'images.html')