diff --git a/searx/engines/google_news.py b/searx/engines/google_news.py index eb114f9c9..3e4371b99 100644 --- a/searx/engines/google_news.py +++ b/searx/engines/google_news.py @@ -20,7 +20,7 @@ language_support = True # engine dependent config url = 'https://ajax.googleapis.com/' -search_url = url + 'ajax/services/search/news?v=2.0&start={offset}&rsz=large&safe=off&filter=off&{query}&hl={language}' # noqa +search_url = url + 'ajax/services/search/news?v=2.0&start={offset}&rsz=large&safe=off&filter=off&{query}&hl={lang}' # do search-request @@ -33,7 +33,7 @@ def request(query, params): params['url'] = search_url.format(offset=offset, query=urlencode({'q': query}), - language=language) + lang=language) return params @@ -52,6 +52,8 @@ def response(resp): for result in search_res['responseData']['results']: # parse publishedDate publishedDate = parser.parse(result['publishedDate']) + if 'url' not in result: + continue # append result results.append({'url': result['unescapedUrl'], diff --git a/searx/tests/engines/test_google_news.py b/searx/tests/engines/test_google_news.py new file mode 100644 index 000000000..31d674121 --- /dev/null +++ b/searx/tests/engines/test_google_news.py @@ -0,0 +1,136 @@ +from collections import defaultdict +import mock +from searx.engines import google_news +from searx.testing import SearxTestCase + + +class TestGoogleNewsEngine(SearxTestCase): + + def test_request(self): + query = 'test_query' + dicto = defaultdict(dict) + dicto['pageno'] = 1 + dicto['language'] = 'fr_FR' + params = google_news.request(query, dicto) + self.assertIn('url', params) + self.assertIn(query, params['url']) + self.assertIn('googleapis.com', params['url']) + self.assertIn('fr', params['url']) + + dicto['language'] = 'all' + params = google_news.request(query, dicto) + self.assertIn('url', params) + self.assertIn('en', params['url']) + + def test_response(self): + self.assertRaises(AttributeError, google_news.response, None) + self.assertRaises(AttributeError, google_news.response, []) + self.assertRaises(AttributeError, google_news.response, '') + self.assertRaises(AttributeError, google_news.response, '[]') + + response = mock.Mock(text='{}') + self.assertEqual(google_news.response(response), []) + + response = mock.Mock(text='{"data": []}') + self.assertEqual(google_news.response(response), []) + + json = """ + { + "responseData": { + "results": [ + { + "GsearchResultClass": "GnewsSearch", + "clusterUrl": "http://news.google.com/news/story?ncl=d2d3t1LMDpNIj2MPPhdTT0ycN4sWM&hl=fr&ned=fr", + "content": "This is the content", + "unescapedUrl": "http://this.is.the.url", + "url": "http://this.is.the.url", + "title": "This is the title", + "titleNoFormatting": "This is the title", + "location": "", + "publisher": "Jeux Actu", + "publishedDate": "Fri, 30 Jan 2015 11:00:25 -0800", + "signedRedirectUrl": "http://news.google.com/", + "language": "fr", + "image": { + "url": "http://i.jeuxactus.com/datas/jeux/d/y/dying-light/vu/dying-light-54cc080b568fb.jpg", + "tbUrl": "http://t1.gstatic.com/images?q=tbn:ANd9GcSF4yYrs9Ycw23DGiOSAZ-5SEPXYwG3LNs", + "originalContextUrl": "http://www.jeuxactu.com/test-dying-light-sur-ps4-97208.htm", + "publisher": "Jeux Actu", + "tbWidth": 80, + "tbHeight": 30 + }, + "relatedStories": [ + { + "unescapedUrl": "http://www.jeuxvideo.com/test/415823/dying-light.htm", + "url": "http%3A%2F%2Fwww.jeuxvideo.com%2Ftest%2F415823%2Fdying-light.htm", + "title": "Test du jeu Dying Light - jeuxvideo.com", + "titleNoFormatting": "Test du jeu Dying Light - jeuxvideo.com", + "location": "", + "publisher": "JeuxVideo.com", + "publishedDate": "Fri, 30 Jan 2015 08:52:30 -0800", + "signedRedirectUrl": "http://news.google.com/news/url?sa=T&", + "language": "fr" + } + ] + } + ] + }, + "responseDetails": null, + "responseStatus": 200 + } + """ + response = mock.Mock(text=json) + results = google_news.response(response) + self.assertEqual(type(results), list) + self.assertEqual(len(results), 1) + self.assertEqual(results[0]['title'], 'This is the title') + self.assertEqual(results[0]['url'], 'http://this.is.the.url') + self.assertEqual(results[0]['content'], 'This is the content') + + json = """ + { + "responseData": { + "results": [ + { + "GsearchResultClass": "GnewsSearch", + "clusterUrl": "http://news.google.com/news/story?ncl=d2d3t1LMDpNIj2MPPhdTT0ycN4sWM&hl=fr&ned=fr", + "content": "This is the content", + "unescapedUrl": "http://this.is.the.url", + "title": "This is the title", + "titleNoFormatting": "This is the title", + "location": "", + "publisher": "Jeux Actu", + "publishedDate": "Fri, 30 Jan 2015 11:00:25 -0800", + "signedRedirectUrl": "http://news.google.com/news/", + "language": "fr", + "image": { + "url": "http://i.jeuxactus.com/datas/jeux/d/y/dying-light/vu/dying-light-54cc080b568fb.jpg", + "tbUrl": "http://t1.gstatic.com/images?q=tbn:b_6f-OSAZ-5SEPXYwG3LNs", + "originalContextUrl": "http://www.jeuxactu.com/test-dying-light-sur-ps4-97208.htm", + "publisher": "Jeux Actu", + "tbWidth": 80, + "tbHeight": 30 + } + } + ] + }, + "responseDetails": null, + "responseStatus": 200 + } + """ + response = mock.Mock(text=json) + results = google_news.response(response) + self.assertEqual(type(results), list) + self.assertEqual(len(results), 0) + + json = """ + { + "responseData": {}, + "responseDetails": null, + "responseStatus": 200 + } + """ + response = mock.Mock(text=json) + results = google_news.response(response) + self.assertEqual(type(results), list) + self.assertEqual(len(results), 0) diff --git a/searx/tests/test_engines.py b/searx/tests/test_engines.py index e609f9a5c..00ac8ffdf 100644 --- a/searx/tests/test_engines.py +++ b/searx/tests/test_engines.py @@ -10,6 +10,7 @@ from searx.tests.engines.test_dummy import * # noqa from searx.tests.engines.test_flickr import * # noqa from searx.tests.engines.test_github import * # noqa from searx.tests.engines.test_google_images import * # noqa +from searx.tests.engines.test_google_news import * # noqa from searx.tests.engines.test_kickass import * # noqa from searx.tests.engines.test_mixcloud import * # noqa from searx.tests.engines.test_searchcode_code import * # noqa