on master help sign in
Merge branch 'master' into patch-1
commit9012ad498cbb13aced7555511c6de6dadc15246a authorAlexandre Flament <alex@al-f.net> dateFri 19 Apr 2019, 09:59 (8 years ago) parent4ca7e68629b36d
3 files changed+29−22
searx/engines/duckduckgo_images.py+7−4 searx/engines/google_images.py+21−17 tests/unit/engines/test_duckduckgo_images.py+1−1
searx/engines/duckduckgo_images.py+7−4
@@ -35,9 +35,12 @@ site_url = 'https://duckduckgo.com/?{query}&iar=images&iax=1&ia=images'
3535
3636# run query in site to get vqd number needed for requesting images
3737# TODO: find a way to get this number without an extra request (is it a hash of the query?)
38−def get_vqd(query):
39− res = get(site_url.format(query=urlencode({'q': query})))
38+def get_vqd(query, headers):
39+ query_url = site_url.format(query=urlencode({'q': query}))
40+ res = get(query_url, headers=headers)
4041 content = res.text
42+ if content.find('vqd=\'') == -1:
43+ raise Exception('Request failed')
4144 vqd = content[content.find('vqd=\'') + 5:]
4245 vqd = vqd[:vqd.find('\'')]
4346 return vqd
@@ -47,7 +50,7 @@ def get_vqd(query):
4750def request(query, params):
4851 # to avoid running actual external requests when testing
4952 if 'is_test' not in params:
50− vqd = get_vqd(query)
53+ vqd = get_vqd(query, params['headers'])
5154 else:
5255 vqd = '12345'
5356
@@ -74,7 +77,7 @@ def response(resp):
7477 try:
7578 res_json = loads(content)
7679 except:
77− return []
80+ raise Exception('Cannot parse results')
7881
7982 # parse results
8083 for result in res_json['results']:
searx/engines/google_images.py+21−17
@@ -11,7 +11,6 @@
1111"""
1212
1313from datetime import date, timedelta
14−from json import loads
1514from lxml import html
1615from searx.url_utils import urlencode, urlparse, parse_qs
1716
@@ -39,7 +38,6 @@ time_range_dict = {'day': 'd',
3938# do search-request
4039def request(query, params):
4140 search_options = {
42− 'ijn': params['pageno'] - 1,
4341 'start': (params['pageno'] - 1) * number_of_results
4442 }
4543
@@ -53,7 +51,7 @@ def request(query, params):
5351 search_options['tbs'] = time_range_custom_attr.format(start=start, end=end)
5452
5553 if safesearch and params['safesearch']:
56− search_options['safe'] = 'on'
54+ search_options['safe'] = 'active'
5755
5856 params['url'] = search_url.format(query=urlencode({'q': query}),
5957 search_options=urlencode(search_options))
@@ -63,24 +61,30 @@ def request(query, params):
6361
6462# get response from search-request
6563def response(resp):
64+ dom = html.fromstring(resp.text)
65+
6666 results = []
67+ for element in dom.xpath('//div[@id="search"] //td'):
68+ link = element.xpath('./a')[0]
6769
68− dom = html.fromstring(resp.text)
70+ google_url = urlparse(link.xpath('.//@href')[0])
71+ query = parse_qs(google_url.query)
72+ source_url = next(iter(query.get('q', [])), None)
6973
70− # parse results
71− for img in dom.xpath('//a'):
72− r = {
73− 'title': u' '.join(img.xpath('.//div[class="rg_ilmbg"]//text()')),
74+ title_parts = element.xpath('./cite//following-sibling::*/text()')
75+ title_parts.extend(element.xpath('./cite//following-sibling::text()')[:-1])
76+
77+ result = {
78+ 'title': ''.join(title_parts),
7479 'content': '',
7580 'template': 'images.html',
81+ 'url': source_url,
82+ 'img_src': source_url,
83+ 'thumbnail_src': next(iter(link.xpath('.//img //@src')), None)
7684 }
77− url = urlparse(img.xpath('.//@href')[0])
78− query = parse_qs(url.query)
79− r['url'] = query['imgrefurl'][0]
80− r['img_src'] = query['imgurl'][0]
81− r['thumbnail_src'] = r['img_src']
82− # append result
83− results.append(r)
84−
85− # return results
85+
86+ if not source_url or not result['thumbnail_src']:
87+ continue
88+
89+ results.append(result)
8690 return results
tests/unit/engines/test_duckduckgo_images.py+1−1
@@ -41,7 +41,7 @@ class TestDuckduckgoImagesEngine(SearxTestCase):
4141 self.assertRaises(AttributeError, duckduckgo_images.response, '[]')
4242
4343 response = mock.Mock(text='If this error persists, please let us know: ops@duckduckgo.com')
44− self.assertEqual(duckduckgo_images.response(response), [])
44+ self.assertRaises(Exception, duckduckgo_images.response, response)
4545
4646 json = u"""
4747 {