git.iko.soy/searx
on master help sign in
add duckduckgo images engine
c65a409marc
searx/engines/duckduckgo.py+35−30
compare with master
unifiedsplit
@@ -41,46 +41,51 @@ title_xpath = './/a[@class="result__a"]'
41content_xpath = './/a[@class="result__snippet"]'
41content_xpath = './/a[@class="result__snippet"]'
42
42
43
43
44−# do search-request
44+# match query's language to a region code that duckduckgo will accept
45−def request(query, params):
45+def get_region_code(lang):
46− if params['time_range'] and params['time_range'] not in time_range_dict:
47− return params
48−
49− offset = 30 + (params['pageno'] - 1) * 50
50− dc_param = offset + 1
51−
52 # custom fixes for languages
46 # custom fixes for languages
53− if params['language'] == 'all':
47+ if lang == 'all':
54− locale = None
48+ region_code = None
55− elif params['language'][:2] == 'ja':
49+ elif lang[:2] == 'ja':
56− locale = 'jp-jp'
50+ region_code = 'jp-jp'
57− elif params['language'][:2] == 'sl':
51+ elif lang[:2] == 'sl':
58− locale = 'sl-sl'
52+ region_code = 'sl-sl'
59− elif params['language'] == 'zh-TW':
53+ elif lang == 'zh-TW':
60− locale = 'tw-tzh'
54+ region_code = 'tw-tzh'
61− elif params['language'] == 'zh-HK':
55+ elif lang == 'zh-HK':
62− locale = 'hk-tzh'
56+ region_code = 'hk-tzh'
63− elif params['language'][-2:] == 'SA':
57+ elif lang[-2:] == 'SA':
64− locale = 'xa-' + params['language'].split('-')[0]
58+ region_code = 'xa-' + lang.split('-')[0]
65− elif params['language'][-2:] == 'GB':
59+ elif lang[-2:] == 'GB':
66− locale = 'uk-' + params['language'].split('-')[0]
60+ region_code = 'uk-' + lang.split('-')[0]
67 else:
61 else:
68− locale = params['language'].split('-')
62+ region_code = lang.split('-')
69− if len(locale) == 2:
63+ if len(region_code) == 2:
70 # country code goes first
64 # country code goes first
71− locale = locale[1].lower() + '-' + locale[0].lower()
65+ region_code = region_code[1].lower() + '-' + region_code[0].lower()
72 else:
66 else:
73 # tries to get a country code from language
67 # tries to get a country code from language
74− locale = locale[0].lower()
68+ region_code = region_code[0].lower()
75 for lc in supported_languages:
69 for lc in supported_languages:
76 lc = lc.split('-')
70 lc = lc.split('-')
77− if locale == lc[0]:
71+ if region_code == lc[0]:
78− locale = lc[1].lower() + '-' + lc[0].lower()
72+ region_code = lc[1].lower() + '-' + lc[0].lower()
79 break
73 break
74+ return region_code
75+
76+
77+# do search-request
78+def request(query, params):
79+ if params['time_range'] and params['time_range'] not in time_range_dict:
80+ return params
81+
82+ offset = 30 + (params['pageno'] - 1) * 50
83+ dc_param = offset + 1
80
84
81− if locale:
85+ region_code = get_region_code(params['language'])
86+ if region_code:
82 params['url'] = url.format(
87 params['url'] = url.format(
83− query=urlencode({'q': query, 'kl': locale}), offset=offset, dc_param=dc_param)
88+ query=urlencode({'q': query, 'kl': region_code}), offset=offset, dc_param=dc_param)
84 else:
89 else:
85 params['url'] = url.format(
90 params['url'] = url.format(
86 query=urlencode({'q': query}), offset=offset, dc_param=dc_param)
91 query=urlencode({'q': query}), offset=offset, dc_param=dc_param)
searx/engines/duckduckgo_images.pynew file+91
@@ -0,0 +1,91 @@
1+"""
2+ DuckDuckGo (Images)
3+
4+ @website https://duckduckgo.com/
5+ @provide-api yes (https://duckduckgo.com/api),
6+ but images are not supported
7+
8+ @using-api no
9+ @results JSON (site requires js to get images)
10+ @stable no (JSON can change)
11+ @parse url, title, img_src
12+
13+ @todo avoid extra request
14+"""
15+
16+from requests import get
17+from json import loads
18+from searx.engines.xpath import extract_text
19+from searx.engines.duckduckgo import _fetch_supported_languages, supported_languages_url, get_region_code
20+from searx.url_utils import urlencode
21+
22+# engine dependent config
23+categories = ['images']
24+paging = True
25+language_support = True
26+safesearch = True
27+
28+# search-url
29+images_url = 'https://duckduckgo.com/i.js?{query}&s={offset}&p={safesearch}&o=json&vqd={vqd}'
30+site_url = 'https://duckduckgo.com/?{query}&iar=images&iax=1&ia=images'
31+
32+
33+# run query in site to get vqd number needed for requesting images
34+# TODO: find a way to get this number without an extra request (is it a hash of the query?)
35+def get_vqd(query):
36+ res = get(site_url.format(query=urlencode({'q': query})))
37+ content = res.text
38+ vqd = content[content.find('vqd=\'') + 5:]
39+ vqd = vqd[:vqd.find('\'')]
40+ return vqd
41+
42+
43+# do search-request
44+def request(query, params):
45+ # to avoid running actual external requests when testing
46+ if 'is_test' not in params:
47+ vqd = get_vqd(query)
48+ else:
49+ vqd = '12345'
50+
51+ offset = (params['pageno'] - 1) * 50
52+
53+ safesearch = params['safesearch'] - 1
54+
55+ region_code = get_region_code(params['language'])
56+ if region_code:
57+ params['url'] = images_url.format(
58+ query=urlencode({'q': query, 'l': region_code}), offset=offset, safesearch=safesearch, vqd=vqd)
59+ else:
60+ params['url'] = images_url.format(
61+ query=urlencode({'q': query}), offset=offset, safesearch=safesearch, vqd=vqd)
62+
63+ return params
64+
65+
66+# get response from search-request
67+def response(resp):
68+ results = []
69+
70+ content = resp.text
71+ try:
72+ res_json = loads(content)
73+ except:
74+ return []
75+
76+ # parse results
77+ for result in res_json['results']:
78+ title = result['title']
79+ url = result['url']
80+ thumbnail = result['thumbnail']
81+ image = result['image']
82+
83+ # append result
84+ results.append({'template': 'images.html',
85+ 'title': title,
86+ 'content': '',
87+ 'thumbnail_src': thumbnail,
88+ 'img_src': image,
89+ 'url': url})
90+
91+ return results
searx/settings.yml+6
@@ -167,6 +167,12 @@ engines:
167 shortcut : ddg
167 shortcut : ddg
168 disabled : True
168 disabled : True
169
169
170+ - name : duckduckgo images
171+ engine : duckduckgo_images
172+ shortcut : ddi
173+ timeout: 3.0
174+ disabled : True
175+
170 - name : etymonline
176 - name : etymonline
171 engine : xpath
177 engine : xpath
172 paging : True
178 paging : True
tests/unit/engines/test_duckduckgo_images.pynew file+72
@@ -0,0 +1,72 @@
1+# -*- coding: utf-8 -*-
2+from collections import defaultdict
3+import mock
4+from searx.engines import duckduckgo_images
5+from searx.testing import SearxTestCase
6+
7+
8+class TestDuckduckgoImagesEngine(SearxTestCase):
9+
10+ def test_request(self):
11+ query = 'test_query'
12+ dicto = defaultdict(dict)
13+ dicto['is_test'] = True
14+ dicto['pageno'] = 1
15+ dicto['safesearch'] = 0
16+ dicto['language'] = 'all'
17+ params = duckduckgo_images.request(query, dicto)
18+ self.assertIn('url', params)
19+ self.assertIn(query, params['url'])
20+ self.assertIn('duckduckgo.com', params['url'])
21+ self.assertIn('s=0', params['url'])
22+ self.assertIn('p=-1', params['url'])
23+ self.assertIn('vqd=12345', params['url'])
24+
25+ # test paging and safe search
26+ dicto['pageno'] = 2
27+ dicto['safesearch'] = 2
28+ params = duckduckgo_images.request(query, dicto)
29+ self.assertIn('url', params)
30+ self.assertIn(query, params['url'])
31+ self.assertIn('s=50', params['url'])
32+ self.assertIn('p=1', params['url'])
33+
34+ def test_response(self):
35+ self.assertRaises(AttributeError, duckduckgo_images.response, None)
36+ self.assertRaises(AttributeError, duckduckgo_images.response, [])
37+ self.assertRaises(AttributeError, duckduckgo_images.response, '')
38+ self.assertRaises(AttributeError, duckduckgo_images.response, '[]')
39+
40+ response = mock.Mock(text='If this error persists, please let us know: ops@duckduckgo.com')
41+ self.assertEqual(duckduckgo_images.response(response), [])
42+
43+ json = u"""
44+ {
45+ "query": "test_query",
46+ "results": [
47+ {
48+ "title": "Result 1",
49+ "url": "https://site1.url",
50+ "thumbnail": "https://thumb1.nail",
51+ "image": "https://image1"
52+ },
53+ {
54+ "title": "Result 2",
55+ "url": "https://site2.url",
56+ "thumbnail": "https://thumb2.nail",
57+ "image": "https://image2"
58+ }
59+ ]
60+ }
61+ """
62+ response = mock.Mock(text=json)
63+ results = duckduckgo_images.response(response)
64+ self.assertEqual(len(results), 2)
65+ self.assertEqual(results[0]['title'], 'Result 1')
66+ self.assertEqual(results[0]['url'], 'https://site1.url')
67+ self.assertEqual(results[0]['thumbnail_src'], 'https://thumb1.nail')
68+ self.assertEqual(results[0]['img_src'], 'https://image1')
69+ self.assertEqual(results[1]['title'], 'Result 2')
70+ self.assertEqual(results[1]['url'], 'https://site2.url')
71+ self.assertEqual(results[1]['thumbnail_src'], 'https://thumb2.nail')
72+ self.assertEqual(results[1]['img_src'], 'https://image2')