free
/
searx
Watch 1
Star 0
Fork 0
Code Issues 0 Pull Requests 0 Releases 11 Wiki Activity
Browse Source

Merge pull request #531 from guyou/add-doku-engine 

Add doku engine
Adam Tauber 9 years ago
parent
11c2262055 51cb832601
commit
5544fdb756
4 changed files with 171 additions and 0 deletions
Split View Show Diff Stats
  1. 1
    0
      AUTHORS.rst
  2. 84
    0
      searx/engines/doku.py
  3. 7
    0
      searx/settings.yml
  4. 79
    0
      tests/unit/engines/test_doku.py

+ 1
- 0
AUTHORS.rst View File 

@@ -42,3 +42,4 @@ generally made searx better:
42 42 
 - Noemi Vanyi
43 43 
 - Kang-min Liu
44 44 
 - Kirill Isakov
45 
+- Guilhem Bonnefille

+ 84
- 0
searx/engines/doku.py View File 

@@ -0,0 +1,84 @@
1 
+# Doku Wiki
2 
+#
3 
+# @website     https://www.dokuwiki.org/
4 
+# @provide-api yes
5 
+#              (https://www.dokuwiki.org/devel:xmlrpc)
6 
+#
7 
+# @using-api   no
8 
+# @results     HTML
9 
+# @stable      yes
10 
+# @parse       (general)    url, title, content
11 
+
12 
+from urllib import urlencode
13 
+from lxml.html import fromstring
14 
+from searx.engines.xpath import extract_text
15 
+
16 
+# engine dependent config
17 
+categories = ['general']  # TODO , 'images', 'music', 'videos', 'files'
18 
+paging = False
19 
+language_support = False
20 
+number_of_results = 5
21 
+
22 
+# search-url
23 
+# Doku is OpenSearch compatible
24 
+base_url = 'http://localhost:8090'
25 
+search_url = '/?do=search'\
26 
+             '&{query}'
27 
+# TODO             '&startRecord={offset}'\
28 
+# TODO             '&maximumRecords={limit}'\
29 
+
30 
+
31 
+# do search-request
32 
+def request(query, params):
33 
+
34 
+    params['url'] = base_url +\
35 
+        search_url.format(query=urlencode({'id': query}))
36 
+
37 
+    return params
38 
+
39 
+
40 
+# get response from search-request
41 
+def response(resp):
42 
+    results = []
43 
+
44 
+    doc = fromstring(resp.text)
45 
+
46 
+    # parse results
47 
+    # Quickhits
48 
+    for r in doc.xpath('//div[@class="search_quickresult"]/ul/li'):
49 
+        try:
50 
+            res_url = r.xpath('.//a[@class="wikilink1"]/@href')[-1]
51 
+        except:
52 
+            continue
53 
+
54 
+        if not res_url:
55 
+            continue
56 
+
57 
+        title = extract_text(r.xpath('.//a[@class="wikilink1"]/@title'))
58 
+
59 
+        # append result
60 
+        results.append({'title': title,
61 
+                        'content': "",
62 
+                        'url': base_url + res_url})
63 
+
64 
+    # Search results
65 
+    for r in doc.xpath('//dl[@class="search_results"]/*'):
66 
+        try:
67 
+            if r.tag == "dt":
68 
+                res_url = r.xpath('.//a[@class="wikilink1"]/@href')[-1]
69 
+                title = extract_text(r.xpath('.//a[@class="wikilink1"]/@title'))
70 
+            elif r.tag == "dd":
71 
+                content = extract_text(r.xpath('.'))
72 
+
73 
+                # append result
74 
+                results.append({'title': title,
75 
+                                'content': content,
76 
+                                'url': base_url + res_url})
77 
+        except:
78 
+            continue
79 
+
80 
+        if not res_url:
81 
+            continue
82 
+
83 
+    # return results
84 
+    return results

+ 7
- 0
searx/settings.yml View File 

@@ -337,6 +337,13 @@ engines:
337 337 
 #    number_of_results : 5
338 338 
 #    timeout : 3.0
339 339
340 
+# Doku engine lets you access to any Doku wiki instance:
341 
+# A public one or a privete/corporate one.
342 
+#  - name : ubuntuwiki
343 
+#    engine : doku
344 
+#    shortcut : uw
345 
+#    base_url : 'http://doc.ubuntu-fr.org'
346 
+
340 347 
 locales:
341 348 
     en : English
342 349 
     bg : Български (Bulgarian)

+ 79
- 0
tests/unit/engines/test_doku.py View File 

@@ -0,0 +1,79 @@
1 
+# -*- coding: utf-8 -*-
2 
+from collections import defaultdict
3 
+import mock
4 
+from searx.engines import doku
5 
+from searx.testing import SearxTestCase
6 
+
7 
+
8 
+class TestDokuEngine(SearxTestCase):
9 
+
10 
+    def test_request(self):
11 
+        query = 'test_query'
12 
+        dicto = defaultdict(dict)
13 
+        params = doku.request(query, dicto)
14 
+        self.assertIn('url', params)
15 
+        self.assertIn(query, params['url'])
16 
+
17 
+    def test_response(self):
18 
+        self.assertRaises(AttributeError, doku.response, None)
19 
+        self.assertRaises(AttributeError, doku.response, [])
20 
+        self.assertRaises(AttributeError, doku.response, '')
21 
+        self.assertRaises(AttributeError, doku.response, '[]')
22 
+
23 
+        response = mock.Mock(text='<html></html>')
24 
+        self.assertEqual(doku.response(response), [])
25 
+
26 
+        html = u"""
27 
+        <div class="search_quickresult">
28 
+            <h3>Pages trouvées :</h3>
29 
+            <ul class="search_quickhits">
30 
+                <li> <a href="/xfconf-query" class="wikilink1" title="xfconf-query">xfconf-query</a></li>
31 
+            </ul>
32 
+            <div class="clearer"></div>
33 
+        </div>
34 
+        """
35 
+        response = mock.Mock(text=html)
36 
+        results = doku.response(response)
37 
+        expected = [{'content': '', 'title': 'xfconf-query', 'url': 'http://localhost:8090/xfconf-query'}]
38 
+        self.assertEqual(doku.response(response), expected)
39 
+
40 
+        html = u"""
41 
+        <dl class="search_results">
42 
+            <dt><a href="/xvnc?s[]=query" class="wikilink1" title="xvnc">xvnc</a>: 40 Occurrences trouvées</dt>
43 
+            <dd>er = /usr/bin/Xvnc
44 
+     server_args = -inetd -<strong class="search_hit">query</strong> localhost -geometry 640x480 ... er = /usr/bin/Xvnc
45 
+     server_args = -inetd -<strong class="search_hit">query</strong> localhost -geometry 800x600 ... er = /usr/bin/Xvnc
46 
+     server_args = -inetd -<strong class="search_hit">query</strong> localhost -geometry 1024x768 ... er = /usr/bin/Xvnc
47 
+     server_args = -inetd -<strong class="search_hit">query</strong> localhost -geometry 1280x1024 -depth 8 -Sec</dd>
48 
+            <dt><a href="/postfix_mysql_tls_sasl_1404?s[]=query"
49 
+                   class="wikilink1"
50 
+                   title="postfix_mysql_tls_sasl_1404">postfix_mysql_tls_sasl_1404</a>: 14 Occurrences trouvées</dt>
51 
+            <dd>tdepasse
52 
+  hosts = 127.0.0.1
53 
+  dbname = postfix
54 
+  <strong class="search_hit">query</strong> = SELECT goto FROM alias WHERE address='%s' AND a... tdepasse
55 
+  hosts = 127.0.0.1
56 
+  dbname = postfix
57 
+  <strong class="search_hit">query</strong> = SELECT domain FROM domain WHERE domain='%s'
58 
+  #optional <strong class="search_hit">query</strong> to use when relaying for backup MX
59 
+  #<strong class="search_hit">query</strong> = SELECT domain FROM domain WHERE domain='%s' and backupmx =</dd>
60 
+          <dt><a href="/bind9?s[]=query" class="wikilink1" title="bind9">bind9</a>: 12 Occurrences trouvées</dt>
61 
+          <dd>  printcmd
62 
+;; Got answer:
63 
+;; -&gt;&gt;HEADER&lt;&lt;- opcode: <strong class="search_hit">QUERY</strong>, status: NOERROR, id: 13427
64 
+;; flags: qr aa rd ra; <strong class="search_hit">QUERY</strong>: 1, ANSWER: 1, AUTHORITY: 1, ADDITIONAL: 1
65 
+
66 
+[...]
67 
+
68 
+;; <strong class="search_hit">Query</strong> time: 1 msec
69 
+;; SERVER: 127.0.0.1#53(127.0.0.1)
70 
+;... par la requête (<strong class="search_hit">Query</strong> time) , entre la première et la deuxième requête.</dd>
71 
+        </dl>
72 
+        """
73 
+        response = mock.Mock(text=html)
74 
+        results = doku.response(response)
75 
+        self.assertEqual(type(results), list)
76 
+        self.assertEqual(len(results), 3)
77 
+        self.assertEqual(results[0]['title'], 'xvnc')
78 
+# FIXME        self.assertEqual(results[0]['url'], u'http://this.should.be.the.link/ű')
79 
+# FIXME        self.assertEqual(results[0]['content'], 'This should be the content.')