Browse Source

[enh] mymemory translated engine added for multi-word translations

potato 8 years ago
parent
commit
ab471fd13b
2 changed files with 71 additions and 0 deletions
  1. 63
    0
      searx/engines/translated.py
  2. 8
    0
      searx/settings.yml

+ 63
- 0
searx/engines/translated.py View File

@@ -0,0 +1,63 @@
1
+import re
2
+from urlparse import urljoin
3
+from lxml import html
4
+from cgi import escape
5
+from searx.engines.xpath import extract_text
6
+from searx.languages import language_codes
7
+
8
+categories = ['general']
9
+url = 'http://api.mymemory.translated.net/get?q={query}&langpair={from_lang}|{to_lang}'
10
+web_url = 'http://mymemory.translated.net/en/{from_lang}/{to_lang}/{query}'
11
+weight = 100
12
+
13
+parser_re = re.compile(u'.*?([a-z]+)-([a-z]+) (.{2,})$', re.I)
14
+
15
+def is_valid_lang(lang):
16
+    is_abbr = (len(lang) == 2)
17
+    if is_abbr:
18
+        for l in language_codes:
19
+            if l[0][:2] == lang.lower():
20
+                return (True, l[0][:2], l[1].lower())
21
+        return False
22
+    else:
23
+        for l in language_codes:
24
+            if l[1].lower() == lang.lower():
25
+                return (True, l[0][:2], l[1].lower())
26
+        return False
27
+
28
+
29
+def request(query, params):
30
+    m = parser_re.match(unicode(query, 'utf8'))
31
+    if not m:
32
+        return params
33
+
34
+    from_lang, to_lang, query = m.groups()
35
+
36
+    from_lang = is_valid_lang(from_lang)
37
+    to_lang = is_valid_lang(to_lang)
38
+
39
+    if not from_lang or not to_lang:
40
+        return params
41
+
42
+    params['url'] = url.format(from_lang=from_lang[1],
43
+                               to_lang=to_lang[1],
44
+                               query=query)
45
+    params['query'] = query
46
+    params['from_lang'] = from_lang
47
+    params['to_lang'] = to_lang
48
+
49
+    return params
50
+
51
+
52
+def response(resp):
53
+    results = []
54
+    results.append({
55
+        'url': escape(web_url.format(from_lang=resp.search_params['from_lang'][2],
56
+                              to_lang=resp.search_params['to_lang'][2],
57
+                              query=resp.search_params['query'])),
58
+        'title': escape('[{0}-{1}] {2}'.format(resp.search_params['from_lang'][1],
59
+                                                          resp.search_params['to_lang'][1],
60
+                                                          resp.search_params['query'])),
61
+        'content': escape(resp.json()['responseData']['translatedText'])
62
+    })
63
+    return results

+ 8
- 0
searx/settings.yml View File

@@ -499,6 +499,14 @@ engines:
499 499
     engine : dictzone
500 500
     shortcut : dc
501 501
 
502
+  - name : mymemory translated
503
+    engine : translated
504
+    shortcut : tl
505
+    timeout : 5.0
506
+    # You can use without an API key, but you are limited to 1000 words/day
507
+    # See : http://mymemory.translated.net/doc/usagelimits.php
508
+    # api_key : ''
509
+
502 510
 #The blekko technology and team have joined IBM Watson! -> https://blekko.com/
503 511
 #  - name : blekko images
504 512
 #    engine : blekko_images