|
|
@@ -2,83 +2,40 @@
|
|
2
|
2
|
|
|
3
|
3
|
# This script generates languages.py from intersecting each engine's supported languages.
|
|
4
|
4
|
#
|
|
5
|
|
-# The country names are obtained from http://api.geonames.org which requires registering as a user.
|
|
6
|
|
-#
|
|
7
|
5
|
# Output files (engines_languages.json and languages.py)
|
|
8
|
6
|
# are written in current directory to avoid overwriting in case something goes wrong.
|
|
9
|
7
|
|
|
10
|
|
-from requests import get
|
|
11
|
|
-from lxml.html import fromstring
|
|
12
|
|
-from json import loads, dump
|
|
|
8
|
+from json import dump
|
|
13
|
9
|
import io
|
|
14
|
10
|
from sys import path
|
|
|
11
|
+from babel import Locale, UnknownLocaleError
|
|
|
12
|
+from babel.languages import get_global
|
|
|
13
|
+
|
|
15
|
14
|
path.append('../searx') # noqa
|
|
16
|
15
|
from searx import settings
|
|
17
|
|
-from searx.url_utils import urlencode
|
|
18
|
16
|
from searx.engines import initialize_engines, engines
|
|
19
|
17
|
|
|
20
|
|
-# Geonames API for country names.
|
|
21
|
|
-geonames_user = '' # ADD USER NAME HERE
|
|
22
|
|
-country_names_url = 'http://api.geonames.org/countryInfoJSON?{parameters}'
|
|
23
|
|
-
|
|
24
|
18
|
# Output files.
|
|
25
|
19
|
engines_languages_file = 'engines_languages.json'
|
|
26
|
20
|
languages_file = 'languages.py'
|
|
27
|
21
|
|
|
28
|
|
-engines_languages = {}
|
|
29
|
|
-
|
|
30
|
|
-
|
|
31
|
|
-# To filter out invalid codes and dialects.
|
|
32
|
|
-def valid_code(lang_code):
|
|
33
|
|
- # filter invalid codes
|
|
34
|
|
- # sl-SL is technically not invalid, but still a mistake
|
|
35
|
|
- invalid_codes = ['sl-SL', 'wt-WT', 'jw']
|
|
36
|
|
- invalid_countries = ['UK', 'XA', 'XL']
|
|
37
|
|
- if lang_code[:2] == 'xx'\
|
|
38
|
|
- or lang_code in invalid_codes\
|
|
39
|
|
- or lang_code[-2:] in invalid_countries\
|
|
40
|
|
- or is_dialect(lang_code):
|
|
41
|
|
- return False
|
|
42
|
|
-
|
|
43
|
|
- return True
|
|
44
|
|
-
|
|
45
|
|
-
|
|
46
|
|
-# Language codes with any additional tags other than language and country.
|
|
47
|
|
-def is_dialect(lang_code):
|
|
48
|
|
- lang_code = lang_code.split('-')
|
|
49
|
|
- if len(lang_code) > 2 or len(lang_code[0]) > 3:
|
|
50
|
|
- return True
|
|
51
|
|
- if len(lang_code) == 2 and len(lang_code[1]) > 2:
|
|
52
|
|
- return True
|
|
53
|
|
-
|
|
54
|
|
- return False
|
|
55
|
|
-
|
|
56
|
|
-
|
|
57
|
|
-# Get country name in specified language.
|
|
58
|
|
-def get_country_name(locale):
|
|
59
|
|
- if geonames_user is '':
|
|
60
|
|
- return ''
|
|
61
|
|
-
|
|
62
|
|
- locale = locale.split('-')
|
|
63
|
|
- if len(locale) != 2:
|
|
64
|
|
- return ''
|
|
65
|
|
-
|
|
66
|
|
- url = country_names_url.format(parameters=urlencode({'lang': locale[0],
|
|
67
|
|
- 'country': locale[1],
|
|
68
|
|
- 'username': geonames_user}))
|
|
69
|
|
- response = get(url)
|
|
70
|
|
- json = loads(response.text)
|
|
71
|
|
- content = json.get('geonames', None)
|
|
72
|
|
- if content is None or len(content) != 1:
|
|
73
|
|
- print("No country name found for " + locale[0] + "-" + locale[1])
|
|
74
|
|
- return ''
|
|
75
|
|
-
|
|
76
|
|
- return content[0].get('countryName', '')
|
|
|
22
|
+# custom fixes for non standard locale codes
|
|
|
23
|
+# sl-SL is technically not invalid, but still a mistake
|
|
|
24
|
+# TODO: move to respective engines
|
|
|
25
|
+locale_fixes = {
|
|
|
26
|
+ 'sl-sl': 'sl-SI',
|
|
|
27
|
+ 'ar-xa': 'ar-SA',
|
|
|
28
|
+ 'es-xl': 'es-419',
|
|
|
29
|
+ 'zh-chs': 'zh-Hans-CN',
|
|
|
30
|
+ 'zh-cht': 'zh-Hant-TW',
|
|
|
31
|
+ 'tzh-tw': 'zh-Hant-TW',
|
|
|
32
|
+ 'tzh-hk': 'zh-Hant-HK'
|
|
|
33
|
+}
|
|
77
|
34
|
|
|
78
|
35
|
|
|
79
|
36
|
# Fetchs supported languages for each engine and writes json file with those.
|
|
80
|
37
|
def fetch_supported_languages():
|
|
81
|
|
- initialize_engines(settings['engines'])
|
|
|
38
|
+ engines_languages = {}
|
|
82
|
39
|
for engine_name in engines:
|
|
83
|
40
|
if hasattr(engines[engine_name], 'fetch_supported_languages'):
|
|
84
|
41
|
try:
|
|
|
@@ -90,81 +47,134 @@ def fetch_supported_languages():
|
|
90
|
47
|
with io.open(engines_languages_file, "w", encoding="utf-8") as f:
|
|
91
|
48
|
dump(engines_languages, f, ensure_ascii=False)
|
|
92
|
49
|
|
|
|
50
|
+ return engines_languages
|
|
|
51
|
+
|
|
|
52
|
+
|
|
|
53
|
+# Get babel Locale object from lang_code if possible.
|
|
|
54
|
+def get_locale(lang_code):
|
|
|
55
|
+ try:
|
|
|
56
|
+ locale = Locale.parse(lang_code, sep='-')
|
|
|
57
|
+ return locale
|
|
|
58
|
+ except (UnknownLocaleError, ValueError):
|
|
|
59
|
+ return None
|
|
|
60
|
+
|
|
|
61
|
+
|
|
|
62
|
+# Append engine_name to list of engines that support locale.
|
|
|
63
|
+def add_engine_counter(lang_code, engine_name, languages):
|
|
|
64
|
+ if lang_code in languages:
|
|
|
65
|
+ if 'counter' not in languages[lang_code]:
|
|
|
66
|
+ languages[lang_code]['counter'] = [engine_name]
|
|
|
67
|
+ elif engine_name not in languages[lang_code]['counter']:
|
|
|
68
|
+ languages[lang_code]['counter'].append(engine_name)
|
|
93
|
69
|
|
|
94
|
|
-# Join all language lists.
|
|
95
|
|
-# Iterate all languages supported by each engine.
|
|
96
|
|
-def join_language_lists():
|
|
97
|
|
- global languages
|
|
98
|
|
- # include wikipedia first for more accurate language names
|
|
99
|
|
- languages = {code: lang for code, lang
|
|
100
|
|
- in engines_languages['wikipedia'].items()
|
|
101
|
|
- if valid_code(code)}
|
|
102
|
70
|
|
|
|
71
|
+# Join all language lists.
|
|
|
72
|
+# TODO: Add language names from engine's language list if name not known by babel.
|
|
|
73
|
+def join_language_lists(engines_languages):
|
|
|
74
|
+ language_list = {}
|
|
103
|
75
|
for engine_name in engines_languages:
|
|
104
|
|
- for locale in engines_languages[engine_name]:
|
|
105
|
|
- if valid_code(locale):
|
|
106
|
|
- # if language is not on list or if it has no name yet
|
|
107
|
|
- if locale not in languages or not languages[locale].get('name'):
|
|
108
|
|
- if isinstance(engines_languages[engine_name], dict):
|
|
109
|
|
- languages[locale] = engines_languages[engine_name][locale]
|
|
110
|
|
- else:
|
|
111
|
|
- languages[locale] = {}
|
|
112
|
|
-
|
|
113
|
|
- # add to counter of engines that support given language
|
|
114
|
|
- lang = locale.split('-')[0]
|
|
115
|
|
- if lang in languages:
|
|
116
|
|
- if 'counter' not in languages[lang]:
|
|
117
|
|
- languages[lang]['counter'] = [engine_name]
|
|
118
|
|
- elif engine_name not in languages[lang]['counter']:
|
|
119
|
|
- languages[lang]['counter'].append(engine_name)
|
|
120
|
|
-
|
|
121
|
|
- # filter list to include only languages supported by most engines
|
|
122
|
|
- min_supported_engines = int(0.70 * len(engines_languages))
|
|
123
|
|
- languages = {code: lang for code, lang
|
|
124
|
|
- in languages.items()
|
|
125
|
|
- if len(lang.get('counter', [])) >= min_supported_engines or
|
|
126
|
|
- len(languages.get(code.split('-')[0], {}).get('counter', [])) >= min_supported_engines}
|
|
127
|
|
-
|
|
128
|
|
- # get locales that have no name or country yet
|
|
129
|
|
- for locale in languages.keys():
|
|
130
|
|
- # try to get language names
|
|
131
|
|
- if not languages[locale].get('name'):
|
|
132
|
|
- name = languages.get(locale.split('-')[0], {}).get('name', None)
|
|
133
|
|
- if name:
|
|
134
|
|
- languages[locale]['name'] = name
|
|
135
|
|
- else:
|
|
136
|
|
- # filter out locales with no name
|
|
137
|
|
- del languages[locale]
|
|
138
|
|
- continue
|
|
139
|
|
-
|
|
140
|
|
- # try to get language name in english
|
|
141
|
|
- if not languages[locale].get('english_name'):
|
|
142
|
|
- languages[locale]['english_name'] = languages.get(locale.split('-')[0], {}).get('english_name', '')
|
|
143
|
|
-
|
|
144
|
|
- # try to get country name
|
|
145
|
|
- if locale.find('-') > 0 and not languages[locale].get('country'):
|
|
146
|
|
- languages[locale]['country'] = get_country_name(locale) or ''
|
|
147
|
|
-
|
|
148
|
|
-
|
|
149
|
|
-# Remove countryless language if language is featured in only one country.
|
|
150
|
|
-def filter_single_country_languages():
|
|
151
|
|
- prev_lang = None
|
|
152
|
|
- prev_code = None
|
|
153
|
|
- for code in sorted(languages):
|
|
154
|
|
- lang = code.split('-')[0]
|
|
155
|
|
- if lang == prev_lang:
|
|
|
76
|
+ for lang_code in engines_languages[engine_name]:
|
|
|
77
|
+
|
|
|
78
|
+ # apply custom fixes if necessary
|
|
|
79
|
+ if lang_code.lower() in locale_fixes:
|
|
|
80
|
+ lang_code = locale_fixes[lang_code.lower()]
|
|
|
81
|
+
|
|
|
82
|
+ locale = get_locale(lang_code)
|
|
|
83
|
+
|
|
|
84
|
+ # ensure that lang_code uses standard language and country codes
|
|
|
85
|
+ if locale and locale.territory:
|
|
|
86
|
+ lang_code = locale.language + '-' + locale.territory
|
|
|
87
|
+
|
|
|
88
|
+ # add locale if it's not in list
|
|
|
89
|
+ if lang_code not in language_list:
|
|
|
90
|
+ if locale:
|
|
|
91
|
+ language_list[lang_code] = {'name': locale.get_language_name().title(),
|
|
|
92
|
+ 'english_name': locale.english_name,
|
|
|
93
|
+ 'country': locale.get_territory_name() or ''}
|
|
|
94
|
+
|
|
|
95
|
+ # also add language without country
|
|
|
96
|
+ if locale.language not in language_list:
|
|
|
97
|
+ language_list[locale.language] = {'name': locale.get_language_name().title(),
|
|
|
98
|
+ 'english_name': locale.english_name}
|
|
|
99
|
+ else:
|
|
|
100
|
+ language_list[lang_code] = {}
|
|
|
101
|
+
|
|
|
102
|
+ # count engine for both language_country combination and language alone
|
|
|
103
|
+ add_engine_counter(lang_code, engine_name, language_list)
|
|
|
104
|
+ add_engine_counter(lang_code.split('-')[0], engine_name, language_list)
|
|
|
105
|
+
|
|
|
106
|
+ return language_list
|
|
|
107
|
+
|
|
|
108
|
+
|
|
|
109
|
+# Filter language list so it only includes the most supported languages and countries.
|
|
|
110
|
+def filter_language_list(all_languages):
|
|
|
111
|
+ min_supported_engines = 10
|
|
|
112
|
+ main_engines = [engine_name for engine_name in engines.keys()
|
|
|
113
|
+ if 'general' in engines[engine_name].categories and
|
|
|
114
|
+ engines[engine_name].supported_languages and
|
|
|
115
|
+ not engines[engine_name].disabled]
|
|
|
116
|
+
|
|
|
117
|
+ # filter list to include only languages supported by most engines or all default general engines
|
|
|
118
|
+ filtered_languages = {code: lang for code, lang
|
|
|
119
|
+ in all_languages.items()
|
|
|
120
|
+ if (len(lang.get('counter', [])) >= min_supported_engines or
|
|
|
121
|
+ all(main_engine in lang.get('counter', [])
|
|
|
122
|
+ for main_engine in main_engines))}
|
|
|
123
|
+
|
|
|
124
|
+ return filtered_languages
|
|
|
125
|
+
|
|
|
126
|
+
|
|
|
127
|
+# Add country codes to languages without one and filter out language codes.
|
|
|
128
|
+def assign_country_codes(filtered_languages, all_languages):
|
|
|
129
|
+ sorted_languages = sorted(all_languages,
|
|
|
130
|
+ key=lambda lang: len(all_languages[lang].get('counter', [])),
|
|
|
131
|
+ reverse=True)
|
|
|
132
|
+ previous_lang = None
|
|
|
133
|
+ previous_code = None
|
|
|
134
|
+ countries = 0
|
|
|
135
|
+ for current_code in sorted(filtered_languages):
|
|
|
136
|
+ current_lang = current_code.split('-')[0]
|
|
|
137
|
+
|
|
|
138
|
+ # count country codes per language
|
|
|
139
|
+ if current_lang == previous_lang:
|
|
156
|
140
|
countries += 1
|
|
|
141
|
+
|
|
157
|
142
|
else:
|
|
158
|
|
- if prev_lang is not None and countries == 1:
|
|
159
|
|
- del languages[prev_lang]
|
|
160
|
|
- languages[prev_code]['country'] = ''
|
|
|
143
|
+ if previous_lang is not None:
|
|
|
144
|
+ # if language has no single country code
|
|
|
145
|
+ if countries == 0:
|
|
|
146
|
+ # try to get country code with most supported engines
|
|
|
147
|
+ for l in sorted_languages:
|
|
|
148
|
+ l_parts = l.split('-')
|
|
|
149
|
+ if len(l_parts) == 2 and l_parts[0] == previous_lang:
|
|
|
150
|
+ filtered_languages[l] = all_languages[l]
|
|
|
151
|
+ filtered_languages[l]['country'] = ''
|
|
|
152
|
+ countries = 1
|
|
|
153
|
+ break
|
|
|
154
|
+
|
|
|
155
|
+ if countries == 0:
|
|
|
156
|
+ # get most likely country code from babel
|
|
|
157
|
+ subtags = get_global('likely_subtags').get(previous_lang)
|
|
|
158
|
+ if subtags:
|
|
|
159
|
+ subtag_parts = subtags.split('_')
|
|
|
160
|
+ new_code = subtag_parts[0] + '-' + subtag_parts[-1]
|
|
|
161
|
+ filtered_languages[new_code] = all_languages[previous_lang]
|
|
|
162
|
+ countries = 1
|
|
|
163
|
+
|
|
|
164
|
+ if countries == 1:
|
|
|
165
|
+ # remove countryless version of language if there's only one country
|
|
|
166
|
+ del filtered_languages[previous_lang]
|
|
|
167
|
+ if previous_code in filtered_languages:
|
|
|
168
|
+ filtered_languages[previous_code]['country'] = ''
|
|
|
169
|
+
|
|
161
|
170
|
countries = 0
|
|
162
|
|
- prev_lang = lang
|
|
163
|
|
- prev_code = code
|
|
|
171
|
+ previous_lang = current_lang
|
|
|
172
|
+
|
|
|
173
|
+ previous_code = current_code
|
|
164
|
174
|
|
|
165
|
175
|
|
|
166
|
176
|
# Write languages.py.
|
|
167
|
|
-def write_languages_file():
|
|
|
177
|
+def write_languages_file(languages):
|
|
168
|
178
|
new_file = open(languages_file, 'wb')
|
|
169
|
179
|
file_content = '# -*- coding: utf-8 -*-\n'\
|
|
170
|
180
|
+ '# list of language codes\n'\
|
|
|
@@ -183,7 +193,9 @@ def write_languages_file():
|
|
183
|
193
|
|
|
184
|
194
|
|
|
185
|
195
|
if __name__ == "__main__":
|
|
186
|
|
- fetch_supported_languages()
|
|
187
|
|
- join_language_lists()
|
|
188
|
|
- filter_single_country_languages()
|
|
189
|
|
- write_languages_file()
|
|
|
196
|
+ initialize_engines(settings['engines'])
|
|
|
197
|
+ engines_languages = fetch_supported_languages()
|
|
|
198
|
+ all_languages = join_language_lists(engines_languages)
|
|
|
199
|
+ filtered_languages = filter_language_list(all_languages)
|
|
|
200
|
+ assign_country_codes(filtered_languages, all_languages)
|
|
|
201
|
+ write_languages_file(filtered_languages)
|
Marc Abonce Seguin
7 年前