git.iko.soy/searx
on master help sign in
Merge pull request #933 from dalf/firefox_version
eea2e8eAdam Tauber
searx/data/useragents.jsonnew file+14
compare with master
unifiedsplit
@@ -0,0 +1,14 @@
1+{
2+ "ua": "Mozilla/5.0 ({os}; rv:{version}) Gecko/20100101 Firefox/{version}",
3+ "versions": [
4+ "61.0.1",
5+ "61.0",
6+ "60.0.2",
7+ "60.0.1",
8+ "60.0"
9+ ],
10+ "os": [
11+ "Windows NT 10; WOW64",
12+ "X11; Linux x86_64"
13+ ]
14+}
searx/utils.py+8−20
@@ -10,8 +10,10 @@ from codecs import getincrementalencoder
10from imp import load_source
10from imp import load_source
11from numbers import Number
11from numbers import Number
12from os.path import splitext, join
12from os.path import splitext, join
13+from io import open
13from random import choice
14from random import choice
14import sys
15import sys
16+import json
15
17
16from searx import settings
18from searx import settings
17from searx.version import VERSION_STRING
19from searx.version import VERSION_STRING
@@ -39,29 +41,11 @@ else:
39
41
40logger = logger.getChild('utils')
42logger = logger.getChild('utils')
41
43
42−ua_versions = ('52.8.1',
43− '53.0',
44− '54.0',
45− '55.0',
46− '56.0',
47− '57.0',
48− '58.0',
49− '59.0',
50− '60.0.2')
51−
52−ua_os = ('Windows NT 6.3; WOW64',
53− 'X11; Linux x86_64',
54− 'X11; Linux x86')
55−
56−ua = "Mozilla/5.0 ({os}; rv:{version}) Gecko/20100101 Firefox/{version}"
57−
58blocked_tags = ('script',
44blocked_tags = ('script',
59 'style')
45 'style')
60
46
61−
47+useragents = json.loads(open(os.path.dirname(os.path.realpath(__file__))
62−def gen_useragent(os=None):
48+ + "/data/useragents.json", 'r', encoding='utf-8').read())
63− # TODO
64− return ua.format(os=os or choice(ua_os), version=choice(ua_versions))
65
49
66
50
67def searx_useragent():
51def searx_useragent():
@@ -70,6 +54,10 @@ def searx_useragent():
70 suffix=settings['outgoing'].get('useragent_suffix', ''))
54 suffix=settings['outgoing'].get('useragent_suffix', ''))
71
55
72
56
57+def gen_useragent(os=None):
58+ return str(useragents['ua'].format(os=os or choice(useragents['os']), version=choice(useragents['versions'])))
59+
60+
73def highlight_content(content, query):
61def highlight_content(content, query):
74
62
75 if not content:
63 if not content:
utils/fetch_firefox_version.pynew file+73
@@ -0,0 +1,73 @@
1+#!/usr/bin/env python
2+
3+# set path
4+from sys import path
5+from os.path import realpath, dirname, join
6+path.append(realpath(dirname(realpath(__file__)) + '/../'))
7+
8+#
9+import json
10+import requests
11+import re
12+from distutils.version import LooseVersion, StrictVersion
13+from lxml import html
14+from searx.url_utils import urlparse, urljoin
15+from searx import searx_dir
16+
17+URL = 'https://ftp.mozilla.org/pub/firefox/releases/'
18+RELEASE_PATH = '/pub/firefox/releases/'
19+
20+NORMAL_REGEX = re.compile('^[0-9]+\.[0-9](\.[0-9])?$')
21+# BETA_REGEX = re.compile('.*[0-9]b([0-9\-a-z]+)$')
22+# ESR_REGEX = re.compile('^[0-9]+\.[0-9](\.[0-9])?esr$')
23+
24+#
25+useragents = {
26+ "versions": (),
27+ "os": ('Windows NT 10; WOW64',
28+ 'X11; Linux x86_64'),
29+ "ua": "Mozilla/5.0 ({os}; rv:{version}) Gecko/20100101 Firefox/{version}"
30+}
31+
32+
33+def fetch_firefox_versions():
34+ resp = requests.get(URL, timeout=2.0)
35+ if resp.status_code != 200:
36+ raise Exception("Error fetching firefox versions, HTTP code " + resp.status_code)
37+ else:
38+ dom = html.fromstring(resp.text)
39+ versions = []
40+
41+ for link in dom.xpath('//a/@href'):
42+ url = urlparse(urljoin(URL, link))
43+ path = url.path
44+ if path.startswith(RELEASE_PATH):
45+ version = path[len(RELEASE_PATH):-1]
46+ if NORMAL_REGEX.match(version):
47+ versions.append(LooseVersion(version))
48+
49+ list.sort(versions, reverse=True)
50+ return versions
51+
52+
53+def fetch_firefox_last_versions():
54+ versions = fetch_firefox_versions()
55+
56+ result = []
57+ major_last = versions[0].version[0]
58+ major_list = (major_last, major_last - 1)
59+ for version in versions:
60+ major_current = version.version[0]
61+ if major_current in major_list:
62+ result.append(version.vstring)
63+
64+ return result
65+
66+
67+def get_useragents_filename():
68+ return join(join(searx_dir, "data"), "useragents.json")
69+
70+
71+useragents["versions"] = fetch_firefox_last_versions()
72+with open(get_useragents_filename(), "w") as f:
73+ json.dump(useragents, f, indent=4, ensure_ascii=False)