on master help sign in
[mod] fetch firefox versions in a standalone script
commit066bd916bf0c0344c978d2ea46cf9e9960841a61 authorAlexandre Flament <alex@al-f.net> dateSun 28 May 2017, 15:46 (9 years ago) parentd51732c
3 files changed+92−20
searx/data/useragents.json+15 searx/utils.py+8−20 utils/fetch_firefox_version.py+69
searx/data/useragents.jsonnew file+15
@@ -0,0 +1,15 @@
1+{
2+ "os": [
3+ "Windows NT 10; WOW64",
4+ "X11; Linux x86_64"
5+ ],
6+ "ua": "Mozilla/5.0 ({os}; rv:{version}) Gecko/20100101 Firefox/{version}",
7+ "versions": [
8+ "59.0.2",
9+ "59.0.1",
10+ "59.0",
11+ "58.0.2",
12+ "58.0.1",
13+ "58.0"
14+ ]
15+}
searx/utils.py+8−20
@@ -10,8 +10,10 @@ from codecs import getincrementalencoder
1010from imp import load_source
1111from numbers import Number
1212from os.path import splitext, join
13+from io import open
1314from random import choice
1415import sys
16+import json
1517
1618from searx import settings
1719from searx.version import VERSION_STRING
@@ -39,29 +41,11 @@ else:
3941
4042logger = logger.getChild('utils')
4143
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−
5844blocked_tags = ('script',
5945 'style')
6046
61−
62−def gen_useragent(os=None):
63− # TODO
64− return ua.format(os=os or choice(ua_os), version=choice(ua_versions))
47+useragents = json.loads(open(os.path.dirname(os.path.realpath(__file__))
48+ + "/data/useragents.json", 'r', encoding='utf-8').read())
6549
6650
6751def searx_useragent():
@@ -70,6 +54,10 @@ def searx_useragent():
7054 suffix=settings['outgoing'].get('useragent_suffix', ''))
7155
7256
57+def gen_useragent():
58+ return str(useragents['ua'].format(os=choice(useragents['os']), version=choice(useragents['versions'])))
59+
60+
7361def highlight_content(content, query):
7462
7563 if not content:
utils/fetch_firefox_version.pynew file+69
@@ -0,0 +1,69 @@
1+#!/usr/bin/env python
2+
3+# set path
4+from sys import path
5+from os.path import realpath, dirname
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+
16+URL = 'https://ftp.mozilla.org/pub/firefox/releases/'
17+RELEASE_PATH = '/pub/firefox/releases/'
18+
19+NORMAL_REGEX = re.compile('^[0-9]+\.[0-9](\.[0-9])?(esr)?$')
20+# BETA_REGEX = re.compile('.*[0-9]b([0-9\-a-z]+)$')
21+# ESR_REGEX = re.compile('^[0-9]+\.[0-9](\.[0-9])?esr$')
22+
23+#
24+useragent = {
25+ "versions": (),
26+ "os": ('Windows NT 10; WOW64',
27+ 'X11; Linux x86_64'),
28+ "ua": "Mozilla/5.0 ({os}; rv:{version}) Gecko/20100101 Firefox/{version}"
29+}
30+
31+
32+def fetch_firefox_versions():
33+ resp = requests.get(URL, timeout=2.0)
34+ if resp.status_code != 200:
35+ raise Exception("Error fetching firefox versions, HTTP code " + resp.status_code)
36+ else:
37+ dom = html.fromstring(resp.text)
38+ versions = []
39+
40+ for link in dom.xpath('//a/@href'):
41+ url = urlparse(urljoin(URL, link))
42+ path = url.path
43+ if path.startswith(RELEASE_PATH):
44+ version = path[len(RELEASE_PATH):-1]
45+ if NORMAL_REGEX.match(version):
46+ versions.append(LooseVersion(version))
47+
48+ list.sort(versions, reverse=True)
49+ return versions
50+
51+
52+def fetch_firefox_last_versions():
53+ versions = fetch_firefox_versions()
54+
55+ result = []
56+ major_last = versions[0].version[0]
57+ major_list = (major_last, major_last - 1)
58+ for version in versions:
59+ major_current = version.version[0]
60+ if major_current in major_list and 'esr' not in version.version:
61+ result.append(version.vstring)
62+
63+ return result
64+
65+
66+useragent["versions"] = fetch_firefox_last_versions()
67+f = open("../searx/data/useragents.json", "wb")
68+json.dump(useragent, f, sort_keys=True, indent=4, ensure_ascii=False, encoding="utf-8")
69+f.close()