1
0
mirror of https://github.com/searx/searx synced 2024-12-02 15:10:21 +01:00
searx/utils/fetch_firefox_version.py
Markus Heiser 697dd6c6d3 [fix] ffox useragent: use Windows NT 10.0 and not Windows NT 10
This is an addittion to PR #1934:

  The .0 change really counts on some engines like Bing which seems to allow
  Windows NT 10.0 but not Windows NT 10.

Signed-off-by: Markus Heiser <markus.heiser@darmarit.de>
2020-04-22 00:43:42 +02:00

74 lines
2.0 KiB
Python
Executable File

#!/usr/bin/env python
# set path
from sys import path
from os.path import realpath, dirname, join
path.append(realpath(dirname(realpath(__file__)) + '/../'))
#
import json
import requests
import re
from distutils.version import LooseVersion, StrictVersion
from lxml import html
from searx.url_utils import urlparse, urljoin
from searx import searx_dir
URL = 'https://ftp.mozilla.org/pub/firefox/releases/'
RELEASE_PATH = '/pub/firefox/releases/'
NORMAL_REGEX = re.compile('^[0-9]+\.[0-9](\.[0-9])?$')
# BETA_REGEX = re.compile('.*[0-9]b([0-9\-a-z]+)$')
# ESR_REGEX = re.compile('^[0-9]+\.[0-9](\.[0-9])?esr$')
#
useragents = {
"versions": (),
"os": ('Windows NT 10.0; WOW64',
'X11; Linux x86_64'),
"ua": "Mozilla/5.0 ({os}; rv:{version}) Gecko/20100101 Firefox/{version}"
}
def fetch_firefox_versions():
resp = requests.get(URL, timeout=2.0)
if resp.status_code != 200:
raise Exception("Error fetching firefox versions, HTTP code " + resp.status_code)
else:
dom = html.fromstring(resp.text)
versions = []
for link in dom.xpath('//a/@href'):
url = urlparse(urljoin(URL, link))
path = url.path
if path.startswith(RELEASE_PATH):
version = path[len(RELEASE_PATH):-1]
if NORMAL_REGEX.match(version):
versions.append(LooseVersion(version))
list.sort(versions, reverse=True)
return versions
def fetch_firefox_last_versions():
versions = fetch_firefox_versions()
result = []
major_last = versions[0].version[0]
major_list = (major_last, major_last - 1)
for version in versions:
major_current = version.version[0]
if major_current in major_list:
result.append(version.vstring)
return result
def get_useragents_filename():
return join(join(searx_dir, "data"), "useragents.json")
useragents["versions"] = fetch_firefox_last_versions()
with open(get_useragents_filename(), "w") as f:
json.dump(useragents, f, indent=4, ensure_ascii=False)