| 
									
										
										
										
											2017-05-28 15:46:45 +02:00
										 |  |  | #!/usr/bin/env python | 
					
						
							| 
									
										
										
										
											2022-01-03 12:58:48 +01:00
										 |  |  | # lint: pylint | 
					
						
							| 
									
										
										
										
											2021-10-03 15:12:09 +02:00
										 |  |  | # SPDX-License-Identifier: AGPL-3.0-or-later | 
					
						
							| 
									
										
										
										
											2017-05-28 15:46:45 +02:00
										 |  |  | 
 | 
					
						
							| 
									
										
										
										
											2022-01-03 12:40:06 +01:00
										 |  |  | """Fetch firefox useragent signatures
 | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | Output file: :origin:`searx/data/useragents.json` (:origin:`CI Update data ... | 
					
						
							|  |  |  | <.github/workflows/data-update.yml>`). | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | """
 | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2017-05-28 15:46:45 +02:00
										 |  |  | import json | 
					
						
							|  |  |  | import re | 
					
						
							| 
									
										
										
										
											2022-01-03 12:58:48 +01:00
										 |  |  | from os.path import join | 
					
						
							| 
									
										
										
										
											2020-08-06 17:42:46 +02:00
										 |  |  | from urllib.parse import urlparse, urljoin | 
					
						
							| 
									
										
										
										
											2022-03-25 08:39:40 +01:00
										 |  |  | from distutils.version import LooseVersion  # pylint: disable=deprecated-module | 
					
						
							| 
									
										
										
										
											2022-01-03 12:58:48 +01:00
										 |  |  | 
 | 
					
						
							|  |  |  | import requests | 
					
						
							| 
									
										
										
										
											2017-05-28 15:46:45 +02:00
										 |  |  | from lxml import html | 
					
						
							| 
									
										
										
										
											2018-08-05 10:55:42 +02:00
										 |  |  | from searx import searx_dir | 
					
						
							| 
									
										
										
										
											2017-05-28 15:46:45 +02:00
										 |  |  | 
 | 
					
						
							|  |  |  | URL = 'https://ftp.mozilla.org/pub/firefox/releases/' | 
					
						
							|  |  |  | RELEASE_PATH = '/pub/firefox/releases/' | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2022-01-03 12:58:48 +01:00
										 |  |  | NORMAL_REGEX = re.compile(r'^[0-9]+\.[0-9](\.[0-9])?$') | 
					
						
							|  |  |  | # BETA_REGEX = re.compile(r'.*[0-9]b([0-9\-a-z]+)$') | 
					
						
							|  |  |  | # ESR_REGEX = re.compile(r'^[0-9]+\.[0-9](\.[0-9])?esr$') | 
					
						
							| 
									
										
										
										
											2017-05-28 15:46:45 +02:00
										 |  |  | 
 | 
					
						
							| 
									
										
										
										
											2021-10-03 15:12:09 +02:00
										 |  |  | # | 
					
						
							| 
									
										
										
										
											2018-08-05 10:55:42 +02:00
										 |  |  | useragents = { | 
					
						
							| 
									
										
										
										
											2021-12-27 09:16:03 +01:00
										 |  |  |     # fmt: off | 
					
						
							| 
									
										
										
										
											2017-05-28 15:46:45 +02:00
										 |  |  |     "versions": (), | 
					
						
							| 
									
										
										
										
											2021-12-16 23:10:39 +01:00
										 |  |  |     "os": ('Windows NT 10.0; Win64; x64', | 
					
						
							| 
									
										
										
										
											2017-05-28 15:46:45 +02:00
										 |  |  |            'X11; Linux x86_64'), | 
					
						
							| 
									
										
										
										
											2021-12-27 09:16:03 +01:00
										 |  |  |     "ua": "Mozilla/5.0 ({os}; rv:{version}) Gecko/20100101 Firefox/{version}", | 
					
						
							|  |  |  |     # fmt: on | 
					
						
							| 
									
										
										
										
											2017-05-28 15:46:45 +02:00
										 |  |  | } | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 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) | 
					
						
							| 
									
										
										
										
											2022-01-03 12:58:48 +01:00
										 |  |  |     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 | 
					
						
							| 
									
										
										
										
											2017-05-28 15:46:45 +02:00
										 |  |  | 
 | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 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] | 
					
						
							| 
									
										
										
										
											2018-08-05 10:55:42 +02:00
										 |  |  |         if major_current in major_list: | 
					
						
							| 
									
										
										
										
											2017-05-28 15:46:45 +02:00
										 |  |  |             result.append(version.vstring) | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |     return result | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2018-08-05 10:55:42 +02:00
										 |  |  | def get_useragents_filename(): | 
					
						
							|  |  |  |     return join(join(searx_dir, "data"), "useragents.json") | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2022-01-03 12:40:06 +01:00
										 |  |  | if __name__ == '__main__': | 
					
						
							|  |  |  |     useragents["versions"] = fetch_firefox_last_versions() | 
					
						
							|  |  |  |     with open(get_useragents_filename(), "w", encoding='utf-8') as f: | 
					
						
							|  |  |  |         json.dump(useragents, f, indent=4, ensure_ascii=False) |