Merge branch 'master' into dsgvo

searx/data/useragents.json

@@ -0,0 +1,14 @@
+{
+    "ua": "Mozilla/5.0 ({os}; rv:{version}) Gecko/20100101 Firefox/{version}",
+    "versions": [
+        "61.0.1",
+        "61.0",
+        "60.0.2",
+        "60.0.1",
+        "60.0"
+    ],
+    "os": [
+        "Windows NT 10; WOW64",
+        "X11; Linux x86_64"
+    ]
+}

searx/utils.py

@@ -10,8 +10,10 @@ from codecs import getincrementalencoder
 from imp import load_source
 from numbers import Number
 from os.path import splitext, join
+from io import open
 from random import choice
 import sys
+import json
 
 from searx import settings
 from searx.version import VERSION_STRING
@@ -39,29 +41,11 @@ else:
 
 logger = logger.getChild('utils')
 
-ua_versions = ('52.8.1',
-               '53.0',
-               '54.0',
-               '55.0',
-               '56.0',
-               '57.0',
-               '58.0',
-               '59.0',
-               '60.0.2')
-
-ua_os = ('Windows NT 6.3; WOW64',
-         'X11; Linux x86_64',
-         'X11; Linux x86')
-
-ua = "Mozilla/5.0 ({os}; rv:{version}) Gecko/20100101 Firefox/{version}"
-
 blocked_tags = ('script',
                 'style')
 
-
-def gen_useragent(os=None):
-    # TODO
-    return ua.format(os=os or choice(ua_os), version=choice(ua_versions))
+useragents = json.loads(open(os.path.dirname(os.path.realpath(__file__))
+                             + "/data/useragents.json", 'r', encoding='utf-8').read())
 
 
 def searx_useragent():
@@ -70,6 +54,10 @@ def searx_useragent():
            suffix=settings['outgoing'].get('useragent_suffix', ''))
 
 
+def gen_useragent(os=None):
+    return str(useragents['ua'].format(os=os or choice(useragents['os']), version=choice(useragents['versions'])))
+
+
 def highlight_content(content, query):
 
     if not content:

utils/fetch_firefox_version.py

@@ -0,0 +1,73 @@
+#!/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; 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)