Get common user agents from the calibre website logs

setup/browser_data.py

@@ -3,19 +3,24 @@
 # License: GPLv3 Copyright: 2017, Kovid Goyal <kovid at kovidgoyal.net>
 
 
+import bz2
 import os
-import json
-import gzip
-import io
+import sys
 from datetime import datetime
-
-from setup import download_securely
+from urllib.request import urlopen
 
 from polyglot.builtins import filter
+from setup import download_securely
 
 is_ci = os.environ.get('CI', '').lower() == 'true'
 
 
+def download_from_calibre_server(url):
+    ca = os.path.join(sys.resources_location, 'calibre-ebook-root-CA.crt')
+    with urlopen(url, cafile=ca) as f:
+        return f.read()
+
+
 def filter_ans(ans):
     return list(filter(None, (x.strip() for x in ans)))
 
@@ -39,18 +44,15 @@ def common_user_agents():
             'Mozilla/5.0 (Windows NT 6.1; WOW64; Trident/7.0; rv:11.0) like Gecko',
         ]
     print('Getting recent UAs...')
-    raw = download_securely(
-        'https://raw.githubusercontent.com/intoli/user-agents/master/src/user-agents.json.gz')
-    data = json.loads(gzip.GzipFile(fileobj=io.BytesIO(raw)).read())
-    uas = []
-    for item in data:
-        ua = item['userAgent']
-        if not ua.startswith('Opera'):
-            uas.append(ua)
-    ans = filter_ans(uas)[:256]
-    if not ans:
-        raise ValueError('Failed to download list of common UAs')
-    return ans
+    raw = download_from_calibre_server('https://code.calibre-ebook.com/ua-popularity')
+    ans = {}
+    for line in bz2.decompress(raw).decode('utf-8').splitlines():
+        count, ua = line.partition(':')[::2]
+        count = int(count.strip())
+        ua = ua.strip()
+        if len(ua) > 20:
+            ans[ua] = count
+    return ans, list(sorted(ans, reverse=True, key=ans.__getitem__))
 
 
 def firefox_versions():
@@ -103,7 +105,7 @@ def chrome_versions():
 def all_desktop_platforms(user_agents):
     ans = set()
     for ua in user_agents:
-        if 'Mobile/' not in ua and ('Firefox/' in ua or 'Chrome/' in ua):
+        if ' Mobile ' not in ua and 'Mobile/' not in ua and ('Firefox/' in ua or 'Chrome/' in ua):
             plat = ua.partition('(')[2].partition(')')[0]
             parts = plat.split(';')
             if 'Firefox/' in ua:
@@ -113,10 +115,13 @@ def all_desktop_platforms(user_agents):
 
 
 def get_data():
+    ua_freq_map, common = common_user_agents()
     ans = {
         'chrome_versions': chrome_versions(),
         'firefox_versions': firefox_versions(),
-        'common_user_agents': common_user_agents(),
+        'common_user_agents': common,
+        'user_agents_popularity': ua_freq_map,
+        'timestamp': datetime.utcnow().isoformat() + '+00:00',
     }
     ans['desktop_platforms'] = list(all_desktop_platforms(ans['common_user_agents']))
     return ans

src/calibre/utils/random_ua.py

@@ -23,62 +23,14 @@ def all_firefox_versions(limit=10):
     return user_agent_data()['firefox_versions'][:limit]
 
 
-def random_firefox_version():
-    return random.choice(all_firefox_versions())
-
-
 def random_desktop_platform():
     return random.choice(user_agent_data()['desktop_platforms'])
 
 
-def render_firefox_ua(platform, version):
-    # https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/User-Agent/Firefox
-    return 'Mozilla/5.0 ({p}; rv:{ver}) Gecko/20100101 Firefox/{ver}'.format(
-        p=platform, ver=version)
-
-
-def random_firefox_ua():
-    render_firefox_ua(random_desktop_platform(), random_firefox_version())
-
-
 def all_chrome_versions(limit=10):
     return user_agent_data()['chrome_versions'][:limit]
 
 
-def random_chrome_version():
-    return random.choice(all_chrome_versions())
-
-
-def render_chrome_ua(platform, version):
-    return 'Mozilla/5.0 ({p}) AppleWebKit/{wv} (KHTML, like Gecko) Chrome/{cv} Safari/{wv}'.format(
-        p=platform, wv=version['webkit_version'], cv=version['chrome_version'])
-
-
-def random_chrome_ua():
-    return render_chrome_ua(random_desktop_platform(), random_chrome_version())
-
-
-def all_user_agents():
-    ans = getattr(all_user_agents, 'ans', None)
-    if ans is None:
-        uas = []
-        g = globals()
-        platforms = user_agent_data()['desktop_platforms']
-        for b in ('chrome', 'firefox'):
-            versions = g['all_%s_versions' % b]()
-            func = g['render_%s_ua' % b]
-            for v in versions:
-                for p in platforms:
-                    uas.append(func(p, v))
-        random.shuffle(uas)
-        ans = all_user_agents.ans = tuple(uas)
-    return ans
-
-
-def random_user_agent():
-    return random.choice(all_user_agents())
-
-
 def accept_header_for_ua(ua):
     if 'Firefox/' in ua:
         return 'text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8'