Issue 5496 - Use omahaproxy to get chrome-versions

globals/get_browser_versions.py

 import re
 import os
 import sys
 import json
 import urllib
 import errno
 import logging
 import time
 from xml.dom import minidom
 
 from jinja2 import contextfunction
 
 BROWSERS = {}
 BASE_URL = 'https://product-details.mozilla.org/1.0'
 FIREFOX_URL = BASE_URL + '/firefox_versions.json'
 THUNDERBIRD_URL = BASE_URL + '/thunderbird_versions.json'
 SEAMONKEY_URL = 'http://www.seamonkey-project.org/seamonkey_versions.json'
 CHROME_URL = 'https://omahaproxy.appspot.com/all.json?os=win'

[saroyanm, 2017/08/14 14:26:10]

The URL mentioned in the Issue tracker doesn't include os=win parameter. Is the
usage intentional ?

[Sebastian Noack, 2017/08/14 14:33:30]

Note that this is consistent with the old implementation, using the update API,
where we set <os platform="win">. Since the requirements page doesn't account
for different Chrome version potentially being released for different operating
systems, we have to abstract that detail away. Using the version numbers of
Chrome for Windows seems reasonable as this is by far the most widely used
operating system. I've updated the issue description.

[saroyanm, 2017/08/14 14:34:48]

Acknowledged.

 
 cache = {}
 
 
 def get_json_versions(product_url):
     response = urllib.urlopen(product_url)
     try:
         doc = json.load(response)
     except json.ValueError:
         print 'URL: %s not returning json object'.format(product_url)
@@ skipping 43 matching lines @@
             seamonkey_versions['LATEST_SEAMONKEY_MILESTONE_VERSION'],
             seamonkey_versions['LATEST_SEAMONKEY_DEVEL_VERSION'],
         ]
     }
 
     return versions
 
 BROWSERS['seamonkey'] = get_seamonkey_versions
 
 
-def get_chrome_version(version_info):
-    return version_info['version'].split('.')[0]
-
-
 def get_chrome_versions():
     response = urllib.urlopen(CHROME_URL)
     try:
         data = json.load(response)
     finally:
         response.close()
 
-    channels = {x['channel']: x for x in data[0]['versions']}

[Sebastian Noack, 2017/08/14 11:54:31]

It seems the logic can be simplified. All data we need for each channel is the
major version number. And if we extract it here already, we wouldn't need the
get_chrome_version() function.

  channels = {x['channel']: x['version'].split('.')[0]
              for x in data[0]['versions']}
  return {
      'current': channels['stable'],
      'unreleased': [channels['beta'], channels['dev']],
  }

[tlucas, 2017/08/14 12:08:28]

Done.

-
-    return {
-        'current': get_chrome_version(channels['stable']),
-        'unreleased': map(
-            get_chrome_version,
-            (channels['beta'], channels['dev'])
-        ),
+    versions = {
+        x['channel']: x['version'].split('.')[0]
+        for x in data[0]['versions']
+    }
+
+    return {
+        'current': versions['stable'],
+        'unreleased': [versions['beta'], versions['dev']],
     }
 
 BROWSERS['chrome'] = get_chrome_versions
 
 
 def get_opera_version(channel):
     response = urllib.urlopen('https://autoupdate.geo.opera.com/netinstaller/' + channel)
     try:
         spec = json.load(response)
     finally:
@@ skipping 120 matching lines @@
         file.seek(0)
         json.dump(persistent_cache, file)
         file.truncate()
 
     if not versions['previous']:
         logging.warning("Couldn't determine previous browser version, "
                         'please set %s.previous in %s', browser, filename)
 
     cache[browser] = versions
     return versions