check what plugins a wordpress site has installed via http

start.py

# prepare before running:
# 1. create folder "plugin_db/"
# 2. chdir to this folder and run this command: `svn list http://plugins.svn.wordpress.org > plugins.txt`
# 3. split plugin.txt in to multiple chunks:
#    plugin.txt: ~50000 lines => 50000/10=5000 => split with 5000 lines. Command: `split -l 5000 plugins.txt`
#    every chunk of the file is a worker

import subprocess, glob, os

os.chdir("plugin_db/")
for file in glob.glob("*"):
    if file == 'plugins.txt':   # ignore src file
        continue
    command = "python /Users/gabriel/tmp/wordpress-plugin-finder/worker.py " + file
    print command
    subprocess.Popen(command, shell=True)

worker.py

import urllib2, sys

pluginDBfile = "/Users/gabriel/tmp/wordpress-plugin-finder/plugin_db/" + sys.argv[1]

with open(pluginDBfile, "r") as f:
  for line in f:
    plugin = line.replace("\n", '')
    url    = 'http://a-wordpress-site.cc/wp-content/plugins/' + plugin

    try:
        connection = urllib2.urlopen(url)
        print connection.getcode()
        connection.close()
    except urllib2.HTTPError, e:
        if e.getcode() == 404:
            print "> NOT FOUND: " + plugin
        else:
            print "FOUND: HTTP(" + str(e.getcode()) + ") | Plugin: " + plugin