Calculates the size of the PHP codebase of all sites.

php-codesize.py

#!/usr/bin/env python
# Calculates the size of the PHP codebase of all sites.

import os
import optparse
import subprocess


def calculate(http_root_dir):
    sites = list_sites(http_root_dir)
    sites_with_sizes = []
    total_size = 0
    for site in sites:
        size = int(get_php_size(site))
        sites_with_sizes.append([site, size])
        total_size += size
    sites_with_sizes.append(["TOTAL", total_size])

    # Sort descendingly on the second item in each element: the size.
    sites_with_sizes.sort(key=lambda a: a[1], reverse=True)

    # Output.
    for (site, size) in sites_with_sizes:
        print "%-45s %5.2f MiB" % (site, size / 1024.0 / 1024.0)


def list_sites(http_root_dir):
    (stdout, stderr) = cmd('find %s -mindepth 1 -maxdepth 1 -type d -not -iname "%s"' % (http_root_dir, "aaaaaa"))
    sites = stdout.split("\n")
    return sites


def get_php_size(dir):
    """get size in bytes of all PHP code in this directory"""
    # Fast.
    (stdout, stderr) = cmd('find -H %s \( -iname "*.inc" -or -iname "*.php" -or -iname "*.module" \) -exec du -bcs "{}" + | tail -1 | awk \'{ print $1 }\' 2>&1 | grep -v "permission denied"' % (dir))
    # Slow.
    # (stdout, stderr) = cmd('find -H %s \( -iname "*.inc" -or -iname "*.php" -or -iname "*.module" \) -exec echo -n -e {}"\\0" \; | du -bcs --files0-from=- | tail -1 | awk \'{ print $1 }\' 2>&1 | grep -v "permission denied"' % (dir))
    if len(stdout) == 0:
        return 0
    else:
        return stdout


def cmd(command):
    """run a command and get (stdout, stderr) back"""
    p = subprocess.Popen(command, shell=True, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
    (stdout, stderr) = p.communicate()
    (stdout, stderr) = (stdout.rstrip(), stderr.rstrip())
    return (stdout, stderr)


def main():
    p = optparse.OptionParser(description="Calculates the size of the PHP codebase of all sites.")
    options, arguments = p.parse_args()
    if len(arguments) == 1:
        calculate(arguments[0])
    else:
        p.print_help()


if __name__ == "__main__":
    main()