A simple script to determine what portions of a web app are HTML/CSS/JS. Uses the tool ``cloc``.

cloc.py

#!/usr/bin/env python

"""This requires Python 2.7. Deal with it."""

import shlex
import sys
import xml.etree.ElementTree as ET
from subprocess import check_output

# Language -> [extension, ...] mapping
FORCE_LANGS = {
    'HTML': ('jade', 'ejs'),
    'Javascript': ('json',)
}
err = lambda msg: sys.stderr.write('%s\n' % msg)


if __name__ == '__main__':
    if len(sys.argv) != 2:
        err('Usage: `%s www`, where www is your web app directory.' % sys.argv[0])
        sys.exit(1)

    force_langs = ' '.join('--force-lang="%s",%s' % (lang, e)
                           for lang in FORCE_LANGS for e in FORCE_LANGS[lang])
    out = check_output(shlex.split('cloc --quiet --xml %s %s' % (
        force_langs, sys.argv[1])))

    # Parse the results, extract HTML, CSS, JS.
    tree = ET.fromstring(out.strip())
    langs = {}
    total = 0
    for lang in ('HTML', 'CSS', 'Javascript'):  # Javascript, sic.
        node = tree.find(r".//language[@name='%s']" % lang)
        if node is not None:
            langs[lang] = int(node.attrib['code'])
            total += int(node.attrib['code'])

    # Calculate, display.
    print "Note: Rounding errors are possible."
    print

    for lang in ('HTML', 'CSS', 'Javascript'):
        count = langs.get(lang) or 0
        if lang == 'Javascript': lang = 'JavaScript'  # Seriously.
        print "%s: %d%%" % (lang, round(float(count) / total * 100))