html_tmpl_count

#!/usr/bin/env python

"""
html_template_count - count physical lines of code in HTML files which may or
 may not be templates.
Usage: html_template_count [-f file] [list_of_files]
 file: file with a list of files to count (if "-", read list from stdin)
 list_of_files: list of files to count
 -f file or list_of_files can be used, or both
"""

# NOTE: This plugin is not very efficient, since it reads whole source files
# into memory and performs multiple regular expression substitutions on those
# source files, instead of incrementally reading and parsing them. It works
# fine for our needs, but keep this in mind.

import re
import sys

COMMENT_PAIRS = [
    # HTML comments
    ('<!--', '-->'),
    # Jinja/Django comments
    ('{#', '#}'),
    ('{%\s*comment\s*%}', '{%\s*endcomment\s*%}'),
    # CSS or JS block comments
    ('/\*', '\*/'),
    # JS line comments
    ('//', None)
    ]

COMMENT_PATTERNS = [
    re.compile(r'%s.*?%s' % (start, end), re.M) if end else
    re.compile(r'%s.*$' % start)
    for start, end in COMMENT_PAIRS]

TOTAL_SLOC = 0

def count_file(path):
    global TOTAL_SLOC
    try:
        f = open(path.strip())
        src = f.read()
    finally:
        f.close()
    for pattern in COMMENT_PATTERNS:
        src = re.sub(pattern, '', src)
    sloc = sum(1 for line in src.splitlines() if line.strip())
    TOTAL_SLOC += sloc
    print '%s %s' % (sloc, path.strip())

def main(argv):
    # Do we have "-f" (read list of files from second argument)?
    if len(argv) > 1 and argv[0] == '-f':
        # Yes, we have -f
        if argv[1] == '-':
            # The list of files is in STDIN
            for path in sys.stdin:
                count_file(path)
        else:
            # The list of files is in the file $ARGV[1]
            try:
                f = open(argv[1])
                lines = f.readlines()
            finally:
                f.close()
            map(count_file, lines)
        argv = argv[2:]
    # Process all (remaining) arguments as file names
    for path in argv:
        count_file(path)

    print 'Total:'
    print TOTAL_SLOC


if __name__ == '__main__':
    sys.exit(main(sys.argv[1:]))