fix-copyright-years

#!/usr/bin/env python

import re
import optparse
import datetime
import subprocess
import multiprocessing

co_line_matcher = re.compile('^.*Copyright (.*) Ettus Research LLC$')

def command(*args): return subprocess.Popen(args, stdout=subprocess.PIPE).communicate()[0]

def get_co_line(lines):
    for i, line in enumerate(lines[:5]):
        if co_line_matcher.match(line): return line, i
    return None, None

def fix_co_years(files, keep_years):
    for file in files:
        print file
        lines = open(file).readlines()
        line, num = get_co_line(lines)
        if line is None: continue

        #extract the years from the git history
        log_years = map(
            lambda l: int(l.split()[-2]),
            filter(
                lambda l: l.startswith('Date'),
                command('git', 'log', file).splitlines(),
            ),
        )
        log_years = min(log_years), max(log_years)

        #extract years from co string
        try:
            co_years_str = co_line_matcher.match(line).groups()[0]
            co_years = map(int, co_years_str.split('-'))
            co_years = min(co_years), max(co_years)
        except Exception, e:
            print '    format error on line %d: "%s"'%(num, line), e
            continue

        #keep years means log years is a superset
        if keep_years: log_years = min(co_years+log_years), max(co_years+log_years)

        if log_years != co_years:
            print '    log years: %s != copyright years: %s'%(log_years, co_years)
            year_now = datetime.datetime.now().year
            all_years = min(log_years), max(list(log_years)+[year_now]) #add the current year
            all_years_str = '%s-%s'%all_years
            if all_years[0] == all_years[1]: all_years_str = str(all_years[0])
            new_text = ''.join(lines[:num] + [line.replace(co_years_str, all_years_str)] + lines[num+1:])
            open(file, 'w').write(new_text)

if __name__ == "__main__":
    parser = optparse.OptionParser(usage="usage: %prog [options] path")
    parser.add_option("-k", "--keep", action="store_true", help="keep copyright years", default=False)
    (options, args) = parser.parse_args()

    #get recursive list of files in the repo
    files = command('git', 'ls-tree', '--name-only', 'HEAD', '-r', args[0]).splitlines()

    #start n+1 processes to handle the files
    num_procs = multiprocessing.cpu_count()
    procs = [multiprocessing.Process(
        target=lambda *files: fix_co_years(files, keep_years=options.keep),
        args=files[num::num_procs],
    ) for num in range(num_procs)]
    map(multiprocessing.Process.start, procs)
    map(multiprocessing.Process.join, procs)