#!/usr/bin/python
# Based on version control history, check that copyright notices mention all
# the years in which files were altered.
# Adam Sampson <ats@offog.org>

import sys, os, re
from offog import warn

def parse_darcs(f):
	"""Parse the output of "darcs log --last N -v"."""

	changes = {}

	year = None
	for l in f.readlines():
		l = l[:-1]

		# Tue May 19 13:58:49 BST 2009  Neil Brown ...
		m = re.match(r'^[A-Z].* (\d\d\d\d)  ', l)
		if m is not None:
			year = m.group(1)
			continue

		#     hunk ./common/Types.hs 204
		m = re.match(r'^ +hunk \./([^ ]*) ', l)
		if m is not None:
			s = changes.setdefault(m.group(1), set())
			s.add(year)

	return changes

def scan_files(fns):
	"""Given a list of filenames (some of which might not exist),
	read their copyright headers."""

	changes = {}

	for fn in fns:
		try:
			f = open(fn)
		except IOError:
			continue

		s = set()
		changes[fn] = s

		n = 0
		for l in f.readlines():
			n += 1
			l = l.strip()

			m = re.match(r'^Copyright \(C\) (.*)  (.*)$', l)
			if m is not None:
				years = re.split(r',\s*', m.group(1))
				if years != sorted(years):
					warn(fn, ":", n, ": years out of order")
				for year in years:
					if not re.match(r'^\d{4}$', year):
						warn(fn, ":", n, ": non-standard year list")
				author = m.group(2)
				for year in years:
					s.add(year)
				continue

			m = re.match(r'^Copyright.*20', l)
			if m is not None:
				warn(fn, ":", n, ": non-standard copyright line: '", l, "'")

		f.close()

	return changes

if __name__ == "__main__":
	history = parse_darcs(sys.stdin)
	fns = history.keys()
	notices = scan_files(fns)

	for fn in sorted(fns):
		if fn not in notices:
			warn(fn, ": no longer exists")
			continue
		if len(notices[fn]) == 0:
			warn(fn, ": no copyright notices")
			continue
		missing = history[fn] - notices[fn]
		if len(missing) != 0:
			s = ", ".join(sorted(missing))
			warn(fn, ": missing years: ", s)
			sys.stdout.write("sed 's/^\(Copyright.*\)\(  University.*\)$/\\1, %s\\2/' %s\n" % (s, fn))