check-changelog

#!/usr/bin/python
import sys
import re

# Match the <day> <month> <year> format we use to find
# where entries begin
date_format="\d{1,2} \w{3} \d{4}"

# Used in version/revision bumps *foobar-1.0 (date)
date_regex = re.compile(date_format)

# Used in entries with the actual change text
date_regex_semicolon = re.compile("(%s);" % date_format)

# + is used in package names (for example gtk+) so try to handle that
ebuild_regex = re.compile("((,|\s)\+[\S\+]+\.ebuild)")

def check_changelog(file):
	f = open(file)
	lines = f.readlines()

	# All *${P} entries as ${P}
	bumps = []

	file_printed = False

	# Scan the whole file for new entries because with for example moves
	# from one category to another we have the ebuilds appearing as new
	# ones again
	for line in lines:
		if line.startswith("*"):
			# Check that the date is in valid form
			if not date_regex.search(line):
				if not file_printed:
					print file + ":"
					file_printed = True
				sys.stdout.write("\tbad date entry for new revision/version entry: " + line)
			else:
				try:
					bumps.append(line[1:line.index(" ")])
				except ValueError:
					sys.stderr.write("Failure in parsing: " + file + "\n")
	n=0
	length = len(lines)
	while n < length:
		line = lines[n]
		# Use dates to find separate ChangeLog entries
		match = date_regex_semicolon.search(line)
		if match:
			date = match.group(1)
			while True:
				for m in ebuild_regex.findall(line):
					# remove + and trailing .ebuild
					to_find = m[0][2:-7]
					if not to_find in bumps:
						if not file_printed:
							print file + ":"
							file_printed = True
						print "\tMissing new ebuild entry *" + to_find +" (%s) for addition on line " % date + str(n+1)
				n+=1
				if line.find(":") >= 0 or n == length:
					break
				line = lines[n]
		n+=1	
	f.close()
	return not file_printed

bad_found = False
for ch in sys.argv[1:]:
	if not check_changelog(ch):
		bad_found = True

if bad_found:
	sys.exit(1)