Applies some of the changes to unified diff necessary to port from H-uru/Plasma to CWE-ou

hurutooupatch.py

#!/usr/bin/env python

import sys
import re
import fileinput
import time
import calendar

def innerspacestotabs(m):
	if m.end() % 4 == 0:
		return (m.end()-m.start()+3)//4*"\t"
	else:
		return m.group()

kStateFirstLine = 0
kStateHeaders = 1
kStateMessage = 2
kStateDiffstat = 3
kStateOutsideHunk = 4
kStateInsideHunk = 5

state = kStateFirstLine
spacestotabs = True
for line in fileinput.input():
	if line.endswith("\r\n"):
		line = line[:-2]
	elif line.endswith("\n") or line.endswith("\r"):
		line = line[:-1]
	
	if state == kStateFirstLine:
		if line.startswith("From "):
			state = kStateHeaders
			metadata = "# HG changeset patch\n"
			message = ""
			continue
		else:
			state = kStateOutsideHunk
	
	if state == kStateHeaders:
		if line.startswith("From: "):
			metadata += "# User %s\n" % line[6:]
		elif line.startswith("Date: "):
			tzoffset = (1 if line[-5] == "-" else -1)*(int(line[-4:-2])*3600 + int(line[-2:])*60)
			metadata += "# Date %d %d\n" % (calendar.timegm(time.strptime(line[6:-6], "%a, %d %b %Y %H:%M:%S")) + tzoffset, tzoffset)
		elif line.startswith("Subject: "):
			message = re.sub(r"^\[PATCH.*\] ", "", line[9:]) + "\n"
		elif len(line) == 0:
			sys.stdout.write(metadata)
			message += "\n"
			state = kStateMessage
		else:
			# junk header, ignore
			pass
	elif state == kStateMessage:
		if line == "---":
			sys.stdout.write(message)
			state = kStateDiffstat
		else:
			message += line + "\n"
	elif state == kStateDiffstat:
		if line.startswith("diff "):
			sys.stdout.write("\n")
			sys.stdout.write(line)
			sys.stdout.write("\n")
			state = kStateOutsideHunk
		else:
			# ignore
			pass
	elif state == kStateOutsideHunk:
		# outside of hunk
		m = re.match(r"^(--- (a/)?|\+\+\+ (b/)?)(.*)$", line)
		if m:
			# file name
			prefix = m.group(1)
			filename = m.group(4)
			if filename.startswith("Sources/"):
				filename = "MOULOpenSourceClientPlugin/Plasma20/" + filename
			spacestotabs = not filename.endswith(".py") and not filename.endswith(".rc") and not filename.endswith("resource.h")
			line = prefix + filename
		else:
			m = re.match(r"^@@ -([0-9]+),([0-9]+) \+([0-9]+),([0-9]+) @", line)
			if m:
				# hunk header
				hunklenold = int(m.group(2))
				hunklennew = int(m.group(4))
				state = kStateInsideHunk
			else:
				# junk
				pass
		sys.stdout.write(line)
		sys.stdout.write("\n")
	else:
		# kStateInsideHunk
		m = re.match(r"([ +-])((    )*)(.*)$", line)
		if m:
			if spacestotabs:
				line = m.group(1) + len(m.group(2))//4*"\t" + re.sub(r" {2,}", innerspacestotabs, m.group(4))
			
			if m.group(1) == " ":
				hunklenold -= 1
				hunklennew -= 1
			elif m.group(1) == "+":
				hunklennew -= 1
			elif m.group(1) == "-":
				hunklenold -= 1
			if hunklennew <= 0 and hunklenold <= 0:
				if hunklennew != hunklenold:
					sys.stderr.write("hunk length mismatch on line %d:\n" % fileinput.lineno())
					sys.stderr.write(line)
					sys.stderr.write("\n")
				# end of hunk
				state = kStateOutsideHunk
		elif line[0] == "\\":
			# "\ No newline at end of file"
			pass
		else:
			sys.stderr.write("malformed line %d:\n" % fileinput.lineno())
			sys.stderr.write(line)
			sys.stderr.write("\n")
			sys.exit(1)
		sys.stdout.write(line)
		sys.stdout.write("\r\n")