given a patch file, extract the files that get changed

filesInPatch.py

#!/usr/bin/env python

# usage: filesInPatch.py foo.patch

import sys
import re

# Initialize defaults
debug = 0
fileNames = {}

# ======================================================================

def printError(msg):
    sys.stderr.write(msg)

# ======================================================================

def parseInput(inputFile):
    global debug, fileNames

    while 1:
        lineBuffer = inputFile.readline()
        if not lineBuffer: break
        # print lineBuffer,
        lineMatch = re.search("^\s*---\s+([^\s@]+)[\s@]+", lineBuffer)
        if not lineMatch:
            lineMatch = re.search("^\s*\+\+\+\s+([^\s@]+)[\s@]+", lineBuffer)
        if lineMatch:
            currFileName = lineMatch.group(1)

            # trim off 'a/' and 'b/' that you will normally see in git output
            #
            if len(currFileName) > 2 and currFileName[1] == '/' and \
                    (currFileName[0] == 'a' or currFileName[0] == 'b'):
                currFileName = currFileName[2:]

            # ignore funny files that git can produce
            #
            if currFileName == '/dev/null':
                continue

            if not currFileName in fileNames:
                # print currFileName
                fileNames[currFileName] = 1
            else:
                fileNames[currFileName] += 1

# ======================================================================

def pruneUnwantedNames():
    global debug, fileNames

    names = fileNames.keys()
    unwantedNames = []

    for currName in names:
        # ignore files that end in '.orig' as long as non-.orig exists
        lineMatch = re.search("^(.+)\.[oO][Rr][iI][gG]$", currName)
        if lineMatch and lineMatch.group(1) in fileNames:
            unwantedNames.append( currName )

    for currName in unwantedNames:
        del fileNames[ currName ]
        if debug:
            print("unwanted name: {}".format(currName))

# ======================================================================

def printFileNames():
    global debug, fileNames

    names = list(fileNames.keys())
    names.sort()

    if debug:
        for currName in names:
            print("{} ==> {}".format(currName, fileNames[currName]))
    else:
        for currName in names:
            print(currName)

# ======================================================================

if __name__ == '__main__':

    if len(sys.argv) == 1:
        parseInput(sys.stdin)
    else:
        for currInputName in sys.argv[1:]:
            try:
                # print currInputName
                currInputFile = open(currInputName, 'r')
                parseInput(currInputFile)
                currInputFile.close()
            except IOError as eStr:
                printError("Cannot open %s: %s\n" % (currInputName, eStr))
                sys.exit(255)

    pruneUnwantedNames()
    printFileNames()
    sys.exit(0)

# ======================================================================