JAR diff

jardiff.py

import argparse
import os
import sets
import subprocess
import sys
import tempfile


"""
Assumes procyon: 

https://bitbucket.org/mstrobel/procyon/downloads/

Sample usage:

python jardiff.py --procyon ~/gg/procyon-decompiler-0.5.36.jar  ~/gg/fe ~/gg/bf

"""

args = None

def main():
    global args
    parser = argparse.ArgumentParser()
    parser.add_argument("--procyon", help="location of procyon JAR")
    parser.add_argument("first")
    parser.add_argument("second")
    args = parser.parse_args()
    if os.path.isdir(args.first):
        dir1 = args.first
    else:
        if not(args.first.endswith(".jar")):
            print "Expected either a JAR or dir as first argument"
            sys.exit(1)
        raise Exceptino("TODO unjar")
    if os.path.isdir(args.second):
        dir2 = args.second
    else:
        if not(args.second.endswith(".jar")):
            print "Expected either a JAR or dir as first argument"
            sys.exit(1)
        raise Exceptino("TODO unjar")
    compare_dirs(dir1, dir2)

def walker(files, dirname, fnames):
    for fname in fnames:
        abs_fname = os.path.join(dirname, fname)
        if os.path.isdir(abs_fname):
            os.path.walk(abs_fname, walker, files)
        else:
            files.append(abs_fname)

def get_all_files(dir):
    print "Getting files from %s" % dir
    files = []
    os.path.walk(dir, walker, files)
    # Make them relative
    files = [f[len(dir)+1:] for f in files]
    return files
    


def compare_dirs(dir1, dir2):
    print "Comparing %s vs %s" % (dir1, dir2)
    files1 = get_all_files(dir1)
    files2 = get_all_files(dir2)
    in2only = sets.Set(files2) - sets.Set(files1)
    if in2only:
        print "Files in %s but not in %s" % (dir2, dir1)
        print "\n\t".join(list(in2only))
        print
    for f in files1:
        if f not in files2:
            print "%s not in %s" % (f, dir2)
            continue
        f1 = os.path.join(dir1, f)
        f2 = os.path.join(dir2, f)
        # TODO really this should check checksum, not size
        s1 = os.stat(f1).st_size
        s2 = os.stat(f2).st_size
        if s1 != s2:
            print "Different: %s (%s vs %s)" % (f, s1, s2)
            if f.endswith(".class"):
                f1java = tempfile.NamedTemporaryFile(suffix='.java', delete=False) 
                f1java.write("")
                f1java.close()
                cmd = "java -jar %s %s > %s" % (args.procyon, f1, f1java.name)
                # print cmd
                subprocess.check_call(cmd, shell=True)
                f2java = tempfile.NamedTemporaryFile(suffix='.java', delete=False)
                f2java.write("")
                f2java.close()
                cmd = "java -jar %s %s > %s" % (args.procyon, f2, f2java.name)
                # print cmd
                subprocess.check_call(cmd, shell=True)
                diff_cmd = ["diff","-Z",f1java.name, f2java.name]
                print "Executing %s" % " ".join(diff_cmd)
                try:
                    subprocess.check_output(diff_cmd)
                except subprocess.CalledProcessError, cpe:
                    if cpe.returncode == 1:
                        # diff will return 1 if files are different
                        diff = cpe.output
                        print diff
                    else:
                        # we already know it will not return 0 because they are different
                        # so it can only be 2 - trouble
                        raise
                os.unlink(f1java.name)
                os.unlink(f2java.name)

            else:
                print "\tDo not know how to diff"


if __name__ == "__main__":
    main()