trimr.py

#!/usr/bin/python

#
# On the Hadoop JIRA bug tracker, sometimes large patches are presented as two
# files: one file which has a list of "svn rm" commands, and another with the
# actual diff.
# This script takes a "normal" patch file generated by git diff, svn diff, or
# the patchr.py script, and turns it into this two file hybrid.
#

import getopt
import os
import re
import stat
import string
import subprocess
import sys

output_re = re.compile("^(.*).patch$")
svn_rm_re = re.compile("svn rm ([^ ]*)")
git_diff_re= re.compile("diff --git ([^ ]*) ([^ ]*)")
deletion_re = re.compile("\+\+\+ /dev/null")

## Parse options
def print_usage():
    print os.path.basename(
sys.argv[0]) + ": a program to remove files from a patch.\n\
The files to be removed must be specified in a special 'svn rm' file.\n\
\n\
Usage: " + os.path.basename(sys.argv[0]) + " [options]\n\
\n\
Options: -h: this help message\n\
         -p <file-name>: specify patch file (required)\n\
         -v: turn on verbose mode (to stderr)\n\
"

try:
    optlist, dirs = getopt.getopt(sys.argv[1:], ':hOp:r:v')
except getopt.GetoptError:
    print_usage()
    sys.exit(1)

in_fname = None
verbose = False
for opt in optlist:
    if opt[0] == '-h':
        print_usage()
        sys.exit(0)
    if opt[0] == '-p':
        in_fname = opt[1]
    if opt[0] == '-v':
        verbose = True
if (in_fname == None):
    print >>sys.stderr, "You must specify a patch name."
    print_usage()
    sys.exit(1)
m = output_re.match(in_fname)
if (not m):
    print >>sys.stderr, "Failed to automatically determine name of \
output(s).  Aborting."
    sys.exit(1)
pout_fname = m.group(1) + ".trimmed.patch"
if (os.path.exists(pout_fname)):
    print >>sys.stderr, "File '" + pout_fname + "' already exists!  \
Aborting"
    sys.exit(1)
poutf = open(pout_fname, "w")
rout_fname = m.group(1) + ".rm.patch"

rms = {}
if (os.path.exists(rout_fname)):
    if (verbose):
        print >>sys.stderr, "Using existing svn rm file."
    f = open(rout_fname, "r")
    try:
        lineno = 0
        for line in f:
            lineno = lineno + 1
            m = svn_rm_re.match(line)
            if (not m):
                raise RuntimeError("failed to parse line " + lineno + " of " +
                    rout_fname)
            rms[str(m.group(1)).rstrip()] = 1
    finally:
        f.close()
else:
    if (verbose):
        print >>sys.stderr, "Creating new svn rm file."
    f = open(in_fname, "r")
    try:
        lineno = 0
        cur_file = None
        for line in f:
            lineno = lineno + 1
            m = deletion_re.match(line)
            if (m):
                if (cur_file == None):
                    raise RuntimeError("saw deletion, but not the file name \
it was associated with!")
                rms[cur_file] = 1
                continue
            m = git_diff_re.match(line)
            if (m):
                cur_file = str(m.group(1))
    finally:
        f.close()
    if len(rms.keys()) != 0:
        f = open(rout_fname, "w")
        try:
            for k in rms.keys():
                print >>f, "svn rm " + k
        finally:
            f.close()

if (len(rms.keys()) == 0):
    print >>sys.stderr, "There were no deleted files found.  The trimmed \
patch would be the same as the original patch."
    sys.exit(0)
if (verbose):
    print >>sys.stderr, "removing " + str(len(rms.keys())) + " files from the patch..."

printing = True
f = open(in_fname, "r")
try:
    for line in f:
        m = git_diff_re.match(line)
        if (not m):
            if (printing):
                print >>poutf, line,
        else:
            fname = str(m.group(1))
            #print >>sys.stderr, "fname = '" + fname + "'"
            if (rms.has_key(fname)):
                if (verbose):
                    print >>sys.stderr, "skipping " + fname
                del rms[fname]
                printing = False
            else:
                printing = True
                print >>poutf, line,
finally:
    f.close()

if (len(rms.keys()) != 0):
    print >>sys.stderr, "ERROR!  Failed to use all svn rm directives!  Unused:"
    for k in rms.keys():
        print >>sys.stderr, str(k)
    raise RuntimeError("failed to use all svn directives")

sys.exit(0)