git-create-review.py

#!/usr/bin/python
#
# git-create-review.py
# based on script created here: http://mojodna.net/2009/02/24/my-work-git-workflow.html
# especially for the command to turn a git diff into a svn diff
import subprocess
from subprocess import Popen, PIPE, STDOUT
import sys

findRevCmd = 'git svn find-rev %s'

#Gets the diff between two revisions and then changes it into an svn diff
gitDiffCmd = """
    git diff --no-prefix %s^..%s -- %s |
    sed -e "/^diff --git [^[:space:]]*/{h;}" \\
    -e "s/^diff --git [^[:space:]]*/Index:/" \\
    -e "s/^index.*/===================================================================/" \\
    -e "/^new file .*/d" \\
    -e "/^--- \/dev\/null/{g;}" \\
    -e "s/^diff --git [^[:space:]]*/-+-/" \\
    -e "s/^+++ .*/&\\t(revision %s)/" -e "s/^--- .*/&\\t(revision %s)/" -e "s/^-+- .*/&\\t(revision 0)/" \\
    -e "s/^-+- [[:space:]]*/--- /"
    """


def getRevNumber(hash):
    '''Given a git hash, find the svn revision number'''
    cmd = findRevCmd % hash
    p = Popen(cmd, shell=True, stdout=PIPE)
    return p.communicate()[0].strip()

    
def getFileHashes(fileName, grepString):
    '''Returns a list of the hashes associated with this file by the grep.
    
    returns: a list of hashes.  If the git log returns an error, then this will return None'''
    cmd1 = 'git log --grep=%s --oneline -i --reverse %s' % (grepString, fileName)
    cmd2 = 'awk \'{print $1;}\'' 
    p1 = Popen(cmd1, shell=True, stdout=PIPE, stderr=subprocess.PIPE)
    p2 = Popen(cmd2, shell=True, stdin=p1.stdout, stdout=PIPE)
    
    p1.wait()
    
    # If return code is not 0, return none
    if(p1.returncode != 0):
        return None

    hashes = p2.communicate()[0].split('\n')
    hashes.remove('')
    return hashes
    
def getFirstAndLastHash(fileName, grepString):
    '''Returns the first and last hash for a grep in a change
    
    fileName - the file to find the hashes for
    grepString - string to grep for in the commit comments
    
    returns: an array with 0 set as the first hash and 1 the last hash'''
    allHashes = getFileHashes(fileName, grepString)
    if(allHashes == None):
        return None
    firstLastHashes = [allHashes[0], allHashes[len(allHashes) - 1]]
    return firstLastHashes


searchString = sys.argv[1]

#get a list of files by the grep statement
cmd = 'git log --grep=%s --pretty="format:" -i --name-only' % searchString
p = Popen(cmd, shell=True, stdout=PIPE)
affectedFiles = p.communicate()[0]
theFiles = affectedFiles.split('\n')
finalFileList = set()

#add to a set to ensure no repeats.
for thing in theFiles:
    if(thing != ''):
        finalFileList.add(thing)
        
#for each file, get the svn diff for it, and output the result
for file in finalFileList:
    hashes = getFirstAndLastHash(file, searchString)
    if(hashes == None):
        continue
    prevRev = getRevNumber("%s^" % hashes[0])
    rev = getRevNumber(hashes[1])
    diffCommand = gitDiffCmd % (hashes[0], hashes[1], file, rev, prevRev)
    finalDiff = Popen(diffCommand, shell=True, stdout=PIPE)
    print finalDiff.communicate()[0]