Generate SVN diffs from git. Checkout should be done via git svn. Dependencies: git subversion python3 (could work with 2 too)

git-svn-diff.py

#!/usr/bin/env python
import subprocess
import re

def getOutput(cmd):
	command_stdout = subprocess.Popen(cmd, stdout=subprocess.PIPE).communicate()[0]
	return command_stdout.decode(encoding='utf-8')

# Find the most recent commit that came from SVN and return the SVN revision.
def getSVN():
	revisions = getOutput(['git',  'rev-list', '--date-order', '--max-count=200', 'HEAD'])
	#find the svn revision
	for rev in revisions.split():
		svnrev = getOutput(['git', 'svn', 'find-rev', rev]).strip()
		if svnrev:
			return [svnrev, rev]
	# if we haven't returned yet we haven't found a svn revision
	# TODO fail if we can't find one
	print('Found no svn revision')


rev = getSVN()
svnrev = rev[0]
gitrev = rev[1]

#Generate a diff from the gitrev and transform it to a svn diff
# We use --no-pager as git diff will use a pager by default
diff = getOutput(['git', '--no-pager', 'diff', '--no-prefix', gitrev])

# For new added files only
newfiles = re.compile(r"""
^diff\s--git\s+([ab]/)?(?P<path>[^\s]*)\s.*\n		# Match the first line of a changed file and get the path (exluding a leading a/ or b/)
new.*\n							# Match optional 'new file mode' line
index.*\n						# Match the index line for replacement with =
---\s/dev/null\n					# Match a new file TODO add group with | for modified files
\+\+\+\s([ab]/)?(?P=path)\n				# Match for new file
""", re.VERBOSE | re.MULTILINE) #TODO add check for content and not ^^diff
diff = newfiles.sub(r"""Index: \g<path>\n"""+67*'='+"""\n--- \g<path>\t(revision 0)\n+++ \g<path>\t(working copy)\n""", diff)

# For modified files
modifiedfiles = re.compile(r"""
^diff\s--git\s+([ab]/)?(?P<path>[^\s]*)\s.*\n		# Match the first line
index.*\n						# =
---\s([ab]/)?(?P=path)\n				# old
\+\+\+\s([ab]/)?(?P=path)\n				# new
""", re.VERBOSE | re.MULTILINE)
diff = modifiedfiles.sub(r"""Index: \g<path>\n"""+67*'='+"""\n--- \g<path>\t(revision """+svnrev+""")\n+++ \g<path>\t(working copy)\n""", diff)

# For binary files
# is the mime-type line needed?
binaryfiles = re.compile(r"""
^diff\s--git\s+([ab]/)?(?P<path>[^\s]*)\s.*\n		# Match the first line
^index.*\n						# =
^Binary\ files.*\n
""", re.VERBOSE | re.MULTILINE)
diff = binaryfiles.sub(r"""Index: \g<path>\n"""+67*'='+"""\nCannot display: file marked as binary type.\nsvn:mime-type = application/octet-stream\n""", diff)


# Remove stuff after the second @@
modifiedblock = re.compile(r"""
^@@\ (?P<modified>[-][0-9]+,[0-9]+\ [+][0-9]+(,[0-9]+)?)\ @@.*		#match the @@ lines
""", re.VERBOSE | re.MULTILINE)
diff = modifiedblock.sub(r"""@@ \g<modified> @@""", diff)

print(diff)