Add option to specify similarity level for git diff operations on commandline

git_cl.py

 #!/usr/bin/env python
 # Copyright (c) 2012 The Chromium Authors. All rights reserved.
 # Use of this source code is governed by a BSD-style license that can be
 # found in the LICENSE file.
 
 # Copyright (C) 2008 Evan Martin <martine@danga.com>
 
 """A git-command for integrating reviews on Rietveld."""
 
 import json
@@ skipping 74 matching lines @@
 
 
 def ask_for_data(prompt):
   try:
     return raw_input(prompt)
   except KeyboardInterrupt:
     # Hide the exception.
     sys.exit(1)
 
 
+def add_git_similarity(parser):
+  parser.add_option(
+      '--similarity', metavar='SIM', type='int', action='store', default=None,
+      help='Sets the percentage that a pair of files need to match in order to'
+           ' be considered copies (default 50)')
+
+  old_parser_args = parser.parse_args
+  def Parse(args):
+    options, args = old_parser_args(args)
+
+    branch = Changelist().GetBranch()
+    key = 'branch.%s.git-cl-similarity' % branch
+    if options.similarity is None:
+      if branch:
+        (_, stdout) = RunGitWithCode(['config', '--int', '--get', key])
+        try:
+          options.similarity = int(stdout.strip())
+        except ValueError:
+          pass
+      options.similarity = options.similarity or 50
+    else:
+      if branch:
+        print('Note: Saving similarity of %d%% in git config.'
+              % options.similarity)
+        RunGit(['config', '--int', key, str(options.similarity)])
+
+    options.similarity = max(1, min(options.similarity, 100))
+
+    print('Using %d%% similarity for rename/copy detection. '
+          'Override with --similarity.' % options.similarity)
+
+    return options, args
+  parser.parse_args = Parse
+
+
 def MatchSvnGlob(url, base_url, glob_spec, allow_wildcards):
   """Return the corresponding git ref if |base_url| together with |glob_spec|
   matches the full |url|.
 
   If |allow_wildcards| is true, |glob_spec| can contain wildcards (see below).
   """
   fetch_suburl, as_ref = glob_spec.split(':')
   if allow_wildcards:
     glob_match = re.match('(.+/)?(\*|{[^/]*})(/.+)?', fetch_suburl)
     if glob_match:
@@ skipping 22 matching lines @@
   # Parse specs like "trunk/src:refs/remotes/origin/trunk".
   if fetch_suburl:
     full_url = base_url + '/' + fetch_suburl
   else:
     full_url = base_url
   if full_url == url:
     return as_ref
   return None
 
 
-def print_stats(args):
+def print_stats(similarity, args):
   """Prints statistics about the change to the user."""
   # --no-ext-diff is broken in some versions of Git, so try to work around
   # this by overriding the environment (but there is still a problem if the
   # git config key "diff.external" is used).
   env = os.environ.copy()
   if 'GIT_EXTERNAL_DIFF' in env:
     del env['GIT_EXTERNAL_DIFF']
   return subprocess2.call(
       ['git', 'diff', '--no-ext-diff', '--stat', '--find-copies-harder',
-        '-l100000'] + args, env=env)
+       '-C%s' % similarity, '-l100000'] + args, env=env)
 
 
 class Settings(object):
   def __init__(self):
     self.default_server = None
     self.cc = None
     self.root = None
     self.is_git_svn = None
     self.svn_branch = None
     self.tree_status_url = None
@@ skipping 872 matching lines @@
     if change_desc.reviewers:
       upload_args.extend(['--reviewers', change_desc.reviewers])
     if options.send_mail:
       if not change_desc.reviewers:
         DieWithError("Must specify reviewers to send email.")
       upload_args.append('--send_mail')
     cc = ','.join(filter(None, (cl.GetCCList(), options.cc)))
     if cc:
       upload_args.extend(['--cc', cc])
 
+  upload_args.extend(['--git_similarity', str(options.similarity)])
+
   # Include the upstream repo's URL in the change -- this is useful for
   # projects that have their source spread across multiple repos.
   remote_url = cl.GetGitBaseUrlFromConfig()
   if not remote_url:
     if settings.GetIsGitSvn():
       # URL is dependent on the current directory.
       data = RunGit(['svn', 'info'], cwd=settings.GetRoot())
       if data:
         keys = dict(line.split(': ', 1) for line in data.splitlines()
                     if ': ' in line)
@@ skipping 46 matching lines @@
   parser.add_option('--send-mail', action='store_true',
                     help='send email to reviewer immediately')
   parser.add_option("--emulate_svn_auto_props", action="store_true",
                     dest="emulate_svn_auto_props",
                     help="Emulate Subversion's auto properties feature.")
   parser.add_option('-c', '--use-commit-queue', action='store_true',
                     help='tell the commit queue to commit this patchset')
   if settings.GetIsGerrit():
     parser.add_option('--target_branch', dest='target_branch', default='master',
                       help='target branch to upload')
+  add_git_similarity(parser)
   (options, args) = parser.parse_args(args)
 
   # Print warning if the user used the -m/--message argument.  This will soon
   # change to -t/--title.
   if options.message:
     print >> sys.stderr, (
         '\nWARNING: Use -t or --title to set the title of the patchset.\n'
         'In the near future, -m or --message will send a message instead.\n'
         'See http://goo.gl/JGg0Z for details.\n')
 
@@ skipping 15 matching lines @@
   if not options.bypass_hooks:
     hook_results = cl.RunHook(committing=False, upstream_branch=base_branch,
                               may_prompt=not options.force,
                               verbose=options.verbose,
                               author=None)
     if not hook_results.should_continue():
       return 1
     if not options.reviewers and hook_results.reviewers:
       options.reviewers = hook_results.reviewers
 
-  print_stats(args)
+  print_stats(options.similarity, args)
   if settings.GetIsGerrit():
     return GerritUpload(options, args, cl)
   return RietveldUpload(options, args, cl)
 
 
 def IsSubmoduleMergeCommit(ref):
   # When submodules are added to the repo, we expect there to be a single
   # non-git-svn merge commit at remote HEAD with a signature comment.
   pattern = '^SVN changes up to revision [0-9]*$'
   cmd = ['rev-list', '--merges', '--grep=%s' % pattern, '%s^!' % ref]
@@ skipping 11 matching lines @@
   parser.add_option('--bypass-hooks', action='store_true', dest='bypass_hooks',
                     help='bypass upload presubmit hook')
   parser.add_option('-m', dest='message',
                     help="override review description")
   parser.add_option('-f', action='store_true', dest='force',
                     help="force yes to questions (don't prompt)")
   parser.add_option('-c', dest='contributor',
                     help="external contributor for patch (appended to " +
                          "description and used as author for git). Should be " +
                          "formatted as 'First Last <email@example.com>'")
+  add_git_similarity(parser)
   (options, args) = parser.parse_args(args)
   cl = Changelist()
 
   if not args or cmd == 'push':
     # Default to merging against our best guess of the upstream branch.
     args = [cl.GetUpstreamBranch()]
 
   if options.contributor:
     if not re.match('^.*\s<\S+@\S+>$', options.contributor):
       print "Please provide contibutor as 'First Last <email@example.com>'"
@@ skipping 81 matching lines @@
 
   if cl.GetIssue():
     description += "\n\nReview URL: %s" % cl.GetIssueURL()
 
   if options.contributor:
     description += "\nPatch from %s." % options.contributor
   print 'Description:', repr(description)
 
   branches = [base_branch, cl.GetBranchRef()]
   if not options.force:
-    print_stats(branches)
+    print_stats(options.similarity, branches)
     ask_for_data('About to commit; enter to confirm.')
 
   # We want to squash all this branch's commits into one commit with the proper
   # description. We do this by doing a "reset --soft" to the base branch (which
   # keeps the working copy the same), then dcommitting that.  If origin/master
   # has a submodule merge commit, we'll also need to cherry-pick the squashed
   # commit onto a branch based on the git-svn head.
   MERGE_BRANCH = 'git-cl-commit'
   CHERRY_PICK_BRANCH = 'git-cl-cherry-pick'
   # Delete the branches if they exist.
@@ skipping 28 matching lines @@
       RunGit(['cherry-pick', cherry_pick_commit])
     if cmd == 'push':
       # push the merge branch.
       remote, branch = cl.FetchUpstreamTuple()
       retcode, output = RunGitWithCode(
           ['push', '--porcelain', remote, 'HEAD:%s' % branch])
       logging.debug(output)
     else:
       # dcommit the merge branch.
       retcode, output = RunGitWithCode(['svn', 'dcommit',
+                                        '-C%s' % options.similarity,
                                         '--no-rebase', '--rmdir'])
   finally:
     # And then swap back to the original branch and clean up.
     RunGit(['checkout', '-q', cl.GetBranch()])
     RunGit(['branch', '-D', MERGE_BRANCH])
     if base_has_submodules:
       RunGit(['branch', '-D', CHERRY_PICK_BRANCH])
 
   if cl.GetIssue():
     if cmd == 'dcommit' and 'Committed r' in output:
@@ skipping 383 matching lines @@
              'and retry or visit go/isgaeup.\n%s') % (e.code, str(e)))
 
   # Not a known command. Default to help.
   GenUsage(parser, 'help')
   return CMDhelp(parser, argv)
 
 
 if __name__ == '__main__':
   fix_encoding.fix_encoding()
   sys.exit(main(sys.argv[1:]))