depot_tools: Run git_common_test in Python 3

Bug: 1009809
Change-Id: Idfcbd26de3420798f092c7fa55a6126d7c389a8c
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/tools/depot_tools/+/1834317
Reviewed-by: Anthony Polito <apolito@google.com>
Commit-Queue: Edward Lesmes <ehmaldonado@chromium.org>

git_common.py

@@ -6,12 +6,13 @@
 # Derived from https://gist.github.com/aljungberg/626518
 
 from __future__ import print_function
+from __future__ import unicode_literals
 
 import multiprocessing.pool
 from multiprocessing.pool import IMapIterator
 def wrapper(func):
   def wrap(self, timeout=None):
-    return func(self, timeout=timeout or 1e100)
+    return func(self, timeout=timeout or 1 << 31)
   return wrap
 IMapIterator.next = wrapper(IMapIterator.next)
 IMapIterator.__next__ = IMapIterator.next
@@ -303,11 +304,17 @@ class ProgressPrinter(object):
 def once(function):
   """@Decorates |function| so that it only performs its action once, no matter
   how many times the decorated |function| is called."""
-  def _inner_gen():
-    yield function()
-    while True:
-      yield
-  return _inner_gen().next
+  has_run = [False]
+  def _wrapper(*args, **kwargs):
+    if not has_run[0]:
+      has_run[0] = True
+      function(*args, **kwargs)
+  return _wrapper
+
+
+def unicode_repr(s):
+  result = repr(s)
+  return result[1:] if result.startswith('u') else result
 
 
 ## Git functions
@@ -597,8 +604,8 @@ def mktree(treedict):
     treedict - { name: (mode, type, ref) }
   """
   with tempfile.TemporaryFile() as f:
-    for name, (mode, typ, ref) in treedict.iteritems():
-      f.write('%s %s %s\t%s\0' % (mode, typ, ref, name))
+    for name, (mode, typ, ref) in treedict.items():
+      f.write(('%s %s %s\t%s\0' % (mode, typ, ref, name)).encode())
     f.seek(0)
     return run('mktree', '-z', stdin=f)
 
@@ -612,7 +619,7 @@ def parse_commitrefs(*commitrefs):
     * 'cool_branch~2'
   """
   try:
-    return map(binascii.unhexlify, hash_multi(*commitrefs))
+    return [binascii.unhexlify(h) for h in hash_multi(*commitrefs)]
   except subprocess2.CalledProcessError:
     raise BadCommitRefException(commitrefs)
 
@@ -751,6 +758,7 @@ def run_with_stderr(*cmd, **kwargs):
   kwargs.setdefault('shell', False)
   autostrip = kwargs.pop('autostrip', True)
   indata = kwargs.pop('indata', None)
+  decode = kwargs.pop('decode', True)
 
   cmd = (GIT_EXE, '-c', 'color.ui=never') + cmd
   proc = subprocess2.Popen(cmd, **kwargs)
@@ -760,8 +768,12 @@ def run_with_stderr(*cmd, **kwargs):
     raise subprocess2.CalledProcessError(retcode, cmd, os.getcwd(), ret, err)
 
   if autostrip:
-    ret = (ret or '').strip()
-    err = (err or '').strip()
+    ret = (ret or b'').strip()
+    err = (err or b'').strip()
+
+  if decode:
+    ret = ret.decode('utf-8', 'replace')
+    err = err.decode('utf-8', 'replace')
 
   return ret, err
 
@@ -810,9 +822,9 @@ def status():
   def tokenizer(stream):
     acc = BytesIO()
     c = None
-    while c != '':
+    while c != b'':
       c = stream.read(1)
-      if c in (None, '', '\0'):
+      if c in (None, b'', b'\0'):
         if len(acc.getvalue()):
           yield acc.getvalue()
           acc = BytesIO()
@@ -821,12 +833,14 @@ def status():
 
   def parser(tokens):
     while True:
-      # Raises StopIteration if it runs out of tokens.
-      status_dest = next(tokens)
+      try:
+        status_dest = next(tokens).decode('utf-8')
+      except StopIteration:
+        return
       stat, dest = status_dest[:2], status_dest[3:]
       lstat, rstat = stat
       if lstat == 'R':
-        src = next(tokens)
+        src = next(tokens).decode('utf-8')
       else:
         src = dest
       yield (dest, stat_entry(lstat, rstat, src))
@@ -848,7 +862,7 @@ def squash_current_branch(header=None, merge_base=None):
     # nothing to commit at this point.
     print('Nothing to commit; squashed branch is empty')
     return False
-  run('commit', '--no-verify', '-a', '-F', '-', indata=log_msg)
+  run('commit', '--no-verify', '-a', '-F', '-', indata=log_msg.encode())
   return True
 
 
@@ -858,7 +872,8 @@ def tags(*args):
 
 def thaw():
   took_action = False
-  for sha in (s.strip() for s in run_stream('rev-list', 'HEAD').xreadlines()):
+  for sha in run_stream('rev-list', 'HEAD').readlines():
+    sha = sha.strip().decode('utf-8')
     msg = run('show', '--format=%f%b', '-s', 'HEAD')
     match = FREEZE_MATCHER.match(msg)
     if not match:
@@ -899,7 +914,7 @@ def topo_iter(branch_tree, top_down=True):
   # TODO(iannucci): There is probably a more efficient way to do these.
   if top_down:
     while branch_tree:
-      this_pass = [(b, p) for b, p in branch_tree.iteritems()
+      this_pass = [(b, p) for b, p in branch_tree.items()
                    if p not in branch_tree]
       assert this_pass, "Branch tree has cycles: %r" % branch_tree
       for branch, parent in sorted(this_pass):
@@ -907,11 +922,11 @@ def topo_iter(branch_tree, top_down=True):
         del branch_tree[branch]
   else:
     parent_to_branches = collections.defaultdict(set)
-    for branch, parent in branch_tree.iteritems():
+    for branch, parent in branch_tree.items():
       parent_to_branches[parent].add(branch)
 
     while branch_tree:
-      this_pass = [(b, p) for b, p in branch_tree.iteritems()
+      this_pass = [(b, p) for b, p in branch_tree.items()
                    if not parent_to_branches[b]]
       assert this_pass, "Branch tree has cycles: %r" % branch_tree
       for branch, parent in sorted(this_pass):
@@ -1014,7 +1029,9 @@ def get_branches_info(include_tracking_status):
     if info.upstream not in info_map and info.upstream not in missing_upstreams:
       missing_upstreams[info.upstream] = None
 
-  return dict(info_map.items() + missing_upstreams.items())
+  result = info_map.copy()
+  result.update(missing_upstreams)
+  return result
 
 
 def make_workdir_common(repository, new_workdir, files_to_symlink,

git_hyper_blame.py

@@ -7,6 +7,7 @@
 """
 
 from __future__ import print_function
+from __future__ import unicode_literals
 
 import argparse
 import collections
@@ -94,11 +95,6 @@ def parse_blame(blameoutput):
     yield BlameLine(commit, context, lineno_then, lineno_now, False)
 
 
-def num_codepoints(s):
-  """Gets the length of a UTF-8 byte string, in Unicode codepoints."""
-  return len(s.decode('utf-8', errors='replace'))
-
-
 def print_table(table, colsep=' ', rowsep='\n', align=None, out=sys.stdout):
   """Print a 2D rectangular array, aligning columns with spaces.
 
@@ -112,10 +108,9 @@ def print_table(table, colsep=' ', rowsep='\n', align=None, out=sys.stdout):
   colwidths = None
   for row in table:
     if colwidths is None:
-      colwidths = [num_codepoints(x) for x in row]
+      colwidths = [len(x) for x in row]
     else:
-      colwidths = [max(colwidths[i], num_codepoints(x))
-                   for i, x in enumerate(row)]
+      colwidths = [max(colwidths[i], len(x)) for i, x in enumerate(row)]
 
   if align is None:  # pragma: no cover
     align = 'l' * len(colwidths)
@@ -123,7 +118,7 @@ def print_table(table, colsep=' ', rowsep='\n', align=None, out=sys.stdout):
   for row in table:
     cells = []
     for i, cell in enumerate(row):
-      padding = ' ' * (colwidths[i] - num_codepoints(cell))
+      padding = ' ' * (colwidths[i] - len(cell))
       if align[i] == 'r':
         cell = padding + cell
       elif i < len(row) - 1:
@@ -279,7 +274,7 @@ def hyper_blame(ignored, filename, revision='HEAD', out=sys.stdout,
   try:
     parsed = cache_blame_from(filename, git_common.hash_one(revision))
   except subprocess2.CalledProcessError as e:
-    err.write(e.stderr)
+    err.write(e.stderr.decode())
     return e.returncode
 
   new_parsed = []
@@ -361,7 +356,7 @@ def main(args, stdout=sys.stdout, stderr=sys.stderr):
   try:
     repo_root = git_common.repo_root()
   except subprocess2.CalledProcessError as e:
-    stderr.write(e.stderr)
+    stderr.write(e.stderr.decode())
     return e.returncode
 
   # Make filename relative to the repository root, and cd to the root dir (so

git_number.py

@@ -75,7 +75,8 @@ def get_number_tree(prefix_bytes):
   ref = '%s:%s' % (REF, pathlify(prefix_bytes))
 
   try:
-    raw = buffer(git.run('cat-file', 'blob', ref, autostrip=False))
+    raw = buffer(
+        git.run('cat-file', 'blob', ref, autostrip=False, decode=False))
     return dict(struct.unpack_from(CHUNK_FMT, raw, i * CHUNK_SIZE)
                 for i in xrange(len(raw) / CHUNK_SIZE))
   except subprocess2.CalledProcessError:

git_rebase_update.py

@@ -320,8 +320,8 @@ def main(args=None):
     else:
       root_branch = git.root()
       if return_branch != 'HEAD':
-        print("%r was merged with its parent, checking out %r instead." %
-              (return_branch, root_branch))
+        print("%s was merged with its parent, checking out %s instead." %
+              (git.unicode_repr(return_branch), git.unicode_repr(root_branch)))
       git.run('checkout', root_branch)
 
     # return_workdir may also not be there any more.

testing_support/git_test_utils.py

@@ -15,13 +15,18 @@ import tempfile
 import unittest
 
 
+if sys.version_info.major == 3:
+  # pylint: disable=redefined-builtin
+  basestring = (str,)
+
+
 def git_hash_data(data, typ='blob'):
   """Calculate the git-style SHA1 for some data.
 
   Only supports 'blob' type data at the moment.
   """
   assert typ == 'blob', 'Only support blobs for now'
-  return hashlib.sha1('blob %s\0%s' % (len(data), data)).hexdigest()
+  return hashlib.sha1(b'blob %d\0%s' % (len(data), data)).hexdigest()
 
 
 class OrderedSet(collections.MutableSet):
@@ -121,7 +126,7 @@ class GitRepoSchema(object):
   This is the repo
 
      A - B -  C - D
-           \ E /
+           \\ E /
 
   Whitespace doesn't matter. Each line is a declaration of which commits come
   before which other commits.
@@ -166,15 +171,15 @@ class GitRepoSchema(object):
     is_root = True
     par_map = copy.deepcopy(self.par_map)
     while par_map:
-      empty_keys = set(k for k, v in par_map.iteritems() if not v)
+      empty_keys = set(k for k, v in par_map.items() if not v)
       assert empty_keys, 'Cycle detected! %s' % par_map
 
       for k in sorted(empty_keys):
         yield self.COMMIT(k, self.par_map[k],
-                          not any(k in v for v in self.par_map.itervalues()),
+                          not any(k in v for v in self.par_map.values()),
                           is_root)
         del par_map[k]
-      for v in par_map.itervalues():
+      for v in par_map.values():
         v.difference_update(empty_keys)
       is_root = False
 
@@ -321,7 +326,7 @@ class GitRepo(object):
 
     env = self.get_git_commit_env(commit_data)
 
-    for fname, file_data in commit_data.iteritems():
+    for fname, file_data in commit_data.items():
       # If it isn't a string, it's one of the special keys.
       if not isinstance(fname, basestring):
         continue
@@ -367,7 +372,9 @@ class GitRepo(object):
           self._date += datetime.timedelta(days=1)
         else:
           val = getattr(self, 'DEFAULT_%s' % singleton)
-        env['GIT_%s' % singleton] = str(val)
+        if not isinstance(val, str) and not isinstance(val, bytes):
+          val = str(val)
+        env['GIT_%s' % singleton] = val
     return env
 
   def git(self, *args, **kwargs):
@@ -377,6 +384,7 @@ class GitRepo(object):
       with open(os.devnull, 'wb') as devnull:
         output = subprocess.check_output(
           ('git',) + args, cwd=self.repo_path, stderr=devnull, **kwargs)
+        output = output.decode('utf-8')
       return self.COMMAND_OUTPUT(0, output)
     except subprocess.CalledProcessError as e:
       return self.COMMAND_OUTPUT(e.returncode, e.output)
@@ -418,8 +426,8 @@ class GitRepo(object):
     stdout = sys.stdout
     stderr = sys.stderr
     try:
-      # "multiple statements on a line" pylint: disable=multiple-statements
-      with tempfile.TemporaryFile() as out, tempfile.TemporaryFile() as err:
+      with tempfile.TemporaryFile('w+') as out:
+        with tempfile.TemporaryFile('w+') as err:
           sys.stdout = out
           sys.stderr = err
           try:

tests/git_common_test.py

@@ -1,4 +1,4 @@
-#!/usr/bin/env python
+#!/usr/bin/env vpython3
 # Copyright 2013 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.
@@ -85,7 +85,7 @@ class Support(GitCommonTestBase):
     add_to_list()
     add_to_list()
 
-    self.assertEquals(testlist, ['dog'])
+    self.assertEqual(testlist, ['dog'])
 
 
 def slow_square(i):
@@ -102,7 +102,7 @@ class ScopedPoolTest(GitCommonTestBase):
   def testThreads(self):
     result = []
     with self.gc.ScopedPool(kind='threads') as pool:
-      result = list(pool.imap(slow_square, xrange(10)))
+      result = list(pool.imap(slow_square, range(10)))
     self.assertEqual([0, 1, 4, 9, 16, 25, 36, 49, 64, 81], result)
 
   def testThreadsCtrlC(self):
@@ -110,7 +110,7 @@ class ScopedPoolTest(GitCommonTestBase):
     with self.assertRaises(KeyboardInterrupt):
       with self.gc.ScopedPool(kind='threads') as pool:
         # Make sure this pool is interrupted in mid-swing
-        for i in pool.imap(slow_square, xrange(20)):
+        for i in pool.imap(slow_square, range(20)):
           if i > 32:
             os.kill(os.getpid(), self.CTRL_C)
           result.append(i)
@@ -119,7 +119,7 @@ class ScopedPoolTest(GitCommonTestBase):
   def testProcs(self):
     result = []
     with self.gc.ScopedPool() as pool:
-      result = list(pool.imap(slow_square, xrange(10)))
+      result = list(pool.imap(slow_square, range(10)))
     self.assertEqual([0, 1, 4, 9, 16, 25, 36, 49, 64, 81], result)
 
   def testProcsCtrlC(self):
@@ -127,7 +127,7 @@ class ScopedPoolTest(GitCommonTestBase):
     with self.assertRaises(KeyboardInterrupt):
       with self.gc.ScopedPool() as pool:
         # Make sure this pool is interrupted in mid-swing
-        for i in pool.imap(slow_square, xrange(20)):
+        for i in pool.imap(slow_square, range(20)):
           if i > 32:
             os.kill(os.getpid(), self.CTRL_C)
           result.append(i)
@@ -146,7 +146,6 @@ class ProgressPrinterTest(GitCommonTestBase):
     def flush(self):
       self.count += 1
 
-  @unittest.expectedFailure
   def testBasic(self):
     """This test is probably racy, but I don't have a better alternative."""
     fmt = '%(count)d/10'
@@ -154,12 +153,12 @@ class ProgressPrinterTest(GitCommonTestBase):
 
     pp = self.gc.ProgressPrinter(fmt, enabled=True, fout=stream, period=0.01)
     with pp as inc:
-      for _ in xrange(10):
+      for _ in range(10):
         time.sleep(0.02)
         inc()
 
     filtered = {x.strip() for x in stream.data}
-    rslt = {fmt % {'count': i} for i in xrange(11)}
+    rslt = {fmt % {'count': i} for i in range(11)}
     self.assertSetEqual(filtered, rslt)
     self.assertGreaterEqual(stream.count, 10)
 
@@ -172,24 +171,24 @@ class GitReadOnlyFunctionsTest(git_test_utils.GitRepoReadOnlyTestBase,
   """
 
   COMMIT_A = {
-    'some/files/file1': {'data': 'file1'},
-    'some/files/file2': {'data': 'file2'},
-    'some/files/file3': {'data': 'file3'},
-    'some/other/file':  {'data': 'otherfile'},
+    'some/files/file1': {'data': b'file1'},
+    'some/files/file2': {'data': b'file2'},
+    'some/files/file3': {'data': b'file3'},
+    'some/other/file':  {'data': b'otherfile'},
   }
 
   COMMIT_C = {
     'some/files/file2': {
       'mode': 0o755,
-      'data': 'file2 - vanilla\n'},
+      'data': b'file2 - vanilla\n'},
   }
 
   COMMIT_E = {
-    'some/files/file2': {'data': 'file2 - merged\n'},
+    'some/files/file2': {'data': b'file2 - merged\n'},
   }
 
   COMMIT_D = {
-    'some/files/file2': {'data': 'file2 - vanilla\nfile2 - merged\n'},
+    'some/files/file2': {'data': b'file2 - vanilla\nfile2 - merged\n'},
   }
 
   def testHashes(self):
@@ -209,7 +208,7 @@ class GitReadOnlyFunctionsTest(git_test_utils.GitRepoReadOnlyTestBase,
       self.repo['E'],
       self.repo['C'],
     ], ret)
-    self.assertEquals(
+    self.assertEqual(
       self.repo.run(self.gc.hash_one, 'branch_D'),
       self.repo['D']
     )
@@ -217,23 +216,23 @@ class GitReadOnlyFunctionsTest(git_test_utils.GitRepoReadOnlyTestBase,
         self.repo.run(self.gc.hash_one, 'branch_D', short=True)))
 
   def testStream(self):
-    items = set(self.repo.commit_map.itervalues())
+    items = set(self.repo.commit_map.values())
 
     def testfn():
-      for line in self.gc.run_stream('log', '--format=%H').xreadlines():
-        line = line.strip()
+      for line in self.gc.run_stream('log', '--format=%H').readlines():
+        line = line.strip().decode('utf-8')
         self.assertIn(line, items)
         items.remove(line)
 
     self.repo.run(testfn)
 
   def testStreamWithRetcode(self):
-    items = set(self.repo.commit_map.itervalues())
+    items = set(self.repo.commit_map.values())
 
     def testfn():
       with self.gc.run_stream_with_retcode('log', '--format=%H') as stdout:
-        for line in stdout.xreadlines():
-          line = line.strip()
+        for line in stdout.readlines():
+          line = line.strip().decode('utf-8')
           self.assertIn(line, items)
           items.remove(line)
 
@@ -326,13 +325,14 @@ class GitReadOnlyFunctionsTest(git_test_utils.GitRepoReadOnlyTestBase,
         'tag_C^{}',
       ]
     )
-    self.assertEqual(ret, map(binascii.unhexlify, [
+    hashes = [
         self.repo['D'],
         self.repo['A'],
         self.repo['B'],
         self.repo['E'],
         self.repo['C'],
-    ]))
+    ]
+    self.assertEqual(ret, [binascii.unhexlify(h) for h in hashes])
 
     with self.assertRaisesRegexp(Exception, r"one of \('master', 'bananas'\)"):
       self.repo.run(self.gc.parse_commitrefs, 'master', 'bananas')
@@ -356,20 +356,20 @@ class GitReadOnlyFunctionsTest(git_test_utils.GitRepoReadOnlyTestBase,
     file1 = self.COMMIT_A['some/files/file1']['data']
     file2 = self.COMMIT_D['some/files/file2']['data']
     file3 = self.COMMIT_A['some/files/file3']['data']
-    self.assertEquals(
+    self.assertEqual(
         tree['file1'],
         ('100644', 'blob', git_test_utils.git_hash_data(file1)))
-    self.assertEquals(
+    self.assertEqual(
         tree['file2'],
         ('100755', 'blob', git_test_utils.git_hash_data(file2)))
-    self.assertEquals(
+    self.assertEqual(
         tree['file3'],
         ('100644', 'blob', git_test_utils.git_hash_data(file3)))
 
     tree = self.repo.run(self.gc.tree, 'master:some')
-    self.assertEquals(len(tree), 2)
+    self.assertEqual(len(tree), 2)
     # Don't check the tree hash because we're lazy :)
-    self.assertEquals(tree['files'][:2], ('040000', 'tree'))
+    self.assertEqual(tree['files'][:2], ('040000', 'tree'))
 
     tree = self.repo.run(self.gc.tree, 'master:wat')
     self.assertEqual(tree, None)
@@ -380,16 +380,16 @@ class GitReadOnlyFunctionsTest(git_test_utils.GitRepoReadOnlyTestBase,
     file2 = self.COMMIT_D['some/files/file2']['data']
     file3 = self.COMMIT_A['some/files/file3']['data']
     other = self.COMMIT_A['some/other/file']['data']
-    self.assertEquals(
+    self.assertEqual(
         tree['files/file1'],
         ('100644', 'blob', git_test_utils.git_hash_data(file1)))
-    self.assertEquals(
+    self.assertEqual(
         tree['files/file2'],
         ('100755', 'blob', git_test_utils.git_hash_data(file2)))
-    self.assertEquals(
+    self.assertEqual(
         tree['files/file3'],
         ('100644', 'blob', git_test_utils.git_hash_data(file3)))
-    self.assertEquals(
+    self.assertEqual(
         tree['other/file'],
         ('100644', 'blob', git_test_utils.git_hash_data(other)))
 
@@ -399,7 +399,7 @@ class GitMutableFunctionsTest(git_test_utils.GitRepoReadWriteTestBase,
   REPO_SCHEMA = ''
 
   def _intern_data(self, data):
-    with tempfile.TemporaryFile() as f:
+    with tempfile.TemporaryFile('w') as f:
       f.write(data)
       f.seek(0)
       return self.repo.run(self.gc.intern_f, f)
@@ -407,8 +407,8 @@ class GitMutableFunctionsTest(git_test_utils.GitRepoReadWriteTestBase,
   def testInternF(self):
     data = 'CoolBobcatsBro'
     data_hash = self._intern_data(data)
-    self.assertEquals(git_test_utils.git_hash_data(data), data_hash)
-    self.assertEquals(data, self.repo.git('cat-file', 'blob', data_hash).stdout)
+    self.assertEqual(git_test_utils.git_hash_data(data.encode()), data_hash)
+    self.assertEqual(data, self.repo.git('cat-file', 'blob', data_hash).stdout)
 
   def testMkTree(self):
     tree = {}
@@ -416,46 +416,46 @@ class GitMutableFunctionsTest(git_test_utils.GitRepoReadWriteTestBase,
       name = 'file%d' % i
       tree[name] = ('100644', 'blob', self._intern_data(name))
     tree_hash = self.repo.run(self.gc.mktree, tree)
-    self.assertEquals('37b61866d6e061c4ba478e7eb525be7b5752737d', tree_hash)
+    self.assertEqual('37b61866d6e061c4ba478e7eb525be7b5752737d', tree_hash)
 
   def testConfig(self):
     self.repo.git('config', '--add', 'happy.derpies', 'food')
-    self.assertEquals(self.repo.run(self.gc.get_config_list, 'happy.derpies'),
+    self.assertEqual(self.repo.run(self.gc.get_config_list, 'happy.derpies'),
                      ['food'])
-    self.assertEquals(self.repo.run(self.gc.get_config_list, 'sad.derpies'), [])
+    self.assertEqual(self.repo.run(self.gc.get_config_list, 'sad.derpies'), [])
 
     self.repo.git('config', '--add', 'happy.derpies', 'cat')
-    self.assertEquals(self.repo.run(self.gc.get_config_list, 'happy.derpies'),
+    self.assertEqual(self.repo.run(self.gc.get_config_list, 'happy.derpies'),
                      ['food', 'cat'])
 
-    self.assertEquals('cat', self.repo.run(self.gc.get_config, 'dude.bob',
-                                           'cat'))
+    self.assertEqual(
+        'cat', self.repo.run(self.gc.get_config, 'dude.bob', 'cat'))
 
     self.repo.run(self.gc.set_config, 'dude.bob', 'dog')
 
-    self.assertEquals('dog', self.repo.run(self.gc.get_config, 'dude.bob',
-                                           'cat'))
+    self.assertEqual(
+        'dog', self.repo.run(self.gc.get_config, 'dude.bob', 'cat'))
 
     self.repo.run(self.gc.del_config, 'dude.bob')
 
     # This should work without raising an exception
     self.repo.run(self.gc.del_config, 'dude.bob')
 
-    self.assertEquals('cat', self.repo.run(self.gc.get_config, 'dude.bob',
-                                           'cat'))
+    self.assertEqual(
+        'cat', self.repo.run(self.gc.get_config, 'dude.bob', 'cat'))
 
-    self.assertEquals('origin/master', self.repo.run(self.gc.root))
+    self.assertEqual('origin/master', self.repo.run(self.gc.root))
 
     self.repo.git('config', 'depot-tools.upstream', 'catfood')
 
-    self.assertEquals('catfood', self.repo.run(self.gc.root))
+    self.assertEqual('catfood', self.repo.run(self.gc.root))
 
   def testUpstream(self):
     self.repo.git('commit', '--allow-empty', '-am', 'foooooo')
-    self.assertEquals(self.repo.run(self.gc.upstream, 'bobly'), None)
-    self.assertEquals(self.repo.run(self.gc.upstream, 'master'), None)
+    self.assertEqual(self.repo.run(self.gc.upstream, 'bobly'), None)
+    self.assertEqual(self.repo.run(self.gc.upstream, 'master'), None)
     self.repo.git('checkout', '-tb', 'happybranch', 'master')
-    self.assertEquals(self.repo.run(self.gc.upstream, 'happybranch'),
+    self.assertEqual(self.repo.run(self.gc.upstream, 'happybranch'),
                      'master')
 
   def testNormalizedVersion(self):
@@ -505,7 +505,7 @@ class GitMutableFunctionsTest(git_test_utils.GitRepoReadWriteTestBase,
         ),
         'to_delete': None
     }
-    self.assertEquals(expected, actual)
+    self.assertEqual(expected, actual)
 
 
 class GitMutableStructuredTest(git_test_utils.GitRepoReadWriteTestBase,
@@ -520,12 +520,12 @@ class GitMutableStructuredTest(git_test_utils.GitRepoReadWriteTestBase,
   CAT DOG
   """
 
-  COMMIT_B = {'file': {'data': 'B'}}
-  COMMIT_H = {'file': {'data': 'H'}}
-  COMMIT_I = {'file': {'data': 'I'}}
-  COMMIT_J = {'file': {'data': 'J'}}
-  COMMIT_K = {'file': {'data': 'K'}}
-  COMMIT_L = {'file': {'data': 'L'}}
+  COMMIT_B = {'file': {'data': b'B'}}
+  COMMIT_H = {'file': {'data': b'H'}}
+  COMMIT_I = {'file': {'data': b'I'}}
+  COMMIT_J = {'file': {'data': b'J'}}
+  COMMIT_K = {'file': {'data': b'K'}}
+  COMMIT_L = {'file': {'data': b'L'}}
 
   def setUp(self):
     super(GitMutableStructuredTest, self).setUp()
@@ -536,7 +536,7 @@ class GitMutableStructuredTest(git_test_utils.GitRepoReadWriteTestBase,
     self.repo.git('branch', '--set-upstream-to', 'root_X', 'root_A')
 
   def testTooManyBranches(self):
-    for i in xrange(30):
+    for i in range(30):
       self.repo.git('branch', 'a'*i)
 
     _, rslt = self.repo.capture_stdio(list, self.gc.branches())
@@ -597,8 +597,8 @@ class GitMutableStructuredTest(git_test_utils.GitRepoReadWriteTestBase,
     self.repo.run(self.gc.remove_merge_base, 'branch_K')
     self.repo.run(self.gc.remove_merge_base, 'branch_L')
 
-    self.assertEqual(None, self.repo.run(self.gc.get_config,
-                                         'branch.branch_K.base'))
+    self.assertEqual(
+        None, self.repo.run(self.gc.get_config, 'branch.branch_K.base'))
 
     self.assertEqual({}, self.repo.run(self.gc.branch_config_map, 'base'))
 
@@ -658,20 +658,20 @@ class GitMutableStructuredTest(git_test_utils.GitRepoReadWriteTestBase,
     self.repo.capture_stdio(
       lambda: retval.append(self.repo.run(self.gc.is_dirty_git_tree, 'foo')))
 
-    self.assertEquals(False, retval[0])
+    self.assertEqual(False, retval[0])
     self.repo.open('test.file', 'w').write('test data')
     self.repo.git('add', 'test.file')
 
     retval = []
     self.repo.capture_stdio(
       lambda: retval.append(self.repo.run(self.gc.is_dirty_git_tree, 'foo')))
-    self.assertEquals(True, retval[0])
+    self.assertEqual(True, retval[0])
 
   def testSquashBranch(self):
     self.repo.git('checkout', 'branch_K')
 
-    self.assertEquals(True, self.repo.run(self.gc.squash_current_branch,
-                                          'cool message'))
+    self.assertEqual(
+        True, self.repo.run(self.gc.squash_current_branch, 'cool message'))
 
     lines = ['cool message', '']
     for l in 'HIJK':
@@ -679,18 +679,18 @@ class GitMutableStructuredTest(git_test_utils.GitRepoReadWriteTestBase,
     lines.pop()
     msg = '\n'.join(lines)
 
-    self.assertEquals(self.repo.run(self.gc.run, 'log', '-n1', '--format=%B'),
+    self.assertEqual(self.repo.run(self.gc.run, 'log', '-n1', '--format=%B'),
                      msg)
 
-    self.assertEquals(
+    self.assertEqual(
       self.repo.git('cat-file', 'blob', 'branch_K:file').stdout,
       'K'
     )
 
   def testSquashBranchDefaultMessage(self):
     self.repo.git('checkout', 'branch_K')
-    self.assertEquals(True, self.repo.run(self.gc.squash_current_branch))
-    self.assertEquals(self.repo.run(self.gc.run, 'log', '-n1', '--format=%s'),
+    self.assertEqual(True, self.repo.run(self.gc.squash_current_branch))
+    self.assertEqual(self.repo.run(self.gc.run, 'log', '-n1', '--format=%s'),
                       'git squash commit for branch_K.')
 
   def testSquashBranchEmpty(self):
@@ -699,7 +699,7 @@ class GitMutableStructuredTest(git_test_utils.GitRepoReadWriteTestBase,
     self.repo.git('commit', '-m', 'revert all changes no branch')
     # Should return False since the quash would result in an empty commit
     stdout = self.repo.capture_stdio(self.gc.squash_current_branch)[0]
-    self.assertEquals(stdout, 'Nothing to commit; squashed branch is empty\n')
+    self.assertEqual(stdout, 'Nothing to commit; squashed branch is empty\n')
 
   def testRebase(self):
     self.assertSchema("""
@@ -782,24 +782,24 @@ class GitFreezeThaw(git_test_utils.GitRepoReadWriteTestBase):
   """
 
   COMMIT_A = {
-    'some/files/file1': {'data': 'file1'},
-    'some/files/file2': {'data': 'file2'},
-    'some/files/file3': {'data': 'file3'},
-    'some/other/file':  {'data': 'otherfile'},
+    'some/files/file1': {'data': b'file1'},
+    'some/files/file2': {'data': b'file2'},
+    'some/files/file3': {'data': b'file3'},
+    'some/other/file':  {'data': b'otherfile'},
   }
 
   COMMIT_C = {
     'some/files/file2': {
       'mode': 0o755,
-      'data': 'file2 - vanilla'},
+      'data': b'file2 - vanilla'},
   }
 
   COMMIT_E = {
-    'some/files/file2': {'data': 'file2 - merged'},
+    'some/files/file2': {'data': b'file2 - merged'},
   }
 
   COMMIT_D = {
-    'some/files/file2': {'data': 'file2 - vanilla\nfile2 - merged'},
+    'some/files/file2': {'data': b'file2 - vanilla\nfile2 - merged'},
   }
 
   def testNothing(self):
@@ -825,23 +825,23 @@ class GitFreezeThaw(git_test_utils.GitRepoReadWriteTestBase):
       self.repo.git('add', 'some/files/file5')
 
       # Freeze group 1
-      self.assertEquals(self.repo.git('status', '--porcelain').stdout, STATUS_1)
+      self.assertEqual(self.repo.git('status', '--porcelain').stdout, STATUS_1)
       self.assertIsNone(self.gc.freeze())
-      self.assertEquals(self.repo.git('status', '--porcelain').stdout, '')
+      self.assertEqual(self.repo.git('status', '--porcelain').stdout, '')
 
       # Freeze group 2
       with open('some/files/file2', 'a') as f2:
         print('new! appended line!', file=f2)
-      self.assertEquals(self.repo.git('status', '--porcelain').stdout,
+      self.assertEqual(self.repo.git('status', '--porcelain').stdout,
                        ' M some/files/file2\n')
       self.assertIsNone(self.gc.freeze())
-      self.assertEquals(self.repo.git('status', '--porcelain').stdout, '')
+      self.assertEqual(self.repo.git('status', '--porcelain').stdout, '')
 
       # Thaw it out!
       self.assertIsNone(self.gc.thaw())
       self.assertIsNotNone(self.gc.thaw())  # One thaw should thaw everything
 
-      self.assertEquals(self.repo.git('status', '--porcelain').stdout, STATUS_1)
+      self.assertEqual(self.repo.git('status', '--porcelain').stdout, STATUS_1)
 
     self.repo.run(inner)
 
@@ -850,7 +850,7 @@ class GitFreezeThaw(git_test_utils.GitRepoReadWriteTestBase):
       self.repo.git('config', 'depot-tools.freeze-size-limit', '1')
       with open('bigfile', 'w') as f:
         chunk = 'NERDFACE' * 1024
-        for _ in xrange(128 * 2 + 1):  # Just over 2 mb
+        for _ in range(128 * 2 + 1):  # Just over 2 mb
           f.write(chunk)
       _, err = self.repo.capture_stdio(self.gc.freeze)
       self.assertIn('too much untracked+unignored', err)
@@ -860,10 +860,10 @@ class GitFreezeThaw(git_test_utils.GitRepoReadWriteTestBase):
   def testTooBigMultipleFiles(self):
     def inner():
       self.repo.git('config', 'depot-tools.freeze-size-limit', '1')
-      for i in xrange(3):
+      for i in range(3):
         with open('file%d' % i, 'w') as f:
           chunk = 'NERDFACE' * 1024
-          for _ in xrange(50):  # About 400k
+          for _ in range(50):  # About 400k
             f.write(chunk)
       _, err = self.repo.capture_stdio(self.gc.freeze)
       self.assertIn('too much untracked+unignored', err)
@@ -896,7 +896,7 @@ class GitFreezeThaw(git_test_utils.GitRepoReadWriteTestBase):
       self.repo.git('checkout', '-b', 'unreadable_file_branch')
       with open('bad_file', 'w') as f:
         f.write('some text')
-      os.chmod('bad_file', 0111)
+      os.chmod('bad_file', 0o0111)
       ret = self.repo.run(self.gc.freeze)
       self.assertIn('Failed to index some unindexed files.', ret)
 
@@ -935,11 +935,11 @@ class GitTestUtilsTest(git_test_utils.GitRepoReadOnlyTestBase):
   """
 
   COMMIT_A = {
-    'file1': {'data': 'file1'},
+    'file1': {'data': b'file1'},
   }
 
   COMMIT_B = {
-    'file1': {'data': 'file1 changed'},
+    'file1': {'data': b'file1 changed'},
   }
 
   # Test special keys (custom commit data).
@@ -952,7 +952,7 @@ class GitTestUtilsTest(git_test_utils.GitRepoReadOnlyTestBase):
     GitRepo.COMMITTER_EMAIL: 'committer@example.com',
     GitRepo.COMMITTER_DATE: datetime.datetime(1990, 4, 5, 6, 7, 8,
                                               tzinfo=git_test_utils.UTC),
-    'file1': {'data': 'file1 changed again'},
+    'file1': {'data': b'file1 changed again'},
   }
 
   def testAutomaticCommitDates(self):
@@ -960,20 +960,20 @@ class GitTestUtilsTest(git_test_utils.GitRepoReadOnlyTestBase):
     # must be in UTC (otherwise the tests are system-dependent, and if your
     # local timezone is positive, timestamps will be <0 which causes bizarre
     # behaviour in Git; http://crbug.com/581895).
-    self.assertEquals('Author McAuthorly 1970-01-01 00:00:00 +0000',
+    self.assertEqual('Author McAuthorly 1970-01-01 00:00:00 +0000',
                      self.repo.show_commit('A', format_string='%an %ai'))
-    self.assertEquals('Charles Committish 1970-01-02 00:00:00 +0000',
+    self.assertEqual('Charles Committish 1970-01-02 00:00:00 +0000',
                      self.repo.show_commit('A', format_string='%cn %ci'))
-    self.assertEquals('Author McAuthorly 1970-01-03 00:00:00 +0000',
+    self.assertEqual('Author McAuthorly 1970-01-03 00:00:00 +0000',
                      self.repo.show_commit('B', format_string='%an %ai'))
-    self.assertEquals('Charles Committish 1970-01-04 00:00:00 +0000',
+    self.assertEqual('Charles Committish 1970-01-04 00:00:00 +0000',
                      self.repo.show_commit('B', format_string='%cn %ci'))
 
   def testCustomCommitData(self):
-    self.assertEquals('Custom Author author@example.com '
+    self.assertEqual('Custom Author author@example.com '
                      '1980-09-08 07:06:05 +0000',
                      self.repo.show_commit('C', format_string='%an %ae %ai'))
-    self.assertEquals('Custom Committer committer@example.com '
+    self.assertEqual('Custom Committer committer@example.com '
                      '1990-04-05 06:07:08 +0000',
                      self.repo.show_commit('C', format_string='%cn %ce %ci'))
 

tests/git_hyper_blame_test.py

@@ -4,15 +4,21 @@
 # found in the LICENSE file.
 """Tests for git_dates."""
 
+from __future__ import unicode_literals
+
 import datetime
 import os
 import re
 import shutil
-import StringIO
 import sys
 import tempfile
 import unittest
 
+if sys.version_info.major == 2:
+  from StringIO import StringIO
+else:
+  from io import StringIO
+
 DEPOT_TOOLS_ROOT = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
 sys.path.insert(0, DEPOT_TOOLS_ROOT)
 
@@ -32,8 +38,8 @@ class GitHyperBlameTestBase(git_test_utils.GitRepoReadOnlyTestBase):
     cls.git_hyper_blame = git_hyper_blame
 
   def run_hyperblame(self, ignored, filename, revision):
-    stdout = StringIO.StringIO()
-    stderr = StringIO.StringIO()
+    stdout = StringIO()
+    stderr = StringIO()
     ignored = [self.repo[c] for c in ignored]
     retval = self.repo.run(self.git_hyper_blame.hyper_blame, ignored, filename,
                            revision=revision, out=stdout, err=stderr)
@@ -61,22 +67,22 @@ class GitHyperBlameMainTest(GitHyperBlameTestBase):
   REPO_SCHEMA = "A B C D"
 
   COMMIT_A = {
-    'some/files/file': {'data': 'line 1\nline 2\n'},
+    'some/files/file': {'data': b'line 1\nline 2\n'},
   }
 
   COMMIT_B = {
-    'some/files/file': {'data': 'line 1\nline 2.1\n'},
+    'some/files/file': {'data': b'line 1\nline 2.1\n'},
   }
 
   COMMIT_C = {
-    'some/files/file': {'data': 'line 1.1\nline 2.1\n'},
+    'some/files/file': {'data': b'line 1.1\nline 2.1\n'},
   }
 
   COMMIT_D = {
     # This file should be automatically considered for ignore.
-    '.git-blame-ignore-revs': {'data': 'tag_C'},
+    '.git-blame-ignore-revs': {'data': b'tag_C'},
     # This file should not be considered.
-    'some/files/.git-blame-ignore-revs': {'data': 'tag_B'},
+    'some/files/.git-blame-ignore-revs': {'data': b'tag_B'},
   }
 
   def setUp(self):
@@ -89,8 +95,8 @@ class GitHyperBlameMainTest(GitHyperBlameTestBase):
     """Tests the main function (simple end-to-end test with no ignores)."""
     expected_output = [self.blame_line('C', '1) line 1.1'),
                        self.blame_line('B', '2) line 2.1')]
-    stdout = StringIO.StringIO()
-    stderr = StringIO.StringIO()
+    stdout = StringIO()
+    stderr = StringIO()
     retval = self.repo.run(self.git_hyper_blame.main,
                            args=['tag_C', 'some/files/file'], stdout=stdout,
                            stderr=stderr)
@@ -102,8 +108,8 @@ class GitHyperBlameMainTest(GitHyperBlameTestBase):
     """Tests the main function (simple end-to-end test with ignores)."""
     expected_output = [self.blame_line('C', ' 1) line 1.1'),
                        self.blame_line('A', '2*) line 2.1')]
-    stdout = StringIO.StringIO()
-    stderr = StringIO.StringIO()
+    stdout = StringIO()
+    stderr = StringIO()
     retval = self.repo.run(self.git_hyper_blame.main,
                            args=['-i', 'tag_B', 'tag_C', 'some/files/file'],
                            stdout=stdout, stderr=stderr)
@@ -118,8 +124,8 @@ class GitHyperBlameMainTest(GitHyperBlameTestBase):
     tempdir = tempfile.mkdtemp(suffix='_nogit', prefix='git_repo')
     try:
       os.chdir(tempdir)
-      stdout = StringIO.StringIO()
-      stderr = StringIO.StringIO()
+      stdout = StringIO()
+      stderr = StringIO()
       retval = self.git_hyper_blame.main(
           args=['-i', 'tag_B', 'tag_C', 'some/files/file'], stdout=stdout,
           stderr=stderr)
@@ -134,8 +140,8 @@ class GitHyperBlameMainTest(GitHyperBlameTestBase):
 
   def testBadFilename(self):
     """Tests the main function (bad filename)."""
-    stdout = StringIO.StringIO()
-    stderr = StringIO.StringIO()
+    stdout = StringIO()
+    stderr = StringIO()
     retval = self.repo.run(self.git_hyper_blame.main,
                            args=['-i', 'tag_B', 'tag_C', 'some/files/xxxx'],
                            stdout=stdout, stderr=stderr)
@@ -150,8 +156,8 @@ class GitHyperBlameMainTest(GitHyperBlameTestBase):
 
   def testBadRevision(self):
     """Tests the main function (bad revision to blame from)."""
-    stdout = StringIO.StringIO()
-    stderr = StringIO.StringIO()
+    stdout = StringIO()
+    stderr = StringIO()
     retval = self.repo.run(self.git_hyper_blame.main,
                            args=['-i', 'tag_B', 'xxxx', 'some/files/file'],
                            stdout=stdout, stderr=stderr)
@@ -165,8 +171,8 @@ class GitHyperBlameMainTest(GitHyperBlameTestBase):
     """Tests the main function (bad revision passed to -i)."""
     expected_output = [self.blame_line('C', '1) line 1.1'),
                        self.blame_line('B', '2) line 2.1')]
-    stdout = StringIO.StringIO()
-    stderr = StringIO.StringIO()
+    stdout = StringIO()
+    stderr = StringIO()
     retval = self.repo.run(self.git_hyper_blame.main,
                            args=['-i', 'xxxx', 'tag_C', 'some/files/file'],
                            stdout=stdout, stderr=stderr)
@@ -178,8 +184,8 @@ class GitHyperBlameMainTest(GitHyperBlameTestBase):
     """Tests passing the ignore list in a file."""
     expected_output = [self.blame_line('C', ' 1) line 1.1'),
                        self.blame_line('A', '2*) line 2.1')]
-    stdout = StringIO.StringIO()
-    stderr = StringIO.StringIO()
+    stdout = StringIO()
+    stderr = StringIO()
 
     with tempfile.NamedTemporaryFile(mode='w+', prefix='ignore') as ignore_file:
       ignore_file.write('# Line comments are allowed.\n')
@@ -205,8 +211,8 @@ class GitHyperBlameMainTest(GitHyperBlameTestBase):
 
     expected_output = [self.blame_line('A', '1*) line 1.1'),
                        self.blame_line('B', ' 2) line 2.1')]
-    stdout = StringIO.StringIO()
-    stderr = StringIO.StringIO()
+    stdout = StringIO()
+    stderr = StringIO()
 
     retval = self.repo.run(self.git_hyper_blame.main,
                            args=['tag_D', 'some/files/file'],
@@ -219,8 +225,8 @@ class GitHyperBlameMainTest(GitHyperBlameTestBase):
     # Test blame from a different revision. Despite the default ignore file
     # *not* being committed at that revision, it should still be picked up
     # because D is currently checked out.
-    stdout = StringIO.StringIO()
-    stderr = StringIO.StringIO()
+    stdout = StringIO()
+    stderr = StringIO()
 
     retval = self.repo.run(self.git_hyper_blame.main,
                            args=['tag_C', 'some/files/file'],
@@ -238,8 +244,8 @@ class GitHyperBlameMainTest(GitHyperBlameTestBase):
 
     expected_output = [self.blame_line('C', '1) line 1.1'),
                        self.blame_line('B', '2) line 2.1')]
-    stdout = StringIO.StringIO()
-    stderr = StringIO.StringIO()
+    stdout = StringIO()
+    stderr = StringIO()
 
     retval = self.repo.run(
         self.git_hyper_blame.main,
@@ -257,44 +263,44 @@ class GitHyperBlameSimpleTest(GitHyperBlameTestBase):
   """
 
   COMMIT_A = {
-    'some/files/file1': {'data': 'file1'},
-    'some/files/file2': {'data': 'file2'},
-    'some/files/empty': {'data': ''},
-    'some/other/file':  {'data': 'otherfile'},
+    'some/files/file1': {'data': b'file1'},
+    'some/files/file2': {'data': b'file2'},
+    'some/files/empty': {'data': b''},
+    'some/other/file':  {'data': b'otherfile'},
   }
 
   COMMIT_B = {
     'some/files/file2': {
       'mode': 0o755,
-      'data': 'file2 - vanilla\n'},
-    'some/files/empty': {'data': 'not anymore'},
-    'some/files/file3': {'data': 'file3'},
+      'data': b'file2 - vanilla\n'},
+    'some/files/empty': {'data': b'not anymore'},
+    'some/files/file3': {'data': b'file3'},
   }
 
   COMMIT_C = {
-    'some/files/file2': {'data': 'file2 - merged\n'},
+    'some/files/file2': {'data': b'file2 - merged\n'},
   }
 
   COMMIT_D = {
-    'some/files/file2': {'data': 'file2 - vanilla\nfile2 - merged\n'},
+    'some/files/file2': {'data': b'file2 - vanilla\nfile2 - merged\n'},
   }
 
   COMMIT_E = {
-    'some/files/file2': {'data': 'file2 - vanilla\nfile_x - merged\n'},
+    'some/files/file2': {'data': b'file2 - vanilla\nfile_x - merged\n'},
   }
 
   COMMIT_F = {
-    'some/files/file2': {'data': 'file2 - vanilla\nfile_y - merged\n'},
+    'some/files/file2': {'data': b'file2 - vanilla\nfile_y - merged\n'},
   }
 
   # Move file2 from files to other.
   COMMIT_G = {
     'some/files/file2': {'data': None},
-    'some/other/file2': {'data': 'file2 - vanilla\nfile_y - merged\n'},
+    'some/other/file2': {'data': b'file2 - vanilla\nfile_y - merged\n'},
   }
 
   COMMIT_H = {
-    'some/other/file2': {'data': 'file2 - vanilla\nfile_z - merged\n'},
+    'some/other/file2': {'data': b'file2 - vanilla\nfile_z - merged\n'},
   }
 
   def testBlameError(self):
@@ -387,34 +393,34 @@ class GitHyperBlameLineMotionTest(GitHyperBlameTestBase):
   """
 
   COMMIT_A = {
-    'file':  {'data': 'A\ngreen\nblue\n'},
+    'file':  {'data': b'A\ngreen\nblue\n'},
   }
 
   # Change "green" to "yellow".
   COMMIT_B = {
-    'file': {'data': 'A\nyellow\nblue\n'},
+    'file': {'data': b'A\nyellow\nblue\n'},
   }
 
   # Insert 2 lines at the top,
   # Change "yellow" to "red".
   # Insert 1 line at the bottom.
   COMMIT_C = {
-    'file': {'data': 'X\nY\nA\nred\nblue\nZ\n'},
+    'file': {'data': b'X\nY\nA\nred\nblue\nZ\n'},
   }
 
   # Insert 2 more lines at the top.
   COMMIT_D = {
-    'file': {'data': 'earth\nfire\nX\nY\nA\nred\nblue\nZ\n'},
+    'file': {'data': b'earth\nfire\nX\nY\nA\nred\nblue\nZ\n'},
   }
 
   # Insert a line before "red", and indent "red" and "blue".
   COMMIT_E = {
-    'file': {'data': 'earth\nfire\nX\nY\nA\ncolors:\n red\n blue\nZ\n'},
+    'file': {'data': b'earth\nfire\nX\nY\nA\ncolors:\n red\n blue\nZ\n'},
   }
 
   # Insert a line between "A" and "colors".
   COMMIT_F = {
-    'file': {'data': 'earth\nfire\nX\nY\nA\nB\ncolors:\n red\n blue\nZ\n'},
+    'file': {'data': b'earth\nfire\nX\nY\nA\nB\ncolors:\n red\n blue\nZ\n'},
   }
 
   def testCacheDiffHunks(self):
@@ -537,22 +543,22 @@ class GitHyperBlameLineNumberTest(GitHyperBlameTestBase):
   """
 
   COMMIT_A = {
-    'file':  {'data': 'red\nblue\n'},
+    'file':  {'data': b'red\nblue\n'},
   }
 
   # Change "blue" to "green".
   COMMIT_B = {
-    'file': {'data': 'red\ngreen\n'},
+    'file': {'data': b'red\ngreen\n'},
   }
 
   # Insert 2 lines at the top,
   COMMIT_C = {
-    'file': {'data': '\n\nred\ngreen\n'},
+    'file': {'data': b'\n\nred\ngreen\n'},
   }
 
   # Change "green" to "yellow".
   COMMIT_D = {
-    'file': {'data': '\n\nred\nyellow\n'},
+    'file': {'data': b'\n\nred\nyellow\n'},
   }
 
   def testTwoChangesWithAddedLines(self):
@@ -584,19 +590,19 @@ class GitHyperBlameUnicodeTest(GitHyperBlameTestBase):
 
   COMMIT_A = {
     GitRepo.AUTHOR_NAME: 'ASCII Author',
-    'file':  {'data': 'red\nblue\n'},
+    'file':  {'data': b'red\nblue\n'},
   }
 
   # Add a line.
   COMMIT_B = {
-    GitRepo.AUTHOR_NAME: u'\u4e2d\u56fd\u4f5c\u8005'.encode('utf-8'),
-    'file': {'data': 'red\ngreen\nblue\n'},
+    GitRepo.AUTHOR_NAME: '\u4e2d\u56fd\u4f5c\u8005'.encode('utf-8'),
+    'file': {'data': b'red\ngreen\nblue\n'},
   }
 
   # Modify a line with non-UTF-8 author and file text.
   COMMIT_C = {
-    GitRepo.AUTHOR_NAME: u'Lat\u00edn-1 Author'.encode('latin-1'),
-    'file': {'data': u'red\ngre\u00e9n\nblue\n'.encode('latin-1')},
+    GitRepo.AUTHOR_NAME: 'Lat\u00edn-1 Author'.encode('latin-1'),
+    'file': {'data': 'red\ngre\u00e9n\nblue\n'.encode('latin-1')},
   }
 
   def testNonASCIIAuthorName(self):
@@ -610,8 +616,8 @@ class GitHyperBlameUnicodeTest(GitHyperBlameTestBase):
     expected_output = [
         self.blame_line('A', '1) red', author='ASCII Author'),
         # Expect 8 spaces, to line up with the other name.
-        self.blame_line('B', '2) green',
-            author=u'\u4e2d\u56fd\u4f5c\u8005        '.encode('utf-8')),
+        self.blame_line(
+            'B', '2) green', author='\u4e2d\u56fd\u4f5c\u8005        '),
         self.blame_line('A', '3) blue', author='ASCII Author'),
     ]
     retval, output = self.run_hyperblame([], 'file', 'tag_B')
@@ -626,9 +632,9 @@ class GitHyperBlameUnicodeTest(GitHyperBlameTestBase):
     """
     expected_output = [
         self.blame_line('A', '1) red', author='ASCII Author  '),
-        # The Author has been re-encoded as UTF-8. The file data is preserved as
-        # raw byte data.
-        self.blame_line('C', '2) gre\xe9n', author='Lat\xc3\xadn-1 Author'),
+        # The Author has been re-encoded as UTF-8. The file data is converted to
+        # UTF8 and unknown characters replaced.
+        self.blame_line('C', '2) gre\ufffdn', author='Lat\xedn-1 Author'),
         self.blame_line('A', '3) blue', author='ASCII Author  '),
     ]
     retval, output = self.run_hyperblame([], 'file', 'tag_C')