Bug 1572333 - Use a normalized str for GIT_AUTHOR_NAME and GIT_AUTHOR…

…_EMAIL in Windows r=glob

Reviewers: glob

Reviewed By: glob

Bug #: 1572333

Differential Revision: https://phabricator.services.mozilla.com/D41149

moz-phab

@@ -36,6 +36,7 @@ import tempfile
 import threading
 import time
 import traceback
+import unicodedata
 import urllib2
 import urlparse
 import uuid
@@ -2694,8 +2695,20 @@ class Git(Repository):
                 ["commit-tree", "-p", parent, "-F", message_file, tree_hash],
                 split=False,
                 extra_env={
-                    "GIT_AUTHOR_NAME": author_name.encode("utf-8"),
-                    "GIT_AUTHOR_EMAIL": author_email.encode("utf-8"),
+                    "GIT_AUTHOR_NAME": (
+                        unicodedata.normalize("NFKD", author_name).encode(
+                            "ascii", "ignore"
+                        )
+                        if IS_WINDOWS
+                        else author_name.encode("utf-8")
+                    ),
+                    "GIT_AUTHOR_EMAIL": (
+                        unicodedata.normalize("NFKD", author_email).encode(
+                            "ascii", "ignore"
+                        )
+                        if IS_WINDOWS
+                        else author_email.encode("utf-8")
+                    ),
                     "GIT_AUTHOR_DATE": author_date,
                 },
             )

tests/test_git.py

@@ -1,3 +1,5 @@
+# coding=utf-8
+
 import imp
 import mock
 import os
@@ -409,3 +411,18 @@ def test_is_cinnabar(m_git_out, git):
     git._cinnabar_installed = None
     assert not git.is_cinnabar_installed
     m_git_out.assert_called_once_with(["--list-cmds=main,others"])
+
+
+@mock.patch("mozphab.Git.git_out")
+def test_unicode_in_windows_env(m_git_out, git, monkeypatch):
+    utf8 = "ćwikła".decode("utf-8")
+    asci = "cwika"
+    monkeypatch.setattr(mozphab, "IS_WINDOWS", True)
+    git._commit_tree("parent", "tree_hash", "message", utf8, utf8, "date")
+    m_git_out.assert_called_once_with(
+        ["commit-tree", "-p", "parent", "-F", mock.ANY, "tree_hash"],
+        split=False,
+        extra_env=dict(
+            GIT_AUTHOR_NAME=asci, GIT_AUTHOR_EMAIL=asci, GIT_AUTHOR_DATE="date"
+        ),
+    )