[grit] always remove ByteOrderMark marks

This CL changes grit MessageNode so that ByteOrderMarks are always
removed from message translations.

This will prevent lint errors due to accidental BOM characters in xtb
files that will break the build on TC imports.

Currently the BOM appears in 8 string translations. This is after
several strings had been re-translated to remove the BOM.

This CL has been tested with all ByteOrderMark suppressions removed
from build/android/lint/suppressions.xml and there are no BOM related
lint errors.

Bug: 1033305
Change-Id: Ia3d7a4e449ed717f19a4eea013e4ca1b4113e5c1
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2039342Reviewed-by: Andrew Grieve <agrieve@chromium.org>
Reviewed-by: Megan Jablonski <megjablon@chromium.org>
Commit-Queue: Andrew Grieve <agrieve@chromium.org>
Cr-Commit-Position: refs/heads/master@{#738821}

tools/grit/grit/node/message.py

@@ -268,6 +268,8 @@ class MessageNode(base.ContentNode):
                                          ).GetRealContent()
     if self._replace_ellipsis:
       msg = _ELLIPSIS_PATTERN.sub(_ELLIPSIS_SYMBOL, msg)
+    # Always remove all byte order marks (\uFEFF) https://crbug.com/1033305
+    msg = msg.replace(u'\uFEFF','')
     return msg.replace('[GRITLANGCODE]', lang)
 
   def NameOrOffset(self):

tools/grit/grit/node/message_unittest.py

@@ -98,6 +98,17 @@ class MessageUnittest(unittest.TestCase):
     content = msg.Translate('en')
     self.failUnlessEqual(u'A...B.... %s\u2026 B\u2026 C\u2026', content)
 
+  def testRemoveByteOrderMark(self):
+    root = util.ParseGrdForUnittest(u'''
+        <messages>
+        <message name="IDS_HAS_BOM" desc="">
+        \uFEFFThis\uFEFF i\uFEFFs OK\uFEFF
+        </message>
+        </messages>''')
+    msg, = root.GetChildrenOfType(message.MessageNode)
+    content = msg.Translate('en')
+    self.failUnlessEqual(u'This is OK', content)
+
   def testPlaceholderHasTooManyExamples(self):
     try:
       util.ParseGrdForUnittest("""\