GitLab

The 'name' field of entries produced by Json5File.load_from_files() was
a str, and this CL makes it a NameStyleConverter object.  The original
string is accessible via |entry['name'].original|, and
|str(entry['name'])|.  This will make filter setup simpler because we
don't need to provide filters for naming style conversion.

* name_style_converter.py:
 - Provide |original| getter, and __str__().
 - Provide __lt__() for sort().
 - Provide __eq__() for groupby().
  Some Jinja2 templates use built-in sort and groupby filters, and it's
  hard to specify key object in templates.

* json5_generator.py
  Sets entry['name'] to a NameStyleConverter object, and remove
  entry['tokenized_name'].

* DEPS
  Clean stale *.pyc in blink/renderer/build/scripts too. Some buildbots
  has media_feature_symbol.pyc in build/scripts/core/css/.

* Other files
 - If code requires a string name, access entry['name'].original.
 - Use entry.name instead of entry.tokenized_name


Bug: 843927
Change-Id: Ic47c90c51c8e88f08307a41b052ecfce971a4545
Reviewed-on: https://chromium-review.googlesource.com/1068575Reviewed-by: Hitoshi Yoshida <peria@chromium.org>
Commit-Queue: Kent Tamura <tkent@chromium.org>
Cr-Commit-Position: refs/heads/master@{#561780}