Sort token encode/decode mappings, so that longer strings are

converted first, which is more likely to give correct results for
text that has multiple possible de/encodings.

0 files changed, 0 insertions, 0 deletions