tree 05fa2e680b9fb6cdf33fe8833e2dc50fe482691c
parent d7be62c714fef2c2bf00bf1e173a2d1fc50fddb5
author bnc <bnc@google.com> 1602006810 -0700
committer Copybara-Service <copybara-worker@google.com> 1602010407 -0700

Change Huffman encoder benchmark to typical character distribution.

Reimplement a Huffman decoder and run it on random binary input to generate
characters from the ideal distribution corresponding to the Huffman code.
This is much closer to real life usage than repeated 'a' characters, giving more
meaningful benchmark results.

PiperOrigin-RevId: 335674758
Change-Id: I7f90439066691b45fea693715091f70b15628920