fix(reduce-idents): improve encode

[defcc]

fix #853

[codecov-io]

Codecov Report

Merging #862 into master will increase coverage by <.01%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master     #862      +/-   ##
==========================================
+ Coverage   97.66%   97.66%   +<.01%     
==========================================
  Files         111      111              
  Lines        3338     3339       +1     
  Branches     1011     1011              
==========================================
+ Hits         3260     3261       +1     
  Misses         71       71              
  Partials        7        7

Impacted Files	Coverage Δ
packages/postcss-reduce-idents/src/lib/encode.js	100% <100%> (ø)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 8243401...d21abbe. Read the comment docs.

[alexander-akait]

/cc @ben-eb can you look at this?

[ben-eb]

Looks great! I think out of the scope of this PR but I'm not a fan of how we test the encoder here. Do you have any suggestions? One I was thinking about recently was having some kind of test "oracle" which would return specific character pairs to capture the edge cases, then test the real implementation against it. Something like:

const oracle = (val, num) => ({ 51: 'Z', 52: 'aa' }[num]);

...

expect(encode(null, num)).toBe(oracle(null, num));

What do you think? Possibly with a few more cases in there? 🙂

[defcc]

Thanks, I totally agree with you. I could add more edge cases according to the encoder spec with hardcode later.

—-
There maybe another way to test it with a decode util，x=decode(encode(x))，but then we have to ensure both two are correct:)

[ben-eb]

Looks great, thanks! 💯