Added support to read and write Decimal128 to Avro

[potter420]

Hi, this is to partially address #733 .

Added read of Decimal128 from both fixed and bytes schema.

Support writing to bytes schema to avro.

[codecov]

Codecov Report

Merging #837 (ce2a5a7) into main (9e6924b) will decrease coverage by 0.12%.
The diff coverage is 29.62%.

@@            Coverage Diff             @@
##             main     #837      +/-   ##
==========================================
- Coverage   71.55%   71.43%   -0.13%     
==========================================
  Files         327      327              
  Lines       17727    17782      +55     
==========================================
+ Hits        12684    12702      +18     
- Misses       5043     5080      +37     

Impacted Files	Coverage Δ
src/io/avro/write/schema.rs	66.66% <0.00%> (-1.91%)	⬇️
src/io/avro/write/serialize.rs	74.28% <0.00%> (-10.13%)	⬇️
src/io/avro/read/deserialize.rs	69.91% <50.00%> (-3.13%)	⬇️
src/bitmap/utils/slice_iterator.rs	86.20% <0.00%> (-1.73%)	⬇️
src/io/parquet/read/dictionary.rs	60.46% <0.00%> (-0.72%)	⬇️
src/bitmap/immutable.rs	87.80% <0.00%> (ø)
src/buffer/immutable.rs	96.42% <0.00%> (ø)
src/array/utf8/mutable.rs	81.03% <0.00%> (ø)
src/io/avro/read/schema.rs	73.41% <0.00%> (+3.79%)	⬆️

---

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 9e6924b...ce2a5a7. Read the comment docs.

[jorgecarleitao]

Thanks a lot, @potter420 . Would it be possible to add 2 tests, one in the read where we show that we can read from avro-rs, and one in write, to show that we can roundtrip decimals?

[potter420]

@jorgecarleitao For some reason I can't pass the ipc and parquet test locally, despite running write_parquet.py.
I used this for the write part using the following:

let len = (x.leading_zeros() / 8) as usize;
util::zigzag_encode((16 - len) as i64, buf).unwrap();
buf.extend_from_slice(&x.to_be_bytes()[len..]);

which is wrong. I fixed it by

let len = ((x.leading_zeros() / 8) - ((x.leading_zeros() / 8) % 2)) as usize;

Did some round trip with spark, the result look fine. The problem is, I can't find a way to stabilize the file name. So for the CI, we have to find the file name.

[jorgecarleitao]

Flawless. Thanks a lot, @potter420 !