serp-zlib

Compression for Serpentine: zlib-format, raw-DEFLATE and gzip streams, plus crc32/adler32 checksums. Runtime-backed ([R]): native builds use a self-contained DEFLATE implementation in the Serpentine runtime; under CPython the same imports delegate to the stdlib zlib/gzip modules.

from serp_zlib import compress, decompress, crc32

data: bytes = b"hello hello hello"
z: bytes = compress(data, 6)
print(decompress(z) == data)   # True
print(crc32(data))

API

• crc32(data, value=0) / adler32(data, value=1) — running-checksum value like stdlib zlib.
• compress(data, level=-1) (zlib format; -1 = default level 6) / decompress(data, wbits=15) — wbits routes the container: 15 zlib, -15 raw DEFLATE, 31 gzip.
• deflate(data, level=-1) / inflate(data) — raw DEFLATE extras (used by serp-zipfile).
• gzip_compress(data, level=-1) / gzip_decompress(data).
• Constants: MAX_WBITS, DEFLATED, Z_DEFAULT_COMPRESSION, Z_BEST_COMPRESSION, Z_BEST_SPEED, Z_NO_COMPRESSION.

Levels are 0 (stored) through 9. Bad levels raise ValueError("Bad compression level"); corrupt streams raise ValueError("invalid compressed data") (CPython raises zlib.error). compressobj/decompressobj are not provided.

Divergence

Compressed output is a valid stream in both worlds and fully interoperable (each world decompresses the other's output), but the exact compressed bytes differ between CPython's zlib and the native runtime's encoder. Only roundtrips, decompressed content and checksums are world-identical.