# ============================================================================
# inflate.ludic — DEFLATE decompression (RFC 1951), written in Ludic.
#
# PNG stores its pixels zlib-compressed, so decoding one means implementing
# inflate. The C runtime linked zlib for this. We don't: zlib is not present by
# default on every target Ludic compiles for (Windows especially), and shipping
# a dependency to read a sprite is a poor trade when the algorithm is this
# small. So it lives here, in the language.
#
# The decoder is the canonical-Huffman formulation from Mark Adler's `puff`: a
# symbol table plus per-length counts. Short codes (<= Z_FAST bits, which is the
# overwhelming majority) resolve in a single lookup out of a 512-entry table built
# with the symbol table; longer ones fall back to puff's walk, one bit at a time.
# The walk alone was fine for sprites, but a PBR scene inflates hundreds of
# megabytes of texture at load, and there the table is worth its 2 KB.
# ============================================================================

# ---- bit reader (DEFLATE packs bits least-significant-first) ---------------
var z_src:    pointer = null
var z_len:    int = 0
var z_pos:    int = 0
var z_bitbuf: int = 0
var z_bitcnt: int = 0
var z_err:    int = 0

function z_start(src: pointer, len: int) -> void {
  z_tables_once()
  z_src = src
  z_len = len
  z_pos = 0
  z_bitbuf = 0
  z_bitcnt = 0
  z_err = 0
}

# Fill the bit buffer to at least `n` bits without consuming any (n <= 16, so the
# buffer never shifts a byte past bit 15 and cannot reach the sign bit).
function z_need(n: int) -> void {
  while z_bitcnt < n {
    if z_pos >= z_len { return }
    z_bitbuf = (z_bitbuf | (z_src[z_pos] << z_bitcnt))
    z_pos += 1
    z_bitcnt += 8
  }
}

function z_bits(need: int) -> int {
  var val = z_bitbuf
  while z_bitcnt < need {
    if z_pos >= z_len {
      z_err = 1
      return 0
    }
    val = (val | (z_src[z_pos] << z_bitcnt))
    z_pos += 1
    z_bitcnt += 8
  }
  z_bitbuf = (val >> need)
  z_bitcnt -= need
  return (val & ((1 << need) - 1))
}

# ---- Huffman tables -------------------------------------------------------
# One buffer per table: 16 length-counts, then a Z_FASTSZ-entry lookup keyed by
# the next Z_FAST bits of the stream, then the symbols in canonical order.
# A lookup entry is (length << 16) | symbol, or 0 when no code that short matches.
# Z_FAST = 10 measured fastest over a 493 MB corpus (9 and 11 are both ~8% slower:
# 9 misses the table more often, 11 spends more clearing it per dynamic block).
const Z_FAST:   int = 10
const Z_FASTSZ: int = 1024          # 1 << Z_FAST
const Z_SYMS:   int = 1040         # 16 + Z_FASTSZ: where the symbols start

function z_table_new(nsym: int) -> words {
  return words((Z_SYMS + nsym))
}

# lengths[i] = code length of symbol i (0 = symbol unused)
function z_table_build(table: words, lengths: words, n: int) -> void {
  for i in 0 .. 16 {
    table[i] = 0
  }
  for s in 0 .. n {
    let l = lengths[s]
    table[l] += 1
  }
  table[0] = 0                 # length 0 means "not present"
  # offset of each length's first symbol
  let offs: words = words(16)
  offs[1] = 0
  for l in 1 .. 15 {
    offs[l + 1] = offs[l] + table[l]
  }
  for s in 0 .. n {
    let l = lengths[s]
    if l != 0 {
      table[Z_SYMS + offs[l]] = s
      offs[l] += 1
    }
  }
  free(offs)

  # ---- the fast lookup ----
  for i in 0 .. Z_FASTSZ {
    table[16 + i] = 0
  }
  # first canonical code of each length
  let firstc: words = words(17)
  var code = 0
  for l in 1 .. 16 {
    code = ((code + table[l - 1]) << 1)
    firstc[l] = code
  }
  var idx = 0
  for l in 1 .. 16 {
    let cnt = table[l]
    var k = 0
    while k < cnt {
      let sym = table[Z_SYMS + idx]
      if l <= Z_FAST {
        # DEFLATE reads a code most-significant-bit first out of a stream packed
        # least-significant-bit first, so the table is keyed by the reversed code
        let c = firstc[l] + k
        var rev = 0
        var b = 0
        while b < l {
          rev = ((rev << 1) | ((c >> b) & 1))
          b += 1
        }
        let entry = ((l << 16) | sym)
        var j = rev
        while j < Z_FASTSZ {
          table[16 + j] = entry
          j += (1 << l)
        }
      }
      idx += 1
      k += 1
    }
  }
  free(firstc)
}

function z_decode(table: words) -> int {
  z_need(Z_FAST)
  if z_bitcnt >= Z_FAST {
    let e = table[16 + (z_bitbuf & (Z_FASTSZ - 1))]
    if e != 0 {
      let l = (e >> 16)
      z_bitbuf = (z_bitbuf >> l)
      z_bitcnt -= l
      return (e & 65535)
    }
  }
  # a code longer than Z_FAST bits (or a stream too short to peek): walk it
  var code = 0
  var first = 0
  var index = 0
  for len in 1 .. 16 {
    code = (code | z_bits(1))
    let count = table[len]
    if code - first < count {
      return table[Z_SYMS + index + (code - first)]
    }
    index += count
    first = ((first + count) << 1)
    code = (code << 1)
  }
  z_err = 1
  return -1
}

# ---- length / distance code tables (RFC 1951 section 3.2.5) ---------------
function z_len_base(sym: int) -> int {
  if sym < 8 { return 3 + sym }
  if sym == 28 { return 258 }
  let extra = (sym - 4) / 4
  let group = (1 << extra)
  return 3 + ((group - 1) << 2) + 4 + (sym - 4 - extra * 4) * group
}

function z_len_extra(sym: int) -> int {
  if sym < 8 { return 0 }
  if sym == 28 { return 0 }
  return (sym - 4) / 4
}

function z_dist_base(sym: int) -> int {
  if sym < 4 { return 1 + sym }
  let extra = (sym - 2) / 2
  let group = (1 << extra)
  return 1 + (group << 1) + (sym - 2 - extra * 2) * group
}

function z_dist_extra(sym: int) -> int {
  if sym < 4 { return 0 }
  return (sym - 2) / 2
}

# The RFC tables above are pure functions of the symbol; compute them once rather
# than dividing per match.
var z_lbase: words = null
var z_lext:  words = null
var z_dbase: words = null
var z_dext:  words = null

function z_tables_once() -> void {
  if z_lbase != null { return }
  z_lbase = words(29)
  z_lext = words(29)
  for s in 0 .. 29 {
    z_lbase[s] = z_len_base(s)
    z_lext[s] = z_len_extra(s)
  }
  z_dbase = words(30)
  z_dext = words(30)
  for s in 0 .. 30 {
    z_dbase[s] = z_dist_base(s)
    z_dext[s] = z_dist_extra(s)
  }
}

# ---- block decoders -------------------------------------------------------
# `out` is the destination window; returns the new write position, or -1.
function z_stored(out: pointer, at: int, cap: int) -> int {
  z_bitbuf = 0
  z_bitcnt = 0                        # stored blocks are byte-aligned
  if z_pos + 4 > z_len { return -1 }
  let n = z_src[z_pos] + (z_src[z_pos + 1] << 8)
  z_pos += 4                   # LEN then its one's complement
  var w = at
  for i in 0 .. n {
    if z_pos >= z_len { return -1 }
    if w >= cap { return -1 }
    out[w] = z_src[z_pos]
    w += 1
    z_pos += 1
  }
  return w
}

function z_codes(out: pointer, at: int, cap: int, lit: words, dist: words) -> int {
  var w = at
  var sym = z_decode(lit)
  while sym != 256 {
    if z_err != 0 { return -1 }
    if sym < 0 { return -1 }
    if sym < 256 {
      if w >= cap { return -1 }
      out[w] = sym
      w += 1
    }
    if sym > 256 {
      let s = sym - 257
      if s >= 29 { return -1 }
      let length = z_lbase[s] + z_bits(z_lext[s])
      let d = z_decode(dist)
      if d < 0 { return -1 }
      if d >= 30 { return -1 }
      let distance = z_dbase[d] + z_bits(z_dext[d])
      if distance > w { return -1 }
      if w + length > cap { return -1 }      # bounds once, not per byte
      var sp = w - distance
      var k = 0
      while k < length {
        out[w] = out[sp]
        w += 1
        sp += 1
        k += 1
      }
    }
    sym = z_decode(lit)
  }
  return w
}

function z_fixed_tables(lit: words, dist: words) -> void {
  let lengths: words = words(288)
  for i in 0 .. 144 { lengths[i] = 8 }
  for i in 144 .. 256 { lengths[i] = 9 }
  for i in 256 .. 280 { lengths[i] = 7 }
  for i in 280 .. 288 { lengths[i] = 8 }
  z_table_build(lit, lengths, 288)
  for i in 0 .. 30 { lengths[i] = 5 }
  z_table_build(dist, lengths, 30)
  free(lengths)
}

function z_dynamic_tables(lit: words, dist: words) -> int {
  let nlen = z_bits(5) + 257
  let ndist = z_bits(5) + 1
  let ncode = z_bits(4) + 4
  if nlen > 286 { return 0 }
  if ndist > 30 { return 0 }

  let lengths: words = words(320)
  for i in 0 .. 19 { lengths[i] = 0 }
  # the code-length alphabet is transmitted in this fixed permutation
  # 16,17,18,0,8,7,9,6,10,5,11,4,12,3,13,2,14,1,15 — biased by '0' so it is one literal
  let order = "@AB08796:5;4<3=2>1?"
  for i in 0 .. ncode {
    lengths[order[i] - 48] = z_bits(3)
  }
  let clen = z_table_new(19)
  z_table_build(clen, lengths, 19)

  var n = 0
  while n < nlen + ndist {
    let sym = z_decode(clen)
    if sym < 0 { return 0 }
    if sym < 16 {
      lengths[n] = sym
      n += 1
    }
    if sym >= 16 {
      var prev = 0
      var rep = 0
      if sym == 16 {
        if n == 0 { return 0 }
        prev = lengths[n - 1]
        rep = 3 + z_bits(2)
      }
      if sym == 17 { rep = 3 + z_bits(3) }
      if sym == 18 { rep = 11 + z_bits(7) }
      for k in 0 .. rep {
        if n < 320 {
          lengths[n] = prev
          n += 1
        }
      }
    }
  }
  z_table_build(lit, lengths, nlen)
  # the distance lengths follow the literal ones in the same buffer
  let dl: words = words(32)
  for i in 0 .. ndist { dl[i] = lengths[nlen + i] }
  z_table_build(dist, dl, ndist)
  free(dl)
  free(lengths)
  free(clen)
  return 1
}

# Inflate a raw DEFLATE stream. Returns bytes written, or -1.
function z_inflate(src: pointer, len: int, out: pointer, cap: int) -> int {
  z_start(src, len)
  let lit = z_table_new(288)
  let dist = z_table_new(30)
  var w = 0
  var final = 0
  while final == 0 {
    final = z_bits(1)
    let btype = z_bits(2)
    if z_err != 0 { return -1 }
    if btype == 0 { w = z_stored(out, w, cap) }
    if btype == 1 {
      z_fixed_tables(lit, dist)
      w = z_codes(out, w, cap, lit, dist)
    }
    if btype == 2 {
      if z_dynamic_tables(lit, dist) == 0 { return -1 }
      w = z_codes(out, w, cap, lit, dist)
    }
    if btype == 3 { return -1 }
    if w < 0 { return -1 }
  }
  free(lit)
  free(dist)
  return w
}

# zlib wrapper (RFC 1950): two header bytes, then DEFLATE, then Adler-32.
function z_uncompress(src: pointer, len: int, out: pointer, cap: int) -> int {
  if len < 2 { return -1 }
  let cmf = src[0]
  if (cmf & 15) != 8 { return -1 }
  return z_inflate(offset(src, 2), len - 2, out, cap)
}

# gzip framing (RFC 1952): a 10-byte header (magic 1f 8b, CM=8, FLG, 4-byte MTIME,
# XFL, OS), optional FEXTRA/FNAME/FCOMMENT/FHCRC fields selected by FLG, then the
# same DEFLATE body zlib carries, then an 8-byte CRC32 + ISIZE trailer. We skip
# the header + optional fields, inflate the body, and ignore the trailer — the
# CRC is a redundancy check, not needed to decode (PNG likewise ignores ancillary
# CRCs). Returns bytes written, or -1.
function z_gunzip(src: pointer, len: int, out: pointer, cap: int) -> int {
  if len < 18 { return -1 }                 # 10 header + 8 trailer minimum
  if src[0] != 31 { return -1 }             # 0x1f
  if src[1] != 139 { return -1 }            # 0x8b
  if src[2] != 8 { return -1 }              # CM must be DEFLATE
  let flg = src[3]
  var pos = 10                              # past the fixed header
  if (flg & 4) != 0 {                       # FEXTRA: 2-byte length then that many bytes
    if pos + 2 > len { return -1 }
    let xlen = src[pos] + (src[pos + 1] << 8)
    pos = pos + 2 + xlen
  }
  if (flg & 8) != 0 {                       # FNAME: NUL-terminated
    while pos < len and src[pos] != 0 { pos += 1 }
    pos += 1
  }
  if (flg & 16) != 0 {                      # FCOMMENT: NUL-terminated
    while pos < len and src[pos] != 0 { pos += 1 }
    pos += 1
  }
  if (flg & 2) != 0 { pos += 2 }       # FHCRC: 2-byte header CRC
  if pos + 8 > len { return -1 }
  return z_inflate(offset(src, pos), len - pos - 8, out, cap)
}