/*-*- mode:c;indent-tabs-mode:nil;c-basic-offset:4;tab-width:8;coding:utf-8 -*-│
│vi: set net ft=c ts=4 sts=4 sw=4 fenc=utf-8                                :vi│
╞══════════════════════════════════════════════════════════════════════════════╡
│ Copyright 1995-2016 Mark Adler                                               │
│ Copyright 2017 The Chromium Authors                                          │
│ Use of this source code is governed by the BSD-style licenses that can       │
│ be found in the third_party/zlib/LICENSE file.                               │
╚─────────────────────────────────────────────────────────────────────────────*/
#include "libc/dce.h"
#include "libc/nexgen32e/x86feature.h"
#include "third_party/zlib/internal.h"
#include "third_party/zlib/zutil.h"

asm(".ident\t\"\\n\\n\
zlib (zlib License)\\n\
Copyright 1995-2017 Jean-loup Gailly and Mark Adler\"");
asm(".include \"libc/disclaimer.inc\"");

#define BASE 65521U /* largest prime smaller than 65536 */
#define NMAX 5552   /* largest n such that 255n(n+1)/2+(n+1)(BASE-1)<=2^32-1 */

#define DO1(buf, i)    \
  {                    \
    adler += (buf)[i]; \
    sum2 += adler;     \
  }
#define DO2(buf, i) \
  DO1(buf, i);      \
  DO1(buf, i + 1);
#define DO4(buf, i) \
  DO2(buf, i);      \
  DO2(buf, i + 2);
#define DO8(buf, i) \
  DO4(buf, i);      \
  DO4(buf, i + 4);
#define DO16(buf) \
  DO8(buf, 0);    \
  DO8(buf, 8);

/* use NO_DIVIDE if your processor does not do division in hardware --
   try it both ways to see which is faster */
#ifdef NO_DIVIDE
/* note that this assumes BASE is 65521, where 65536 % 65521 == 15
   (thank you to John Reiser for pointing this out) */
#define CHOP(a)                  \
  do {                           \
    unsigned long tmp = a >> 16; \
    a &= 0xffffUL;               \
    a += (tmp << 4) - tmp;       \
  } while (0)
#define MOD28(a)              \
  do {                        \
    CHOP(a);                  \
    if (a >= BASE) a -= BASE; \
  } while (0)
#define MOD(a) \
  do {         \
    CHOP(a);   \
    MOD28(a);  \
  } while (0)
#define MOD63(a)                            \
  do { /* this assumes a is not negative */ \
    int64_t tmp = a >> 32;                  \
    a &= 0xffffffffL;                       \
    a += (tmp << 8) - (tmp << 5) + tmp;     \
    tmp = a >> 16;                          \
    a &= 0xffffL;                           \
    a += (tmp << 4) - tmp;                  \
    tmp = a >> 16;                          \
    a &= 0xffffL;                           \
    a += (tmp << 4) - tmp;                  \
    if (a >= BASE) a -= BASE;               \
  } while (0)
#else
#define MOD(a)   a %= BASE
#define MOD28(a) a %= BASE
#define MOD63(a) a %= BASE
#endif

uLong adler32_z(uLong adler, const Bytef *buf, size_t len) {
  return adler32(adler, buf, len);
}

uLong adler32(uLong adler, const Bytef *buf, uInt len) {
  unsigned long sum2;
  unsigned n;
  if (!IsTiny() && X86_HAVE(SSSE3) && buf && len >= 64) {
    return adler32_simd_(adler, buf, len);
  }
  /* split Adler-32 into component sums */
  sum2 = (adler >> 16) & 0xffff;
  adler &= 0xffff;
  /* in case user likes doing a byte at a time, keep it fast */
  if (len == 1) {
    adler += buf[0];
    if (adler >= BASE) adler -= BASE;
    sum2 += adler;
    if (sum2 >= BASE) sum2 -= BASE;
    return adler | (sum2 << 16);
  }
  /* initial Adler-32 value (deferred check for len == 1 speed) */
  if (buf == Z_NULL) return 1L;
  /* in case short lengths are provided, keep it somewhat fast */
  if (len < 16) {
    while (len--) {
      adler += *buf++;
      sum2 += adler;
    }
    if (adler >= BASE) adler -= BASE;
    MOD28(sum2); /* only added so many BASE's */
    return adler | (sum2 << 16);
  }
  /* do length NMAX blocks -- requires just one modulo operation */
  while (len >= NMAX) {
    len -= NMAX;
    n = NMAX / 16; /* NMAX is divisible by 16 */
    do {
      DO16(buf); /* 16 sums unrolled */
      buf += 16;
    } while (--n);
    MOD(adler);
    MOD(sum2);
  }
  /* do remaining bytes (less than NMAX, still just one modulo) */
  if (len) { /* avoid modulos if none remaining */
    while (len >= 16) {
      len -= 16;
      DO16(buf);
      buf += 16;
    }
    while (len--) {
      adler += *buf++;
      sum2 += adler;
    }
    MOD(adler);
    MOD(sum2);
  }
  /* return recombined sums */
  return adler | (sum2 << 16);
}

uLong adler32_combine(uLong adler1, uLong adler2, int64_t len2) {
  unsigned long sum1;
  unsigned long sum2;
  unsigned rem;
  /* for negative len, return invalid adler32 as a clue for debugging */
  if (len2 < 0) return 0xffffffffUL;
  /* the derivation of this formula is left as an exercise for the reader */
  MOD63(len2); /* assumes len2 >= 0 */
  rem = (unsigned)len2;
  sum1 = adler1 & 0xffff;
  sum2 = rem * sum1;
  MOD(sum2);
  sum1 += (adler2 & 0xffff) + BASE - 1;
  sum2 += ((adler1 >> 16) & 0xffff) + ((adler2 >> 16) & 0xffff) + BASE - rem;
  if (sum1 >= BASE) sum1 -= BASE;
  if (sum1 >= BASE) sum1 -= BASE;
  if (sum2 >= ((unsigned long)BASE << 1)) sum2 -= ((unsigned long)BASE << 1);
  if (sum2 >= BASE) sum2 -= BASE;
  return sum1 | (sum2 << 16);
}