mirror of https://github.com/tongzx/nt5src
You can not select more than 25 topics
Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
1293 lines
28 KiB
1293 lines
28 KiB
/* -------------------------------------------------------------------- */
|
|
/* */
|
|
/* Copyright (c) 1991-1999 by Andrew Kadatch */
|
|
/* */
|
|
/* -------------------------------------------------------------------- */
|
|
|
|
#include "xprs.h"
|
|
|
|
#define MAX_CHAIN 9
|
|
|
|
|
|
// Zobrist hashing
|
|
#define Z_HASH_SIZE_LOG (BUFF_SIZE_LOG - 1)
|
|
#define Z_HASH_SIZE (1 << Z_HASH_SIZE_LOG)
|
|
#define Z_HASH_SUM(b) (z_hash_map[0][b[0]] ^ z_hash_map[1][b[1]] ^ z_hash_map[2][b[2]])
|
|
|
|
// quick hashing
|
|
#define Q_HASH_SH1 3
|
|
#define Q_HASH_SH2 (Q_HASH_SH1 >> 1)
|
|
#define Q_HASH_SUM3(c1,c2,c3) (((c1) << Q_HASH_SH1) + ((c2) << Q_HASH_SH2) + (c3))
|
|
#define Q_HASH_SUM(b) Q_HASH_SUM3 (b[0], b[1], b[2])
|
|
#define Q_HASH_SIZE (Q_HASH_SUM3 (255, 255, 255) + 1)
|
|
|
|
#define z_hash_t uint16
|
|
#define z_index_t uint16
|
|
|
|
#if CODING & (CODING_HUFF_LEN | CODING_HUFF_PTR | CODING_HUFF_ALL)
|
|
typedef struct huff_node_t huff_node;
|
|
|
|
struct huff_node_t
|
|
{
|
|
huff_node *son[2];
|
|
uxint freq;
|
|
uint16 ch;
|
|
uint16 bits;
|
|
};
|
|
|
|
typedef struct
|
|
{
|
|
huff_node buff[2 * HUFF_SIZE], *head[256], **link[256];
|
|
} huff_info;
|
|
#endif
|
|
|
|
typedef struct
|
|
{
|
|
struct
|
|
{
|
|
#if CODING == CODING_BY_BIT
|
|
xint bits;
|
|
uchar *ptr;
|
|
#elif CODING & (CODING_DIRECT | CODING_DIRECT2)
|
|
uchar *ptr;
|
|
#elif CODING & (CODING_HUFF_LEN | CODING_HUFF_PTR | CODING_HUFF_ALL)
|
|
uxint freq[HUFF_SIZE];
|
|
uxint mask[HUFF_SIZE];
|
|
uchar bits[HUFF_SIZE];
|
|
huff_info info;
|
|
uxint pointers;
|
|
uxint extra;
|
|
uxint masks;
|
|
#endif /* CODING */
|
|
} stat;
|
|
xint chain;
|
|
xint max_size;
|
|
struct
|
|
{
|
|
uchar *beg;
|
|
uchar *ptr;
|
|
} comp;
|
|
struct
|
|
{
|
|
z_index_t *hash;
|
|
uchar *beg;
|
|
uchar *ptr;
|
|
uchar *tag_ptr;
|
|
tag_t tag_mask;
|
|
} temp;
|
|
struct
|
|
{
|
|
xint len;
|
|
xint pos;
|
|
} match;
|
|
struct
|
|
{
|
|
xint pos;
|
|
xint size;
|
|
xint stop;
|
|
xint progress;
|
|
const uchar *ptr;
|
|
const uchar *end;
|
|
const uchar *end_16;
|
|
const uchar *end_3;
|
|
const uchar *ptr_stop;
|
|
} orig;
|
|
} context;
|
|
|
|
|
|
#define v p[-1].c
|
|
|
|
typedef struct
|
|
{
|
|
union
|
|
{
|
|
z_index_t z_next[16];
|
|
const uchar *q_last[16];
|
|
} x;
|
|
|
|
context c;
|
|
} prs;
|
|
|
|
#define MAGIC_ENCODE 0x53E7C0DE
|
|
|
|
typedef struct
|
|
{
|
|
int magic;
|
|
void *memory;
|
|
prs *p;
|
|
uchar *temp;
|
|
unsigned max_size;
|
|
int chain;
|
|
} xpress_info;
|
|
|
|
|
|
#if MAX_CHAIN >= 1
|
|
static z_hash_t z_hash_map[MIN_MATCH][256];
|
|
static int z_hash_map_initialized = 0;
|
|
|
|
static void z_hash_map_init (void)
|
|
{
|
|
long v1, v2;
|
|
z_hash_t *m;
|
|
|
|
if (z_hash_map_initialized)
|
|
return;
|
|
|
|
v1 = 0x13579bdfL; v2 = 0x87654321L;
|
|
for (m = z_hash_map[0]; m < z_hash_map[0] + sizeof (z_hash_map) / sizeof (z_hash_map[0][0]); ++m)
|
|
{
|
|
long vv1 = v2, vv2 = v1, d = 0;
|
|
xint i = 32;
|
|
do
|
|
{
|
|
d += 0x9e3779b9L; vv1 += d; vv2 += d;
|
|
v1 += ((v2<<3) + vv1) ^ (v2 + d) ^ ((v2>>5) + vv2);
|
|
v2 += ((v1<<3) + vv2) ^ (v1 + d) ^ ((v1>>5) + vv1);
|
|
--i;
|
|
}
|
|
while (i);
|
|
*m = (z_hash_t) ((v1 += v2) & (Z_HASH_SIZE - 1));
|
|
}
|
|
|
|
z_hash_map_initialized = 1;
|
|
}
|
|
|
|
static void z_hash_insert (prs *p)
|
|
{
|
|
const uchar *b, *e;
|
|
xint n, h;
|
|
z_index_t *hash = v.temp.hash;
|
|
|
|
e = v.orig.end - (MIN_MATCH-1);
|
|
b = v.orig.ptr;
|
|
n = 0;
|
|
for (; b < e; ++b, ++n)
|
|
{
|
|
h = Z_HASH_SUM (b);
|
|
p->x.z_next[n] = hash[h];
|
|
hash[h] = (z_index_t) n;
|
|
}
|
|
e += MIN_MATCH-1;
|
|
for (; b < e; ++b, ++n)
|
|
p->x.z_next[n] = 0;
|
|
}
|
|
#endif
|
|
|
|
#if CODING != CODING_BY_BIT
|
|
|
|
static void tag_write_start (prs *p)
|
|
{
|
|
v.temp.tag_ptr = v.temp.ptr;
|
|
v.temp.ptr += sizeof (v.temp.tag_mask);
|
|
v.temp.tag_mask = 1;
|
|
}
|
|
|
|
#if CODING == CODING_HUFF_ALL
|
|
#define INC_MASKS ++v.stat.masks
|
|
#else
|
|
#define INC_MASKS
|
|
#endif
|
|
|
|
#define tag_write(p,ptr,n) \
|
|
{ \
|
|
tag_t __n = n | (v.temp.tag_mask << 1); \
|
|
if (v.temp.tag_mask < 0) \
|
|
{ \
|
|
*(__unaligned tag_t *) v.temp.tag_ptr = __n; \
|
|
v.temp.tag_ptr = ptr; \
|
|
ptr += sizeof (v.temp.tag_mask); \
|
|
INC_MASKS; \
|
|
__n = 1; \
|
|
} \
|
|
v.temp.tag_mask = __n; \
|
|
}
|
|
|
|
static void tag_write_finish (prs *p)
|
|
{
|
|
uchar *ptr = v.temp.ptr;
|
|
do
|
|
{
|
|
tag_write (p, ptr, 1);
|
|
}
|
|
while (ptr == v.temp.ptr);
|
|
}
|
|
|
|
#elif CODING == CODING_BY_BIT
|
|
|
|
static void tag_write_start (prs *p)
|
|
{
|
|
v.temp.tag_ptr = (uchar *) &v.temp.tag_mask;
|
|
v.temp.tag_mask = 0;
|
|
v.stat.bits = 0;
|
|
}
|
|
|
|
#define tag_write(p,ptr,n) do { \
|
|
if (--v.stat.bits < 0) \
|
|
{ \
|
|
*(__unaligned tag_t *)v.temp.tag_ptr = v.temp.tag_mask; \
|
|
v.temp.tag_mask = n; \
|
|
v.stat.bits = 8 * sizeof (v.temp.tag_mask) - 1; \
|
|
v.temp.tag_ptr = ptr; \
|
|
ptr += sizeof (v.temp.tag_mask); \
|
|
} \
|
|
v.temp.tag_mask = (v.temp.tag_mask << 1) + (n); \
|
|
} while (0)
|
|
|
|
#define tag_write_mask(p,ptr,n,b) do { \
|
|
if ((v.stat.bits -= (b)) < 0) \
|
|
{ \
|
|
*(__unaligned tag_t *)v.temp.tag_ptr = (v.temp.tag_mask << ((b) + v.stat.bits)) \
|
|
+ ((n) >> (-v.stat.bits)); \
|
|
v.stat.bits += 8 * sizeof (v.temp.tag_mask); \
|
|
v.temp.tag_mask = (n); \
|
|
v.temp.tag_ptr = ptr; \
|
|
ptr += sizeof (v.temp.tag_mask); \
|
|
} \
|
|
else \
|
|
v.temp.tag_mask = (v.temp.tag_mask << (b)) + (n); \
|
|
} while (0);
|
|
|
|
static void tag_write_finish (prs *p)
|
|
{
|
|
do
|
|
tag_write (p, v.temp.ptr, 1);
|
|
while (v.stat.bits != 0);
|
|
*(__unaligned tag_t *)v.temp.tag_ptr = v.temp.tag_mask;
|
|
}
|
|
|
|
#define write_lit(p,ptr,ch) do { \
|
|
tag_write (p, ptr, 0); \
|
|
*ptr++ = (ch); \
|
|
} while (0)
|
|
|
|
INLINE uchar *write_ptr (prs *p, uchar *ptr, xint offset, xint length)
|
|
{
|
|
uxint k;
|
|
|
|
--offset;
|
|
|
|
k = 2;
|
|
if (offset > 255) k = 3;
|
|
tag_write_mask (p, ptr, k, 2);
|
|
*ptr++ = (uchar) offset;
|
|
if (offset > 255) *ptr++ = (uchar) (offset >>= 8);
|
|
|
|
if (length <= 8)
|
|
{
|
|
length -= MIN_MATCH - 1;
|
|
tag_write_mask (p, ptr, 1, length);
|
|
}
|
|
else
|
|
{
|
|
tag_write_mask (p, ptr, 0, (9 - MIN_MATCH));
|
|
if ((length -= 9) < 15)
|
|
{
|
|
if (v.stat.ptr == 0)
|
|
{
|
|
v.stat.ptr = ptr;
|
|
*ptr++ = (uchar) length;
|
|
}
|
|
else
|
|
{
|
|
v.stat.ptr[0] |= length << 4;
|
|
v.stat.ptr = 0;
|
|
}
|
|
}
|
|
else
|
|
{
|
|
length -= 15;
|
|
if (v.stat.ptr == 0)
|
|
{
|
|
v.stat.ptr = ptr;
|
|
*ptr++ = 15;
|
|
}
|
|
else
|
|
{
|
|
v.stat.ptr[0] += 0xf0;
|
|
v.stat.ptr = 0;
|
|
}
|
|
*ptr++ = (uchar) length;
|
|
if (length >= 255)
|
|
{
|
|
ptr[-1] = 255;
|
|
ptr[0] = (uchar) length;
|
|
ptr[1] = (uchar) (length >>= 8);
|
|
ptr += 2;
|
|
}
|
|
}
|
|
}
|
|
|
|
return (ptr);
|
|
}
|
|
|
|
#endif
|
|
|
|
#if CODING & (CODING_DIRECT | CODING_DIRECT2)
|
|
|
|
#define write_lit(p,ptr,ch) do { \
|
|
*ptr++ = (ch); \
|
|
tag_write (p, ptr, 0); \
|
|
} while (0)
|
|
|
|
INLINE uchar *write_ptr (prs *p, uchar *ptr, int offset, int length)
|
|
{
|
|
length -= MIN_MATCH;
|
|
--offset;
|
|
|
|
#if CODING == CODING_DIRECT2
|
|
offset <<= DIRECT2_LEN_LOG;
|
|
if (length < DIRECT2_MAX_LEN)
|
|
{
|
|
offset |= length;
|
|
ptr[0] = (uchar) offset;
|
|
ptr[1] = (uchar) (offset >>= 8);
|
|
ptr += 2;
|
|
}
|
|
else
|
|
{
|
|
offset |= DIRECT2_MAX_LEN;
|
|
length -= DIRECT2_MAX_LEN;
|
|
ptr[0] = (uchar) offset;
|
|
ptr[1] = (uchar) (offset >>= 8);
|
|
ptr += 2;
|
|
if (v.stat.ptr == 0)
|
|
{
|
|
v.stat.ptr = ptr;
|
|
*ptr++ = (uchar) (length < 15 ? length : 15);
|
|
}
|
|
else
|
|
{
|
|
v.stat.ptr[0] |= (uchar) ((length < 15 ? length : 15) << 4);
|
|
v.stat.ptr = 0;
|
|
}
|
|
if ((length -= 15) >= 0)
|
|
{
|
|
*ptr++ = (uchar) length;
|
|
if (length >= 255)
|
|
{
|
|
ptr[-1] = 255;
|
|
length += DIRECT2_MAX_LEN + 15;
|
|
ptr[0] = (uchar) length;
|
|
ptr[1] = (uchar) (length >>= 8);
|
|
ptr += 2;
|
|
}
|
|
}
|
|
}
|
|
#elif CODING == CODING_DIRECT
|
|
if (v.stat.ptr == 0)
|
|
{
|
|
if (length < 7)
|
|
{
|
|
length <<= 5;
|
|
v.stat.ptr = ptr;
|
|
short_len:
|
|
ptr[0] = (uchar) length;
|
|
ptr[1] = (uchar) offset;
|
|
ptr += 2;
|
|
if (offset > 255)
|
|
{
|
|
ptr[0] = (uchar) (offset >>= 8);
|
|
ptr[-2] = (uchar) (length += 16);
|
|
ptr += 1;
|
|
}
|
|
}
|
|
else if (length < 15 + 7)
|
|
{
|
|
length += (14 << 4) - 7;
|
|
goto short_len;
|
|
}
|
|
else
|
|
{
|
|
if (offset > 255)
|
|
ptr[0] = 0xff;
|
|
else
|
|
ptr[0] = 0xef;
|
|
long_len:
|
|
ptr[1] = (uchar) (length -= (7 + 15));
|
|
ptr += 2;
|
|
if (length >= 255)
|
|
{
|
|
length += 7 + 15;
|
|
ptr[-1] = 255;
|
|
ptr[0] = (uchar) (length);
|
|
ptr[1] = (uchar) (length >>= 8);
|
|
ptr += 2;
|
|
}
|
|
*ptr++ = (uchar) offset;
|
|
if (offset > 255)
|
|
*ptr++ = (uchar) (offset >>= 8);
|
|
}
|
|
}
|
|
else
|
|
{
|
|
if (length < 7)
|
|
{
|
|
length |= v.stat.ptr[0];
|
|
*ptr++ = (uchar) offset;
|
|
if (offset > 255)
|
|
{
|
|
*ptr++ = (uchar) (offset >>= 8);
|
|
length |= 8;
|
|
}
|
|
v.stat.ptr[0] = (uchar) length;
|
|
v.stat.ptr = 0;
|
|
}
|
|
else if (length < 15 + 7)
|
|
{
|
|
length -= 7;
|
|
ptr[1] = (uchar) offset;
|
|
ptr[0] = (uchar) (length <<= 4);
|
|
if (offset > 255)
|
|
{
|
|
v.stat.ptr[0] |= 15;
|
|
v.stat.ptr = ptr;
|
|
ptr[2] = (uchar) (offset >>= 8);
|
|
ptr += 3;
|
|
}
|
|
else
|
|
{
|
|
v.stat.ptr[0] |= 7;
|
|
v.stat.ptr = ptr;
|
|
ptr += 2;
|
|
}
|
|
}
|
|
else
|
|
{
|
|
if (offset > 255)
|
|
v.stat.ptr[0] |= 15;
|
|
else
|
|
v.stat.ptr[0] |= 7;
|
|
v.stat.ptr = ptr;
|
|
ptr[0] = 15 << 4;
|
|
goto long_len;
|
|
}
|
|
}
|
|
#endif /* CODING */
|
|
|
|
tag_write (p, ptr, 1);
|
|
|
|
return (ptr);
|
|
}
|
|
|
|
|
|
#elif CODING & (CODING_HUFF_LEN | CODING_HUFF_PTR | CODING_HUFF_ALL)
|
|
|
|
#if CODING == CODING_HUFF_ALL
|
|
#define write_lit(p,ptr,ch) do { \
|
|
++v.stat.freq[*ptr++ = (ch)]; \
|
|
tag_write (p, ptr, 0); \
|
|
} while (0)
|
|
#else
|
|
#define write_lit(p,ptr,ch) do { \
|
|
*ptr++ = (ch); \
|
|
tag_write (p, ptr, 0); \
|
|
} while (0)
|
|
#endif
|
|
|
|
#define BIOWR(mask,bits) { \
|
|
assert (((mask) >> (bits)) == 0); \
|
|
if ((Bits -= (bits)) < 0) \
|
|
{ \
|
|
*(__unaligned bitmask2 *)Ptr1 = (bitmask2) ((Mask << (Bits + (bits))) \
|
|
+ ((mask) >> (-Bits))); \
|
|
Mask = (mask); \
|
|
Bits += sizeof (ubitmask2) * 8; \
|
|
Ptr1 = Ptr2; \
|
|
Ptr2 = (ubitmask2 *) ptr; \
|
|
ptr += sizeof (ubitmask2); \
|
|
} \
|
|
else \
|
|
Mask = (Mask << (bits)) + (mask); \
|
|
}
|
|
|
|
|
|
#if CODING == CODING_HUFF_LEN
|
|
|
|
INLINE uchar *write_ptr (prs *p, uchar *ptr, int offset, int length)
|
|
{
|
|
xint k;
|
|
|
|
++v.stat.pointers;
|
|
|
|
length -= MIN_MATCH;
|
|
--offset;
|
|
|
|
k = 0; if (offset > 255) k = 1;
|
|
|
|
if (length < MAX_LENGTH - 1)
|
|
++v.stat.freq[*ptr++ = (uchar) (k |= length << 1)];
|
|
else
|
|
{
|
|
length -= MAX_LENGTH - 1;
|
|
++v.stat.freq[ptr[0] = (uchar) (k |= (MAX_LENGTH - 1) << 1)];
|
|
ptr[1] = (uchar) length;
|
|
ptr += 2;
|
|
if (length >= 255)
|
|
{
|
|
ptr[-1] = 255;
|
|
length += MAX_LENGTH - 1;
|
|
ptr[0] = (uchar) length;
|
|
ptr[1] = (uchar) (length >>= 8);
|
|
ptr += 2;
|
|
}
|
|
}
|
|
|
|
*ptr++ = (uchar) offset;
|
|
if (offset > 255)
|
|
*ptr++ = (uchar) (offset >>= 8);
|
|
|
|
tag_write (p, ptr, 1);
|
|
|
|
return (ptr);
|
|
}
|
|
|
|
static void encode_pass2 (prs *p)
|
|
{
|
|
xint Bits;
|
|
ubitmask4 Mask;
|
|
ubitmask2 *Ptr1, *Ptr2;
|
|
tag_t bmask;
|
|
uchar *src = v.temp.beg;
|
|
uchar *ptr = v.comp.ptr;
|
|
uxint k;
|
|
|
|
Ptr1 = (ubitmask2 *) ptr; ptr += sizeof (ubitmask2);
|
|
Ptr2 = (ubitmask2 *) ptr; ptr += sizeof (ubitmask2);
|
|
Mask = 0;
|
|
Bits = 8 * sizeof (ubitmask2);
|
|
bmask = 0;
|
|
goto start;
|
|
|
|
next:
|
|
if (bmask >= 0)
|
|
{
|
|
bmask <<= 1;
|
|
copy_byte:
|
|
*ptr++ = *src++;
|
|
goto next;
|
|
}
|
|
|
|
if ((bmask <<= 1) == 0)
|
|
{
|
|
start:
|
|
*(__unaligned tag_t *)ptr = bmask = *(__unaligned tag_t *)src;
|
|
src += sizeof (tag_t);
|
|
ptr += sizeof (tag_t);
|
|
if (bmask >= 0)
|
|
{
|
|
bmask = (bmask << 1) + 1;
|
|
goto copy_byte;
|
|
}
|
|
bmask = (bmask << 1) + 1;
|
|
}
|
|
|
|
if (src >= v.temp.ptr)
|
|
goto done;
|
|
|
|
k = *src++;
|
|
assert (k < HUFF_SIZE);
|
|
BIOWR (v.stat.mask[k], v.stat.bits[k]);
|
|
|
|
if (k >= ((MAX_LENGTH - 1) << 1))
|
|
{
|
|
if ((*ptr++ = *src++) == 255)
|
|
{
|
|
ptr[0] = src[0];
|
|
ptr[1] = src[1];
|
|
src += 2;
|
|
ptr += 2;
|
|
}
|
|
}
|
|
|
|
*ptr++ = *src++;
|
|
if (k & 1) *ptr++ = *src++;
|
|
goto next;
|
|
|
|
done:
|
|
*Ptr1 = (ubitmask2) (Mask <<= Bits);
|
|
*Ptr2 = 0;
|
|
v.comp.ptr = ptr;
|
|
assert (src == v.temp.ptr);
|
|
}
|
|
|
|
#elif CODING & (CODING_HUFF_PTR | CODING_HUFF_ALL)
|
|
|
|
#define MAX_BITNO_LOG 8
|
|
#define MAX_BITNO (1 << MAX_BITNO_LOG)
|
|
|
|
static uchar bitno_table[MAX_BITNO];
|
|
static int bitno_table_initialized = 0;
|
|
|
|
static void bitno_init (void)
|
|
{
|
|
int i, k, n;
|
|
if (bitno_table_initialized)
|
|
return;
|
|
bitno_table[0] = 255;
|
|
for (i = 0; i < MAX_BITNO_LOG; ++i)
|
|
{
|
|
for (n = (k = 1<<i) << 1; k < n; ++k)
|
|
bitno_table[k] = (uchar) i;
|
|
}
|
|
bitno_table_initialized = 1;
|
|
}
|
|
|
|
static int bitno (uxint n)
|
|
{
|
|
assert (n != 0 && (n >> (MAX_BITNO_LOG*2)) == 0);
|
|
if (n >= MAX_BITNO)
|
|
return (bitno_table[n >> MAX_BITNO_LOG] + MAX_BITNO_LOG);
|
|
return (bitno_table[n]);
|
|
}
|
|
|
|
#if CODING == CODING_HUFF_ALL
|
|
#define CODING_ADJUST(n) (256 + (n))
|
|
#else
|
|
#define CODING_ADJUST(n) (n)
|
|
#endif
|
|
|
|
INLINE uchar *write_ptr (prs *p, uchar *ptr, int offset, int length)
|
|
{
|
|
xint k;
|
|
|
|
k = bitno (offset);
|
|
length -= MIN_MATCH;
|
|
offset ^= 1 << k;
|
|
|
|
v.stat.pointers += 2;
|
|
v.stat.extra += k;
|
|
|
|
k <<= MAX_LENGTH_LOG;
|
|
if (length < MAX_LENGTH - 1)
|
|
{
|
|
k |= length;
|
|
*ptr++ = (uchar) k;
|
|
++v.stat.freq[CODING_ADJUST (k)];
|
|
}
|
|
else
|
|
{
|
|
k |= MAX_LENGTH - 1;
|
|
length -= MAX_LENGTH - 1;
|
|
ptr[0] = (uchar) k;
|
|
++v.stat.freq[CODING_ADJUST (k)];
|
|
ptr[1] = (uchar) length;
|
|
ptr += 2;
|
|
if (length >= 255)
|
|
{
|
|
length += MAX_LENGTH - 1;
|
|
ptr[-1] = 255;
|
|
ptr[0] = (uchar) length;
|
|
ptr[1] = (uchar) (length >>= 8);
|
|
ptr += 2;
|
|
}
|
|
}
|
|
|
|
*ptr++ = (uchar) offset;
|
|
if (k >= (9 << MAX_LENGTH_LOG))
|
|
{
|
|
v.stat.pointers += 1;
|
|
*ptr++ = (uchar) (offset >>= 8);
|
|
}
|
|
|
|
tag_write (p, ptr, 1);
|
|
|
|
return (ptr);
|
|
}
|
|
|
|
static void encode_pass2 (prs *p)
|
|
{
|
|
xint Bits;
|
|
uxint Mask;
|
|
ubitmask2 *Ptr1, *Ptr2;
|
|
tag_t bmask;
|
|
uchar *src = v.temp.beg;
|
|
uchar *ptr = v.comp.ptr;
|
|
uxint k;
|
|
|
|
Ptr1 = (ubitmask2 *) ptr; ptr += sizeof (ubitmask2);
|
|
Ptr2 = (ubitmask2 *) ptr; ptr += sizeof (ubitmask2);
|
|
Mask = 0;
|
|
Bits = 8 * sizeof (ubitmask2);
|
|
bmask = 0;
|
|
goto start;
|
|
|
|
next:
|
|
if (bmask >= 0)
|
|
{
|
|
bmask <<= 1;
|
|
copy_byte:
|
|
#if CODING == CODING_HUFF_ALL
|
|
k = *src++;
|
|
BIOWR (v.stat.mask[k], v.stat.bits[k]);
|
|
#elif CODING == CODING_HUFF_PTR
|
|
*ptr++ = *src++;
|
|
#endif
|
|
goto next;
|
|
}
|
|
|
|
if ((bmask <<= 1) == 0)
|
|
{
|
|
start:
|
|
bmask = *(__unaligned tag_t *)src;
|
|
src += sizeof (tag_t);
|
|
#if CODING == CODING_HUFF_PTR
|
|
*(__unaligned tag_t *)ptr = bmask;
|
|
ptr += sizeof (tag_t);
|
|
#endif
|
|
if (bmask >= 0)
|
|
{
|
|
bmask = (bmask << 1) + 1;
|
|
goto copy_byte;
|
|
}
|
|
bmask = (bmask << 1) + 1;
|
|
}
|
|
|
|
if (src >= v.temp.ptr)
|
|
goto done;
|
|
|
|
k = *src++;
|
|
assert (k < HUFF_SIZE);
|
|
BIOWR (v.stat.mask[CODING_ADJUST (k)], v.stat.bits[CODING_ADJUST (k)]);
|
|
|
|
if ((k & (MAX_LENGTH - 1)) == MAX_LENGTH - 1)
|
|
{
|
|
if ((*ptr++ = *src++) == 255)
|
|
{
|
|
ptr[0] = src[0];
|
|
ptr[1] = src[1];
|
|
src += 2;
|
|
ptr += 2;
|
|
}
|
|
}
|
|
|
|
k >>= MAX_LENGTH_LOG;
|
|
{
|
|
uxint m = *src++;
|
|
if (k > 8)
|
|
m += *src++ << 8;
|
|
BIOWR (m, k);
|
|
}
|
|
goto next;
|
|
|
|
done:
|
|
#if CODING == CODING_HUFF_ALL
|
|
BIOWR (v.stat.mask[CODING_ADJUST(0)], v.stat.bits[CODING_ADJUST(0)]);
|
|
#endif
|
|
*Ptr1 = (ubitmask2) (Mask <<= Bits);
|
|
*Ptr2 = 0;
|
|
v.comp.ptr = ptr;
|
|
assert (src == v.temp.ptr);
|
|
}
|
|
|
|
#endif
|
|
|
|
|
|
/* ------------------ Create canonical Huffman code ------------------- */
|
|
/* ----------------------------- */
|
|
|
|
#define MAX_ALPHABET HUFF_SIZE
|
|
static void huffman_create_codes (huff_info *info, uxint *freq, xint n, uxint *mask, uchar *length, uxint maxbits, uchar *encoded, uxint *total)
|
|
{
|
|
huff_node
|
|
*p, *q, *r,
|
|
*first_sorted, *first_free;
|
|
xint i, k;
|
|
|
|
assert ((uxint) (n-1) <= (MAX_ALPHABET-1));
|
|
|
|
/* honestly it is easy enough to create Huffman code in-place */
|
|
/* but the use of explicit data structures makes code simpler */
|
|
|
|
/* clean everything up */
|
|
memset (length, 0, sizeof (length[0]) * n);
|
|
memset (encoded, 0, (n + 1) >> 1);
|
|
|
|
if (mask != 0 && mask != freq)
|
|
memset (mask, 0, sizeof (mask[0]) * n);
|
|
|
|
/* store frequencies */
|
|
p = info->buff;
|
|
for (i = 0; i < n; ++i)
|
|
{
|
|
if ((p->freq = freq[i]) != 0)
|
|
{
|
|
p->son[0] = p+1; p->son[1] = 0;
|
|
p->ch = (uint16) i;
|
|
++p;
|
|
}
|
|
}
|
|
|
|
/* handle simple case */
|
|
*total = 0;
|
|
if (p <= info->buff + 1)
|
|
{
|
|
if (p == info->buff) /* if no symbols do nothing */
|
|
return;
|
|
i = p[-1].ch; /* single symbol code */
|
|
mask[i] = 0;
|
|
encoded[i >> 1] = 0x11; /* two symbols has 1-bit length */
|
|
return;
|
|
}
|
|
|
|
first_free = p; /* store location of first unused node */
|
|
|
|
p[-1].son[0] = 0; /* terminate the list */
|
|
/* radix sort the list by frequency */
|
|
p = info->buff; /* head of the list */
|
|
/* initialize */
|
|
for (n = 0; n < 256; ++n)
|
|
*(info->link[n] = info->head + n) = 0;
|
|
for (i = 0; i < (BUFF_SIZE_LOG <= 16 ? 16 : 32); i += 8)
|
|
{
|
|
/* link node to the end of respective bucket */
|
|
do
|
|
{
|
|
n = (p->freq >> i) & 0xff;
|
|
info->link[n][0] = p; info->link[n] = p->son;
|
|
}
|
|
while ((p = p->son[0]) != 0);
|
|
|
|
/* merge buckets into single list */
|
|
n = 0;
|
|
while (info->head[n] == 0) ++n;
|
|
p = info->head[n]; info->head[k = n] = 0;
|
|
while (++n < 256)
|
|
{
|
|
if (info->head[n] == 0) continue;
|
|
info->link[k][0] = info->head[n]; info->link[k] = info->head + k; info->head[n] = 0;
|
|
k = n;
|
|
}
|
|
info->link[k][0] = 0; info->link[k] = info->head + k;
|
|
}
|
|
first_sorted = p; /* store head of sorted symbol's list */
|
|
|
|
restart:
|
|
assert (p == first_sorted);
|
|
q = first_free;
|
|
r = q - 1;
|
|
while (p != 0 || q != r)
|
|
{
|
|
++r;
|
|
|
|
/* select left subtree */
|
|
assert (q <= r && (p != 0 || q != r));
|
|
if (p == 0 || (q != r && p->freq > q->freq))
|
|
{
|
|
r->son[0] = q; r->freq = q->freq; ++q;
|
|
}
|
|
else
|
|
{
|
|
r->son[0] = p; r->freq = p->freq; p = p->son[0];
|
|
}
|
|
|
|
/* select right subtree */
|
|
assert (q <= r && (p != 0 || q != r));
|
|
if (p == 0 || (q != r && p->freq > q->freq))
|
|
{
|
|
r->son[1] = q; r->freq += q->freq; ++q;
|
|
}
|
|
else
|
|
{
|
|
r->son[1] = p; r->freq += p->freq; p = p->son[0];
|
|
}
|
|
}
|
|
|
|
/* evaluate codewords' length */
|
|
i = -1; /* stack pointer */
|
|
n = 0; /* current tree depth */
|
|
p = r; /* current subtree root */
|
|
for (;;)
|
|
{
|
|
while (p->son[1] != 0)
|
|
{
|
|
/* put right son into stack and set up its depth */
|
|
(info->head[++i] = p->son[1])->bits = (uint16) (++n);
|
|
(p = p->son[0])->bits = (uint16) n;
|
|
}
|
|
length[p->ch] = (uchar) n;
|
|
|
|
if (i < 0) break; /* nothing's in stack */
|
|
n = (p = info->head[i--])->bits;
|
|
}
|
|
|
|
p = first_sorted;
|
|
#if DEBUG
|
|
for (q = p; (r = q->son[0]) != 0; q = r)
|
|
assert (q->bits >= r->bits);
|
|
#endif
|
|
if (p->bits > maxbits)
|
|
{
|
|
assert (p == first_sorted);
|
|
q = p;
|
|
do
|
|
q->freq = (q->freq + 1) >> 1;
|
|
while ((q = q->son[0]) != 0);
|
|
goto restart;
|
|
}
|
|
|
|
/* now sort symbols in a stable way by increasing codeword length */
|
|
/* initialize */
|
|
memset (info->head, 0, sizeof (info->head[0]) * 32);
|
|
for (n = 0; n < 32; ++n)
|
|
info->link[n] = info->head + n;
|
|
|
|
/* link node to the end of respective bucket */
|
|
p = info->buff;
|
|
do
|
|
{
|
|
n = p->bits;
|
|
info->link[n][0] = p; info->link[n] = p->son;
|
|
}
|
|
while (++p != first_free);
|
|
|
|
/* merge buckets into single list */
|
|
n = 0;
|
|
while (info->head[n] == 0) ++n;
|
|
p = info->head[n]; k = n;
|
|
while (++n < 32)
|
|
{
|
|
if (info->head[n] == 0) continue;
|
|
info->link[k][0] = info->head[n];
|
|
k = n;
|
|
}
|
|
info->link[k][0] = 0;
|
|
|
|
#if DEBUG
|
|
for (q = p; (r = q->son[0]) != 0; q = r)
|
|
assert (r->bits > q->bits || (r->bits == q->bits && r->ch > q->ch));
|
|
#endif
|
|
|
|
/* set up code masks */
|
|
if (mask == freq)
|
|
memset (mask, 0, sizeof (mask[0]) * n);
|
|
|
|
n = 0; /* mask */
|
|
i = 1; /* bit length */
|
|
k = 1; /* first index */
|
|
do
|
|
{
|
|
/* sum a[i] * b[i] may be evaluated without multiplications */
|
|
/* using O(B) memory and O(N+B) time if 0 <= b[i] < B */
|
|
*total += freq[p->ch] * p->bits;
|
|
encoded[p->ch >> 1] |= p->bits << (p->ch & 1 ? 4 : 0);
|
|
mask[p->ch] = (n <<= p->bits - i);
|
|
i = p->bits;
|
|
++n;
|
|
}
|
|
while ((p = p->son[0]) != 0);
|
|
}
|
|
|
|
#endif /* CODING */
|
|
|
|
#define CHAIN 0
|
|
#define encode_pass1 encode0_pass1
|
|
#include "xencode.i"
|
|
|
|
#if MAX_CHAIN >= 1
|
|
#define CHAIN 1
|
|
#define encode_pass1 encode1_pass1
|
|
#define find_match find_match1
|
|
#include "xencode.i"
|
|
#endif
|
|
|
|
#if MAX_CHAIN >= 2
|
|
#define CHAIN 2
|
|
#define encode_pass1 encode2_pass1
|
|
#define find_match find_match2
|
|
#include "xencode.i"
|
|
#endif
|
|
|
|
#if MAX_CHAIN >= 3
|
|
#define CHAIN 3
|
|
#define encode_pass1 encodeN_pass1
|
|
#define find_match find_matchN
|
|
#include "xencode.i"
|
|
#endif
|
|
|
|
typedef void encode_pass1_proc (prs *p);
|
|
|
|
static void encode_pass1_progress (
|
|
prs *p,
|
|
encode_pass1_proc *encode_pass1,
|
|
XpressProgressFn *ProgressFn, // NULL or progress callback
|
|
void *ProgressContext, // user-defined context that will be passed to ProgressFn
|
|
int ProgressSize // call ProgressFn each time ProgressSize bytes processed
|
|
)
|
|
{
|
|
xint stop;
|
|
|
|
if (ProgressFn == 0)
|
|
{
|
|
encode_pass1 (p);
|
|
return;
|
|
}
|
|
|
|
stop = v.orig.stop;
|
|
for (;;)
|
|
{
|
|
if (v.orig.pos - v.orig.progress >= ProgressSize)
|
|
{
|
|
#if CODING & (CODING_HUFF_LEN | CODING_HUFF_PTR | CODING_HUFF_ALL)
|
|
ProgressFn (ProgressContext, v.orig.ptr);
|
|
#else
|
|
ProgressFn (ProgressContext, (v.orig.pos * 15) >> 4);
|
|
#endif
|
|
v.orig.progress = v.orig.pos;
|
|
}
|
|
|
|
v.orig.stop = stop;
|
|
if (v.orig.pos >= stop)
|
|
break;
|
|
|
|
if (stop - v.orig.progress > ProgressSize)
|
|
v.orig.stop = v.orig.progress + ProgressSize;
|
|
assert (v.orig.stop > v.orig.pos);
|
|
|
|
encode_pass1 (p);
|
|
}
|
|
}
|
|
|
|
|
|
int XPRESS_CALL XpressEncode
|
|
(
|
|
XpressEncodeStream stream,
|
|
void *comp,
|
|
int comp_size,
|
|
const void *orig,
|
|
int orig_size,
|
|
XpressProgressFn *ProgressFn, // NULL or progress callback
|
|
void *ProgressContext, // user-defined context that will be passed to ProgressFn
|
|
int ProgressSize // call ProgressFn each time ProgressSize bytes processed
|
|
)
|
|
{
|
|
#if CODING & (CODING_HUFF_LEN | CODING_HUFF_PTR | CODING_HUFF_ALL)
|
|
uchar huff_buff [HUFF_SIZE >> 1];
|
|
uxint huff_total;
|
|
#endif
|
|
uxint c_size;
|
|
prs *p;
|
|
xpress_info *info = (xpress_info *) stream;
|
|
encode_pass1_proc *encode_pass1;
|
|
|
|
if (info == 0 || info->magic != MAGIC_ENCODE)
|
|
return (0);
|
|
|
|
if ((unsigned) (orig_size-1) > info->max_size
|
|
|| orig_size <= MIN_SIZE
|
|
|| comp_size < MIN_SIZE
|
|
|| comp == 0
|
|
|| orig == 0)
|
|
{
|
|
return (orig_size);
|
|
}
|
|
|
|
p = info->p;
|
|
memset (&v, 0, sizeof (v));
|
|
v.temp.hash = (z_index_t *) (v.temp.ptr = info->temp);
|
|
v.chain = info->chain;
|
|
|
|
v.orig.end = (v.orig.ptr = orig) + (v.orig.size = v.orig.stop = orig_size);
|
|
v.orig.end_16 = v.orig.end - 16;
|
|
v.orig.end_3 = v.orig.end - MIN_MATCH;
|
|
v.comp.ptr = v.comp.beg = comp;
|
|
v.orig.pos = 0;
|
|
v.temp.beg = v.temp.ptr;
|
|
|
|
encode_pass1 = encode0_pass1;
|
|
if (v.chain <= 0)
|
|
memset ((void *) (&p->x.q_last[0]), 0, Q_HASH_SIZE * sizeof (p->x.q_last[0]));
|
|
#if MAX_CHAIN >= 1
|
|
else if (v.chain == 1)
|
|
{
|
|
encode_pass1 = encode1_pass1;
|
|
memset ((void *) (&p->x.z_next[0]), 0, Z_HASH_SIZE * sizeof (p->x.z_next[0]));
|
|
}
|
|
#if MAX_CHAIN >= 2
|
|
else
|
|
{
|
|
encode_pass1 = encode2_pass1;
|
|
#if MAX_CHAIN >= 3
|
|
if (v.chain >= 3)
|
|
encode_pass1 = encodeN_pass1;
|
|
#endif
|
|
memset (v.temp.hash, 0, sizeof (v.temp.hash[0]) * Z_HASH_SIZE);
|
|
z_hash_insert (p);
|
|
}
|
|
#endif
|
|
#endif
|
|
|
|
if (ProgressSize <= 0 || ProgressSize > orig_size)
|
|
ProgressSize = orig_size;
|
|
|
|
if (ProgressFn != 0)
|
|
ProgressFn (ProgressContext, v.orig.progress = 0);
|
|
|
|
#if CODING & (CODING_DIRECT | CODING_DIRECT2 | CODING_BY_BIT)
|
|
v.temp.beg = v.temp.ptr = v.comp.ptr;
|
|
tag_write_start (p);
|
|
for (;;)
|
|
{
|
|
xint rest = comp_size - (xint) (v.temp.ptr - v.comp.beg) - 2 * sizeof (tag_t);
|
|
rest -= (rest + 7) >> 3;
|
|
if (rest <= (xint) (2 * sizeof (tag_t) + 8))
|
|
goto no_compression;
|
|
if (v.orig.pos >= v.orig.size)
|
|
break;
|
|
v.orig.stop = v.orig.pos + rest;
|
|
if (v.orig.stop > v.orig.size - 8)
|
|
{
|
|
v.orig.stop = v.orig.size - 8;
|
|
if (v.orig.pos >= v.orig.stop)
|
|
break;
|
|
}
|
|
encode_pass1_progress (p, encode_pass1, ProgressFn, ProgressContext, ProgressSize);
|
|
}
|
|
#else
|
|
v.orig.stop -= 7;
|
|
tag_write_start (p);
|
|
encode_pass1_progress (p, encode_pass1, ProgressFn, ProgressContext, ProgressSize);
|
|
#endif
|
|
while (v.orig.pos < v.orig.size)
|
|
{
|
|
write_lit (p, v.temp.ptr, v.orig.ptr[v.orig.pos]);
|
|
++v.orig.pos;
|
|
}
|
|
tag_write_finish (p);
|
|
|
|
#if CODING & (CODING_DIRECT | CODING_DIRECT2 | CODING_BY_BIT)
|
|
c_size = (xint) (v.temp.ptr - v.temp.beg);
|
|
#elif CODING & (CODING_HUFF_LEN | CODING_HUFF_PTR | CODING_HUFF_ALL)
|
|
if (v.stat.pointers == 0)
|
|
goto no_compression;
|
|
#if CODING == CODING_HUFF_ALL
|
|
++v.stat.freq[CODING_ADJUST(0)];
|
|
#endif
|
|
huffman_create_codes (
|
|
&v.stat.info,
|
|
v.stat.freq, HUFF_SIZE,
|
|
v.stat.mask, v.stat.bits, 15, huff_buff, &huff_total);
|
|
|
|
c_size = huff_total;
|
|
#if CODING & (CODING_HUFF_PTR | CODING_HUFF_ALL)
|
|
c_size += v.stat.extra;
|
|
#endif
|
|
if (c_size == 0) c_size = 1;
|
|
c_size = (((c_size - 1) & ~(sizeof (bitmask2) * 8 - 1)) >> 3);
|
|
c_size += (int) (v.temp.ptr - v.temp.beg) - v.stat.pointers + 4 + sizeof (huff_buff);
|
|
#if CODING == CODING_HUFF_ALL
|
|
for (huff_total = 0; huff_total < 256; ++huff_total)
|
|
c_size -= v.stat.freq[huff_total];
|
|
c_size -= v.stat.masks * sizeof (tag_t);
|
|
#endif
|
|
#endif /* CODING */
|
|
|
|
if (c_size >= (uxint) comp_size)
|
|
{
|
|
no_compression:
|
|
comp_size = orig_size;
|
|
}
|
|
else
|
|
{
|
|
#if CODING & (CODING_HUFF_LEN | CODING_HUFF_PTR | CODING_HUFF_ALL)
|
|
memcpy (v.comp.ptr, huff_buff, sizeof (huff_buff));
|
|
v.comp.ptr += sizeof (huff_buff);
|
|
encode_pass2 (p);
|
|
#elif CODING & (CODING_BY_BIT | CODING_DIRECT | CODING_DIRECT2)
|
|
v.comp.ptr += c_size;
|
|
#else
|
|
#error Unknown CODING
|
|
#endif
|
|
|
|
comp_size = (int) (v.comp.ptr - v.comp.beg);
|
|
|
|
#if DEBUG
|
|
if (c_size != (uxint) comp_size)
|
|
printf ("error: c_size = %d, comp_size = %d\n", c_size, comp_size);
|
|
#endif
|
|
}
|
|
|
|
if (ProgressFn != 0)
|
|
ProgressFn (ProgressContext, orig_size);
|
|
|
|
return (comp_size);
|
|
}
|
|
|
|
#define ALIGN 32
|
|
|
|
XpressEncodeStream XPRESS_CALL XpressEncodeCreate
|
|
(int orig_size, void *context, XpressAllocFn *AllocFn, int chain)
|
|
{
|
|
xpress_info *info;
|
|
prs *p;
|
|
uchar *b;
|
|
int temp_size;
|
|
int alloc_size;
|
|
|
|
if (AllocFn == 0 || (unsigned) (orig_size-1) > (BUFF_SIZE-1))
|
|
return (0);
|
|
|
|
#if CODING & (CODING_DIRECT | CODING_DIRECT2 | CODING_BY_BIT)
|
|
temp_size = 0;
|
|
#else
|
|
temp_size = orig_size + ((orig_size + 7) >> 3);
|
|
#endif
|
|
|
|
alloc_size = sizeof (p->x.q_last[0]) * Q_HASH_SIZE;
|
|
#if MAX_CHAIN <= 0
|
|
chain = 0;
|
|
#else
|
|
if (chain > MAX_CHAIN)
|
|
chain = MAX_CHAIN;
|
|
if (chain >= 1)
|
|
{
|
|
alloc_size = sizeof (p->x.z_next[0]) * Z_HASH_SIZE;
|
|
#if MAX_CHAIN >= 2
|
|
if (chain >= 2)
|
|
{
|
|
alloc_size = sizeof (p->x.z_next[0]) * orig_size;
|
|
if (temp_size < sizeof (v.temp.hash[0]) * Z_HASH_SIZE)
|
|
temp_size = sizeof (v.temp.hash[0]) * Z_HASH_SIZE;
|
|
}
|
|
#endif
|
|
}
|
|
#endif
|
|
|
|
temp_size = (temp_size + 256 + sizeof (*p)) & ~(ALIGN-1);
|
|
alloc_size += temp_size + sizeof (*info) + ALIGN;
|
|
|
|
b = AllocFn (context, alloc_size);
|
|
if (b == 0)
|
|
return (0);
|
|
|
|
info = (xpress_info *) b;
|
|
info->max_size = orig_size - 1;
|
|
info->chain = chain;
|
|
info->magic = MAGIC_ENCODE;
|
|
info->memory = b;
|
|
|
|
b = (uchar *) (info + 1);
|
|
b += ALIGN - (((__int64) b) & (ALIGN-1));
|
|
info->p = p = ((prs *) (b + temp_size));
|
|
info->temp = b;
|
|
|
|
#if MAX_CHAIN >= 1
|
|
z_hash_map_init ();
|
|
#endif
|
|
|
|
#if CODING & (CODING_HUFF_PTR | CODING_HUFF_ALL)
|
|
bitno_init ();
|
|
#endif
|
|
|
|
return ((XpressEncodeStream) info);
|
|
}
|
|
|
|
void XPRESS_CALL XpressEncodeClose
|
|
(XpressEncodeStream stream, void *context, XpressFreeFn *FreeFn)
|
|
{
|
|
xpress_info *info = (xpress_info *) stream;
|
|
if (info != 0 && FreeFn != 0 && info->magic == MAGIC_ENCODE)
|
|
{
|
|
info->magic = 0;
|
|
FreeFn (context, info->memory);
|
|
}
|
|
}
|