encode_one_block (working_state *state, JCOEFPTR block, int last_dc_val,
c_derived_tbl *dctbl, c_derived_tbl *actbl)
{
int temp, temp2, temp3;
int nbits;
int r, code, size;
JOCTET _buffer[BUFSIZE], *buffer;
size_t put_buffer; int put_bits;
int code_0xf0 = actbl->ehufco[0xf0], size_0xf0 = actbl->ehufsi[0xf0];
size_t bytes, bytestocopy; int localbuf = 0;
put_buffer = state->cur.put_buffer;
put_bits = state->cur.put_bits;
LOAD_BUFFER()
/* Encode the DC coefficient difference per section F.1.2.1 */
temp = temp2 = block[0] - last_dc_val;
/* This is a well-known technique for obtaining the absolute value without a
* branch. It is derived from an assembly language technique presented in
* "How to Optimize for the Pentium Processors", Copyright (c) 1996, 1997 by
* Agner Fog.
*/
temp3 = temp >> (CHAR_BIT * sizeof(int) - 1);
temp ^= temp3;
temp -= temp3;
/* For a negative input, want temp2 = bitwise complement of abs(input) */
/* This code assumes we are on a two's complement machine */
temp2 += temp3;
/* Find the number of bits needed for the magnitude of the coefficient */
nbits = JPEG_NBITS(temp);
/* Emit the Huffman-coded symbol for the number of bits */
code = dctbl->ehufco[nbits];
size = dctbl->ehufsi[nbits];
EMIT_BITS(code, size)
/* Mask off any extra bits in code */
temp2 &= (((JLONG) 1)<<nbits) - 1;
/* Emit that number of bits of the value, if positive, */
/* or the complement of its magnitude, if negative. */
EMIT_BITS(temp2, nbits)
/* Encode the AC coefficients per section F.1.2.2 */
r = 0; /* r = run length of zeros */
/* Manually unroll the k loop to eliminate the counter variable. This
* improves performance greatly on systems with a limited number of
* registers (such as x86.)
*/
#define kloop(jpeg_natural_order_of_k) { \
if ((temp = block[jpeg_natural_order_of_k]) == 0) { \
r++; \
} else { \
temp2 = temp; \
/* Branch-less absolute value, bitwise complement, etc., same as above */ \
temp3 = temp >> (CHAR_BIT * sizeof(int) - 1); \
temp ^= temp3; \
temp -= temp3; \
temp2 += temp3; \
nbits = JPEG_NBITS_NONZERO(temp); \
/* if run length > 15, must emit special run-length-16 codes (0xF0) */ \
while (r > 15) { \
EMIT_BITS(code_0xf0, size_0xf0) \
r -= 16; \
} \
/* Emit Huffman symbol for run length / number of bits */ \
temp3 = (r << 4) + nbits; \
code = actbl->ehufco[temp3]; \
size = actbl->ehufsi[temp3]; \
EMIT_CODE(code, size) \
r = 0; \
} \
}
/* One iteration for each value in jpeg_natural_order[] */
kloop(1); kloop(8); kloop(16); kloop(9); kloop(2); kloop(3);
kloop(10); kloop(17); kloop(24); kloop(32); kloop(25); kloop(18);
kloop(11); kloop(4); kloop(5); kloop(12); kloop(19); kloop(26);
kloop(33); kloop(40); kloop(48); kloop(41); kloop(34); kloop(27);
kloop(20); kloop(13); kloop(6); kloop(7); kloop(14); kloop(21);
kloop(28); kloop(35); kloop(42); kloop(49); kloop(56); kloop(57);
kloop(50); kloop(43); kloop(36); kloop(29); kloop(22); kloop(15);
kloop(23); kloop(30); kloop(37); kloop(44); kloop(51); kloop(58);
kloop(59); kloop(52); kloop(45); kloop(38); kloop(31); kloop(39);
kloop(46); kloop(53); kloop(60); kloop(61); kloop(54); kloop(47);
kloop(55); kloop(62); kloop(63);
/* If the last coef(s) were zero, emit an end-of-block code */
if (r > 0) {
code = actbl->ehufco[0];
size = actbl->ehufsi[0];
EMIT_BITS(code, size)
}
encode_mcu_DC_first(j_compress_ptr cinfo, JBLOCKROW *MCU_data)
{
phuff_entropy_ptr entropy = (phuff_entropy_ptr)cinfo->entropy;
register int temp, temp2, temp3;
register int nbits;
int blkn, ci;
int Al = cinfo->Al;
JBLOCKROW block;
jpeg_component_info *compptr;
ISHIFT_TEMPS
entropy->next_output_byte = cinfo->dest->next_output_byte;
entropy->free_in_buffer = cinfo->dest->free_in_buffer;
/* Emit restart marker if needed */
if (cinfo->restart_interval)
if (entropy->restarts_to_go == 0)
emit_restart(entropy, entropy->next_restart_num);
/* Encode the MCU data blocks */
for (blkn = 0; blkn < cinfo->blocks_in_MCU; blkn++) {
block = MCU_data[blkn];
ci = cinfo->MCU_membership[blkn];
compptr = cinfo->cur_comp_info[ci];
/* Compute the DC value after the required point transform by Al.
* This is simply an arithmetic right shift.
*/
temp2 = IRIGHT_SHIFT((int)((*block)[0]), Al);
/* DC differences are figured on the point-transformed values. */
temp = temp2 - entropy->last_dc_val[ci];
entropy->last_dc_val[ci] = temp2;
/* Encode the DC coefficient difference per section G.1.2.1 */
/* This is a well-known technique for obtaining the absolute value without
* a branch. It is derived from an assembly language technique presented
* in "How to Optimize for the Pentium Processors", Copyright (c) 1996,
* 1997 by Agner Fog.
*/
temp3 = temp >> (CHAR_BIT * sizeof(int) - 1);
temp ^= temp3;
temp -= temp3; /* temp is abs value of input */
/* For a negative input, want temp2 = bitwise complement of abs(input) */
temp2 = temp ^ temp3;
/* Find the number of bits needed for the magnitude of the coefficient */
nbits = JPEG_NBITS(temp);
/* Check for out-of-range coefficient values.
* Since we're encoding a difference, the range limit is twice as much.
*/
if (nbits > MAX_COEF_BITS + 1)
ERREXIT(cinfo, JERR_BAD_DCT_COEF);
/* Count/emit the Huffman-coded symbol for the number of bits */
emit_symbol(entropy, compptr->dc_tbl_no, nbits);
/* Emit that number of bits of the value, if positive, */
/* or the complement of its magnitude, if negative. */
if (nbits) /* emit_bits rejects calls with size 0 */
emit_bits(entropy, (unsigned int)temp2, nbits);
}
cinfo->dest->next_output_byte = entropy->next_output_byte;
cinfo->dest->free_in_buffer = entropy->free_in_buffer;
/* Update restart-interval state too */
if (cinfo->restart_interval) {
if (entropy->restarts_to_go == 0) {
entropy->restarts_to_go = cinfo->restart_interval;
entropy->next_restart_num++;
entropy->next_restart_num &= 7;
}
entropy->restarts_to_go--;
}
return TRUE;
}
