/* MP3 bitstream Output interface for LAME
* Copyright 1999-2000 Mark Taylor
* Copyright 1999-2002 Takehiro Tominaga
*/
#ifdef HAVE_CONFIG_H
#include <config.h>
#endif
#include <stdlib.h>
#include <stdio.h>
#include "lame.h"
#include "machine.h"
#include "encoder.h"
#include "util.h"
#include "tables.h"
#include "quantize_pvt.h"
#include "lame_global_flags.h"
#include "gain_analysis.h"
#include "VbrTag.h"
#include "bitstream.h"
#include "tables.h"
/* unsigned int is at least this large: */
/* we work with ints, so when doing bit manipulation, we limit
* ourselves to MAX_LENGTH-2 just to be on the safe side */
#define MAX_LENGTH 32
#ifdef DEBUG
static int hogege;
#endif
int SessionConfig_t::calcFrameLength( int kbps, int pad)const{
return 8 * ((version + 1) * 72000 * kbps / samplerate_out + pad);
}
/***********************************************************************
* compute bitsperframe and mean_bits for a layer III frame
**********************************************************************/
int lame_internal_flags::getframebits()const{
// SessionConfig_t const & cfg = gfc.cfg;
EncResult_t const & eov = ov_enc;
int bit_rate;
/* get bitrate in kbps [?] */
if (eov.bitrate_index)
bit_rate = bitrate_table[cfg.version][eov.bitrate_index];
else
bit_rate = cfg.avg_bitrate;
/*assert(bit_rate <= 550); */
assert(8 <= bit_rate && bit_rate <= 640);
/* main encoding routine toggles padding on and off */
/* one Layer3 Slot consists of 8 bits */
return cfg.calcFrameLength(bit_rate, eov.padding);
}
int SessionConfig_t::get_max_frame_buffer_size_by_constraint(int constraint)const{
int maxmp3buf = 0;
if (avg_bitrate > 320) {
/* in freeformat the buffer is constant */
if (constraint == MDB_STRICT_ISO) {
maxmp3buf = calcFrameLength(avg_bitrate, 0);
}
else {
/* maximum allowed bits per granule are 7680 */
maxmp3buf = 7680 * (version + 1);
}
}
else {
int max_kbps;
if (samplerate_out < 16000) {
max_kbps = bitrate_table[version][8]; /* default: allow 64 kbps (MPEG-2.5) */
}
else {
max_kbps = bitrate_table[version][14];
}
switch (constraint)
{
default:
case MDB_DEFAULT:
/* Bouvigne suggests this more lax interpretation of the ISO doc instead of using 8*960. */
/* All mp3 decoders should have enough buffer to handle this value: size of a 320kbps 32kHz frame */
maxmp3buf = 8 * 1440;
break;
case MDB_STRICT_ISO:
maxmp3buf = calcFrameLength(max_kbps, 0);
break;
case MDB_MAXIMUM:
maxmp3buf = 7680 * (version + 1);
break;
}
}
return maxmp3buf;
}
void lame_internal_flags::putheader_bits() {
// SessionConfig_t const & cfg = gfc.cfg;
EncStateVar_t & esv = sv_enc;
// Bit_stream_struc & bs = gfc.bs;
#ifdef DEBUG
hogege += cfg.sideinfo_len * 8;
#endif
memcpy(bs.buf+bs.buf_byte_idx,esv.header[esv.w_ptr].buf,cfg.sideinfo_len);
bs.buf_byte_idx += cfg.sideinfo_len;
bs.totbit += cfg.sideinfo_len * 8;
esv.w_ptr = (esv.w_ptr + 1) & (MAX_HEADER_BUF - 1);
}
/*write j bits into the bit stream */
void lame_internal_flags::putbits2(int val, int j) {
EncStateVar_t const & esv = sv_enc;
// Bit_stream_struc &bs = gfc.bs;
assert(j < MAX_LENGTH - 2);
while (j > 0) {
int k;
if (bs.buf_bit_idx == 0) {
bs.buf_bit_idx = 8;
bs.buf_byte_idx++;
assert(bs.buf_byte_idx < BUFFER_SIZE);
assert(esv.header[esv.w_ptr].write_timing >= bs.totbit);
if (esv.header[esv.w_ptr].write_timing == bs.totbit) {
putheader_bits();
}
bs.buf[bs.buf_byte_idx] = 0;
}
k = Min(j, bs.buf_bit_idx);
j -= k;
bs.buf_bit_idx -= k;
assert(j < MAX_LENGTH); /* 32 too large on 32 bit machines */
assert(bs.buf_bit_idx < MAX_LENGTH);
bs.buf[bs.buf_byte_idx] |= ((val >> j) << bs.buf_bit_idx);
bs.totbit += k;
}
}
/*write j bits into the bit stream, ignoring frame headers */
void lame_internal_flags::putbits_noheaders(int val, int j) {
// Bit_stream_struc &bs = gfc.bs;
assert(j < MAX_LENGTH - 2);
while (j > 0) {
int k;
if (bs.buf_bit_idx == 0) {
bs.buf_bit_idx = 8;
bs.buf_byte_idx++;
assert(bs.buf_byte_idx < BUFFER_SIZE);
bs.buf[bs.buf_byte_idx] = 0;
}
k = Min(j, bs.buf_bit_idx);
j -= k;
bs.buf_bit_idx -= k;
assert(j < MAX_LENGTH); /* 32 too large on 32 bit machines */
assert(bs.buf_bit_idx < MAX_LENGTH);
bs.buf[bs.buf_byte_idx] |= ((val >> j) << bs.buf_bit_idx);
bs.totbit += k;
}
}
/*
Some combinations of bitrate, Fs, and stereo make it impossible to stuff
out a frame using just main_data, due to the limited number of bits to
indicate main_data_length. In these situations, we put stuffing bits into
the ancillary data...
*/
void lame_internal_flags::drain_into_ancillary(int remainingBits) {
// SessionConfig_t const & cfg = gfc.cfg;
EncStateVar_t & esv = sv_enc;
int i;
assert(remainingBits >= 0);
if (remainingBits >= 8) {
putbits2(0x4c, 8);
remainingBits -= 8;
}
if (remainingBits >= 8) {
putbits2(0x41, 8);
remainingBits -= 8;
}
if (remainingBits >= 8) {
putbits2(0x4d, 8);
remainingBits -= 8;
}
if (remainingBits >= 8) {
putbits2(0x45, 8);
remainingBits -= 8;
}
if (remainingBits >= 32) {
const char *const version = get_lame_short_version();
if (remainingBits >= 32)
for (i = 0; i < (int) strlen(version) && remainingBits >= 8; ++i) {
remainingBits -= 8;
putbits2(version[i], 8);
}
}
for (; remainingBits >= 1; remainingBits -= 1) {
putbits2(esv.ancillary_flag, 1);
esv.ancillary_flag ^= !cfg.disable_reservoir;
}
assert(remainingBits == 0);
}
/*write N bits into the header */
void lame_internal_flags::writeheader(int val, int j) {
EncStateVar_t & esv = sv_enc;
int ptr = esv.header[esv.h_ptr].ptr;
while (j > 0) {
int const k = Min(j, 8 - (ptr & 7));
j -= k;
assert(j < MAX_LENGTH); /* >> 32 too large for 32 bit machines */
esv.header[esv.h_ptr].buf[ptr >> 3]
|= ((val >> j)) << (8 - (ptr & 7) - k);
ptr += k;
}
esv.header[esv.h_ptr].ptr = ptr;
}
static int CRC_update(int value, int crc) {
int i;
value <<= 8;
for (i = 0; i < 8; i++) {
value <<= 1;
crc <<= 1;
if (((crc ^ value) & 0x10000)) crc ^= CRC16_POLYNOMIAL;
}
return crc;
}
void lame_internal_flags::CRC_writeheader(char *header) const{
// SessionConfig_t const & cfg = gfc.cfg;
int crc = 0xffff; /* (jo) init crc16 for error_protection */
int i;
crc = CRC_update(((unsigned char *) header)[2], crc);
crc = CRC_update(((unsigned char *) header)[3], crc);
for (i = 6; i < cfg.sideinfo_len; i++) {
crc = CRC_update(((unsigned char *) header)[i], crc);
}
header[4] = crc >> 8;
header[5] = crc & 255;
}
void lame_internal_flags::encodeSideInfo2(int bitsPerFrame) {
// SessionConfig_t const & cfg = gfc.cfg;
EncResult_t const & eov = ov_enc;
EncStateVar_t & esv = sv_enc;
// III_side_info_t &l3_side=gfc.l3_side;
esv.header[esv.h_ptr].ptr = 0;
memset(esv.header[esv.h_ptr].buf, 0, cfg.sideinfo_len);
if (cfg.samplerate_out < 16000) writeheader(0xffe, 12);
else writeheader(0xfff, 12);
writeheader((cfg.version), 1);
writeheader(4 - 3, 2);
writeheader((!cfg.error_protection), 1);
writeheader((eov.bitrate_index), 4);
writeheader((cfg.samplerate_index), 2);
writeheader((eov.padding), 1);
writeheader((cfg.extension), 1);
writeheader((cfg.mode), 2);
writeheader((eov.mode_ext), 2);
writeheader((cfg.copyright), 1);
writeheader((cfg.original), 1);
writeheader((cfg.emphasis), 2);
if (cfg.error_protection) {
writeheader(0, 16); /* dummy */
}
if (cfg.version == 1) {
/* MPEG1 */
assert(l3_side.main_data_begin >= 0);
writeheader((l3_side.main_data_begin), 9);
if (cfg.channels_out == 2)
writeheader(l3_side.private_bits, 3);
else
writeheader(l3_side.private_bits, 5);
for (int ch = 0; ch < cfg.channels_out; ch++) {
for (int band = 0; band < 4; band++) {
writeheader(l3_side.scfsi[ch][band], 1);
}
}
for (int gr = 0; gr < 2; gr++) {
for (int ch = 0; ch < cfg.channels_out; ch++) {
gr_info & gi = l3_side.tt[gr][ch];
writeheader(gi.part2_3_length + gi.part2_length, 12);
writeheader(gi.big_values / 2, 9);
writeheader(gi.global_gain, 8);
writeheader(gi.scalefac_compress, 4);
if (gi.block_type != NORM_TYPE) {
writeheader(1, 1); /* window_switching_flag */
writeheader(gi.block_type, 2);
writeheader(gi.mixed_block_flag, 1);
if (gi.table_select[0] == 14)
gi.table_select[0] = 16;
writeheader(gi.table_select[0], 5);
if (gi.table_select[1] == 14)
gi.table_select[1] = 16;
writeheader(gi.table_select[1], 5);
writeheader(gi.subblock_gain[0], 3);
writeheader(gi.subblock_gain[1], 3);
writeheader(gi.subblock_gain[2], 3);
}else{
writeheader(0, 1); /* window_switching_flag */
if (gi.table_select[0] == 14)
gi.table_select[0] = 16;
writeheader(gi.table_select[0], 5);
if (gi.table_select[1] == 14)
gi.table_select[1] = 16;
writeheader(gi.table_select[1], 5);
if (gi.table_select[2] == 14)
gi.table_select[2] = 16;
writeheader(gi.table_select[2], 5);
assert(0 <= gi.region0_count && gi.region0_count < 16);
assert(0 <= gi.region1_count && gi.region1_count < 8);
writeheader(gi.region0_count, 4);
writeheader(gi.region1_count, 3);
}
writeheader(gi.preflag, 1);
writeheader(gi.scalefac_scale, 1);
writeheader(gi.count1table_select, 1);
}
}
}else{
/* MPEG2 */
assert(l3_side.main_data_begin >= 0);
writeheader((l3_side.main_data_begin), 8);
writeheader(l3_side.private_bits, cfg.channels_out);
int gr = 0;
for (int ch = 0; ch < cfg.channels_out; ch++) {
gr_info & gi = l3_side.tt[gr][ch];
writeheader(gi.part2_3_length + gi.part2_length, 12);
writeheader(gi.big_values / 2, 9);
writeheader(gi.global_gain, 8);
writeheader(gi.scalefac_compress, 9);
if (gi.block_type != NORM_TYPE) {
writeheader(1, 1); /* window_switching_flag */
writeheader(gi.block_type, 2);
writeheader(gi.mixed_block_flag, 1);
if (gi.table_select[0] == 14)
gi.table_select[0] = 16;
writeheader(gi.table_select[0], 5);
if (gi.table_select[1] == 14)
gi.table_select[1] = 16;
writeheader(gi.table_select[1], 5);
writeheader(gi.subblock_gain[0], 3);
writeheader(gi.subblock_gain[1], 3);
writeheader(gi.subblock_gain[2], 3);
}else{
writeheader(0, 1); /* window_switching_flag */
if (gi.table_select[0] == 14)
gi.table_select[0] = 16;
writeheader(gi.table_select[0], 5);
if (gi.table_select[1] == 14)
gi.table_select[1] = 16;
writeheader(gi.table_select[1], 5);
if (gi.table_select[2] == 14)
gi.table_select[2] = 16;
writeheader(gi.table_select[2], 5);
assert(0 <= gi.region0_count && gi.region0_count < 16);
assert(0 <= gi.region1_count && gi.region1_count < 8);
writeheader(gi.region0_count, 4);
writeheader(gi.region1_count, 3);
}
writeheader(gi.scalefac_scale, 1);
writeheader(gi.count1table_select, 1);
}
}
if (cfg.error_protection) {
/* (jo) error_protection: add crc16 information to header */
CRC_writeheader(esv.header[esv.h_ptr].buf);
}
{
int const old = esv.h_ptr;
assert(esv.header[old].ptr == cfg.sideinfo_len * 8);
esv.h_ptr = (old + 1) & (MAX_HEADER_BUF - 1);
esv.header[esv.h_ptr].write_timing = esv.header[old].write_timing + bitsPerFrame;
if (esv.h_ptr == esv.w_ptr) {
/* yikes! we are out of header buffer space */
ERRORF(*this, "Error: MAX_HEADER_BUF too small in bitstream.c \n");
}
}
}
int lame_internal_flags::huffman_coder_count1(gr_info const & gi) {
/* Write count1 area */
struct huffcodetab const &h = ht[gi.count1table_select + 32];
int i, bits = 0;
#ifdef DEBUG
int gegebo = gfc.bs.totbit;
#endif
int const *ix = &gi.l3_enc[gi.big_values];
FLOAT const *xr = &gi.xr[gi.big_values];
assert(gi.count1table_select < 2);
for (i = (gi.count1 - gi.big_values) / 4; i > 0; --i) {
int huffbits = 0;
int p = 0, v;
v = ix[0];
if (v) {
p += 8;
if (xr[0] < 0.0f)
huffbits++;
assert(v <= 1);
}
v = ix[1];
if (v) {
p += 4;
huffbits *= 2;
if (xr[1] < 0.0f)
huffbits++;
assert(v <= 1);
}
v = ix[2];
if (v) {
p += 2;
huffbits *= 2;
if (xr[2] < 0.0f)
huffbits++;
assert(v <= 1);
}
v = ix[3];
if (v) {
p++;
huffbits *= 2;
if (xr[3] < 0.0f)
huffbits++;
assert(v <= 1);
}
ix += 4;
xr += 4;
putbits2(huffbits + h.table[p], h.hlen[p]);
bits += h.hlen[p];
}
#ifdef DEBUG
DEBUGF(gfc, "count1: real: %ld counted:%d (bigv %d count1len %d)\n",
gfc.bs.totbit - gegebo, gi.count1bits, gi.big_values, gi.count1);
#endif
return bits;
}
/*
Implements the pseudocode of page 98 of the IS
*/
int lame_internal_flags::Huffmancode(const unsigned tableindex,
int start, int end, gr_info const &gi) {
struct huffcodetab const & h = ht[tableindex];
unsigned int const linbits = h.xlen;
int i, bits = 0;
assert(tableindex < 32u);
if (!tableindex)
return bits;
for (i = start; i < end; i += 2) {
int16_t cbits = 0;
uint16_t xbits = 0;
unsigned int xlen = h.xlen;
unsigned int ext = 0;
unsigned int x1 = gi.l3_enc[i];
unsigned int x2 = gi.l3_enc[i + 1];
assert(gi.l3_enc[i] >= 0);
assert(gi.l3_enc[i+1] >= 0);
if (x1 != 0u) {
if (gi.xr[i] < 0.0f)
ext++;
cbits--;
}
if (tableindex > 15u) {
/* use ESC-words */
if (x1 >= 15u) {
uint16_t const linbits_x1 = x1 - 15u;
assert(linbits_x1 <= h.linmax);
ext |= linbits_x1 << 1u;
xbits = linbits;
x1 = 15u;
}
if (x2 >= 15u) {
uint16_t const linbits_x2 = x2 - 15u;
assert(linbits_x2 <= h.linmax);
ext <<= linbits;
ext |= linbits_x2;
xbits += linbits;
x2 = 15u;
}
xlen = 16;
}
if (x2 != 0u) {
ext <<= 1;
if (gi.xr[i + 1] < 0.0f)
ext++;
cbits--;
}
assert((x1 | x2) < 16u);
x1 = x1 * xlen + x2;
xbits -= cbits;
cbits += h.hlen[x1];
assert(cbits <= MAX_LENGTH);
assert(xbits <= MAX_LENGTH);
putbits2(h.table[x1], cbits);
putbits2((int)ext, xbits);
bits += cbits + xbits;
}
return bits;
}
/*
Note the discussion of huffmancodebits() on pages 28
and 29 of the IS, as well as the definitions of the side
information on pages 26 and 27.
*/
int lame_internal_flags::ShortHuffmancodebits(gr_info const &gi) {
int bits;
int region1Start;
region1Start = 3 * scalefac_band.s[3];
if (region1Start > gi.big_values)
region1Start = gi.big_values;
/* short blocks do not have a region2 */
bits = Huffmancode(gi.table_select[0], 0, region1Start, gi);
bits += Huffmancode(gi.table_select[1], region1Start, gi.big_values, gi);
return bits;
}
int lame_internal_flags::LongHuffmancodebits(gr_info const &gi) {
unsigned int i;
int bigvalues, bits;
int region1Start, region2Start;
bigvalues = gi.big_values;
assert(0 <= bigvalues && bigvalues <= 576);
assert(gi.region0_count >= -1);
assert(gi.region1_count >= -1);
i = gi.region0_count + 1;
assert((size_t) i < elemof(scalefac_band.l));
region1Start = scalefac_band.l[i];
i += gi.region1_count + 1;
assert((size_t) i < elemof(scalefac_band.l));
region2Start = scalefac_band.l[i];
if (region1Start > bigvalues)
region1Start = bigvalues;
if (region2Start > bigvalues)
region2Start = bigvalues;
bits = Huffmancode(gi.table_select[0], 0, region1Start, gi);
bits += Huffmancode(gi.table_select[1], region1Start, region2Start, gi);
bits += Huffmancode(gi.table_select[2], region2Start, bigvalues, gi);
return bits;
}
int lame_internal_flags::writeMainData() {
// SessionConfig_t const & cfg = gfc.cfg;
// III_side_info_t const & l3_side = gfc.l3_side;
int gr, ch, sfb, data_bits, tot_bits = 0;
if (cfg.version == 1) {
/* MPEG 1 */
for (gr = 0; gr < 2; gr++) {
for (ch = 0; ch < cfg.channels_out; ch++) {
gr_info const & gi = l3_side.tt[gr][ch];
int const slen1 = slen1_tab[gi.scalefac_compress];
int const slen2 = slen2_tab[gi.scalefac_compress];
data_bits = 0;
#ifdef DEBUG
hogege = bs.totbit;
#endif
for (sfb = 0; sfb < gi.sfbdivide; sfb++) {
if (gi.scalefac[sfb] == -1)
continue; /* scfsi is used */
putbits2(gi.scalefac[sfb], slen1);
data_bits += slen1;
}
for (; sfb < gi.sfbmax; sfb++) {
if (gi.scalefac[sfb] == -1)
continue; /* scfsi is used */
putbits2(gi.scalefac[sfb], slen2);
data_bits += slen2;
}
assert(data_bits == gi.part2_length);
if (gi.block_type == SHORT_TYPE) {
data_bits += ShortHuffmancodebits(gi);
}
else {
data_bits += LongHuffmancodebits(gi);
}
data_bits += huffman_coder_count1(gi);
#ifdef DEBUG
DEBUGF(this, "<%ld> ", bs.totbit - hogege);
#endif
/* does bitcount in quantize.c agree with actual bit count? */
assert(data_bits == gi.part2_3_length + gi.part2_length);
tot_bits += data_bits;
} /* for ch */
} /* for gr */
}
else {
/* MPEG 2 */
gr = 0;
for (ch = 0; ch < cfg.channels_out; ch++) {
gr_info const & gi = l3_side.tt[gr][ch];
int i, sfb_partition, scale_bits = 0;
assert(gi.sfb_partition_table);
data_bits = 0;
#ifdef DEBUG
hogege = bs.totbit;
#endif
sfb = 0;
sfb_partition = 0;
if (gi.block_type == SHORT_TYPE) {
for (; sfb_partition < 4; sfb_partition++) {
int const sfbs = gi.sfb_partition_table[sfb_partition] / 3;
int const slen = gi.slen[sfb_partition];
for (i = 0; i < sfbs; i++, sfb++) {
putbits2(Max(gi.scalefac[sfb * 3 + 0], 0), slen);
putbits2(Max(gi.scalefac[sfb * 3 + 1], 0), slen);
putbits2(Max(gi.scalefac[sfb * 3 + 2], 0), slen);
scale_bits += 3 * slen;
}
}
data_bits += ShortHuffmancodebits(gi);
}
else {
for (; sfb_partition < 4; sfb_partition++) {
int const sfbs = gi.sfb_partition_table[sfb_partition];
int const slen = gi.slen[sfb_partition];
for (i = 0; i < sfbs; i++, sfb++) {
putbits2(Max(gi.scalefac[sfb], 0), slen);
scale_bits += slen;
}
}
data_bits += LongHuffmancodebits(gi);
}
data_bits += huffman_coder_count1(gi);
#ifdef DEBUG
DEBUGF(this, "<%ld> ", bs.totbit - hogege);
#endif
/* does bitcount in quantize.c agree with actual bit count? */
assert(data_bits == gi.part2_3_length);
assert(scale_bits == gi.part2_length);
tot_bits += scale_bits + data_bits;
} /* for ch */
} /* for gf */
return tot_bits;
} /* main_data */
/* compute the number of bits required to flush all mp3 frames
currently in the buffer. This should be the same as the
reservoir size. Only call this routine between frames - i.e.
only after all headers and data have been added to the buffer
by format_bitstream().
Also compute total_bits_output =
size of mp3 buffer (including frame headers which may not
have yet been send to the mp3 buffer) +
number of bits needed to flush all mp3 frames.
total_bytes_output is the size of the mp3 output buffer if
lame_encode_flush_nogap() was called right now.
*/
int lame_internal_flags::compute_flushbits(int *total_bytes_output)const{
// SessionConfig_t const & cfg = gfc.cfg;
EncStateVar_t const & esv = sv_enc;
int flushbits, remaining_headers;
int bitsPerFrame;
int last_ptr, first_ptr;
first_ptr = esv.w_ptr; /* first header to add to bitstream */
last_ptr = esv.h_ptr - 1; /* last header to add to bitstream */
if (last_ptr == -1)
last_ptr = MAX_HEADER_BUF - 1;
/* add this many bits to bitstream so we can flush all headers */
flushbits = esv.header[last_ptr].write_timing - bs.totbit;
*total_bytes_output = flushbits;
if (flushbits >= 0) {
/* if flushbits >= 0, some headers have not yet been written */
/* reduce flushbits by the size of the headers */
remaining_headers = 1 + last_ptr - first_ptr;
if (last_ptr < first_ptr)
remaining_headers = 1 + last_ptr - first_ptr + MAX_HEADER_BUF;
flushbits -= remaining_headers * 8 * cfg.sideinfo_len;
}
/* finally, add some bits so that the last frame is complete
* these bits are not necessary to decode the last frame, but
* some decoders will ignore last frame if these bits are missing
*/
bitsPerFrame = getframebits();
flushbits += bitsPerFrame;
*total_bytes_output += bitsPerFrame;
/* round up: */
if (*total_bytes_output % 8)
*total_bytes_output = 1 + (*total_bytes_output / 8);
else
*total_bytes_output = (*total_bytes_output / 8);
*total_bytes_output += bs.buf_byte_idx + 1;
if (flushbits < 0) {
#if 0
/* if flushbits < 0, this would mean that the buffer looks like:
* (data...) last_header (data...) (extra data that should not be here...)
*/
DEBUGF(*this, "last header write_timing = %i \n", esv.header[last_ptr].write_timing);
DEBUGF(*this, "first header write_timing = %i \n", esv.header[first_ptr].write_timing);
DEBUGF(*this, "bs.totbit: %i \n", gfc.bs.totbit);
DEBUGF(*this, "first_ptr, last_ptr %i %i \n", first_ptr, last_ptr);
DEBUGF(*this, "remaining_headers = %i \n", remaining_headers);
DEBUGF(*this, "bitsperframe: %i \n", bitsPerFrame);
DEBUGF(*this, "sidelen: %i \n", cfg.sideinfo_len);
#endif
ERRORF(*this, "strange error flushing buffer ... \n");
}
return flushbits;
}
void lame_internal_flags::flush_bitstream() {
EncStateVar_t & esv = sv_enc;
int nbytes;
int flushbits;
int last_ptr = esv.h_ptr - 1; /* last header to add to bitstream */
if (last_ptr == -1)
last_ptr = MAX_HEADER_BUF - 1;
// III_side_info_t & l3_side = gfc.l3_side;
if ((flushbits = compute_flushbits(&nbytes)) < 0) return;
drain_into_ancillary( flushbits);
/* check that the 100% of the last frame has been written to bitstream */
assert(esv.header[last_ptr].write_timing + getframebits()
== bs.totbit);
/* we have padded out all frames with ancillary data, which is the
same as filling the bitreservoir with ancillary data, so : */
esv.ResvSize = 0;
l3_side.main_data_begin = 0;
}
void lame_internal_flags::add_dummy_byte(unsigned char val, unsigned int n) {
EncStateVar_t & esv = sv_enc;
while (n-- > 0u) {
putbits_noheaders(val, 8);
for (int i = 0; i < MAX_HEADER_BUF; ++i)
esv.header[i].write_timing += 8;
}
}
/*
format_bitstream()
This is called after a frame of audio has been quantized and coded.
It will write the encoded audio to the bitstream. Note that
from a layer3 encoder's perspective the bit stream is primarily
a series of main_data() blocks, with header and side information
inserted at the proper locations to maintain framing. (See Figure A.7
in the IS).
*/
int lame_internal_flags::format_bitstream() {
// SessionConfig_t const & cfg = gfc.cfg;
EncStateVar_t & esv = sv_enc;
int bits, nbytes;
int bitsPerFrame;
// III_side_info_t & l3_side = gfc.l3_side;
bitsPerFrame = getframebits();
drain_into_ancillary(l3_side.resvDrain_pre);
encodeSideInfo2(bitsPerFrame);
bits = 8 * cfg.sideinfo_len;
bits += writeMainData();
drain_into_ancillary(l3_side.resvDrain_post);
bits += l3_side.resvDrain_post;
l3_side.main_data_begin += (bitsPerFrame - bits) / 8;
/* compare number of bits needed to clear all buffered mp3 frames
* with what we think the resvsize is: */
if (compute_flushbits(&nbytes) != esv.ResvSize) {
ERRORF(*this, "Internal buffer inconsistency. flushbits <> ResvSize");
}
/* compare main_data_begin for the next frame with what we
* think the resvsize is: */
if ((l3_side.main_data_begin * 8) != esv.ResvSize) {
ERRORF(*this, "bit reservoir error: \n"
"l3_side.main_data_begin: %i \n"
"Resvoir size: %i \n"
"resv drain (post) %i \n"
"resv drain (pre) %i \n"
"header and sideinfo: %i \n"
"data bits: %i \n"
"total bits: %i (remainder: %i) \n"
"bitsperframe: %i \n",
8 * l3_side.main_data_begin,
esv.ResvSize,
l3_side.resvDrain_post,
l3_side.resvDrain_pre,
8 * cfg.sideinfo_len,
bits - l3_side.resvDrain_post - 8 * cfg.sideinfo_len,
bits, bits % 8, bitsPerFrame);
ERRORF(*this, "This is a fatal error. It has several possible causes:");
ERRORF(*this, "90%% LAME compiled with buggy version of gcc using advanced optimizations");
ERRORF(*this, " 9%% Your system is overclocked");
ERRORF(*this, " 1%% bug in LAME encoding library");
esv.ResvSize = l3_side.main_data_begin * 8;
};
assert(bs.totbit % 8 == 0);
if (bs.totbit > 1000000000) {
/* to avoid totbit overflow, (at 8h encoding at 128kbs) lets reset bit counter */
int i;
for (i = 0; i < MAX_HEADER_BUF; ++i)
esv.header[i].write_timing -= bs.totbit;
bs.totbit = 0;
}
return 0;
}
int lame_internal_flags::do_gain_analysis(unsigned char* buffer, int minimum) {
// SessionConfig_t const & cfg = gfc.cfg;
RpgStateVar_t const & rsv = sv_rpg;
RpgResult_t & rov = ov_rpg;
#ifdef DECODE_ON_THE_FLY
if (cfg.decode_on_the_fly) { /* decode the frame */
sample_t pcm_buf[2][1152];
int mp3_in = minimum;
int samples_out = -1;
/* re-synthesis to pcm. Repeat until we get a samples_out=0 */
while (samples_out != 0) {
samples_out = hip_decode1_unclipped(hip, buffer, mp3_in, pcm_buf[0], pcm_buf[1]);
/* samples_out = 0: need more data to decode
* samples_out = -1: error. Lets assume 0 pcm output
* samples_out = number of samples output */
/* set the lenght of the mp3 input buffer to zero, so that in the
* next iteration of the loop we will be querying mpglib about
* buffered data */
mp3_in = 0;
if (samples_out == -1) {
/* error decoding. Not fatal, but might screw up
* the ReplayGain tag. What should we do? Ignore for now */
samples_out = 0;
}
if (samples_out > 0) {
/* process the PCM data */
/* this should not be possible, and indicates we have
* overflown the pcm_buf buffer */
assert(samples_out <= 1152);
if (cfg.findPeakSample) {
int i;
/* FIXME: is this correct? maybe Max(fabs(pcm),PeakSample) */
for (i = 0; i < samples_out; i++) {
if (pcm_buf[0][i] > rov->PeakSample)
rov->PeakSample = pcm_buf[0][i];
else if (-pcm_buf[0][i] > rov->PeakSample)
rov->PeakSample = -pcm_buf[0][i];
}
if (cfg.channels_out > 1)
for (i = 0; i < samples_out; i++) {
if (pcm_buf[1][i] > rov->PeakSample)
rov->PeakSample = pcm_buf[1][i];
else if (-pcm_buf[1][i] > rov->PeakSample)
rov->PeakSample = -pcm_buf[1][i];
}
}
if (cfg.findReplayGain)
if (AnalyzeSamples
(rsv->rgdata, pcm_buf[0], pcm_buf[1], samples_out,
cfg.channels_out) == GAIN_ANALYSIS_ERROR)
return -6;
} /* if (samples_out>0) */
} /* while (samples_out!=0) */
} /* if (decode_on_the_fly) */
#endif
return minimum;
}
int lame_internal_flags::do_copy_buffer(unsigned char *buffer, int size) {
// Bit_stream_struc & bs = gfc.bs;
int const minimum = bs.buf_byte_idx + 1;
if (minimum <= 0) return 0;
if (minimum > size) return -1; /* buffer is too small */
memcpy(buffer, bs.buf, minimum);
bs.buf_byte_idx = -1;
bs.buf_bit_idx = 0;
return minimum;
}
/* copy data out of the internal MP3 bit buffer into a user supplied
unsigned char buffer.
mp3data=0 indicates data in buffer is an id3tags and VBR tags
mp3data=1 data is real mp3 frame data.
*/
int lame_internal_flags::copy_buffer(unsigned char *buffer, int size, bool mp3data) {
int const minimum = do_copy_buffer(buffer, size);
if (minimum > 0 && mp3data) {
UpdateMusicCRC(&nMusicCRC, buffer, minimum);
/** sum number of bytes belonging to the mp3 stream
* this info will be written into the Xing/LAME header for seeking
*/
VBR_seek_table.nBytesWritten += minimum;
return do_gain_analysis(buffer, minimum);
} /* if (mp3data) */
return minimum;
}
void lame_internal_flags::init_bit_stream_w() {
EncStateVar_t & esv = sv_enc;
esv.h_ptr = esv.w_ptr = 0;
esv.header[esv.h_ptr].write_timing = 0;
bs.buf = lame_calloc(unsigned char, BUFFER_SIZE);
bs.buf_size = BUFFER_SIZE;
bs.buf_byte_idx = -1;
bs.buf_bit_idx = 0;
bs.totbit = 0;
}
Detected encoding: ASCII (7 bit) | 2
|