axfer: add support for a container of Creative Tech. voice format
This commit adds support for data of Creative Tech. voice format. In this
data format, values in each of field are represented in little-endian byte
order and available formats of data sample are restricted in little-endian
byte order.
In version 1.10 of this format, sampling rate is represented with
reciprocal number of the rate, thus we cannot calculate original sampling
rate precisely just from its header. For example at 44.1kHz, file header
includes 233 (=256-1,000,000/44,100), but we cannot recover the value just
from the code (43478.2...). For my convenience, this commit adds a
pre-computed table and lookup major rates from the table.
Additionally, this format can includes several blocks with different
sample format. When handling this type of file, we need to start/stop
substream for each of the block, while this brings complicated code.
This type of format is enough ancient and presently quite minor. This
commit takes a compromise and handles a first sample block only.
Signed-off-by: Takashi Sakamoto <o-takashi@sakamocchi.jp>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
2018-11-13 07:41:18 +01:00
|
|
|
// SPDX-License-Identifier: GPL-2.0
|
|
|
|
//
|
|
|
|
// container-voc.c - a parser/builder for a container of Creative Voice File.
|
|
|
|
//
|
|
|
|
// Copyright (c) 2018 Takashi Sakamoto <o-takashi@sakamocchi.jp>
|
|
|
|
//
|
|
|
|
// Licensed under the terms of the GNU General Public License, version 2.
|
|
|
|
|
|
|
|
#include "container.h"
|
|
|
|
#include "misc.h"
|
|
|
|
|
|
|
|
// Not portable to all of UNIX platforms.
|
|
|
|
#include <endian.h>
|
|
|
|
|
|
|
|
// References:
|
|
|
|
// - http://sox.sourceforge.net/
|
|
|
|
|
|
|
|
#define VOC_MAGIC "Creative Voice File\x1a"
|
|
|
|
#define VOC_VERSION_1_10 0x010a
|
|
|
|
#define VOC_VERSION_1_20 0x0114
|
|
|
|
|
|
|
|
enum block_type {
|
|
|
|
BLOCK_TYPE_TERMINATOR = 0x00,
|
|
|
|
BLOCK_TYPE_V110_DATA = 0x01,
|
|
|
|
BLOCK_TYPE_CONTINUOUS_DATA = 0x02,
|
|
|
|
BLOCK_TYPE_SILENCE = 0x03,
|
|
|
|
BLOCK_TYPE_MARKER = 0x04,
|
|
|
|
BLOCK_TYPE_STRING = 0x05,
|
|
|
|
BLOCK_TYPE_REPEAT_START = 0x06,
|
|
|
|
BLOCK_TYPE_REPEAT_END = 0x07,
|
|
|
|
BLOCK_TYPE_EXTENDED_V110_FORMAT = 0x08,
|
|
|
|
BLOCK_TYPE_V120_DATA = 0x09,
|
|
|
|
};
|
|
|
|
|
|
|
|
enum code_id {
|
|
|
|
// Version 1.10.
|
|
|
|
CODE_ID_GENERIC_MBLA_U8 = 0x00,
|
|
|
|
CODE_ID_CREATIVE_ADPCM_8BIT_TO_4BIT_LE = 0x01,
|
|
|
|
CODE_ID_CREATIVE_ADPCM_8BIT_TO_3BIT_LE = 0x02,
|
|
|
|
CODE_ID_CREATIVE_ADPCM_8BIT_TO_2BIT_LE = 0x03,
|
|
|
|
// Version 1.20.
|
|
|
|
CODE_ID_GENERIC_MBLA_S16_LE = 0x04,
|
|
|
|
CODE_ID_CCIT_A_LAW_LE = 0x06,
|
|
|
|
CODE_ID_CCIT_MU_LAW_LE = 0x07,
|
|
|
|
CODE_ID_CREATIVE_ADPCM_16BIT_TO_4BIT_LE = 0x2000,
|
|
|
|
};
|
|
|
|
|
|
|
|
struct format_map {
|
|
|
|
unsigned int minimal_version;
|
|
|
|
enum code_id code_id;
|
|
|
|
snd_pcm_format_t format;
|
|
|
|
};
|
|
|
|
|
|
|
|
static const struct format_map format_maps[] = {
|
|
|
|
{VOC_VERSION_1_10, CODE_ID_GENERIC_MBLA_U8, SND_PCM_FORMAT_U8},
|
|
|
|
{VOC_VERSION_1_20, CODE_ID_GENERIC_MBLA_S16_LE, SND_PCM_FORMAT_S16_LE},
|
|
|
|
{VOC_VERSION_1_20, CODE_ID_CCIT_A_LAW_LE, SND_PCM_FORMAT_A_LAW},
|
|
|
|
{VOC_VERSION_1_20, CODE_ID_CCIT_MU_LAW_LE, SND_PCM_FORMAT_MU_LAW},
|
|
|
|
// The other formats are not supported by ALSA.
|
|
|
|
};
|
|
|
|
|
|
|
|
struct container_header {
|
|
|
|
uint8_t magic[20];
|
|
|
|
uint16_t hdr_size;
|
|
|
|
uint16_t version;
|
|
|
|
uint16_t version_compr;
|
|
|
|
};
|
|
|
|
|
|
|
|
// A format for data blocks except for terminator type.
|
|
|
|
struct block_header {
|
|
|
|
uint8_t type;
|
|
|
|
uint8_t size[3];
|
|
|
|
|
|
|
|
uint8_t data[0];
|
|
|
|
};
|
|
|
|
|
|
|
|
// Data block for terminator type has an exceptional format.
|
|
|
|
struct block_terminator {
|
|
|
|
uint8_t type;
|
|
|
|
};
|
|
|
|
|
|
|
|
struct time_const {
|
|
|
|
unsigned int frames_per_second;
|
|
|
|
uint16_t code;
|
|
|
|
};
|
|
|
|
|
|
|
|
static const struct time_const v110_time_consts[] = {
|
|
|
|
{5512, 74},
|
|
|
|
{8000, 130},
|
|
|
|
{11025, 165},
|
|
|
|
{16000, 193},
|
|
|
|
{22050, 210},
|
|
|
|
{32000, 224},
|
|
|
|
{44100, 233},
|
|
|
|
{48000, 235},
|
|
|
|
{64000, 240},
|
|
|
|
// Time constant for the upper sampling rate is not identical.
|
|
|
|
};
|
|
|
|
|
|
|
|
static const struct time_const ex_v110_time_consts[] = {
|
|
|
|
{5512, 19092},
|
|
|
|
{8000, 33536},
|
|
|
|
{11025, 42317},
|
|
|
|
{16000, 49536},
|
|
|
|
{22050, 53927},
|
|
|
|
{32000, 57536},
|
|
|
|
{44100, 59732},
|
|
|
|
{48000, 60203},
|
|
|
|
{64000, 61536},
|
|
|
|
{88200, 62634},
|
|
|
|
{96000, 62870},
|
|
|
|
{176400, 64085},
|
|
|
|
{192000, 64203},
|
|
|
|
// This support up to 192.0 kHz. The rest is for cases with 2ch.
|
|
|
|
{352800, 64811},
|
|
|
|
{384000, 64870},
|
|
|
|
};
|
|
|
|
|
|
|
|
// v1.10 format:
|
|
|
|
// - monaural.
|
|
|
|
// - frames_per_second = 1,000,000 / (256 - time_const)
|
|
|
|
struct block_v110_data {
|
|
|
|
uint8_t type;
|
|
|
|
uint8_t size[3]; // Equals to (2 + the size of frames).
|
|
|
|
|
|
|
|
uint8_t time_const;
|
|
|
|
uint8_t code_id;
|
|
|
|
uint8_t frames[0]; // Aligned to little-endian.
|
|
|
|
};
|
|
|
|
|
|
|
|
struct block_continuous_data {
|
|
|
|
uint8_t type;
|
|
|
|
uint8_t size[3]; // Equals to the size of frames.
|
|
|
|
|
|
|
|
uint8_t frames[0]; // Aligned to little-endian.
|
|
|
|
};
|
|
|
|
|
|
|
|
// v1.10 format:
|
|
|
|
// - monaural.
|
|
|
|
// - frames_per_second = 1,000,000 / (256 - time_const).
|
|
|
|
struct block_silence {
|
|
|
|
uint8_t type;
|
|
|
|
uint8_t size[3]; // Equals to 3.
|
|
|
|
|
|
|
|
uint16_t frame_count;
|
|
|
|
uint8_t time_const;
|
|
|
|
};
|
|
|
|
|
|
|
|
struct block_marker {
|
|
|
|
uint8_t type;
|
|
|
|
uint8_t size[3]; // Equals to 2.
|
|
|
|
|
|
|
|
uint16_t mark;
|
|
|
|
};
|
|
|
|
|
|
|
|
struct block_string {
|
|
|
|
uint8_t type;
|
|
|
|
uint8_t size[3]; // Equals to the length of string with 0x00.
|
|
|
|
|
|
|
|
uint8_t chars[0];
|
|
|
|
};
|
|
|
|
|
|
|
|
struct block_repeat_start {
|
|
|
|
uint8_t type;
|
|
|
|
uint8_t size[3]; // Equals to 2.
|
|
|
|
|
|
|
|
uint16_t count;
|
|
|
|
};
|
|
|
|
|
|
|
|
struct block_repeat_end {
|
|
|
|
uint8_t type;
|
|
|
|
uint8_t size[3]; // Equals to 0.
|
|
|
|
};
|
|
|
|
|
|
|
|
// Extended v1.10 format:
|
|
|
|
// - manaural/stereo.
|
|
|
|
// - frames_per_second =
|
|
|
|
// 256,000,000 / (samples_per_frame * (65536 - time_const)).
|
|
|
|
// - Appear just before v110_data block.
|
|
|
|
struct block_extended_v110_format {
|
|
|
|
uint8_t type;
|
|
|
|
uint8_t size[3]; // Equals to 4.
|
|
|
|
|
|
|
|
uint16_t time_const;
|
|
|
|
uint8_t code_id;
|
|
|
|
uint8_t ch_mode; // 0 is monaural, 1 is stereo.
|
|
|
|
};
|
|
|
|
|
|
|
|
// v1.20 format:
|
|
|
|
// - monaural/stereo.
|
|
|
|
// - 8/16 bits_per_sample.
|
|
|
|
// - time_const is not used.
|
|
|
|
// - code_id is extended.
|
|
|
|
struct block_v120_format {
|
|
|
|
uint8_t type;
|
|
|
|
uint8_t size[3]; // Equals to (12 + ).
|
|
|
|
|
|
|
|
uint32_t frames_per_second;
|
|
|
|
uint8_t bits_per_sample;
|
|
|
|
uint8_t samples_per_frame;
|
|
|
|
uint16_t code_id;
|
|
|
|
uint8_t reserved[4];
|
|
|
|
|
|
|
|
uint8_t frames[0]; // Aligned to little-endian.
|
|
|
|
};
|
|
|
|
|
|
|
|
// Aligned to little endian order but 24 bits field.
|
|
|
|
static uint32_t parse_block_data_size(uint8_t fields[3])
|
|
|
|
{
|
|
|
|
return (fields[2] << 16) | (fields[1] << 8) | fields[0];
|
|
|
|
}
|
|
|
|
|
|
|
|
static void build_block_data_size(uint8_t fields[3], unsigned int size)
|
|
|
|
{
|
|
|
|
fields[0] = (size & 0x0000ff);
|
|
|
|
fields[1] = (size & 0x00ff00) >> 8;
|
|
|
|
fields[2] = (size & 0xff0000) >> 16;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int build_time_constant(unsigned int frames_per_second,
|
|
|
|
unsigned int samples_per_frame, uint16_t *code,
|
|
|
|
bool extended)
|
|
|
|
{
|
|
|
|
int i;
|
|
|
|
|
|
|
|
// 16 bits are available for this purpose.
|
|
|
|
if (extended) {
|
|
|
|
if (samples_per_frame > 2)
|
|
|
|
return -EINVAL;
|
|
|
|
frames_per_second *= samples_per_frame;
|
|
|
|
|
|
|
|
for (i = 0; i < ARRAY_SIZE(ex_v110_time_consts); ++i) {
|
|
|
|
if (ex_v110_time_consts[i].frames_per_second ==
|
|
|
|
frames_per_second)
|
|
|
|
break;
|
|
|
|
}
|
2019-03-13 14:19:12 +01:00
|
|
|
if (i < ARRAY_SIZE(ex_v110_time_consts) &&
|
axfer: add support for a container of Creative Tech. voice format
This commit adds support for data of Creative Tech. voice format. In this
data format, values in each of field are represented in little-endian byte
order and available formats of data sample are restricted in little-endian
byte order.
In version 1.10 of this format, sampling rate is represented with
reciprocal number of the rate, thus we cannot calculate original sampling
rate precisely just from its header. For example at 44.1kHz, file header
includes 233 (=256-1,000,000/44,100), but we cannot recover the value just
from the code (43478.2...). For my convenience, this commit adds a
pre-computed table and lookup major rates from the table.
Additionally, this format can includes several blocks with different
sample format. When handling this type of file, we need to start/stop
substream for each of the block, while this brings complicated code.
This type of format is enough ancient and presently quite minor. This
commit takes a compromise and handles a first sample block only.
Signed-off-by: Takashi Sakamoto <o-takashi@sakamocchi.jp>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
2018-11-13 07:41:18 +01:00
|
|
|
frames_per_second <= 192000) {
|
|
|
|
*code = ex_v110_time_consts[i].code;
|
|
|
|
} else {
|
|
|
|
*code = 65536 - 256000000 / frames_per_second;
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
if (samples_per_frame != 1)
|
|
|
|
return -EINVAL;
|
|
|
|
|
|
|
|
for (i = 0; i < ARRAY_SIZE(v110_time_consts); ++i) {
|
|
|
|
if (v110_time_consts[i].frames_per_second ==
|
|
|
|
frames_per_second)
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
// Should be within 8 bit.
|
|
|
|
if (i < ARRAY_SIZE(v110_time_consts))
|
|
|
|
*code = (uint8_t)v110_time_consts[i].code;
|
|
|
|
else
|
|
|
|
*code = 256 - 1000000 / frames_per_second;
|
|
|
|
}
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
static unsigned int parse_time_constant(uint16_t code,
|
|
|
|
unsigned int samples_per_frame,
|
|
|
|
unsigned int *frames_per_second,
|
|
|
|
bool extended)
|
|
|
|
{
|
|
|
|
int i;
|
|
|
|
|
|
|
|
if (extended) {
|
|
|
|
if (samples_per_frame > 2)
|
|
|
|
return -EINVAL;
|
|
|
|
|
|
|
|
for (i = 0; i < ARRAY_SIZE(ex_v110_time_consts); ++i) {
|
|
|
|
if (ex_v110_time_consts[i].code == code ||
|
|
|
|
ex_v110_time_consts[i].code - 1 == code)
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
if (i < ARRAY_SIZE(ex_v110_time_consts)) {
|
|
|
|
*frames_per_second =
|
|
|
|
ex_v110_time_consts[i].frames_per_second /
|
|
|
|
samples_per_frame;
|
|
|
|
} else {
|
|
|
|
*frames_per_second = 256000000 / samples_per_frame /
|
|
|
|
(65536 - code);
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
if (samples_per_frame != 1)
|
|
|
|
return -EINVAL;
|
|
|
|
|
|
|
|
for (i = 0; i < ARRAY_SIZE(v110_time_consts); ++i) {
|
|
|
|
if (v110_time_consts[i].code == code ||
|
|
|
|
v110_time_consts[i].code - 1 == code)
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
if (i < ARRAY_SIZE(v110_time_consts)) {
|
|
|
|
*frames_per_second =
|
|
|
|
v110_time_consts[i].frames_per_second;
|
|
|
|
} else {
|
|
|
|
*frames_per_second = 1000000 / (256 - code);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
struct parser_state {
|
|
|
|
unsigned int version;
|
|
|
|
bool extended;
|
|
|
|
|
|
|
|
unsigned int frames_per_second;
|
|
|
|
unsigned int samples_per_frame;
|
|
|
|
unsigned int bytes_per_sample;
|
|
|
|
enum code_id code_id;
|
|
|
|
uint32_t byte_count;
|
|
|
|
};
|
|
|
|
|
|
|
|
static int parse_container_header(struct parser_state *state,
|
|
|
|
struct container_header *header)
|
|
|
|
{
|
|
|
|
uint16_t hdr_size;
|
|
|
|
uint16_t version;
|
|
|
|
uint16_t version_compr;
|
|
|
|
|
|
|
|
hdr_size = le16toh(header->hdr_size);
|
|
|
|
version = le16toh(header->version);
|
|
|
|
version_compr = le16toh(header->version_compr);
|
|
|
|
|
|
|
|
if (memcmp(header->magic, VOC_MAGIC, sizeof(header->magic)))
|
|
|
|
return -EIO;
|
|
|
|
|
|
|
|
if (hdr_size != sizeof(*header))
|
|
|
|
return -EIO;
|
|
|
|
|
|
|
|
if (version_compr != 0x1234 + ~version)
|
|
|
|
return -EIO;
|
|
|
|
|
|
|
|
if (version != VOC_VERSION_1_10 && version != VOC_VERSION_1_20)
|
|
|
|
return -EIO;
|
|
|
|
|
|
|
|
state->version = version;
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
static bool check_code_id(uint8_t code_id, unsigned int version)
|
|
|
|
{
|
|
|
|
int i;
|
|
|
|
|
|
|
|
for (i = 0; i < ARRAY_SIZE(format_maps); ++i) {
|
|
|
|
if (code_id != format_maps[i].code_id)
|
|
|
|
continue;
|
|
|
|
if (version >= format_maps[i].minimal_version)
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int parse_v120_format_block(struct parser_state *state,
|
|
|
|
struct block_v120_format *block)
|
|
|
|
{
|
|
|
|
state->frames_per_second = le32toh(block->frames_per_second);
|
|
|
|
state->bytes_per_sample = block->bits_per_sample / 8;
|
|
|
|
state->samples_per_frame = block->samples_per_frame;
|
|
|
|
state->code_id = le16toh(block->code_id);
|
|
|
|
state->byte_count = parse_block_data_size(block->size) - 12;
|
|
|
|
|
|
|
|
if (!check_code_id(state->code_id, VOC_VERSION_1_20))
|
|
|
|
return -EIO;
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int parse_extended_v110_format(struct parser_state *state,
|
|
|
|
struct block_extended_v110_format *block)
|
|
|
|
{
|
|
|
|
unsigned int time_const;
|
|
|
|
unsigned int frames_per_second;
|
|
|
|
int err;
|
|
|
|
|
|
|
|
state->code_id = block->code_id;
|
|
|
|
if (!check_code_id(state->code_id, VOC_VERSION_1_10))
|
|
|
|
return -EIO;
|
|
|
|
|
|
|
|
if (block->ch_mode == 0)
|
|
|
|
state->samples_per_frame = 1;
|
|
|
|
else if (block->ch_mode == 1)
|
|
|
|
state->samples_per_frame = 2;
|
|
|
|
else
|
|
|
|
return -EIO;
|
|
|
|
|
|
|
|
time_const = le16toh(block->time_const);
|
|
|
|
err = parse_time_constant(time_const, state->samples_per_frame,
|
|
|
|
&frames_per_second, true);
|
|
|
|
if (err < 0)
|
|
|
|
return err;
|
|
|
|
state->frames_per_second = frames_per_second;
|
|
|
|
|
|
|
|
state->extended = true;
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int parse_v110_data(struct parser_state *state,
|
|
|
|
struct block_v110_data *block)
|
|
|
|
{
|
|
|
|
unsigned int time_const;
|
|
|
|
unsigned int frames_per_second;
|
|
|
|
int err;
|
|
|
|
|
|
|
|
if (!state->extended) {
|
|
|
|
state->code_id = block->code_id;
|
|
|
|
if (!check_code_id(state->code_id, VOC_VERSION_1_10))
|
|
|
|
return -EIO;
|
|
|
|
|
|
|
|
time_const = block->time_const;
|
|
|
|
err = parse_time_constant(time_const, 1, &frames_per_second,
|
|
|
|
false);
|
|
|
|
if (err < 0)
|
|
|
|
return err;
|
|
|
|
state->frames_per_second = frames_per_second;
|
|
|
|
state->samples_per_frame = 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
state->bytes_per_sample = 1;
|
|
|
|
state->byte_count = parse_block_data_size(block->size) - 2;
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int detect_container_version(struct container_context *cntr)
|
|
|
|
{
|
|
|
|
struct parser_state *state = cntr->private_data;
|
|
|
|
struct container_header header = {0};
|
|
|
|
int err;
|
|
|
|
|
|
|
|
// 4 bytes were alread read to detect container type.
|
|
|
|
memcpy(&header.magic, cntr->magic, sizeof(cntr->magic));
|
|
|
|
err = container_recursive_read(cntr,
|
|
|
|
(char *)&header + sizeof(cntr->magic),
|
|
|
|
sizeof(header) - sizeof(cntr->magic));
|
|
|
|
if (err < 0)
|
|
|
|
return err;
|
|
|
|
if (cntr->eof)
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
return parse_container_header(state, &header);
|
|
|
|
}
|
|
|
|
|
|
|
|
static int allocate_for_block_cache(struct container_context *cntr,
|
|
|
|
struct block_header *header, void **buf)
|
|
|
|
{
|
|
|
|
uint32_t block_size;
|
|
|
|
char *cache;
|
|
|
|
int err;
|
|
|
|
|
|
|
|
if (header->type == BLOCK_TYPE_V110_DATA)
|
|
|
|
block_size = sizeof(struct block_v110_data);
|
|
|
|
else if (header->type == BLOCK_TYPE_CONTINUOUS_DATA)
|
|
|
|
block_size = sizeof(struct block_continuous_data);
|
|
|
|
else if (header->type == BLOCK_TYPE_EXTENDED_V110_FORMAT)
|
|
|
|
block_size = sizeof(struct block_extended_v110_format);
|
|
|
|
else if (header->type == BLOCK_TYPE_V120_DATA)
|
|
|
|
block_size = sizeof(struct block_v120_format);
|
|
|
|
else
|
|
|
|
block_size = parse_block_data_size(header->size);
|
|
|
|
|
|
|
|
cache = malloc(block_size);
|
|
|
|
if (cache == NULL)
|
|
|
|
return -ENOMEM;
|
|
|
|
memset(cache, 0, block_size);
|
|
|
|
|
|
|
|
memcpy(cache, header, sizeof(*header));
|
|
|
|
err = container_recursive_read(cntr, cache + sizeof(*header),
|
|
|
|
block_size - sizeof(*header));
|
|
|
|
if (err < 0) {
|
|
|
|
free(cache);
|
|
|
|
return err;
|
|
|
|
}
|
|
|
|
if (cntr->eof) {
|
|
|
|
free(cache);
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
*buf = cache;
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int cache_data_block(struct container_context *cntr,
|
|
|
|
struct block_header *header, void **buf)
|
|
|
|
{
|
|
|
|
int err;
|
|
|
|
|
|
|
|
// Check type of this block.
|
|
|
|
err = container_recursive_read(cntr, &header->type,
|
|
|
|
sizeof(header->type));
|
|
|
|
if (err < 0)
|
|
|
|
return err;
|
|
|
|
if (cntr->eof)
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
if (header->type > BLOCK_TYPE_V120_DATA)
|
|
|
|
return -EIO;
|
|
|
|
if (header->type == BLOCK_TYPE_TERMINATOR)
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
// Check size of this block. If the block includes a batch of data,
|
|
|
|
err = container_recursive_read(cntr, &header->size,
|
|
|
|
sizeof(header->size));
|
|
|
|
if (err < 0)
|
|
|
|
return err;
|
|
|
|
if (cntr->eof)
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
return allocate_for_block_cache(cntr, header, buf);
|
|
|
|
}
|
|
|
|
|
|
|
|
static int detect_format_block(struct container_context *cntr)
|
|
|
|
{
|
|
|
|
struct parser_state *state = cntr->private_data;
|
|
|
|
struct block_header header;
|
2019-03-13 14:19:12 +01:00
|
|
|
void *buf;
|
axfer: add support for a container of Creative Tech. voice format
This commit adds support for data of Creative Tech. voice format. In this
data format, values in each of field are represented in little-endian byte
order and available formats of data sample are restricted in little-endian
byte order.
In version 1.10 of this format, sampling rate is represented with
reciprocal number of the rate, thus we cannot calculate original sampling
rate precisely just from its header. For example at 44.1kHz, file header
includes 233 (=256-1,000,000/44,100), but we cannot recover the value just
from the code (43478.2...). For my convenience, this commit adds a
pre-computed table and lookup major rates from the table.
Additionally, this format can includes several blocks with different
sample format. When handling this type of file, we need to start/stop
substream for each of the block, while this brings complicated code.
This type of format is enough ancient and presently quite minor. This
commit takes a compromise and handles a first sample block only.
Signed-off-by: Takashi Sakamoto <o-takashi@sakamocchi.jp>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
2018-11-13 07:41:18 +01:00
|
|
|
int err;
|
|
|
|
|
|
|
|
again:
|
2019-03-13 14:19:12 +01:00
|
|
|
buf = NULL;
|
axfer: add support for a container of Creative Tech. voice format
This commit adds support for data of Creative Tech. voice format. In this
data format, values in each of field are represented in little-endian byte
order and available formats of data sample are restricted in little-endian
byte order.
In version 1.10 of this format, sampling rate is represented with
reciprocal number of the rate, thus we cannot calculate original sampling
rate precisely just from its header. For example at 44.1kHz, file header
includes 233 (=256-1,000,000/44,100), but we cannot recover the value just
from the code (43478.2...). For my convenience, this commit adds a
pre-computed table and lookup major rates from the table.
Additionally, this format can includes several blocks with different
sample format. When handling this type of file, we need to start/stop
substream for each of the block, while this brings complicated code.
This type of format is enough ancient and presently quite minor. This
commit takes a compromise and handles a first sample block only.
Signed-off-by: Takashi Sakamoto <o-takashi@sakamocchi.jp>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
2018-11-13 07:41:18 +01:00
|
|
|
err = cache_data_block(cntr, &header, &buf);
|
|
|
|
if (err < 0)
|
|
|
|
return err;
|
2019-03-13 14:19:12 +01:00
|
|
|
if (buf) {
|
|
|
|
if (header.type == BLOCK_TYPE_EXTENDED_V110_FORMAT) {
|
|
|
|
err = parse_extended_v110_format(state, buf);
|
|
|
|
} else if (header.type == BLOCK_TYPE_V120_DATA) {
|
|
|
|
err = parse_v120_format_block(state, buf);
|
|
|
|
} else if (header.type == BLOCK_TYPE_V110_DATA) {
|
|
|
|
err = parse_v110_data(state, buf);
|
|
|
|
} else {
|
|
|
|
free(buf);
|
|
|
|
goto again;
|
|
|
|
}
|
axfer: add support for a container of Creative Tech. voice format
This commit adds support for data of Creative Tech. voice format. In this
data format, values in each of field are represented in little-endian byte
order and available formats of data sample are restricted in little-endian
byte order.
In version 1.10 of this format, sampling rate is represented with
reciprocal number of the rate, thus we cannot calculate original sampling
rate precisely just from its header. For example at 44.1kHz, file header
includes 233 (=256-1,000,000/44,100), but we cannot recover the value just
from the code (43478.2...). For my convenience, this commit adds a
pre-computed table and lookup major rates from the table.
Additionally, this format can includes several blocks with different
sample format. When handling this type of file, we need to start/stop
substream for each of the block, while this brings complicated code.
This type of format is enough ancient and presently quite minor. This
commit takes a compromise and handles a first sample block only.
Signed-off-by: Takashi Sakamoto <o-takashi@sakamocchi.jp>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
2018-11-13 07:41:18 +01:00
|
|
|
|
|
|
|
free(buf);
|
|
|
|
|
2019-03-13 14:19:12 +01:00
|
|
|
if (err < 0)
|
|
|
|
return err;
|
|
|
|
}
|
axfer: add support for a container of Creative Tech. voice format
This commit adds support for data of Creative Tech. voice format. In this
data format, values in each of field are represented in little-endian byte
order and available formats of data sample are restricted in little-endian
byte order.
In version 1.10 of this format, sampling rate is represented with
reciprocal number of the rate, thus we cannot calculate original sampling
rate precisely just from its header. For example at 44.1kHz, file header
includes 233 (=256-1,000,000/44,100), but we cannot recover the value just
from the code (43478.2...). For my convenience, this commit adds a
pre-computed table and lookup major rates from the table.
Additionally, this format can includes several blocks with different
sample format. When handling this type of file, we need to start/stop
substream for each of the block, while this brings complicated code.
This type of format is enough ancient and presently quite minor. This
commit takes a compromise and handles a first sample block only.
Signed-off-by: Takashi Sakamoto <o-takashi@sakamocchi.jp>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
2018-11-13 07:41:18 +01:00
|
|
|
|
|
|
|
// Expect to detect block_v110_data.
|
|
|
|
if (header.type == BLOCK_TYPE_EXTENDED_V110_FORMAT)
|
|
|
|
goto again;
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int voc_parser_pre_process(struct container_context *cntr,
|
|
|
|
snd_pcm_format_t *format,
|
|
|
|
unsigned int *samples_per_frame,
|
|
|
|
unsigned int *frames_per_second,
|
|
|
|
uint64_t *byte_count)
|
|
|
|
{
|
|
|
|
struct parser_state *state = cntr->private_data;
|
|
|
|
int i;
|
|
|
|
int err;
|
|
|
|
|
|
|
|
err = detect_container_version(cntr);
|
|
|
|
if (err < 0)
|
|
|
|
return err;
|
|
|
|
|
|
|
|
err = detect_format_block(cntr);
|
|
|
|
if (err < 0)
|
|
|
|
return err;
|
|
|
|
|
|
|
|
for (i = 0; i < ARRAY_SIZE(format_maps); ++i) {
|
|
|
|
if (format_maps[i].code_id == state->code_id)
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
if (i == ARRAY_SIZE(format_maps))
|
|
|
|
return -EINVAL;
|
|
|
|
|
|
|
|
*format = format_maps[i].format;
|
|
|
|
*samples_per_frame = state->samples_per_frame;
|
|
|
|
*frames_per_second = state->frames_per_second;
|
|
|
|
|
|
|
|
// This program handles PCM frames in this data block only.
|
|
|
|
*byte_count = state->byte_count;
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
struct builder_state {
|
|
|
|
unsigned int version;
|
|
|
|
bool extended;
|
|
|
|
enum code_id code_id;
|
|
|
|
|
|
|
|
unsigned int samples_per_frame;
|
|
|
|
unsigned int bytes_per_sample;
|
|
|
|
};
|
|
|
|
|
|
|
|
static int write_container_header(struct container_context *cntr,
|
|
|
|
struct container_header *header)
|
|
|
|
{
|
|
|
|
struct builder_state *state = cntr->private_data;
|
|
|
|
|
|
|
|
// Process container header.
|
|
|
|
memcpy(header->magic, VOC_MAGIC, sizeof(header->magic));
|
|
|
|
header->hdr_size = htole16(sizeof(*header));
|
|
|
|
header->version = htole16(state->version);
|
|
|
|
header->version_compr = htole16(0x1234 + ~state->version);
|
|
|
|
|
|
|
|
return container_recursive_write(cntr, header, sizeof(*header));
|
|
|
|
}
|
|
|
|
|
|
|
|
static int write_v120_format_block(struct container_context *cntr,
|
|
|
|
struct block_v120_format *block,
|
|
|
|
unsigned int frames_per_second,
|
|
|
|
uint64_t byte_count)
|
|
|
|
{
|
|
|
|
struct builder_state *state = cntr->private_data;
|
|
|
|
|
|
|
|
block->type = BLOCK_TYPE_V120_DATA;
|
|
|
|
build_block_data_size(block->size, 12 + byte_count);
|
|
|
|
|
|
|
|
block->frames_per_second = htole32(frames_per_second);
|
2019-04-21 02:15:54 +02:00
|
|
|
block->bits_per_sample = state->bytes_per_sample * 8;
|
|
|
|
block->samples_per_frame = state->samples_per_frame;
|
axfer: add support for a container of Creative Tech. voice format
This commit adds support for data of Creative Tech. voice format. In this
data format, values in each of field are represented in little-endian byte
order and available formats of data sample are restricted in little-endian
byte order.
In version 1.10 of this format, sampling rate is represented with
reciprocal number of the rate, thus we cannot calculate original sampling
rate precisely just from its header. For example at 44.1kHz, file header
includes 233 (=256-1,000,000/44,100), but we cannot recover the value just
from the code (43478.2...). For my convenience, this commit adds a
pre-computed table and lookup major rates from the table.
Additionally, this format can includes several blocks with different
sample format. When handling this type of file, we need to start/stop
substream for each of the block, while this brings complicated code.
This type of format is enough ancient and presently quite minor. This
commit takes a compromise and handles a first sample block only.
Signed-off-by: Takashi Sakamoto <o-takashi@sakamocchi.jp>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
2018-11-13 07:41:18 +01:00
|
|
|
block->code_id = htole16(state->code_id);
|
|
|
|
|
|
|
|
return container_recursive_write(cntr, block, sizeof(*block));
|
|
|
|
}
|
|
|
|
|
|
|
|
static int write_extended_v110_format_block(struct container_context *cntr,
|
|
|
|
unsigned int frames_per_second,
|
|
|
|
struct block_extended_v110_format *block)
|
|
|
|
{
|
|
|
|
struct builder_state *state = cntr->private_data;
|
|
|
|
uint16_t time_const;
|
|
|
|
int err;
|
|
|
|
|
|
|
|
block->type = BLOCK_TYPE_EXTENDED_V110_FORMAT;
|
|
|
|
build_block_data_size(block->size, 4);
|
|
|
|
|
|
|
|
// 16 bits are available for this purpose.
|
|
|
|
err = build_time_constant(frames_per_second, state->samples_per_frame,
|
|
|
|
&time_const, true);
|
|
|
|
if (err < 0)
|
|
|
|
return err;
|
|
|
|
block->time_const = htole16(time_const);
|
|
|
|
block->code_id = htole16(state->code_id);
|
|
|
|
|
|
|
|
if (state->samples_per_frame == 1)
|
|
|
|
block->ch_mode = 0;
|
|
|
|
else
|
|
|
|
block->ch_mode = 1;
|
|
|
|
|
|
|
|
return container_recursive_write(cntr, block, sizeof(*block));
|
|
|
|
}
|
|
|
|
|
|
|
|
static int write_v110_format_block(struct container_context *cntr,
|
|
|
|
struct block_v110_data *block,
|
|
|
|
unsigned int frames_per_second,
|
|
|
|
uint64_t byte_count)
|
|
|
|
{
|
|
|
|
struct builder_state *state = cntr->private_data;
|
|
|
|
uint16_t time_const;
|
|
|
|
int err;
|
|
|
|
|
|
|
|
block->type = BLOCK_TYPE_V110_DATA;
|
|
|
|
build_block_data_size(block->size, 2 + byte_count);
|
|
|
|
|
|
|
|
// These fields were obsoleted by extension.
|
|
|
|
err = build_time_constant(frames_per_second, 1, &time_const, false);
|
|
|
|
if (err < 0)
|
|
|
|
return err;
|
|
|
|
block->time_const = (uint8_t)time_const;
|
|
|
|
block->code_id = state->code_id;
|
|
|
|
return container_recursive_write(cntr, block, sizeof(*block));
|
|
|
|
}
|
|
|
|
|
|
|
|
static int write_data_blocks(struct container_context *cntr,
|
|
|
|
unsigned int frames_per_second,
|
|
|
|
uint64_t byte_count)
|
|
|
|
{
|
|
|
|
union {
|
|
|
|
struct container_header header;
|
|
|
|
struct block_v110_data v110_data;
|
|
|
|
struct block_extended_v110_format extended_v110_format;
|
|
|
|
struct block_v120_format v120_format;
|
|
|
|
} buf = {0};
|
|
|
|
struct builder_state *state = cntr->private_data;
|
|
|
|
int err;
|
|
|
|
|
|
|
|
err = write_container_header(cntr, &buf.header);
|
|
|
|
if (err < 0)
|
|
|
|
return err;
|
|
|
|
|
|
|
|
if (state->version == VOC_VERSION_1_20) {
|
|
|
|
err = write_v120_format_block(cntr, &buf.v120_format,
|
|
|
|
frames_per_second, byte_count);
|
|
|
|
} else {
|
|
|
|
if (state->extended) {
|
|
|
|
err = write_extended_v110_format_block(cntr,
|
|
|
|
frames_per_second,
|
|
|
|
&buf.extended_v110_format);
|
|
|
|
if (err < 0)
|
|
|
|
return err;
|
|
|
|
}
|
|
|
|
err = write_v110_format_block(cntr, &buf.v110_data,
|
|
|
|
frames_per_second, byte_count);
|
|
|
|
}
|
|
|
|
|
|
|
|
return err;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int voc_builder_pre_process(struct container_context *cntr,
|
|
|
|
snd_pcm_format_t *format,
|
|
|
|
unsigned int *samples_per_frame,
|
|
|
|
unsigned int *frames_per_second,
|
|
|
|
uint64_t *byte_count)
|
|
|
|
{
|
|
|
|
struct builder_state *state = cntr->private_data;
|
|
|
|
int i;
|
|
|
|
|
|
|
|
// Validate parameters.
|
|
|
|
for (i = 0; i < ARRAY_SIZE(format_maps); ++i) {
|
|
|
|
if (format_maps[i].format == *format)
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
if (i == ARRAY_SIZE(format_maps))
|
|
|
|
return -EINVAL;
|
|
|
|
state->code_id = format_maps[i].code_id;
|
|
|
|
|
|
|
|
// Decide container version.
|
|
|
|
if (*samples_per_frame > 2)
|
|
|
|
state->version = VOC_VERSION_1_20;
|
|
|
|
else
|
|
|
|
state->version = format_maps[i].minimal_version;
|
|
|
|
if (state->version == VOC_VERSION_1_10) {
|
|
|
|
if (*samples_per_frame == 2) {
|
|
|
|
for (i = 0;
|
|
|
|
i < ARRAY_SIZE(ex_v110_time_consts); ++i) {
|
|
|
|
if (ex_v110_time_consts[i].frames_per_second ==
|
|
|
|
*frames_per_second)
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
if (i == ARRAY_SIZE(ex_v110_time_consts))
|
|
|
|
state->version = VOC_VERSION_1_20;
|
|
|
|
else
|
|
|
|
state->extended = true;
|
|
|
|
} else {
|
|
|
|
for (i = 0; i < ARRAY_SIZE(v110_time_consts); ++i) {
|
|
|
|
if (v110_time_consts[i].frames_per_second ==
|
|
|
|
*frames_per_second)
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
if (i == ARRAY_SIZE(v110_time_consts))
|
|
|
|
state->version = VOC_VERSION_1_20;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
state->bytes_per_sample = snd_pcm_format_physical_width(*format) / 8;
|
|
|
|
state->samples_per_frame = *samples_per_frame;
|
|
|
|
|
|
|
|
return write_data_blocks(cntr, *frames_per_second, *byte_count);
|
|
|
|
}
|
|
|
|
|
|
|
|
static int write_block_terminator(struct container_context *cntr)
|
|
|
|
{
|
|
|
|
struct block_terminator block = {0};
|
|
|
|
|
|
|
|
block.type = BLOCK_TYPE_TERMINATOR;
|
|
|
|
return container_recursive_write(cntr, &block, sizeof(block));
|
|
|
|
}
|
|
|
|
|
|
|
|
static int write_data_size(struct container_context *cntr, uint64_t byte_count)
|
|
|
|
{
|
|
|
|
struct builder_state *state = cntr->private_data;
|
|
|
|
off64_t offset;
|
|
|
|
uint8_t size_field[3];
|
|
|
|
int err;
|
|
|
|
|
|
|
|
offset = sizeof(struct container_header) + sizeof(uint8_t);
|
|
|
|
if (state->version == VOC_VERSION_1_10 && state->extended)
|
|
|
|
offset += sizeof(struct block_extended_v110_format);
|
|
|
|
err = container_seek_offset(cntr, offset);
|
|
|
|
if (err < 0)
|
|
|
|
return err;
|
|
|
|
|
|
|
|
if (state->version == VOC_VERSION_1_10)
|
|
|
|
offset = 2;
|
|
|
|
else
|
|
|
|
offset = 12;
|
|
|
|
|
|
|
|
if (byte_count > cntr->max_size - offset)
|
|
|
|
byte_count = cntr->max_size;
|
|
|
|
else
|
|
|
|
byte_count += offset;
|
|
|
|
build_block_data_size(size_field, byte_count);
|
|
|
|
|
|
|
|
return container_recursive_write(cntr, &size_field, sizeof(size_field));
|
|
|
|
}
|
|
|
|
|
|
|
|
static int voc_builder_post_process(struct container_context *cntr,
|
|
|
|
uint64_t handled_byte_count)
|
|
|
|
{
|
|
|
|
int err;
|
|
|
|
|
|
|
|
err = write_block_terminator(cntr);
|
|
|
|
if (err < 0)
|
|
|
|
return err;
|
|
|
|
|
|
|
|
return write_data_size(cntr, handled_byte_count);
|
|
|
|
}
|
|
|
|
|
|
|
|
const struct container_parser container_parser_voc = {
|
|
|
|
.format = CONTAINER_FORMAT_VOC,
|
|
|
|
.magic = VOC_MAGIC,
|
|
|
|
.max_size = 0xffffff - // = UINT24_MAX.
|
|
|
|
sizeof(struct block_terminator),
|
|
|
|
.ops = {
|
|
|
|
.pre_process = voc_parser_pre_process,
|
|
|
|
},
|
|
|
|
.private_size = sizeof(struct parser_state),
|
|
|
|
};
|
|
|
|
|
|
|
|
const struct container_builder container_builder_voc = {
|
|
|
|
.format = CONTAINER_FORMAT_VOC,
|
|
|
|
.max_size = 0xffffff - // = UINT24_MAX.
|
|
|
|
sizeof(struct block_terminator),
|
|
|
|
.ops = {
|
|
|
|
.pre_process = voc_builder_pre_process,
|
|
|
|
.post_process = voc_builder_post_process,
|
|
|
|
},
|
|
|
|
.private_size = sizeof(struct builder_state),
|
|
|
|
};
|