/*
* (Re)fill the stream buffer that is to be decoded. If any data
* still exists in the buffer then they are first shifted to be
* front of the stream buffer.
*/
static int sox_mp3_input(sox_format_t * ft)
{
priv_t *p = (priv_t *) ft->priv;
size_t bytes_read;
size_t remaining;
remaining = p->Stream.bufend - p->Stream.next_frame;
/* libmad does not consume all the buffer it's given. Some
* data, part of a truncated frame, is left unused at the
* end of the buffer. That data must be put back at the
* beginning of the buffer and taken in account for
* refilling the buffer. This means that the input buffer
* must be large enough to hold a complete frame at the
* highest observable bit-rate (currently 448 kb/s).
* TODO: Is 2016 bytes the size of the largest frame?
* (448000*(1152/32000))/8
*/
memmove(p->mp3_buffer, p->Stream.next_frame, remaining);
bytes_read = lsx_readbuf(ft, p->mp3_buffer+remaining,
p->mp3_buffer_size-remaining);
if (bytes_read == 0)
{
return SOX_EOF;
}
p->mad_stream_buffer(&p->Stream, p->mp3_buffer, bytes_read+remaining);
p->Stream.error = 0;
return SOX_SUCCESS;
}
static int dvms_read_header(sox_format_t * ft, struct dvms_header *hdr)
{
unsigned char hdrbuf[DVMS_HEADER_LEN];
unsigned char *pch = hdrbuf;
int i;
unsigned sum;
if (lsx_readbuf(ft, hdrbuf, sizeof(hdrbuf)) != sizeof(hdrbuf))
{
return (SOX_EOF);
}
for(i = sizeof(hdrbuf), sum = 0; i > /*2*/3; i--) /* Deti bug */
sum += *pch++;
pch = hdrbuf;
memcpy(hdr->Filename, pch, sizeof(hdr->Filename));
pch += sizeof(hdr->Filename);
hdr->Id = get16_le(&pch);
hdr->State = get16_le(&pch);
hdr->Unixtime = get32_le(&pch);
hdr->Usender = get16_le(&pch);
hdr->Ureceiver = get16_le(&pch);
hdr->Length = get32_le(&pch);
hdr->Srate = get16_le(&pch);
hdr->Days = get16_le(&pch);
hdr->Custom1 = get16_le(&pch);
hdr->Custom2 = get16_le(&pch);
memcpy(hdr->Info, pch, sizeof(hdr->Info));
pch += sizeof(hdr->Info);
memcpy(hdr->extend, pch, sizeof(hdr->extend));
pch += sizeof(hdr->extend);
hdr->Crc = get16_le(&pch);
if (sum != hdr->Crc)
{
lsx_report("DVMS header checksum error, read %u, calculated %u",
hdr->Crc, sum);
return (SOX_EOF);
}
return (SOX_SUCCESS);
}
static int callback_read(void* ft_data, unsigned char* ptr, int nbytes)
{
sox_format_t* ft = (sox_format_t*)ft_data;
return lsx_readbuf(ft, ptr, (size_t)nbytes);
}
static int startread(sox_format_t * ft)
{
priv_t *p = (priv_t *) ft->priv;
size_t ReadSize;
sox_bool ignore_length = ft->signal.length == SOX_IGNORE_LENGTH;
int open_library_result;
LSX_DLLIBRARY_OPEN(
p,
mad_dl,
MAD_FUNC_ENTRIES,
"MAD decoder library",
mad_library_names,
open_library_result);
if (open_library_result)
return SOX_EOF;
p->mp3_buffer_size = sox_globals.bufsiz;
p->mp3_buffer = lsx_malloc(p->mp3_buffer_size);
ft->signal.length = SOX_UNSPEC;
if (ft->seekable) {
#ifdef USING_ID3TAG
read_comments(ft);
rewind((FILE*)ft->fp);
if (!ft->signal.length)
#endif
if (!ignore_length)
ft->signal.length = mp3_duration_ms(ft);
}
p->mad_stream_init(&p->Stream);
p->mad_frame_init(&p->Frame);
p->mad_synth_init(&p->Synth);
mad_timer_reset(&p->Timer);
ft->encoding.encoding = SOX_ENCODING_MP3;
/* Decode at least one valid frame to find out the input
* format. The decoded frame will be saved off so that it
* can be processed later.
*/
ReadSize = lsx_readbuf(ft, p->mp3_buffer, p->mp3_buffer_size);
if (ReadSize != p->mp3_buffer_size && ferror((FILE*)ft->fp))
return SOX_EOF;
p->mad_stream_buffer(&p->Stream, p->mp3_buffer, ReadSize);
/* Find a valid frame before starting up. This makes sure
* that we have a valid MP3 and also skips past ID3v2 tags
* at the beginning of the audio file.
*/
p->Stream.error = 0;
while (p->mad_frame_decode(&p->Frame,&p->Stream))
{
/* check whether input buffer needs a refill */
if (p->Stream.error == MAD_ERROR_BUFLEN)
{
if (sox_mp3_input(ft) == SOX_EOF)
return SOX_EOF;
continue;
}
/* Consume any ID3 tags */
sox_mp3_inputtag(ft);
/* FIXME: We should probably detect when we've read
* a bunch of non-ID3 data and still haven't found a
* frame. In that case we can abort early without
* scanning the whole file.
*/
p->Stream.error = 0;
}
if (p->Stream.error)
{
lsx_fail_errno(ft,SOX_EOF,"No valid MP3 frame found");
return SOX_EOF;
}
switch(p->Frame.header.mode)
{
case MAD_MODE_SINGLE_CHANNEL:
case MAD_MODE_DUAL_CHANNEL:
case MAD_MODE_JOINT_STEREO:
case MAD_MODE_STEREO:
ft->signal.channels = MAD_NCHANNELS(&p->Frame.header);
break;
default:
lsx_fail_errno(ft, SOX_EFMT, "Cannot determine number of channels");
return SOX_EOF;
}
p->FrameCount=1;
p->mad_timer_add(&p->Timer,p->Frame.header.duration);
p->mad_synth_frame(&p->Synth,&p->Frame);
ft->signal.precision = MP3_MAD_PRECISION;
ft->signal.rate=p->Synth.pcm.samplerate;
if (ignore_length)
ft->signal.length = SOX_UNSPEC;
else {
ft->signal.length = (uint64_t)(ft->signal.length * .001 * ft->signal.rate + .5);
ft->signal.length *= ft->signal.channels; /* Keep separate from line above! */
}
p->cursamp = 0;
return SOX_SUCCESS;
}
static char const * auto_detect_format(sox_format_t * ft, char const * ext)
{
char data[AUTO_DETECT_SIZE];
size_t len = lsx_readbuf(ft, data, ft->seekable? sizeof(data) : PIPE_AUTO_DETECT_SIZE);
#define CHECK(type, p2, l2, d2, p1, l1, d1) if (len >= p1 + l1 && \
!memcmp(data + p1, d1, (size_t)l1) && !memcmp(data + p2, d2, (size_t)l2)) return #type;
CHECK(voc , 0, 0, "" , 0, 20, "Creative Voice File\x1a")
CHECK(smp , 0, 0, "" , 0, 17, "SOUND SAMPLE DATA")
CHECK(wve , 0, 0, "" , 0, 15, "ALawSoundFile**")
CHECK(gsrt , 0, 0, "" , 16, 9, "ring.bin")
CHECK(amr-wb, 0, 0, "" , 0, 9, "#!AMR-WB\n")
CHECK(prc , 0, 0, "" , 0, 8, "\x37\x00\x00\x10\x6d\x00\x00\x10")
CHECK(sph , 0, 0, "" , 0, 7, "NIST_1A")
CHECK(amr-nb, 0, 0, "" , 0, 6, "#!AMR\n")
CHECK(txw , 0, 0, "" , 0, 6, "LM8953")
CHECK(sndt , 0, 0, "" , 0, 6, "SOUND\x1a")
CHECK(vorbis, 0, 4, "OggS" , 29, 6, "vorbis")
CHECK(opus , 0, 4, "OggS" , 28, 8, "OpusHead")
CHECK(speex , 0, 4, "OggS" , 28, 6, "Speex")
CHECK(hcom ,65, 4, "FSSD" , 128,4, "HCOM")
CHECK(wav , 0, 4, "RIFF" , 8, 4, "WAVE")
CHECK(wav , 0, 4, "RIFX" , 8, 4, "WAVE")
CHECK(wav , 0, 4, "RF64" , 8, 4, "WAVE")
CHECK(aiff , 0, 4, "FORM" , 8, 4, "AIFF")
CHECK(aifc , 0, 4, "FORM" , 8, 4, "AIFC")
CHECK(8svx , 0, 4, "FORM" , 8, 4, "8SVX")
CHECK(maud , 0, 4, "FORM" , 8, 4, "MAUD")
CHECK(xa , 0, 0, "" , 0, 4, "XA\0\0")
CHECK(xa , 0, 0, "" , 0, 4, "XAI\0")
CHECK(xa , 0, 0, "" , 0, 4, "XAJ\0")
CHECK(au , 0, 0, "" , 0, 4, ".snd")
CHECK(au , 0, 0, "" , 0, 4, "dns.")
CHECK(au , 0, 0, "" , 0, 4, "\0ds.")
CHECK(au , 0, 0, "" , 0, 4, ".sd\0")
CHECK(flac , 0, 0, "" , 0, 4, "fLaC")
CHECK(avr , 0, 0, "" , 0, 4, "2BIT")
CHECK(caf , 0, 0, "" , 0, 4, "caff")
CHECK(wv , 0, 0, "" , 0, 4, "wvpk")
CHECK(paf , 0, 0, "" , 0, 4, " paf")
CHECK(sf , 0, 0, "" , 0, 4, "\144\243\001\0")
CHECK(sf , 0, 0, "" , 0, 4, "\0\001\243\144")
CHECK(sf , 0, 0, "" , 0, 4, "\144\243\002\0")
CHECK(sf , 0, 0, "" , 0, 4, "\0\002\243\144")
CHECK(sf , 0, 0, "" , 0, 4, "\144\243\003\0")
CHECK(sf , 0, 0, "" , 0, 4, "\0\003\243\144")
CHECK(sf , 0, 0, "" , 0, 4, "\144\243\004\0")
CHECK(sox , 0, 0, "" , 0, 4, ".SoX")
CHECK(sox , 0, 0, "" , 0, 4, "XoS.")
if (ext && !strcasecmp(ext, "snd"))
CHECK(sndr , 7, 1, "" , 0, 2, "\0")
#undef CHECK
#if HAVE_MAGIC
if (sox_globals.use_magic) {
static magic_t magic;
char const * filetype = NULL;
if (!magic) {
magic = magic_open(MAGIC_MIME | MAGIC_SYMLINK);
if (magic)
magic_load(magic, NULL);
}
if (magic)
filetype = magic_buffer(magic, data, len);
if (filetype && strncmp(filetype, "application/octet-stream", (size_t)24) &&
!lsx_strends(filetype, "/unknown") &&
strncmp(filetype, "text/plain", (size_t)10) )
return filetype;
else if (filetype)
lsx_debug("libmagic detected %s", filetype);
}
#endif
return NULL;
}
static int startread(sox_format_t * ft)
{
priv_t * avr = (priv_t *)ft->priv;
int rc;
lsx_reads(ft, avr->magic, (size_t)4);
if (strncmp (avr->magic, AVR_MAGIC, (size_t)4)) {
lsx_fail_errno(ft,SOX_EHDR,"AVR: unknown header");
return(SOX_EOF);
}
lsx_readbuf(ft, avr->name, sizeof(avr->name));
lsx_readw (ft, &(avr->mono));
if (avr->mono) {
ft->signal.channels = 2;
}
else {
ft->signal.channels = 1;
}
lsx_readw (ft, &(avr->rez));
if (avr->rez == 8) {
ft->encoding.bits_per_sample = 8;
}
else if (avr->rez == 16) {
ft->encoding.bits_per_sample = 16;
}
else {
lsx_fail_errno(ft,SOX_EFMT,"AVR: unsupported sample resolution");
return(SOX_EOF);
}
lsx_readw (ft, &(avr->sign));
if (avr->sign) {
ft->encoding.encoding = SOX_ENCODING_SIGN2;
}
else {
ft->encoding.encoding = SOX_ENCODING_UNSIGNED;
}
lsx_readw (ft, &(avr->loop));
lsx_readw (ft, &(avr->midi));
lsx_readdw (ft, &(avr->rate));
/*
* No support for AVRs created by ST-Replay,
* Replay Proffesional and PRO-Series 12.
*
* Just masking the upper byte out.
*/
ft->signal.rate = (avr->rate & 0x00ffffff);
lsx_readdw (ft, &(avr->size));
lsx_readdw (ft, &(avr->lbeg));
lsx_readdw (ft, &(avr->lend));
lsx_readw (ft, &(avr->res1));
lsx_readw (ft, &(avr->res2));
lsx_readw (ft, &(avr->res3));
lsx_readbuf(ft, avr->ext, sizeof(avr->ext));
lsx_readbuf(ft, avr->user, sizeof(avr->user));
rc = lsx_rawstartread (ft);
if (rc)
return rc;
return(SOX_SUCCESS);
}
/*-----------------------------------------------------------------
* startread() -- start reading a VOC file
*-----------------------------------------------------------------*/
static int startread(sox_format_t * ft)
{
int rtn = SOX_SUCCESS;
char header[20];
priv_t * v = (priv_t *) ft->priv;
unsigned short sbseek;
int rc;
int ii; /* for getting rid of lseek */
unsigned char uc;
if (lsx_readbuf(ft, header, (size_t)20) != 20) {
lsx_fail_errno(ft, SOX_EHDR, "unexpected EOF in VOC header");
return (SOX_EOF);
}
if (strncmp(header, "Creative Voice File\032", (size_t)19)) {
lsx_fail_errno(ft, SOX_EHDR, "VOC file header incorrect");
return (SOX_EOF);
}
/* read the offset to data, from start of file */
/* after this read we have read 20 bytes of header + 2 */
lsx_readw(ft, &sbseek);
/* ANN: read to skip the header, instead of lseek */
/* this should allow use with pipes.... */
for (ii = 22; ii < sbseek; ii++)
lsx_readb(ft, &uc);
v->rate = -1;
v->block_remaining = 0;
v->total_size = 0; /* ANN added */
v->extended = 0;
/* read until we get the format information.... */
rc = getblock(ft);
if (rc)
return rc;
/* get rate of data */
if (v->rate == -1) {
lsx_fail_errno(ft, SOX_EOF, "Input .voc file had no sound!");
return (SOX_EOF);
}
/* setup word length of data */
/* ANN: Check VOC format and map to the proper libSoX format value */
switch (v->format) {
case VOC_FMT_LIN8U: /* 0 8 bit unsigned linear PCM */
ft->encoding.encoding = SOX_ENCODING_UNSIGNED;
v->size = 8;
break;
case VOC_FMT_CRLADPCM4: /* 1 Creative 8-bit to 4-bit ADPCM */
ft->encoding.encoding = SOX_ENCODING_CL_ADPCM;
v->size = 4;
break;
case VOC_FMT_CRLADPCM3: /* 2 Creative 8-bit to 3-bit ADPCM */
ft->encoding.encoding = SOX_ENCODING_CL_ADPCM;
v->size = 3;
break;
case VOC_FMT_CRLADPCM2: /* 3 Creative 8-bit to 2-bit ADPCM */
ft->encoding.encoding = SOX_ENCODING_CL_ADPCM;
v->size = 2;
break;
case VOC_FMT_LIN16: /* 4 16-bit signed PCM */
ft->encoding.encoding = SOX_ENCODING_SIGN2;
v->size = 16;
break;
case VOC_FMT_ALAW: /* 6 CCITT a-Law 8-bit PCM */
ft->encoding.encoding = SOX_ENCODING_ALAW;
v->size = 8;
break;
case VOC_FMT_MU255: /* 7 CCITT u-Law 8-bit PCM */
ft->encoding.encoding = SOX_ENCODING_ULAW;
v->size = 8;
break;
case VOC_FMT_CRLADPCM4A: /*0x200 Creative 16-bit to 4-bit ADPCM */
ft->encoding.encoding = SOX_ENCODING_CL_ADPCM16;
v->size = 4;
break;
default:
lsx_fail("Unknown VOC format %d", v->format);
rtn = SOX_EOF;
break;
}
ft->encoding.bits_per_sample = v->size;
/* setup number of channels */
if (ft->signal.channels == 0)
ft->signal.channels = v->channels;
return (SOX_SUCCESS);
}
static int sox_mp3seek(sox_format_t * ft, uint64_t offset)
{
priv_t * p = (priv_t *) ft->priv;
size_t initial_bitrate = p->Frame.header.bitrate;
size_t tagsize = 0, consumed = 0;
sox_bool vbr = sox_false; /* Variable Bit Rate */
sox_bool depadded = sox_false;
uint64_t to_skip_samples = 0;
/* Reset all */
lsx_rewind(ft);
mad_timer_reset(&p->Timer);
p->FrameCount = 0;
/* They where opened in startread */
mad_synth_finish(&p->Synth);
p->mad_frame_finish(&p->Frame);
p->mad_stream_finish(&p->Stream);
p->mad_stream_init(&p->Stream);
p->mad_frame_init(&p->Frame);
p->mad_synth_init(&p->Synth);
offset /= ft->signal.channels;
to_skip_samples = offset;
while(sox_true) { /* Read data from the MP3 file */
size_t padding = 0;
size_t read;
size_t leftover = p->Stream.bufend - p->Stream.next_frame;
memcpy(p->mp3_buffer, p->Stream.this_frame, leftover);
read = lsx_readbuf(ft, p->mp3_buffer + leftover, p->mp3_buffer_size - leftover);
if (read == 0) {
lsx_debug("seek failure. unexpected EOF (frames=%" PRIuPTR " leftover=%" PRIuPTR ")", p->FrameCount, leftover);
break;
}
for (; !depadded && padding < read && !p->mp3_buffer[padding]; ++padding);
depadded = sox_true;
p->mad_stream_buffer(&p->Stream, p->mp3_buffer + padding, leftover + read - padding);
while (sox_true) { /* Decode frame headers */
static unsigned short samples;
p->Stream.error = MAD_ERROR_NONE;
/* Not an audio frame */
if (p->mad_header_decode(&p->Frame.header, &p->Stream) == -1) {
if (p->Stream.error == MAD_ERROR_BUFLEN)
break; /* Normal behaviour; get some more data from the file */
if (!MAD_RECOVERABLE(p->Stream.error)) {
lsx_warn("unrecoverable MAD error");
break;
}
if (p->Stream.error == MAD_ERROR_LOSTSYNC) {
unsigned available = (p->Stream.bufend - p->Stream.this_frame);
tagsize = tagtype(p->Stream.this_frame, (size_t) available);
if (tagsize) { /* It's some ID3 tags, so just skip */
if (tagsize >= available) {
lsx_seeki(ft, (off_t)(tagsize - available), SEEK_CUR);
depadded = sox_false;
}
p->mad_stream_skip(&p->Stream, min(tagsize, available));
}
else lsx_warn("MAD lost sync");
}
else lsx_warn("recoverable MAD error");
continue;
}
consumed += p->Stream.next_frame - p->Stream.this_frame;
vbr |= (p->Frame.header.bitrate != initial_bitrate);
samples = 32 * MAD_NSBSAMPLES(&p->Frame.header);
p->FrameCount++;
p->mad_timer_add(&p->Timer, p->Frame.header.duration);
if(to_skip_samples <= samples)
{
p->mad_frame_decode(&p->Frame,&p->Stream);
p->mad_synth_frame(&p->Synth, &p->Frame);
p->cursamp = to_skip_samples;
return SOX_SUCCESS;
}
else to_skip_samples -= samples;
/* If not VBR, we can extrapolate frame size */
if (p->FrameCount == 64 && !vbr) {
p->FrameCount = offset / samples;
to_skip_samples = offset % samples;
if (SOX_SUCCESS != lsx_seeki(ft, (off_t)(p->FrameCount * consumed / 64 + tagsize), SEEK_SET))
return SOX_EOF;
/* Reset Stream for refilling buffer */
p->mad_stream_finish(&p->Stream);
p->mad_stream_init(&p->Stream);
break;
}
}
};
return SOX_EOF;
}
static int start_read(sox_format_t * ft)
{
unsigned long header_size_ul = 0, num_samples_ul = 0;
sox_encoding_t encoding = SOX_ENCODING_SIGN2;
size_t header_size, bytes_read;
size_t num_samples = 0;
unsigned bytes_per_sample = 0;
unsigned channels = 1;
unsigned rate = 16000;
char fldname[64], fldtype[16], fldsval[128];
char * buf;
/* Magic header */
if (lsx_reads(ft, fldname, (size_t)8) || strncmp(fldname, "NIST_1A", (size_t)7) != 0) {
lsx_fail_errno(ft, SOX_EHDR, "Sphere header does not begin with magic word `NIST_1A'");
return (SOX_EOF);
}
if (lsx_reads(ft, fldsval, (size_t)8)) {
lsx_fail_errno(ft, SOX_EHDR, "Error reading Sphere header");
return (SOX_EOF);
}
/* Determine header size, and allocate a buffer large enough to hold it. */
sscanf(fldsval, "%lu", &header_size_ul);
buf = lsx_malloc(header_size = header_size_ul);
/* Skip what we have read so far */
header_size -= 16;
if (lsx_reads(ft, buf, header_size) == SOX_EOF) {
lsx_fail_errno(ft, SOX_EHDR, "Error reading Sphere header");
free(buf);
return (SOX_EOF);
}
header_size -= (strlen(buf) + 1);
while (strncmp(buf, "end_head", (size_t)8) != 0) {
if (strncmp(buf, "sample_n_bytes", (size_t)14) == 0)
sscanf(buf, "%63s %15s %u", fldname, fldtype, &bytes_per_sample);
else if (strncmp(buf, "channel_count", (size_t)13) == 0)
sscanf(buf, "%63s %15s %u", fldname, fldtype, &channels);
else if (strncmp(buf, "sample_count ", (size_t)13) == 0)
sscanf(buf, "%53s %15s %lu", fldname, fldtype, &num_samples_ul);
else if (strncmp(buf, "sample_rate ", (size_t)12) == 0)
sscanf(buf, "%53s %15s %u", fldname, fldtype, &rate);
else if (strncmp(buf, "sample_coding", (size_t)13) == 0) {
sscanf(buf, "%63s %15s %127s", fldname, fldtype, fldsval);
if (!strcasecmp(fldsval, "ulaw") || !strcasecmp(fldsval, "mu-law"))
encoding = SOX_ENCODING_ULAW;
else if (!strcasecmp(fldsval, "pcm"))
encoding = SOX_ENCODING_SIGN2;
else {
lsx_fail_errno(ft, SOX_EFMT, "sph: unsupported coding `%s'", fldsval);
free(buf);
return SOX_EOF;
}
}
else if (strncmp(buf, "sample_byte_format", (size_t)18) == 0) {
sscanf(buf, "%53s %15s %127s", fldname, fldtype, fldsval);
if (strcmp(fldsval, "01") == 0) /* Data is little endian. */
ft->encoding.reverse_bytes = MACHINE_IS_BIGENDIAN;
else if (strcmp(fldsval, "10") == 0) /* Data is big endian. */
ft->encoding.reverse_bytes = MACHINE_IS_LITTLEENDIAN;
else if (strcmp(fldsval, "1")) {
lsx_fail_errno(ft, SOX_EFMT, "sph: unsupported coding `%s'", fldsval);
free(buf);
return SOX_EOF;
}
}
if (lsx_reads(ft, buf, header_size) == SOX_EOF) {
lsx_fail_errno(ft, SOX_EHDR, "Error reading Sphere header");
free(buf);
return (SOX_EOF);
}
header_size -= (strlen(buf) + 1);
}
if (!bytes_per_sample)
bytes_per_sample = encoding == SOX_ENCODING_ULAW? 1 : 2;
while (header_size) {
bytes_read = lsx_readbuf(ft, buf, header_size);
if (bytes_read == 0) {
free(buf);
return (SOX_EOF);
}
header_size -= bytes_read;
}
free(buf);
if (ft->seekable) {
/* Check first four bytes of data to see if it's shorten compressed. */
char shorten_check[4];
if (lsx_readchars(ft, shorten_check, sizeof(shorten_check)))
return SOX_EOF;
lsx_seeki(ft, -(off_t)sizeof(shorten_check), SEEK_CUR);
if (!memcmp(shorten_check, "ajkg", sizeof(shorten_check))) {
lsx_fail_errno(ft, SOX_EFMT,
"File uses shorten compression, cannot handle this.");
return (SOX_EOF);
}
}
num_samples = num_samples_ul;
return lsx_check_read_params(ft, channels, (sox_rate_t)rate, encoding,
bytes_per_sample << 3, (uint64_t)num_samples * channels, sox_true);
}
