static
rc_t read_spot_coord(IlluminaFileInfo* file, const char* data, size_t data_sz, const char** tail)
{
rc_t rc = 0;
const char* t, *str = data, *end = data + data_sz;
int tabs = 0;
if( tail ) {
*tail = NULL;
}
do {
if( (t = memchr(str, '\t', end - str)) != NULL ) {
switch(++tabs) {
case 1:
errno = 0;
file->coord[0] = strtol(str, NULL, 10);
if( errno != 0 ) {
file->coord[0] = 0;
}
rc = pstring_assign(&file->name, str, t - str);
break;
case 2:
case 3:
case 4:
errno = 0;
file->coord[tabs - 1] = strtol(str, NULL, 10);
if( errno != 0 ) {
file->coord[tabs - 1] = 0;
}
if( (rc = pstring_append(&file->name, ":", 1)) == 0 ) {
rc = pstring_append(&file->name, str, t - str);
}
if( tail ) {
*tail = t + 1;
}
break;
}
str = ++t;
}
} while( rc == 0 && t != NULL && str < end && tabs < 4 );
if( tabs < 4 ) {
rc = RC(rcSRA, rcFormatter, rcReading, rcData, rcTooShort);
}
return rc;
}
/*
* read fasta or quality, which maybe wrapped on 70th column width,
* into asciiZ buffer
*/
static
rc_t read_multiline_seq_or_qual(FastqFileInfo* file, const char stop, pstring* str)
{
rc_t rc = 0;
bool append = false, optional = false;
while( rc == 0 ) {
if( (rc = file_read_line(file, optional)) == 0 ) {
if( optional && (file->line == NULL || (file->line_len > 0 && file->line[0] == stop)) ) {
/* eof or next line is defline -> stop, line stays in buffer */
break;
}
if( append && memchr(str->data, ' ', str->len) != NULL ) {
rc = pstring_append(str, " ", 1);
}
if( rc == 0 && (rc = pstring_append(str, file->line, file->line_len)) == 0 ) {
file->line = NULL; /* line processed */
optional = true;
}
append = true;
}
}
return rc;
}
static
rc_t fe_new_read(fe_context_t *self, int flags, pstring *readId )
{
rc_t rc;
char *suffix;
pstring readName, spotGroup;
static IlluminaSpot spot;
/* look for spot group */
suffix = strchr(readId->data, '#');
if( suffix != NULL ) {
readId->len = suffix++ - readId->data;
if( (rc = pstring_assign(&spotGroup, suffix, strlen(suffix))) != 0 ) {
SRALoaderFile_LOG(self->ctx.file, klogInt, rc,
"extracting barcode from spot '$(spotname)'", "spotname=%s", readId->data);
return rc;
}
} else {
pstring_clear(&spotGroup);
}
/* build the read name from prefix (self->name_prefix) and read id */
if(self->name_prefix.len > 0 ) {
if( (rc = pstring_copy(&readName, &self->name_prefix)) == 0 ) {
if( isdigit(readName.data[readName.len - 1]) ) {
rc = pstring_append(&readName, ":", 1);
}
if( rc == 0 ) {
rc = pstring_concat(&readName, readId);
}
}
} else {
rc = pstring_copy(&readName, readId);
}
if( rc != 0 ) {
SRALoaderFile_LOG(self->ctx.file, klogErr, rc,
"preparing spot name $(spotname)", "spotname=%s", readId->data);
return rc;
}
SRF_set_read_filter(&self->read.filter, flags);
IlluminaSpot_Init(&spot);
if( (rc = IlluminaSpot_Add(&spot, &readName, &spotGroup, &self->read)) == 0 ) {
rc = SRAWriterIllumina_Write(self->writer, self->ctx.file, &spot);
}
return rc;
}
rc_t SRAWriteAbsolid_MakeName(const pstring* prefix, const pstring* suffix, pstring* name)
{
rc_t rc = 0;
if( prefix == NULL || name == NULL ) {
rc = RC(rcSRA, rcFormatter, rcParsing, rcParam, rcNull);
} else if( (rc = pstring_copy(name, prefix)) == 0 ) {
if( suffix && suffix->len > 0 ) {
if( name->len > 0 && name->data[name->len - 1] != '_' && suffix->data[0] != '_' ) {
rc = pstring_append(name, "_", 1);
}
if( rc == 0 ) {
pstring_concat(name, suffix);
}
}
}
if( rc != 0 ) {
LOGERR(klogErr, rc, "preparing spot name");
}
return rc;
}
rc_t pstring_concat(pstring* dst, const pstring* src)
{
return pstring_append(dst, src->data, src->len);
}
/* reads from a file data for a sinlge spot, data may be partial */
static
rc_t read_next_spot(const char* blk_pfx, IlluminaFileInfo* file)
{
rc_t rc = 0;
const char* tail = file->line;
if( file->ready ) {
/* data still not used */
return 0;
}
IlluminaFileInfo_init(file);
if( (rc = file_read_line(file, true)) != 0 ) {
return SRALoaderFile_LOG(file->file, klogErr, rc, "$(msg)", "msg=reading more data");
} else if( file->line == NULL ) {
return 0; /* eof */
}
switch( file->type ) {
case eIlluminaNativeFileTypeQSeq:
if( (rc = parse_qseq(file, file->line, file->line_len)) != 0 ) {
return SRALoaderFile_LOG(file->file, klogErr, rc, "$(msg)", "msg=reading qseq");
}
break;
case eIlluminaNativeFileTypeFasta:
case eIlluminaNativeFileTypeNoise:
case eIlluminaNativeFileTypeIntensity:
case eIlluminaNativeFileTypeSignal:
{{
/* read only common first 4 coords into name and prepend with DATA_BLOCK/@name */
if( (rc = read_spot_coord(file, file->line, file->line_len, &tail)) == 0 ) {
if( blk_pfx != NULL ) {
pstring tmp_name;
if( (rc = pstring_copy(&tmp_name, &file->name)) == 0 &&
(rc = pstring_assign(&file->name, blk_pfx, strlen(blk_pfx))) == 0 &&
(rc = pstring_append(&file->name, ":", 1)) == 0 ) {
rc = pstring_concat(&file->name, &tmp_name);
}
}
}
if( rc != 0 ) {
return SRALoaderFile_LOG(file->file, klogErr, rc, "$(msg)", "msg=reading spot coord");
}
break;
}}
case eIlluminaNativeFileTypeQuality4:
if( (rc = read_quality(file->line, file->line_len, &file->read)) != 0 ) {
return SRALoaderFile_LOG(file->file, klogErr, rc, "$(msg)", "msg=reading quality");
} else if( (rc = pstring_assign(&file->name, blk_pfx, strlen(blk_pfx))) != 0 ) {
return SRALoaderFile_LOG(file->file, klogErr, rc, "$(msg)", "msg=name for quality 4");
}
break;
default:
rc = RC(rcSRA, rcFormatter, rcReading, rcFileFormat, rcUnknown);
return SRALoaderFile_LOG(file->file, klogErr, rc, "$(msg)", "msg=processing data line");
break;
}
/* process tail (after coords) for some file types */
file->line_len -= tail - file->line; /* length of tail */
switch( file->type ) {
case eIlluminaNativeFileTypeQSeq:
case eIlluminaNativeFileTypeQuality4:
default:
/* completely processed before */
break;
case eIlluminaNativeFileTypeFasta:
if( (rc = pstring_assign(&file->read.seq, tail, file->line_len)) != 0 ||
!pstring_is_fasta(&file->read.seq) ) {
rc = RC(rcSRA, rcFormatter, rcReading, rcData, rcCorrupt);
return SRALoaderFile_LOG(file->file, klogErr, rc, "$(msg)", "msg=reading fasta");
}
break;
case eIlluminaNativeFileTypeNoise:
if( (rc = read_signal(tail, file->line_len, &file->read.noise)) != 0 ) {
return SRALoaderFile_LOG(file->file, klogErr, rc, "$(msg)", "msg=converting noise");
}
break;
case eIlluminaNativeFileTypeIntensity:
if( (rc = read_signal(tail, file->line_len, &file->read.intensity)) != 0 ) {
return SRALoaderFile_LOG(file->file, klogErr, rc, "$(msg)", "msg=converting intensity");
}
break;
case eIlluminaNativeFileTypeSignal:
if( (rc = read_signal(tail, file->line_len, &file->read.signal)) != 0 ) {
return SRALoaderFile_LOG(file->file, klogErr, rc, "$(msg)", "msg=converting signal");
}
break;
}
file->line = NULL;
file->ready = true;
#if _DEBUGGING
DEBUG_MSG(3, ("name:'%s' [%li:%li:%li:%li]\n", file->name.data,
file->coord[0], file->coord[1], file->coord[2], file->coord[3]));
if( file->read.seq.len ) {
DEBUG_MSG(3, ("seq:'%.*s'\n", file->read.seq.len, file->read.seq.data));
}
if( file->read.qual.len ) {
DEBUG_MSG(3, ("qual{0x%x}: %u bytes\n", file->read.qual_type, file->read.qual.len));
}
#endif
return 0;
}
/*
* assumes tab separated file:
* first 2 postiions concatinated with "_" into spot prefix
* nextg 4 postiions concatinated with ":" into spot id: lane:tile:x:y
* 7th (index) ignored
* 8th is read id
* 9th fasta
* 10th quality
* 11th (optional) read filter
*/
static
rc_t parse_qseq(IlluminaFileInfo* file, const char* data, size_t data_sz)
{
rc_t rc = 0;
const char* t, *str = data, *end = data + data_sz;
int tabs = 0;
do {
if( (t = memchr(str, '\t', end - str)) != NULL ) {
switch(++tabs) {
case 1:
rc = pstring_assign(&file->name, str, t - str);
break;
case 2:
if( (rc = pstring_append(&file->name, "_", 1)) == 0 ) {
rc = pstring_append(&file->name, str, t - str);
}
break;
case 3:
case 4:
case 5:
case 6:
errno = 0;
file->coord[tabs - 3] = strtol(str, NULL, 10);
if( errno != 0 ) {
file->coord[tabs - 3] = 0;
}
if( (rc = pstring_append(&file->name, ":", 1)) == 0 ) {
rc = pstring_append(&file->name, str, t - str);
}
break;
case 7:
if( t - str != 1 || (*str != '0' && *str != '1') ) {
rc = pstring_assign(&file->barcode, str, t - str);
}
break;
case 8:
if( t - str != 1 || !isdigit(*str) ) {
rc = RC(rcSRA, rcFormatter, rcReading, rcData, rcInvalid);
} else {
file->read.read_id = *str - '0';
if( file->read.read_id == 0 ) {
file->read.read_id = ILLUMINAWRITER_READID_NONE;
}
}
break;
case 9:
rc = pstring_assign(&file->read.seq, str, t - str);
break;
case 10:
file->read.qual_type = ILLUMINAWRITER_COLMASK_QUALITY_PHRED;
rc = pstring_assign(&file->read.qual, str, t - str);
break;
}
str = ++t;
}
} while( rc == 0 && t != NULL && str < end );
if( rc == 0 ) {
if( tabs == 9 ) {
file->read.qual_type = ILLUMINAWRITER_COLMASK_QUALITY_PHRED;
rc = pstring_assign(&file->read.qual, str, end - str);
} else if( tabs == 10 ) {
if( end - str != 1 ) {
rc = RC(rcSRA, rcFormatter, rcReading, rcData, rcInvalid);
} else if( *str == '1' ) {
file->read.filter = SRA_READ_FILTER_PASS;
} else if( *str == '0' ) {
file->read.filter = SRA_READ_FILTER_REJECT;
} else {
rc = RC(rcSRA, rcFormatter, rcReading, rcData, rcInvalid);
}
} else {
rc = RC(rcSRA, rcFormatter, rcReading, rcData, rcInvalid);
}
if( rc == 0 ) {
if( file->read.seq.len != file->read.qual.len ) {
rc = RC(rcSRA, rcFormatter, rcReading, rcData, rcInconsistent);
} else {
rc = pstring_quality_convert(&file->read.qual, eExperimentQualityEncoding_Ascii, 64, 0, 0x7F);
}
}
}
return rc;
}