static void cvt_string(nvlist_t *nvl, nvpair_t *pair, data_type_t tgt)
{
const char *name = nvpair_name(pair);
uint64_t intval;
char *str;
int ret;
ASSERT3U(nvpair_type(pair), ==, DATA_TYPE_STRING);
ret = nvpair_value_string(pair, &str);
if (ret)
return;
switch (tgt) {
case DATA_TYPE_UINT64:
ret = str2u64(str, &intval);
if (!ret)
nvl_set_int(nvl, name, intval);
break;
default:
panic("unexpected target type");
}
}
static
rc_t CC CGReads15_Header(const CGReads15* cself, const char* buf, const size_t len)
{
rc_t rc = 0;
size_t slen;
CGReads15* self = (CGReads15*)cself;
if( strncmp("ASSEMBLY_ID\t", buf, slen = 12) == 0 ) {
rc = str2buf(&buf[slen], len - slen, self->assembly_id, sizeof(self->assembly_id));
} else if( strncmp("BATCH_FILE_NUMBER\t", buf, slen = 18) == 0 ) {
rc = str2u32(&buf[slen], len - slen, &self->batch_file_number);
if( self->batch_file_number < 1 ) {
rc = RC(rcRuntime, rcFile, rcConstructing, rcItem, rcOutofrange);
}
} else if( strncmp("BATCH_OFFSET\t", buf, slen = 13) == 0 ) {
rc = str2u64(&buf[slen], len - slen, &self->batch_offset);
} else if( strncmp("FIELD_SIZE\t", buf, slen = 11) == 0 ) {
rc = str2u32(&buf[slen], len - slen, &self->field_size);
} else if( strncmp("GENERATED_AT\t", buf, slen = 13) == 0 ) {
rc = str2buf(&buf[slen], len - slen, self->generated_at, sizeof(self->generated_at));
} else if( strncmp("GENERATED_BY\t", buf, slen = 13) == 0 ) {
rc = str2buf(&buf[slen], len - slen, self->generated_by, sizeof(self->generated_by));
} else if( strncmp("LANE\t", buf, slen = 5) == 0 ) {
rc = str2buf(&buf[slen], len - slen, self->lane, sizeof(self->lane));
} else if( strncmp("LIBRARY\t", buf, slen = 8) == 0 ) {
rc = str2buf(&buf[slen], len - slen, self->library, sizeof(self->library));
} else if( strncmp("SAMPLE\t", buf, slen = 7) == 0 ) {
rc = str2buf(&buf[slen], len - slen, self->sample, sizeof(self->sample));
} else if( strncmp("SLIDE\t", buf, slen = 6) == 0 ) {
rc = str2buf(&buf[slen], len - slen, self->slide, sizeof(self->slide));
} else if( strncmp("SOFTWARE_VERSION\t", buf, slen = 17) == 0 ) {
rc = str2buf(&buf[slen], len - slen, self->software_version, sizeof(self->software_version));
} else {
rc = RC(rcRuntime, rcFile, rcConstructing, rcName, rcUnrecognized);
}
return rc;
}
static rc_t CC CGReads25_Header(const CGReads15* cself,
const char* buf, const size_t len)
{
rc_t rc = 0;
size_t slen = 0;
CGReads15* self = (CGReads15*)cself;
/* from SRA-2617 files */
if (strncmp("APPROVAL\t", buf, slen = 9) == 0) {
}
else if (strncmp("TITLE\t", buf, slen = 6) == 0) {
}
else if (strncmp("ADDRESS\t", buf, slen = 8) == 0) {
}
/* From Table 1: Header Metadata Present in all Data Files */
else if (strncmp("CUSTOMER_SAMPLE_ID\t", buf, slen = 19) == 0) {
}
else if (strncmp("SAMPLE_SOURCE\t", buf, slen = 14) == 0) {
}
else if (strncmp("REPORTED_GENDER\t", buf, slen = 16) == 0) {
}
else if (strncmp("CALLED_GENDER\t", buf, slen = 14) == 0) {
}
else if (strncmp("TUMOR_STATUS\t", buf, slen = 13) == 0) {
}
else if (strncmp("LIBRARY_TYPE\t", buf, slen = 13) == 0) {
}
else if (strncmp("LIBRARY_SOURCE\t", buf, slen = 13) == 0) {
}
else if (strncmp("ASSEMBLY_ID\t", buf, slen = 12) == 0) {
rc = str2buf(&buf[slen], len - slen,
self->assembly_id, sizeof(self->assembly_id));
}
else if (strncmp("BATCH_FILE_NUMBER\t", buf, slen = 18) == 0) {
rc = str2u32(&buf[slen], len - slen, &self->batch_file_number);
if (self->batch_file_number < 1) {
rc = RC(rcRuntime, rcFile, rcConstructing, rcItem, rcOutofrange);
}
}
else if (strncmp("BATCH_OFFSET\t", buf, slen = 13) == 0) {
rc = str2u64(&buf[slen], len - slen, &self->batch_offset);
}
else if (strncmp("FIELD_SIZE\t", buf, slen = 11) == 0) {
rc = str2u32(&buf[slen], len - slen, &self->field_size);
}
else if (strncmp("GENERATED_AT\t", buf, slen = 13) == 0) {
rc = str2buf(&buf[slen], len - slen,
self->generated_at, sizeof(self->generated_at));
}
else if (strncmp("GENERATED_BY\t", buf, slen = 13) == 0) {
rc = str2buf(&buf[slen], len - slen,
self->generated_by, sizeof(self->generated_by));
}
else if (strncmp("LANE\t", buf, slen = 5) == 0) {
rc = str2buf(&buf[slen], len - slen,
self->lane, sizeof(self->lane));
}
else if (strncmp("LIBRARY\t", buf, slen = 8) == 0) {
rc = str2buf(&buf[slen], len - slen,
self->library, sizeof(self->library));
}
else if (strncmp("SAMPLE\t", buf, slen = 7) == 0) {
rc = str2buf(&buf[slen], len - slen,
self->sample, sizeof(self->sample));
}
else if (strncmp("SLIDE\t", buf, slen = 6) == 0) {
rc = str2buf(&buf[slen], len - slen, self->slide, sizeof(self->slide));
}
else if (strncmp("SOFTWARE_VERSION\t", buf, slen = 17) == 0) {
rc = str2buf(&buf[slen], len - slen,
self->software_version, sizeof(self->software_version));
}
else {
rc = RC(rcRuntime, rcFile, rcConstructing, rcName, rcUnrecognized);
}
return rc;
}
static
rc_t CC CGEvidenceDnbs_Read(const CGEvidenceDnbs15* cself, const char* interval_id, TEvidenceDnbsData* data, int score_allele_num)
{
rc_t rc = 0;
TEvidenceDnbsData_dnb* m = NULL;
static TEvidenceDnbsData_dnb next_rec;
static char next_interval_id[32] = "";
/* local copy of unused TEvidenceDnbsData_dnb struct elements */
char reference_alignment[CG_EVDNC_ALLELE_CIGAR_LEN];
INSDC_coord_zero mate_offset_in_reference;
char mate_reference_alignment[CG_EVDNC_ALLELE_CIGAR_LEN];
uint16_t score_allele[4] = {0, 0, 0, 0}; /* v1.5 has ScoreAllele[012]; v2.0 - [0123] */
char qual[CG_EVDNC_SPOT_LEN];
strcpy(data->interval_id, interval_id);
data->qty = 0;
/* already read one rec for this interval_id */
if( next_interval_id[0] != '\0' ) {
if( strcmp(next_interval_id, interval_id) != 0 ) {
/* nothing todo since next interval id is different */
return rc;
}
m = &data->dnbs[data->qty++];
memcpy(m, &next_rec, sizeof(next_rec));
DEBUG_MSG(10, ("%3u evidenceDnbs: '%s'\t'%s'\t'%s'\t'%s'\t%u\t%lu\t%hu\t%c\t%c\t%i\t'%.*s'"
"\t%i\tnot_used\t0\tnot_used\t%c\t0\t0\t0\t'%.*s'\t'--'\n",
data->qty, next_interval_id, m->chr, m->slide, m->lane, m->file_num_in_lane,
m->dnb_offset_in_lane_file, m->allele_index, m->side, m->strand, m->offset_in_allele,
m->allele_alignment_length, m->allele_alignment, m->offset_in_reference,
m->mapping_quality, m->read_len, m->read));
}
do {
int i = 0;
char tmp[2];
CG_LINE_START(cself->file, b, len, p);
if( b == NULL || len == 0 ) {
next_interval_id[0] = '\0';
break; /* EOF */
}
if( data->qty >= data->max_qty ) {
TEvidenceDnbsData_dnb* x;
data->max_qty += 100;
x = realloc(data->dnbs, sizeof(*(data->dnbs)) * data->max_qty);
if( x == NULL ) {
rc = RC(rcRuntime, rcFile, rcReading, rcMemory, rcExhausted);
break;
}
data->dnbs = x;
}
m = &data->dnbs[data->qty++];
/*DEBUG_MSG(10, ("%2hu evidenceDnbs: '%.*s'\n", data->qty, len, b));*/
CG_LINE_NEXT_FIELD(b, len, p);
rc = str2buf(b, p - b, next_interval_id, sizeof(next_interval_id));
CG_LINE_NEXT_FIELD(b, len, p);
rc = str2buf(b, p - b, m->chr, sizeof(m->chr));
CG_LINE_NEXT_FIELD(b, len, p);
rc = str2buf(b, p - b, m->slide, sizeof(m->slide));
CG_LINE_NEXT_FIELD(b, len, p);
rc = str2buf(b, p - b, m->lane, sizeof(m->lane));
CG_LINE_NEXT_FIELD(b, len, p);
rc = str2u32(b, p - b, &m->file_num_in_lane);
CG_LINE_NEXT_FIELD(b, len, p);
rc = str2u64(b, p - b, &m->dnb_offset_in_lane_file);
CG_LINE_NEXT_FIELD(b, len, p);
rc = str2u16(b, p - b, &m->allele_index);
CG_LINE_NEXT_FIELD(b, len, p);
rc = str2buf(b, p - b, tmp, sizeof(tmp));
if( tmp[0] != 'L' && tmp[0] != 'R' ) {
rc = RC(rcRuntime, rcFile, rcReading, rcData, rcOutofrange);
}
m->side = tmp[0];
CG_LINE_NEXT_FIELD(b, len, p);
rc = str2buf(b, p - b, tmp, sizeof(tmp));
if( tmp[0] != '+' && tmp[0] != '-' ) {
rc = RC(rcRuntime, rcFile, rcReading, rcData, rcOutofrange);
}
m->strand = tmp[0];
CG_LINE_NEXT_FIELD(b, len, p);
rc = str2i32(b, p - b, &m->offset_in_allele);
CG_LINE_NEXT_FIELD(b, len, p);
rc = str2buf(b, p - b, m->allele_alignment, sizeof(m->allele_alignment));
m->allele_alignment_length = p - b;
CG_LINE_NEXT_FIELD(b, len, p);
rc = str2i32(b, p - b, &m->offset_in_reference);
CG_LINE_NEXT_FIELD(b, len, p);
rc = str2buf(b, p - b, reference_alignment, sizeof(reference_alignment));
CG_LINE_NEXT_FIELD(b, len, p);
rc = str2i32(b, p - b, &mate_offset_in_reference);
CG_LINE_NEXT_FIELD(b, len, p);
rc = str2buf(b, p - b, mate_reference_alignment, sizeof(mate_reference_alignment));
CG_LINE_NEXT_FIELD(b, len, p);
rc = str2buf(b, p - b, tmp, sizeof(tmp));
if( tmp[0] < 33 || tmp[0] > 126 ) {
rc = RC(rcRuntime, rcFile, rcReading, rcData, rcOutofrange);
}
m->mapping_quality = tmp[0];
for (i = 0; i < score_allele_num; ++i) {
CG_LINE_NEXT_FIELD(b, len, p);
rc = str2u16(b, p - b, &score_allele[i]);
if(rc){
score_allele[i] =0;
rc =0;
}
}
CG_LINE_NEXT_FIELD(b, len, p);
m->read_len = p - b;
rc = str2buf(b, m->read_len, m->read, sizeof(m->read));
CG_LINE_LAST_FIELD(b, len, p);
if( m->read_len != p - b ) {
rc = RC(rcRuntime, rcFile, rcReading, rcData, rcInconsistent);
} else {
rc = str2buf(b, p - b, qual, sizeof(qual));
}
((CGEvidenceDnbs15*)cself)->records++;
if( strcmp(next_interval_id, data->interval_id) != 0 ) {
if (score_allele_num == 3) {
DEBUG_MSG(10, ("%3u evidenceDnbs: '%s'\t'%s'\t'%s'\t'%s'\t%u\t%lu\t%hu\t%c\t%c\t%i\t'%.*s'"
"\t%i\t'%s'\t%i\t'%s'\t%c\t%hu\t%hu\t%hu\t'%.*s'\t'%s'\n",
data->qty, next_interval_id, m->chr, m->slide, m->lane, m->file_num_in_lane,
m->dnb_offset_in_lane_file, m->allele_index, m->side, m->strand, m->offset_in_allele,
m->allele_alignment_length, m->allele_alignment, m->offset_in_reference,
reference_alignment, mate_offset_in_reference, mate_reference_alignment,
m->mapping_quality, score_allele[0], score_allele[1], score_allele[2], m->read_len, m->read, qual));
}
else if (score_allele_num == 4) {
DEBUG_MSG(10, ("%3u evidenceDnbs: '%s'\t'%s'\t'%s'\t'%s'\t%u\t%lu\t%hu\t%c\t%c\t%i\t'%.*s'"
"\t%i\t'%s'\t%i\t'%s'\t%c\t%hu\t%hu\t%hu\t%hu\t'%.*s'\t'%s'\n",
data->qty, next_interval_id, m->chr, m->slide, m->lane, m->file_num_in_lane,
m->dnb_offset_in_lane_file, m->allele_index, m->side, m->strand, m->offset_in_allele,
m->allele_alignment_length, m->allele_alignment, m->offset_in_reference,
reference_alignment, mate_offset_in_reference, mate_reference_alignment,
m->mapping_quality, score_allele[0], score_allele[1], score_allele[2], score_allele[3], m->read_len, m->read, qual));
}
else { assert(0); }
}
CG_LINE_END();
if( next_interval_id[0] == '\0' ) {
break;
}
if( strcmp(next_interval_id, data->interval_id) != 0 ) {
/* next record is from next interval, remeber it and stop */
memcpy(&next_rec, m, sizeof(next_rec));
data->qty--;
break;
}
} while( rc == 0 );
return rc;
}
