void core_fasta_input_init(struct biosal_input_format *input)
{
char *file;
struct core_fasta_input *fasta;
uint64_t offset;
fasta = (struct core_fasta_input *)biosal_input_format_implementation(input);
file = biosal_input_format_file(input);
CORE_DEBUGGER_ASSERT(input->operations != NULL);
#if 0
printf("DEBUG BEFORE faulty call.\n");
#endif
offset = biosal_input_format_start_offset(input);
core_buffered_reader_init(&fasta->reader, file, offset);
fasta->buffer = NULL;
fasta->next_header = NULL;
fasta->has_header = 0;
fasta->has_first = 0;
}
uint64_t core_fasta_input_get_offset(struct biosal_input_format *self)
{
struct core_fasta_input *fasta;
fasta = (struct core_fasta_input *)biosal_input_format_implementation(self);
return core_buffered_reader_get_offset(&fasta->reader);
}
void core_fastq_input_destroy(struct biosal_input_format *input)
{
struct core_fastq_input *fastq;
fastq = (struct core_fastq_input *)biosal_input_format_implementation(input);
core_buffered_reader_destroy(&fastq->reader);
if (fastq->buffer != NULL) {
core_memory_free(fastq->buffer, MEMORY_FASTQ);
fastq->buffer = NULL;
}
}
int core_fastq_input_is_identifier(struct biosal_input_format *self, const char *line)
{
int length;
char buffer[2];
int read;
struct core_fastq_input *fastq;
fastq = (struct core_fastq_input *)biosal_input_format_implementation(self);
length = strlen(line);
if (length < 1) {
return 0;
}
if (line[0] != '@') {
return 0;
}
/*
* Now, the line is either a quality string
* or an identifier string since it starts with a @.
*/
read = -1;
read = core_buffered_reader_get_previous_bytes(&fastq->reader,
buffer, 3);
/*
* This is an identifier if nothing is available before.
*/
if (read == 0) {
return 1;
}
/*
* Operation not supported by the driver.
*/
if (read < 0) {
return core_fastq_input_is_identifier_mock(self, line);
}
/*
* Fall back on this method call.
*/
return core_fastq_input_is_identifier_mock(self, line);
}
void core_fastq_input_init(struct biosal_input_format *input)
{
char *file;
struct core_fastq_input *fastq;
uint64_t offset;
file = biosal_input_format_file(input);
offset = biosal_input_format_start_offset(input);
#ifdef BIOSAL_FASTQ_INPUT_DEBUG
printf("DEBUG core_fastq_input_init %s\n",
file);
#endif
fastq = (struct core_fastq_input *)biosal_input_format_implementation(input);
core_buffered_reader_init(&fastq->reader, file, offset);
fastq->buffer = NULL;
fastq->has_first = 0;
}
uint64_t core_fasta_input_get_sequence(struct biosal_input_format *input,
char *sequence)
{
struct core_fasta_input *fasta;
/* TODO use a dynamic buffer to accept long reads... */
int maximum_sequence_length = BIOSAL_INPUT_MAXIMUM_SEQUENCE_LENGTH;
int value;
int lines;
int total;
int position_in_sequence;
int is_header;
int block_length;
fasta = (struct core_fasta_input *)biosal_input_format_implementation(input);
if (fasta->buffer == NULL) {
fasta->buffer = core_memory_allocate(maximum_sequence_length + 1, MEMORY_FASTA);
fasta->next_header= core_memory_allocate(maximum_sequence_length + 1, MEMORY_FASTA);
fasta->buffer[0] = '\0';
fasta->next_header[0] = '\0';
}
value = 0;
total = 0;
lines = 0;
/*
* Read name
*/
if (fasta->has_header) {
strcpy(fasta->buffer, fasta->next_header);
value = strlen(fasta->buffer);
fasta->has_header = 0;
} else {
value = core_buffered_reader_read_line(&fasta->reader, fasta->buffer,
maximum_sequence_length);
/* Make sure that this is an identifier.
*/
if (!fasta->has_first) {
while (!core_fasta_input_check_header(input, fasta->buffer)) {
value = core_buffered_reader_read_line(&fasta->reader, fasta->buffer,
maximum_sequence_length);
}
fasta->has_first = 1;
}
}
/*
* Add new line.
*/
if (value) {
++lines;
}
total += value;
/*
* Read sequence.
*
* Discard any new line symbol too.
*/
position_in_sequence = 0;
while (1) {
value = core_buffered_reader_read_line(&fasta->reader, fasta->buffer,
maximum_sequence_length);
if (value == 0) {
break;
}
is_header = 0;
if (strlen(fasta->buffer) > 0
&& fasta->buffer[0] == '>') {
is_header = 1;
}
if (is_header) {
sequence[position_in_sequence] = '\0';
strcpy(fasta->next_header, fasta->buffer);
fasta->has_header = 1;
break;
}
/*
* Otherwise, add the sequence.
*/
if (value) {
++lines;
}
block_length = strlen(fasta->buffer);
/*
* Remove the new line.
*/
if (fasta->buffer[block_length - 1] == '\n') {
--block_length;
}
core_memory_copy(sequence + position_in_sequence,
fasta->buffer,
block_length);
position_in_sequence += block_length;
}
return total;
}
uint64_t core_fastq_input_get_sequence(struct biosal_input_format *input,
char *sequence)
{
struct core_fastq_input *fastq;
/*
* Input sequence has at least BIOSAL_INPUT_MAXIMUM_SEQUENCE_LENGTH
* which is currently 512k
*/
/* TODO use a dynamic buffer to accept long reads... */
int maximum_sequence_length = BIOSAL_INPUT_MAXIMUM_SEQUENCE_LENGTH;
int value;
int length;
fastq = (struct core_fastq_input *)biosal_input_format_implementation(input);
if (fastq->buffer == NULL) {
fastq->buffer = (char *)core_memory_allocate(maximum_sequence_length + 1, MEMORY_FASTQ);
}
value = 0;
/*
* Read name
*/
value += core_buffered_reader_read_line(&fastq->reader, fastq->buffer,
maximum_sequence_length);
#ifdef FIND_IDENTIFIER
/*
* If we do not have the first entry yet,
* make sure that the line is a good line.
*/
if (!fastq->has_first) {
while (!core_fastq_input_is_identifier(input, fastq->buffer)) {
value += core_buffered_reader_read_line(&fastq->reader, fastq->buffer,
maximum_sequence_length);
}
fastq->has_first = 1;
}
#endif
/*
* Read DNA sequence
*/
length = core_buffered_reader_read_line(&fastq->reader, sequence,
maximum_sequence_length);
#ifdef BIOSAL_FASTQ_INPUT_DEBUG_READ_LINE
printf("FASTQ ReadLine <<%s>>\n", sequence);
#endif
if (sequence[length - 1] == '\n') {
/*
* Remove new line symbol.
*/
sequence[length - 1] = '\0';
}
value += length;
#ifdef BIOSAL_FASTQ_INPUT_DEBUG2
printf("DEBUG core_fastq_input_get_sequence %s\n", buffer);
#endif
/*
* Read the + symbol
*/
value += core_buffered_reader_read_line(&fastq->reader, fastq->buffer,
maximum_sequence_length);
/*
* Read quality string.
*/
value += core_buffered_reader_read_line(&fastq->reader, fastq->buffer,
maximum_sequence_length);
return value;
}
