GT_INLINE gt_compact_dna_string* gt_segmented_sequence_get_block(gt_segmented_sequence* const sequence,const uint64_t position) {
const uint64_t num_block = position/GT_SEQ_ARCHIVE_BLOCK_SIZE;
const uint64_t blocks_used = gt_vector_get_used(sequence->blocks);
// Allocate new blocks (if needed)
if (num_block>=blocks_used) {
uint64_t i;
for (i=blocks_used;i<num_block;++i) {
gt_vector_insert(sequence->blocks,NULL,gt_compact_dna_string*);
}
gt_compact_dna_string* const block = gt_cdna_string_new(GT_SEQ_ARCHIVE_BLOCK_SIZE);
gt_vector_insert(sequence->blocks,block,gt_compact_dna_string*);
return block;
} else {
/*
* Basic line functions
*/
GT_INLINE size_t gt_input_file_dump_to_buffer(gt_input_file* const input_file,gt_vector* const buffer_dst) { // FIXME: If mmap file, internal buffer is just pointers to mem
GT_INPUT_FILE_CHECK(input_file);
// Copy internal file buffer to buffer_dst
const uint64_t chunk_size = input_file->buffer_pos-input_file->buffer_begin;
if (gt_expect_false(chunk_size==0)) return 0;
gt_vector_reserve_additional(buffer_dst,chunk_size);
memcpy(gt_vector_get_mem(buffer_dst,uint8_t)+gt_vector_get_used(buffer_dst),
input_file->file_buffer+input_file->buffer_begin,chunk_size);
gt_vector_add_used(buffer_dst,chunk_size);
// Update position
input_file->buffer_begin=input_file->buffer_pos;
// Return number of written bytes
return chunk_size;
}
GT_INLINE size_t gt_input_file_next_record(
gt_input_file* const input_file,gt_vector* const buffer_dst,gt_string* const first_field,
uint64_t* const num_blocks,uint64_t* const num_tabs) {
GT_INPUT_FILE_CHECK(input_file);
GT_VECTOR_CHECK(buffer_dst);
GT_INPUT_FILE_CHECK_BUFFER__DUMP(input_file,buffer_dst);
if (input_file->eof) return GT_INPUT_FILE_EOF;
// Read line
uint64_t const begin_line_pos_at_file = input_file->buffer_pos;
uint64_t const begin_line_pos_at_buffer = gt_vector_get_used(buffer_dst);
uint64_t current_pfield = 0, length_first_field = 0;
while (gt_expect_true(!input_file->eof &&
GT_INPUT_FILE_CURRENT_CHAR(input_file)!=EOL &&
GT_INPUT_FILE_CURRENT_CHAR(input_file)!=DOS_EOL)) {
if (current_pfield==0) {
if (gt_expect_false(GT_INPUT_FILE_CURRENT_CHAR(input_file)==TAB)) {
++current_pfield;
++(*num_tabs);
} else {
++length_first_field;
}
} else if (current_pfield==1) {
if (gt_expect_false(GT_INPUT_FILE_CURRENT_CHAR(input_file)==SPACE)) {
++(*num_blocks);
} else if (gt_expect_false(GT_INPUT_FILE_CURRENT_CHAR(input_file)==TAB)) {
++current_pfield;
++(*num_tabs);
++(*num_blocks);
}
} else {
if (gt_expect_false(GT_INPUT_FILE_CURRENT_CHAR(input_file)==TAB)) {
++current_pfield;
++(*num_tabs);
}
}
GT_INPUT_FILE_NEXT_CHAR__DUMP(input_file,buffer_dst);
}
// Handle EOL
GT_INPUT_FILE_HANDLE_EOL(input_file,buffer_dst);
// Set first field (from the input_file_buffer or the buffer_dst)
if (first_field) {
char* first_field_begin;
if (input_file->buffer_pos <= begin_line_pos_at_file) {
gt_input_file_dump_to_buffer(input_file,buffer_dst); // Forced to dump to buffer
first_field_begin = gt_vector_get_elm(buffer_dst,begin_line_pos_at_buffer,char);
} else {
GT_INLINE void gt_sam_header_add_sequence_record(gt_sam_headers* const sam_headers,gt_sam_header_record* const header_record) {
GT_SAM_HEADERS_CHECK(sam_headers);
gt_string *sn_tag=gt_sam_header_record_get_tag(header_record,"SN");
gt_cond_error(!sn_tag,PARSE_SAM_HEADER_MISSING_TAG,"SQ","SN");
gt_cond_error(!gt_sam_header_record_get_tag(header_record,"LN"),PARSE_SAM_HEADER_MISSING_TAG,"SQ","LN");
if(sn_tag) {
if(!sam_headers->sequence_dictionary_sn_hash) sam_headers->sequence_dictionary_sn_hash=gt_shash_new();
char *sn_str=gt_string_get_string(sn_tag);
size_t* ix=gt_shash_get_element(sam_headers->sequence_dictionary_sn_hash,sn_str);
// If SN Tag already exists, overwrite.
if(ix) {
gt_sam_header_record_delete(*(gt_sam_header_record **)gt_vector_get_elm(sam_headers->sequence_dictionary,*ix,gt_sam_header_record*));
gt_vector_set_elm(sam_headers->sequence_dictionary,*ix,gt_sam_header_record*,header_record);
gt_error(PARSE_SAM_HEADER_DUPLICATE_TAG,"SQ","SN",sn_str);
} else {
ix=gt_alloc(size_t);
*ix=gt_vector_get_used(sam_headers->sequence_dictionary);
gt_shash_insert(sam_headers->sequence_dictionary_sn_hash,sn_str,ix,size_t*);
gt_vector_insert(sam_headers->sequence_dictionary,header_record,gt_sam_header_record*);
}
}
GT_INLINE bool gt_buffered_input_file_eob(gt_buffered_input_file* const buffered_input_file) {
GT_BUFFERED_INPUT_FILE_CHECK(buffered_input_file);
return gt_buffered_input_file_get_cursor_pos(buffered_input_file) >= gt_vector_get_used(buffered_input_file->block_buffer);
}
