GT_INLINE gt_status gt_buffered_input_file_add_lines_to_block(
gt_buffered_input_file* const buffered_input_file,const uint64_t num_lines) {
GT_BUFFERED_INPUT_FILE_CHECK(buffered_input_file);
register gt_input_file* const input_file = buffered_input_file->input_file;
// Read lines
if (input_file->eof) return GT_BMI_EOF;
register const uint64_t current_position =
buffered_input_file->cursor - gt_vector_get_mem(buffered_input_file->block_buffer,char);
register const uint64_t lines_added =
gt_input_file_get_lines(input_file,buffered_input_file->block_buffer,
gt_expect_true(num_lines)?num_lines:GT_BMI_NUM_LINES);
buffered_input_file->lines_in_buffer += lines_added;
buffered_input_file->cursor = gt_vector_get_elm(buffered_input_file->block_buffer,current_position,char);
return lines_added;
}
GT_INLINE size_t gt_input_file_next_line(gt_input_file* const input_file,gt_vector* const buffer_dst) {
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
while (gt_expect_true(!input_file->eof &&
GT_INPUT_FILE_CURRENT_CHAR(input_file)!=EOL &&
GT_INPUT_FILE_CURRENT_CHAR(input_file)!=DOS_EOL)) {
GT_INPUT_FILE_NEXT_CHAR__DUMP(input_file,buffer_dst);
}
// Handle EOL
GT_INPUT_FILE_HANDLE_EOL(input_file,buffer_dst);
return GT_INPUT_FILE_LINE_READ;
}
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 {
/*
* Handlers
*/
GT_INLINE void gt_cdna_allocate__init_blocks(gt_compact_dna_string* const cdna_string,const uint64_t pos) {
GT_COMPACT_DNA_STRING_CHECK(cdna_string);
if (pos >= cdna_string->length) {
// Check allocated blocks
if (pos >= cdna_string->allocated) gt_cdna_string_resize(cdna_string,pos+(GT_CDNA_BLOCK_CHARS*10));
// Initialize new accessed blocks
register const uint64_t next_block_num = gt_expect_true(cdna_string->length>0) ? ((cdna_string->length-1)/GT_CDNA_BLOCK_CHARS)+1 : 1;
register const uint64_t top_block_num = pos/GT_CDNA_BLOCK_CHARS;
if (next_block_num <= top_block_num) {
register uint64_t i;
register uint64_t* block_mem = GT_CDNA_GET_MEM_BLOCK(cdna_string->bitmaps,next_block_num);
for (i=next_block_num; i<=top_block_num; ++i) {
GT_CDNA_INIT_BLOCK(block_mem);
block_mem+=GT_CDNA_BLOCK_BITMAPS;
}
}
// Update total length
cdna_string->length = pos+1;
}
}
GT_INLINE size_t gt_input_file_fill_buffer(gt_input_file* const input_file) {
#ifdef HAVE_BZLIB
int bzerr;
#endif
GT_INPUT_FILE_CHECK(input_file);
input_file->global_pos += input_file->buffer_size;
input_file->buffer_pos = 0;
input_file->buffer_begin = 0;
if (gt_expect_true(
(input_file->file_type==STREAM && !feof(input_file->file)) ||
(input_file->file_type==REGULAR_FILE && !feof(input_file->file)))) {
input_file->buffer_size =
fread(input_file->file_buffer,sizeof(uint8_t),GT_INPUT_BUFFER_SIZE,input_file->file);
if (input_file->buffer_size==0) {
input_file->eof = true;
}
return input_file->buffer_size;
} else if (input_file->file_type==MAPPED_FILE && input_file->global_pos < input_file->file_size) {
input_file->buffer_size = input_file->file_size-input_file->global_pos;
return input_file->buffer_size;
#ifdef HAVE_ZLIB
} else if (input_file->file_type==GZIPPED_FILE && !gzeof((gzFile)input_file->file)) {
input_file->buffer_size = gzread((gzFile)input_file->file,input_file->file_buffer,GT_INPUT_BUFFER_SIZE);
if (input_file->buffer_size==0) {
input_file->eof = true;
}
return input_file->buffer_size;
#endif
#ifdef HAVE_BZLIB
} else if (input_file->file_type==BZIPPED_FILE) {
input_file->buffer_size = BZ2_bzRead(&bzerr,input_file->file,input_file->file_buffer,GT_INPUT_BUFFER_SIZE);
if(input_file->buffer_size==0) {
input_file->eof=true;
}
return input_file->buffer_size;
#endif
} else {
input_file->eof = true;
return 0;
}
}
GT_INLINE gt_status gt_buffered_input_file_get_block(
gt_buffered_input_file* const buffered_input_file,const uint64_t num_lines) {
GT_BUFFERED_INPUT_FILE_CHECK(buffered_input_file);
register gt_input_file* const input_file = buffered_input_file->input_file;
// Read lines
if (input_file->eof) return GT_BMI_EOF;
gt_input_file_lock(input_file);
if (input_file->eof) {
gt_input_file_unlock(input_file);
return GT_BMI_EOF;
}
buffered_input_file->block_id = gt_input_file_next_id(input_file) % UINT32_MAX;
buffered_input_file->current_line_num = input_file->processed_lines+1;
buffered_input_file->lines_in_buffer =
gt_input_file_get_lines(input_file,buffered_input_file->block_buffer,
gt_expect_true(num_lines)?num_lines:GT_BMI_NUM_LINES);
gt_input_file_unlock(input_file);
// Setup the block
buffered_input_file->cursor = gt_vector_get_mem(buffered_input_file->block_buffer,char);
return buffered_input_file->lines_in_buffer;
}