fq_count_exact_buff.go

// CountFastxRecords counts the number of records in a fastx file.
func CountFastxRecords(fh *os.File, buffSize int, format string) int {
	// Start the timer:
	start := time.Now()
	pattern := []byte{'\n', '+', '\n'} // Pattern to look for.
	// Create new buffered reader:
	reader := bufio.NewReaderSize(fh, buffSize)
	// Byte buffer:
	buffer := make([]byte, buffSize)
	// Auxiliary buffer:
	auxBuff := make([]byte, 0, 2)
	// Variable for counting records:
	var records int

	for {
		// Read next line:
		nrBytes, err := reader.Read(buffer)
		// Anything else went wrong?
		if err != io.EOF {
			checkError(err)
		}
		// Count the number of fastq separator lines in the buffer:
		records += bytes.Count(buffer[:nrBytes], pattern)

		// We have bytes across the buffer boundary:
		if len(auxBuff) > 0 && nrBytes > 0 {
			// Do we have a record separator line across the buffer boundaries (|)?
			auxLast := len(auxBuff) - 1
			if (auxBuff[auxLast] == pattern[0]) && (bytes.Equal(buffer[0:2], pattern[1:3])) {
				// \n|+\n
				records++
			}
			if len(auxBuff) < 2 {
				continue
			}
			if bytes.Equal(auxBuff[auxLast-1:], pattern[:2]) && buffer[0] == pattern[2] {
				// \n+|\n
				records++
			}
		}
		// Save the last bytes:
		saveLen := 2
		if nrBytes < saveLen {
			saveLen = nrBytes
		}
		auxBuff = auxBuff[:0]
		auxBuff = append(auxBuff, buffer[nrBytes-saveLen:nrBytes]...)
		// Are we at the end of the file?
		if err == io.EOF {
			break
		}
	}
	// Print out the time needed for counting:
	fmt.Fprintf(os.Stderr, "%d\n", time.Since(start).Microseconds())
	// Return the number of records:
	return records
}