path: root/wavextract.c

#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <errno.h>
#include <stdint.h>
#include <unistd.h>
#include <fcntl.h>
#include <inttypes.h>

#define VERSION "0.2"

#ifdef _WIN32
	#define WIN32_LEAN_AND_MEAN
	#include <windows.h>
	#define PRIsize_t "llu"
	#define htole16(x) (x)
	#define htole32(x) (x)
#else
	#define PRIsize_t "lu"
#endif

#pragma pack(push, 1)
typedef struct /*_chunk_riff*/ {
	char magic_riff[4];
	uint32_t size;
	char magic_wave[4];	
} hdr_riff_t;

typedef struct /*_chunk_fmt_base*/ {
	char magic[4];
	uint32_t size;
	uint16_t fmt;
	uint16_t channels;
	uint32_t sample_rate;
	uint32_t byte_rate; //(samplerate * sampleBits * channels) / 8.
	uint16_t bitdepth; //(sampleBits * channels) / 8
	uint16_t samplebits;
	//uint16_t extsize; //only present if fmt != 1
} hdr_fmt_t;

typedef struct /*_chunk_fact*/ {
	char magic[4];
	uint32_t size;
	uint32_t samples;
} hdr_fact_t;

typedef struct /*_chunk_data*/ {
	char magic[4];
	uint32_t size;
} hdr_data_t;

typedef union {
	int32_t i32;
	int16_t i16;
	float f32;
	double f64;
	uint8_t bytes[8];
} sample_t;
#pragma pack(pop)

static int usage(char* fn){
	fprintf(stdout, "wavextract " VERSION " - Convert WAVE files to CSV sample data\n");
	fprintf(stdout, "\tUsage: %s <file.wav>\n\n", fn);
	fprintf(stdout, "Supported wave formats: PCM s16le, s32le, f32, f64\n");
	return EXIT_FAILURE;
}

int wave_verify_riff(int fd, hdr_riff_t* hdr){
	if(read(fd, hdr, sizeof(hdr_riff_t)) != sizeof(hdr_riff_t)){
		fprintf(stderr, "Failed to read input data: %s\n", strerror(errno));
		return 1;
	}

	if(memcmp(hdr->magic_riff, "RIFF", 4) || memcmp(hdr->magic_wave, "WAVE", 4)){
		fprintf(stderr, "Invalid RIFF/WAVE magic - probably not a WAVE file\n");
		return 1;
	}

	fprintf(stderr, "RIFF chunk size: %d\n", hdr->size);
	return 0;
}

int wave_verify_fmt(int fd, hdr_fmt_t* hdr){
	uint16_t extra_data;
	uint8_t skip_byte;
	size_t n;

	if(read(fd, hdr, sizeof(hdr_fmt_t)) != sizeof(hdr_fmt_t)){
		fprintf(stderr, "Failed to read input data: %s\n", strerror(errno));
		return 1;
	}

	if(memcmp(hdr->magic, "fmt ", 4)){
		fprintf(stderr, "Invalid format header magic - format probably incompatible\n");
		return 1;
	}

	fprintf(stderr, "Format chunk size: %d\n", hdr->size);
	fprintf(stderr, "Format: %d\n", hdr->fmt);
	fprintf(stderr, "Channels: %d\n", hdr->channels);
	fprintf(stderr, "Samplerate: %d\n", hdr->sample_rate);
	fprintf(stderr, "Byterate: %d\n", hdr->byte_rate);
	fprintf(stderr, "Align: %d\n", hdr->bitdepth);
	fprintf(stderr, "Samplebits: %d\n", hdr->samplebits);

	if(hdr->fmt != 1){
		if(read(fd, &extra_data, 2) != 2){
			fprintf(stderr, "Failed to read input data: %s\n", strerror(errno));
			return 1;
		}
	}
	else if(hdr->size > 16 /* length of fmt header as counted from after the size member */){
		fprintf(stderr, "Skipping %d additional header bytes:", hdr->size - 16);
		for(n = 16; n < hdr->size; n++){
			if(read(fd, &skip_byte, 1) != 1){
				fprintf(stderr, "Failed to read input data: %s\n", strerror(errno));
				return 1;
			}
			fprintf(stderr, " %02X", skip_byte);
		}
		fprintf(stderr, "\n");
	}

	return 0;
}

int wave_verify_fact(int fd, hdr_fact_t* hdr){
	if(read(fd, hdr, sizeof(hdr_fact_t)) != sizeof(hdr_fact_t)){
		fprintf(stderr, "Failed to read input data: %s\n", strerror(errno));
		return 1;
	}

	if(memcmp(hdr->magic, "fact", 4)){
		fprintf(stderr, "Invalid fact magic\n");
		return 1;
	}

	fprintf(stderr, "Fact chunk size: %d\n", hdr->size);
	fprintf(stderr, "%d samples per channel\n", hdr->samples);
	return 0;
}

int wave_verify_data(int fd, hdr_data_t* hdr, size_t sample_bits, size_t channels){
	if(read(fd, hdr, sizeof(hdr_data_t)) != sizeof(hdr_data_t)){
		fprintf(stderr, "Failed to read input data: %s\n", strerror(errno));
		return 1;
	}

	if(memcmp(hdr->magic, "data", 4)){
		fprintf(stderr, "Invalid data magic\n");
		return 1;
	}

	fprintf(stderr, "%d bytes of sample data in file\n", hdr->size);
	if(hdr->size % (sample_bits / 8 * channels)){
		fprintf(stderr, "Sample data unaligned, unable to decode\n");
		return 1;
	}
	fprintf(stderr, "Data chunk maps to %" PRIsize_t " samples of data per channel\n", hdr->size / (sample_bits / 8 * channels));
	return 0;
}

int print_sample(sample_t sample, uint16_t format, uint16_t samplebits){
	if(format == 1){
		if(samplebits == 16){
			printf(",%d", sample.i16);
			return 0;
		}
		else if(samplebits == 32){
			printf(",%d", sample.i32);
			return 0;
		}
	}
	else if(format == 3){
		if(samplebits == 32){
			printf(",%.9g", sample.f32);
			return 0;
		}
		else if(samplebits == 64){
			printf(",%.17g", sample.f64);
			return 0;
		}
	}
	return 1;
}

int main(int argc, char** argv){
	#ifdef _WIN32
	_fmode = _O_BINARY;
	#endif

	if(argc < 2){
		return usage(argv[0]);
	}

	fprintf(stderr, "Opening wave input file %s\n", argv[1]);
	int source_fd = open(argv[1], O_RDONLY);
	if(source_fd < 0){
		fprintf(stderr, "Failed to open input file %s: %s\n", argv[1], strerror(errno));
		return EXIT_FAILURE;
	}

	hdr_riff_t riff;
	hdr_fmt_t fmt;
	hdr_fact_t fact;
	hdr_data_t data;

	if(wave_verify_riff(source_fd, &riff) || wave_verify_fmt(source_fd, &fmt)){
		close(source_fd);
		return EXIT_FAILURE;
	}

	//non-pcm formats have an additional fact header
	if(fmt.fmt != 1 && wave_verify_fact(source_fd, &fact)){
		close(source_fd);
		return EXIT_FAILURE;
	}

	if(wave_verify_data(source_fd, &data, fmt.samplebits, fmt.channels)){
		close(source_fd);
		return EXIT_FAILURE;
	}

	//compression currently unsupported
	if((fmt.fmt != 1 && fmt.fmt != 3) || fmt.samplebits / 8 > sizeof(sample_t)){
		close(source_fd);
		fprintf(stderr, "This format is currently not supported\n");
		return EXIT_FAILURE;
	}

	sample_t sample;
	size_t channel = 0;
	size_t sample_offset = 0;

	for(sample_offset = 0; sample_offset < data.size / (fmt.samplebits / 8 * fmt.channels); sample_offset++){
		//read sample
		//TODO endianness conversion
		if(read(source_fd, &sample, fmt.samplebits / 8) != fmt.samplebits / 8){
			close(source_fd);
			fprintf(stderr, "Failed to read input data at sample %" PRIsize_t ": %s\n", sample_offset, strerror(errno));
			return EXIT_FAILURE;	
		}

		if(channel == 0){
			printf("%" PRIsize_t "", sample_offset);
		}

		print_sample(sample, fmt.fmt, fmt.samplebits);

		if(channel == fmt.channels - 1){
			printf("\n");
		}

		channel += 1;
		channel %= fmt.channels;
	}

	close(source_fd);
	fprintf(stderr, "Done, exported %" PRIsize_t " frames of data for %d channels, processed %" PRIsize_t " samples\n", sample_offset, fmt.channels, fmt.channels * sample_offset);
	return EXIT_SUCCESS;
}