findhex

findhex.c (1668B)

#include <err.h>
#include <string.h>
#include <stdint.h>
#include <stdio.h>
#include <stdlib.h>

int
main(int argc, const char **argv)
{
	uint8_t *pattern, *pattern_mask;
	size_t patlen;
	uint8_t *buf;
	size_t buflen;
	ssize_t nread;
	size_t i, k;
	size_t pos, off;
	size_t arglen;
	FILE *file;

	setvbuf(stdout, NULL, _IONBF, 0);

	if (argc < 2 || argc > 3) {
		printf("Usage: findhex PATTERN [FILE]\n");
		return 0;
	}

	arglen = strlen(argv[1]);
	if (!arglen || arglen % 2 != 0) {
		fprintf(stderr, "Pattern length invalid\n");
		return 1;
	}
	patlen = arglen / 2;

	pattern = malloc(patlen);
	if (!pattern) err(1, "malloc");

	pattern_mask = malloc(patlen);
	if (!pattern_mask) err(1, "malloc");

	buflen = 16 * 1024;
	if (patlen * 2 > buflen)
		buflen = patlen * 2;

	buf = malloc(buflen);
	if (!buf) err(1, "malloc");

	for (i = 0; i < arglen; i += 2) {
		if (!strncmp(argv[1] + i, "__", 2))
			continue;
		if (!sscanf(argv[1] + i, "%02hhx", &pattern[i / 2])) {
			fprintf(stderr, "Invalid hex in pattern\n");
			return 1;
		}
		pattern_mask[i / 2] = 0xff;
	}

	if (argc == 3) {
		file = fopen(argv[2], "r");
		if (!file) err(1, "fopen");
	} else {
		file = stdin;
	}

	off = patlen > 0 ? patlen - 1 : 0;
	if (fread(buf, 1, off, file) != off)
		exit(0);

	pos = 0;
	while (1) {
		nread = fread(buf + off, 1, buflen - off, file);
		if (nread <= 0) break;

		for (i = 0; i <= nread + off - patlen; i++) {
			for (k = 0; k < patlen; k++) {
				if (!pattern_mask[k])
					continue;
				if (buf[i+k] != pattern[k])
					break;
			}
			if (k == patlen)
				printf("%016lX\n", pos + i);
		}

		pos += nread;
		memcpy(buf, buf + buflen - off, off);
	}

	fclose(file);
}