bgrep

bgrep.c (3981B)

#define PCRE2_CODE_UNIT_WIDTH 8

#include <pcre2.h>

#include <unistd.h>

#include <stdio.h>
#include <stdarg.h>
#include <string.h>
#include <stdbool.h>
#include <stdlib.h>

static const char *marker = "";
static const char *pattern = NULL;
static bool overlap = false;
static size_t group = 0;

static void
die(const char *fmt, ...)
{
	va_list ap;

	va_start(ap, fmt);
	fputs("bgrep: ", stderr);
	vfprintf(stderr, fmt, ap);
	if (fmt[strlen(fmt)-1] == ':') {
		perror(NULL);
	} else {
		putc('\n', stderr);
	}
	va_end(ap);

	exit(1);
}

static void
pcre2_die(int rc, const char *fmt, ...)
{
	char error_buf[256];
	va_list ap;

	pcre2_get_error_message(rc, (uint8_t *) error_buf, sizeof(error_buf));

	va_start(ap, fmt);
	fputs("bgrep: ", stderr);
	vfprintf(stderr, fmt, ap);
	fprintf(stderr, "%s\n", error_buf);
	va_end(ap);

	exit(1);
}

static void
writeall(uint8_t *data, size_t size)
{
	size_t sent;
	ssize_t n;

	sent = 0;
	while (sent != size) {
		n = write(1, data + sent, size - sent);
		if (n < 0) die("write:");
		if (!n) break;
		sent += (size_t) n;
	}
}

static void
writef(const char *fmt, ...)
{
	static void *buf = NULL;
	static size_t bufsize = 0;
	va_list ap;
	va_list cpy;
	size_t size;

	va_start(ap, fmt);
	va_copy(cpy, ap);
	size = vsnprintf(NULL, 0, fmt, ap);
	if (!bufsize || size + 1 > bufsize) {
		bufsize = size + 1;
		buf = realloc(buf, bufsize);
		if (!buf) die("realloc:");
	}
	va_end(ap);

	va_start(cpy, fmt);
	vsnprintf(buf, bufsize, fmt, cpy);
	va_end(cpy);

	writeall(buf, size);
}

static void
parse(int argc, const char **argv)
{
	const char **arg;

	if (argc < 2) goto usage;

	for (arg = argv + 1; *arg; arg++) {
		if (!strcmp(*arg, "-h") || !strcmp(*arg, "--help")) {
			goto usage;
		} else if (!strcmp(*arg, "-g") || !strcmp(*arg, "--group")) {
			group = strtoul(*++arg, NULL, 0);
		} else if (!strcmp(*arg, "-m") || !strcmp(*arg, "--marker")) {
			marker = *++arg;
		} else if (!strcmp(*arg, "-o") || !strcmp(*arg, "--overlap")) {
			overlap = true;
		} else {
			if (pattern) goto usage;
			pattern = *arg;
		}
	}
	return;

usage:
	fprintf(stderr, "Usage: bgrep [-g GROUP] [-m MARKER] REGEX\n");
	exit(1);
}

int
main(int argc, const char **argv)
{
	pcre2_match_data *match_data;
	pcre2_code *regex = NULL;
	uint8_t *match;
	uint8_t *file;
	size_t *ovector;
	size_t filesize;
	size_t filecap;
	size_t match_size;
	size_t error_offset = 0;
	size_t offset;
	size_t length;
	size_t pos;
	size_t lastmatch;
	ssize_t n;
	int rc;

	parse(argc, argv);

	regex = pcre2_compile((PCRE2_SPTR8)pattern, strlen(pattern),
		PCRE2_MATCH_INVALID_UTF, &rc, &error_offset, NULL);
	if (!regex) pcre2_die(rc, "bad regex (..'%s')", argv[argc-1] + error_offset);

	match_data = pcre2_match_data_create_from_pattern(regex, NULL);
	if (!match_data) die("pcre2_match_data_create_from_pattern");

	filecap = BUFSIZ;
	filesize = 0;
	file = malloc(filecap);
	if (!file) die("malloc:");

	pos = 0;
	lastmatch = 0;
	while (true) {
		if (BUFSIZ > filecap - filesize) {
			filecap *= 2;
			file = realloc(file, filecap);
			if (!file) die("realloc:");
		}

		n = read(0, file + filesize, BUFSIZ);
		if (n < 0) die("read:");
		if (!n) break;

		length = (size_t) n;
		offset = 0;
		while (offset != length) {
			rc = pcre2_match(regex, file, filesize + length, lastmatch,
					PCRE2_NOTEMPTY, match_data, NULL);
			if (rc < 0 && rc != PCRE2_ERROR_NOMATCH)
				pcre2_die(rc, "pcre2_match: ");

			if (rc != PCRE2_ERROR_NOMATCH) {
				rc = pcre2_substring_get_bynumber(match_data,
					group, &match, &match_size);
				if (rc) pcre2_die(rc, "pcre2_substring_get_byname: ");

				ovector = pcre2_get_ovector_pointer(match_data);
				if (overlap) {
					lastmatch = ovector[0] + 1;
				} else {
					lastmatch = ovector[1];
				}

				writef("%s:0x%zx:", marker, filesize + ovector[0]); 
				writeall(match, match_size);
				writef("\n");
			} else {
				break;
			}
		}

		filesize += length;
	}

	pcre2_match_data_free(match_data);
	pcre2_code_free(regex);

	return 0;
}