micro-utils/src/grep.c

#include <stdio.h>
#include <string.h>
#include <stdlib.h>
#include <unistd.h>
#include <errno.h>
#include <regex.h>
#include <sys/types.h>

static struct pattern {
	char *str;

	regex_t reg;
	char reg_set;
} **ptrns;
static size_t ptrns_size = 0;

static char i_flag;
static char F_flag;
static char H_flag;
static char E_flag;
static char v_flag;
static char x_flag;
static char w_flag;
static char mode_flag;
static int reg_flag;

static void free_patterns(void) {
	if (ptrns == NULL)
		return;

	for (size_t i = 0; i < ptrns_size; i++) {
		if (ptrns[i] == NULL)
			continue;

		if (ptrns[i]->str != NULL)
			free(ptrns[i]->str);

		if (ptrns[i]->reg_set)
			regfree(&ptrns[i]->reg);

		free(ptrns[i]);
	}

	free(ptrns);
	ptrns = NULL;
}

static int addpattern(const char *str, const size_t size) {
	if (ptrns == NULL) {
		ptrns = malloc(sizeof(struct pattern *));
		if (ptrns == NULL) {
			fprintf(stderr, "grep: malloc: %s\n", strerror(errno));
			return 1;
		}
	}

	char reg_error_flag = 0;
	struct pattern **bckp = realloc(ptrns, sizeof(struct pattern *) * (ptrns_size + 1));
	if (bckp == NULL)
		goto ADDP_ERROR;
	ptrns = bckp;

	ptrns[ptrns_size] = malloc(sizeof(struct pattern));
	if (ptrns[ptrns_size] == NULL)
		goto ADDP_ERROR;

	ptrns[ptrns_size]->str = strdup(str);
	if (ptrns[ptrns_size]->str == NULL)
		goto ADDP_ERROR;

	if (!F_flag) {
		char *reg_str = ptrns[ptrns_size]->str;
		size_t rs_size = size;

		char bol = (ptrns[ptrns_size]->str[0] == '^');
		char eol = (ptrns[ptrns_size]->str[size - 1] == '^');

		if (x_flag || w_flag) {
			if (w_flag)
				rs_size += 5 + ((E_flag) ? 2 : 4);

			reg_str = malloc(rs_size + 4);
			if (reg_str == NULL)
				goto ADDP_ERROR;
		}

		if (x_flag)
			snprintf(reg_str, rs_size + 4, "%s%s%s", (bol) ? "" : "^", ptrns[ptrns_size]->str, (eol) ? "" : "$");

		else if (w_flag)
			snprintf(reg_str, rs_size + 4, "%s\\<%s%.*s%s\\>%s", (bol) ? "^" : "", (E_flag) ? "(" : "\\(", (int)size - bol - eol, ptrns[ptrns_size]->str + bol, (E_flag) ? ")" : "\\)", (eol) ? "$" : "");

		if (regcomp(&ptrns[ptrns_size]->reg, reg_str, reg_flag) != 0)
			reg_error_flag = 1;

		ptrns[ptrns_size]->reg_set = 1;
		if (x_flag || w_flag)
			free(reg_str);

	}

	if (reg_error_flag)
		goto ADDP_ERROR;

	ptrns_size++;
	return 0;

ADDP_ERROR:
	ptrns_size++;
	free_patterns();
	fprintf(stderr, "grep: %s\n", (reg_error_flag) ? "bad regex" : strerror(errno));
	return 1;
}

static int addpattern_file(const char *file) {
	FILE *fp = fopen(file, "r");
	if (fp == NULL) {
		fprintf(stderr, "grep: %s: %s\n", file, strerror(errno));
		return 1;
	}

	int ret = 0;

	char *buf = NULL;
	size_t n = 0;

	ssize_t size = 0;
	while ((size = getline(&buf, &n, fp)) > 0) {
		if (size && buf[size - 1] == '\n')
			buf[size - 1] = '\0';

		if (addpattern(buf, (size_t)size)) {
			ret = 1;
			break;
		}
	}

	if (buf != NULL)
		free(buf);

	fclose(fp);
	return ret;
}

static char *my_strcasestr(const char *s1, const char *s2) {
	size_t len = strlen(s2);
	for (; *s1; s1++)
		if (!strncasecmp(s1, s2, len))
			return (char *)s1;

	return NULL;
}

static int cmp(const char *str1, const char *str2) {
	if (x_flag)
		return !((i_flag) ? strcasecmp : strcmp)(str1, str2);

	else
		return ((i_flag) ? my_strcasestr : strstr)(str1, str2) != NULL;
}

static int grep(FILE *fp, const char *file) {
	int ret = 1;
	size_t matched_files = 0;
	regmatch_t m;

	char *buf = NULL;
	size_t n = 0;

	ssize_t size = 0;
	while ((size = getline(&buf, &n, fp)) > 0) {
		if (size && buf[size - 1] == '\n')
			buf[size - 1] = '\0';

		char match = 0;
		size_t i = 0;
		for (; i < ptrns_size; i++) {
			if (F_flag && cmp(buf, ptrns[i]->str)) {
				match = 1;
				break;
			}

			else if (regexec(&ptrns[i]->reg, buf, 1, &m, 0) == 0) {
				match = 1;
				break;
			}
		}

		if (match != v_flag) {
			ret = 0;
			switch (mode_flag) {
				case 'q':
					break;

				case 'o':
					if (ptrns[i]->reg_set) {
						unsigned int start = m.rm_so;
						unsigned int finish = m.rm_eo;
						printf("%.*s\n", finish - start, buf + start);
					}

					else
						puts(ptrns[i]->str);

					break;

				case 'c':
					matched_files++;
					break;

				default:
					if (H_flag)
						printf("%s: ", file);

					puts(buf);
			}
		}
	}

	if (mode_flag == 'c')
		printf("%zu\n", matched_files);

	if (buf != NULL)
		free(buf);

	return ret;
}

int main(int argc, char **argv) {
	int opt;
	while ((opt = getopt(argc, argv, "e:f:iFHEvxwqoc")) != -1) {
		switch (opt) {
			case 'e':
				if (addpattern(optarg, strlen(optarg)))
					return 1;

				break;

			case 'f':
				if (addpattern_file(optarg))
					return 1;

				break;

			case 'i':
				i_flag = 1;
				reg_flag |= REG_ICASE;
				break;

			case 'F':
				F_flag = 1;
				E_flag = 0;
				reg_flag &= ~REG_EXTENDED;
				break;

			case 'H':
				H_flag = 1;
				break;

			case 'E':
				E_flag = 1;
				F_flag = 0;
				reg_flag |= REG_EXTENDED;
				break;

			case 'v':
				v_flag = 1;
				break;

			case 'x':
				x_flag = 1;
				break;

			case 'w':
				w_flag = 1;
				break;

			case 'q':
				mode_flag = 'q';
				break;

			case 'o':
				mode_flag = 'o';
				break;

			case 'c':
				mode_flag = 'c';
				break;

			default:
				puts("grep [efiFHvxwqoc] [FILE]\n\t-e PTRN Pattern to match\n\t-f FILE Read pattern from file\n\t-i Ignore case\n\t-H Add 'filename:' prefix\n\t-F PATTERN is a literal (not regexp)\n\t-E PATTERN is an extended regexp\n\t-v Select non-matching lines\n\t-x Match whole lines only\n\t-w Match whole words only\n\t-q Quiet\n\t-o Show only the matching part of line\n\t-c Show only count of matching lines");
				return 0;
		}
	}

	argv += optind;
	argc -= optind;
	if (ptrns_size == 0) {
		fprintf(stderr, "grep: no patterns specified\n");
		return 1;
	}

	int ret = 0;
	if (argc == 0)
		ret = grep(stdin, "-");

	else {
		for (int i = 0; i < argc; i++) {
			FILE *fp = stdin;
			if (strcmp(argv[i], "-")) {
				fp = fopen(argv[i], "r");
				if (fp == NULL) {
					ret = 1;
					fprintf(stderr, "grep: %s: %s\n", argv[i], strerror(errno));
					continue;
				}
			}

			if (grep(fp, argv[i]))
				ret = 1;

			if (fp != stdin)
				fclose(fp);
		}
	}

	free_patterns();
	return ret;
}