lbbs/src/bwf.c

/* SPDX-License-Identifier: GPL-3.0-or-later */
/*
 * bwf
 *   - bad words filter
 *
 * Copyright (C) 2004-2025  Leaflet <leaflet@leafok.com>
 */

#define PCRE2_CODE_UNIT_WIDTH 8

#include "bwf.h"
#include "log.h"
#include <errno.h>
#include <pcre2.h>
#include <stdio.h>
#include <string.h>

enum _bwf_constant_t
{
        BWF_max_pattern_len = 64,
        BWF_max_pattern_cnt = 1000,
};

static char bwf_pattern_str[(BWF_max_pattern_len + 1) * BWF_max_pattern_cnt];
static pcre2_code *bwf_code = NULL;

int bwf_load(const char *filename)
{
        FILE *fp;
        char line[BWF_max_pattern_len + 1];
        size_t len_line;
        char *p = bwf_pattern_str;
        int line_id = 0;
        int errorcode;
        PCRE2_SIZE erroroffset;

        if (filename == NULL)
        {
                log_error("NULL pointer error\n");
                return -1;
        }

        if ((fp = fopen(filename, "r")) == NULL)
        {
                log_error("fopen(%s) error: %d\n", filename, errno);
                return -2;
        }

        while (fgets(line, sizeof(line), fp) != NULL)
        {
                len_line = strnlen(line, sizeof(line) - 1);
                if (!feof(fp) && line[len_line - 1] != '\n')
                {
                        log_error("Data line %d (len=%d) is truncated\n", line_id, len_line);
                        bwf_pattern_str[0] = '\0';
                        return -3;
                }

                while (len_line > 0 && (line[len_line - 1] == '\n' || line[len_line - 1] == '\r'))
                {
                        line[len_line - 1] = '\0';
                        len_line--;
                }

                if (len_line == 0)
                {
                        continue;
                }

                if (p > bwf_pattern_str)
                {
                        *p++ = '|';
                }

                if (len_line + 2 > sizeof(bwf_pattern_str) - 1 - (size_t)(p - bwf_pattern_str))
                {
                        log_error("Data in %s exceed length limit %d\n", filename, sizeof(bwf_pattern_str) - 1);
                        bwf_pattern_str[0] = '\0';
                        return -3;
                }

                *p++ = '(';
                p = mempcpy(p, line, strnlen(line, sizeof(bwf_pattern_str) - 1 - (size_t)(p - bwf_pattern_str)));
                *p++ = ')';
                line_id++;
        }

        *p = '\0';

        fclose(fp);

#ifdef _DEBUG
        log_error("Debug: bwf_pattern_str: %s\n", bwf_pattern_str);
#endif

        bwf_unload();

        bwf_code = pcre2_compile((PCRE2_SPTR)bwf_pattern_str, PCRE2_ZERO_TERMINATED, PCRE2_CASELESS, &errorcode, &erroroffset, NULL);
        if (bwf_code == NULL)
        {
                log_error("pcre2_compile() error: %d", errorcode);
                return -4;
        }

        return 0;
}

void bwf_unload(void)
{
        if (bwf_code != NULL)
        {
                pcre2_code_free(bwf_code);
                bwf_code = NULL;
        }
}

int check_badwords(char *str, char c_mask)
{
        pcre2_match_data *match_data;
        PCRE2_SIZE startoffset = 0;
        PCRE2_SIZE *ovector;
        uint32_t match_count;
        int ret;
        int i;
        int total_match_count = 0;

        if (bwf_code == NULL)
        {
                log_error("BWF not loaded\n");
                return -1;
        }

        match_data = pcre2_match_data_create_from_pattern(bwf_code, NULL);

        while (1)
        {
                ret = pcre2_match(bwf_code, (PCRE2_SPTR)str, PCRE2_ZERO_TERMINATED, startoffset, 0, match_data, NULL);
                if (ret == PCRE2_ERROR_NOMATCH)
                {
                        ret = total_match_count;
                        break;
                }
                else if (ret < 0)
                {
                        log_error("pcre2_match() error: %d\n", ret);
                }
                else if (ret == 0)
                {
                        log_error("Vector of offsets is too small\n");
                }
                else // ret >= 1
                {
                        ovector = pcre2_get_ovector_pointer(match_data);
                        match_count = pcre2_get_ovector_count(match_data);

                        i = ret - 1;
                        if (ovector[i * 2] == -1 || ovector[i * 2 + 1] == -1)
                        {
                                log_error("Bug: match pattern #%d of %d with invalid offsets [%d, %d)",
                                                  i, match_count, ovector[i * 2], ovector[i * 2 + 1]);
                                ret = -2;
                        }
                        else
                        {
#ifdef _DEBUG
                                log_error("Debug: match pattern #%d of %d at offsets [%d, %d]\n",
                                                  i, match_count, ovector[i * 2], ovector[i * 2 + 1] - ovector[i * 2]);
#endif
                                memset(str + ovector[i * 2], c_mask, ovector[i * 2 + 1] - ovector[i * 2]);
                                total_match_count++;
                                startoffset = ovector[i * 2 + 1];
                        }
                }
        }

        pcre2_match_data_free(match_data);

        return ret;
}
1	sysadm	1.1	/* SPDX-License-Identifier: GPL-3.0-or-later */
2			/*
3			* bwf
4			* - bad words filter
5			*
6			* Copyright (C) 2004-2025 Leaflet <leaflet@leafok.com>
7			*/
8
9			#define PCRE2_CODE_UNIT_WIDTH 8
10
11			#include "bwf.h"
12			#include "log.h"
13			#include <errno.h>
14			#include <pcre2.h>
15			#include <stdio.h>
16			#include <string.h>
17
18			enum _bwf_constant_t
19			{
20			BWF_max_pattern_len = 64,
21			BWF_max_pattern_cnt = 1000,
22			};
23
24			static char bwf_pattern_str[(BWF_max_pattern_len + 1) * BWF_max_pattern_cnt];
25			static pcre2_code *bwf_code = NULL;
26
27			int bwf_load(const char *filename)
28			{
29			FILE *fp;
30			char line[BWF_max_pattern_len + 1];
31			size_t len_line;
32			char *p = bwf_pattern_str;
33			int line_id = 0;
34			int errorcode;
35			PCRE2_SIZE erroroffset;
36
37			if (filename == NULL)
38			{
39			log_error("NULL pointer error\n");
40			return -1;
41			}
42
43			if ((fp = fopen(filename, "r")) == NULL)
44			{
45			log_error("fopen(%s) error: %d\n", filename, errno);
46			return -2;
47			}
48
49			while (fgets(line, sizeof(line), fp) != NULL)
50			{
51			len_line = strnlen(line, sizeof(line) - 1);
52			if (!feof(fp) && line[len_line - 1] != '\n')
53			{
54			log_error("Data line %d (len=%d) is truncated\n", line_id, len_line);
55			bwf_pattern_str[0] = '\0';
56			return -3;
57			}
58
59	sysadm	1.4	while (len_line > 0 && (line[len_line - 1] == '\n' \|\| line[len_line - 1] == '\r'))
60	sysadm	1.1	{
61			line[len_line - 1] = '\0';
62	sysadm	1.4	len_line--;
63			}
64
65			if (len_line == 0)
66			{
67			continue;
68	sysadm	1.1	}
69
70			if (p > bwf_pattern_str)
71			{
72			*p++ = '\|';
73			}
74
75			if (len_line + 2 > sizeof(bwf_pattern_str) - 1 - (size_t)(p - bwf_pattern_str))
76			{
77			log_error("Data in %s exceed length limit %d\n", filename, sizeof(bwf_pattern_str) - 1);
78			bwf_pattern_str[0] = '\0';
79			return -3;
80			}
81
82			*p++ = '(';
83			p = mempcpy(p, line, strnlen(line, sizeof(bwf_pattern_str) - 1 - (size_t)(p - bwf_pattern_str)));
84			*p++ = ')';
85			line_id++;
86			}
87
88			*p = '\0';
89
90			fclose(fp);
91
92			#ifdef _DEBUG
93			log_error("Debug: bwf_pattern_str: %s\n", bwf_pattern_str);
94			#endif
95
96	sysadm	1.2	bwf_unload();
97	sysadm	1.3
98	sysadm	1.1	bwf_code = pcre2_compile((PCRE2_SPTR)bwf_pattern_str, PCRE2_ZERO_TERMINATED, PCRE2_CASELESS, &errorcode, &erroroffset, NULL);
99			if (bwf_code == NULL)
100			{
101			log_error("pcre2_compile() error: %d", errorcode);
102			return -4;
103			}
104
105			return 0;
106			}
107
108			void bwf_unload(void)
109			{
110			if (bwf_code != NULL)
111			{
112			pcre2_code_free(bwf_code);
113			bwf_code = NULL;
114			}
115			}
116
117			int check_badwords(char *str, char c_mask)
118			{
119			pcre2_match_data *match_data;
120	sysadm	1.3	PCRE2_SIZE startoffset = 0;
121	sysadm	1.1	PCRE2_SIZE *ovector;
122	sysadm	1.3	uint32_t match_count;
123	sysadm	1.1	int ret;
124			int i;
125	sysadm	1.3	int total_match_count = 0;
126	sysadm	1.1
127			if (bwf_code == NULL)
128			{
129			log_error("BWF not loaded\n");
130			return -1;
131			}
132
133			match_data = pcre2_match_data_create_from_pattern(bwf_code, NULL);
134
135	sysadm	1.3	while (1)
136	sysadm	1.1	{
137	sysadm	1.3	ret = pcre2_match(bwf_code, (PCRE2_SPTR)str, PCRE2_ZERO_TERMINATED, startoffset, 0, match_data, NULL);
138			if (ret == PCRE2_ERROR_NOMATCH)
139			{
140			ret = total_match_count;
141			break;
142			}
143			else if (ret < 0)
144			{
145			log_error("pcre2_match() error: %d\n", ret);
146			}
147			else if (ret == 0)
148	sysadm	1.1	{
149	sysadm	1.3	log_error("Vector of offsets is too small\n");
150	sysadm	1.1	}
151	sysadm	1.3	else // ret >= 1
152	sysadm	1.1	{
153	sysadm	1.3	ovector = pcre2_get_ovector_pointer(match_data);
154			match_count = pcre2_get_ovector_count(match_data);
155
156			i = ret - 1;
157			if (ovector[i * 2] == -1 \|\| ovector[i * 2 + 1] == -1)
158			{
159			log_error("Bug: match pattern #%d of %d with invalid offsets [%d, %d)",
160			i, match_count, ovector[i * 2], ovector[i * 2 + 1]);
161			ret = -2;
162			}
163			else
164			{
165	sysadm	1.1	#ifdef _DEBUG
166	sysadm	1.3	log_error("Debug: match pattern #%d of %d at offsets [%d, %d]\n",
167			i, match_count, ovector[i * 2], ovector[i * 2 + 1] - ovector[i * 2]);
168	sysadm	1.1	#endif
169	sysadm	1.3	memset(str + ovector[i * 2], c_mask, ovector[i * 2 + 1] - ovector[i * 2]);
170			total_match_count++;
171			startoffset = ovector[i * 2 + 1];
172			}
173	sysadm	1.1	}
174			}
175
176			pcre2_match_data_free(match_data);
177
178			return ret;
179			}
webmaster@leafok.com	ViewVC Help
Powered by ViewVC 1.3.0-beta1