| 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 |
#ifdef HAVE_CONFIG_H
|
| 10 |
#include "config.h"
|
| 11 |
#endif
|
| 12 |
|
| 13 |
#define PCRE2_CODE_UNIT_WIDTH 8
|
| 14 |
|
| 15 |
#include "bwf.h"
|
| 16 |
#include "log.h"
|
| 17 |
#include <errno.h>
|
| 18 |
#include <pcre2.h>
|
| 19 |
#include <stdio.h>
|
| 20 |
#include <string.h>
|
| 21 |
|
| 22 |
enum _bwf_constant_t
|
| 23 |
{
|
| 24 |
BWF_max_pattern_len = 64,
|
| 25 |
BWF_max_pattern_cnt = 1000,
|
| 26 |
};
|
| 27 |
|
| 28 |
static char bwf_pattern_str[(BWF_max_pattern_len + 1) * BWF_max_pattern_cnt];
|
| 29 |
static pcre2_code *bwf_code = NULL;
|
| 30 |
|
| 31 |
int bwf_load(const char *filename)
|
| 32 |
{
|
| 33 |
FILE *fp;
|
| 34 |
char line[BWF_max_pattern_len + 1];
|
| 35 |
size_t len_line;
|
| 36 |
char *p = bwf_pattern_str;
|
| 37 |
int line_id = 0;
|
| 38 |
int errorcode;
|
| 39 |
PCRE2_SIZE erroroffset;
|
| 40 |
|
| 41 |
if (filename == NULL)
|
| 42 |
{
|
| 43 |
log_error("NULL pointer error\n");
|
| 44 |
return -1;
|
| 45 |
}
|
| 46 |
|
| 47 |
if ((fp = fopen(filename, "r")) == NULL)
|
| 48 |
{
|
| 49 |
log_error("fopen(%s) error: %d\n", filename, errno);
|
| 50 |
return -2;
|
| 51 |
}
|
| 52 |
|
| 53 |
while (fgets(line, sizeof(line), fp) != NULL)
|
| 54 |
{
|
| 55 |
len_line = strnlen(line, sizeof(line) - 1);
|
| 56 |
if (!feof(fp) && line[len_line - 1] != '\n')
|
| 57 |
{
|
| 58 |
log_error("Data line %d (len=%d) is truncated\n", line_id, len_line);
|
| 59 |
bwf_pattern_str[0] = '\0';
|
| 60 |
return -3;
|
| 61 |
}
|
| 62 |
|
| 63 |
while (len_line > 0 && (line[len_line - 1] == '\n' || line[len_line - 1] == '\r'))
|
| 64 |
{
|
| 65 |
line[len_line - 1] = '\0';
|
| 66 |
len_line--;
|
| 67 |
}
|
| 68 |
|
| 69 |
if (len_line == 0)
|
| 70 |
{
|
| 71 |
continue;
|
| 72 |
}
|
| 73 |
|
| 74 |
if (p > bwf_pattern_str)
|
| 75 |
{
|
| 76 |
*p++ = '|';
|
| 77 |
}
|
| 78 |
|
| 79 |
if (len_line + 2 > sizeof(bwf_pattern_str) - 1 - (size_t)(p - bwf_pattern_str))
|
| 80 |
{
|
| 81 |
log_error("Data in %s exceed length limit %d\n", filename, sizeof(bwf_pattern_str) - 1);
|
| 82 |
bwf_pattern_str[0] = '\0';
|
| 83 |
return -3;
|
| 84 |
}
|
| 85 |
|
| 86 |
*p++ = '(';
|
| 87 |
p = mempcpy(p, line, strnlen(line, sizeof(bwf_pattern_str) - 1 - (size_t)(p - bwf_pattern_str)));
|
| 88 |
*p++ = ')';
|
| 89 |
line_id++;
|
| 90 |
}
|
| 91 |
|
| 92 |
*p = '\0';
|
| 93 |
|
| 94 |
fclose(fp);
|
| 95 |
|
| 96 |
#ifdef _DEBUG
|
| 97 |
log_error("Debug: bwf_pattern_str: %s\n", bwf_pattern_str);
|
| 98 |
#endif
|
| 99 |
|
| 100 |
bwf_unload();
|
| 101 |
|
| 102 |
bwf_code = pcre2_compile((PCRE2_SPTR)bwf_pattern_str, PCRE2_ZERO_TERMINATED, PCRE2_CASELESS, &errorcode, &erroroffset, NULL);
|
| 103 |
if (bwf_code == NULL)
|
| 104 |
{
|
| 105 |
log_error("pcre2_compile() error: %d", errorcode);
|
| 106 |
return -4;
|
| 107 |
}
|
| 108 |
|
| 109 |
return 0;
|
| 110 |
}
|
| 111 |
|
| 112 |
void bwf_unload(void)
|
| 113 |
{
|
| 114 |
if (bwf_code != NULL)
|
| 115 |
{
|
| 116 |
pcre2_code_free(bwf_code);
|
| 117 |
bwf_code = NULL;
|
| 118 |
}
|
| 119 |
}
|
| 120 |
|
| 121 |
int check_badwords(char *str, char c_mask)
|
| 122 |
{
|
| 123 |
pcre2_match_data *match_data;
|
| 124 |
PCRE2_SIZE startoffset = 0;
|
| 125 |
PCRE2_SIZE *ovector;
|
| 126 |
uint32_t match_count;
|
| 127 |
int ret;
|
| 128 |
int i;
|
| 129 |
int total_match_count = 0;
|
| 130 |
|
| 131 |
if (bwf_code == NULL)
|
| 132 |
{
|
| 133 |
log_error("BWF not loaded\n");
|
| 134 |
return -1;
|
| 135 |
}
|
| 136 |
|
| 137 |
match_data = pcre2_match_data_create_from_pattern(bwf_code, NULL);
|
| 138 |
|
| 139 |
while (1)
|
| 140 |
{
|
| 141 |
ret = pcre2_match(bwf_code, (PCRE2_SPTR)str, PCRE2_ZERO_TERMINATED, startoffset, 0, match_data, NULL);
|
| 142 |
if (ret == PCRE2_ERROR_NOMATCH)
|
| 143 |
{
|
| 144 |
ret = total_match_count;
|
| 145 |
break;
|
| 146 |
}
|
| 147 |
else if (ret < 0)
|
| 148 |
{
|
| 149 |
log_error("pcre2_match() error: %d\n", ret);
|
| 150 |
}
|
| 151 |
else if (ret == 0)
|
| 152 |
{
|
| 153 |
log_error("Vector of offsets is too small\n");
|
| 154 |
}
|
| 155 |
else // ret >= 1
|
| 156 |
{
|
| 157 |
ovector = pcre2_get_ovector_pointer(match_data);
|
| 158 |
match_count = pcre2_get_ovector_count(match_data);
|
| 159 |
|
| 160 |
i = ret - 1;
|
| 161 |
if (ovector[i * 2] == -1 || ovector[i * 2 + 1] == -1)
|
| 162 |
{
|
| 163 |
log_error("Bug: match pattern #%d of %d with invalid offsets [%d, %d)",
|
| 164 |
i, match_count, ovector[i * 2], ovector[i * 2 + 1]);
|
| 165 |
ret = -2;
|
| 166 |
}
|
| 167 |
else
|
| 168 |
{
|
| 169 |
#ifdef _DEBUG
|
| 170 |
log_error("Debug: match pattern #%d of %d at offsets [%d, %d]\n",
|
| 171 |
i, match_count, ovector[i * 2], ovector[i * 2 + 1] - ovector[i * 2]);
|
| 172 |
#endif
|
| 173 |
memset(str + ovector[i * 2], c_mask, ovector[i * 2 + 1] - ovector[i * 2]);
|
| 174 |
total_match_count++;
|
| 175 |
startoffset = ovector[i * 2 + 1];
|
| 176 |
}
|
| 177 |
}
|
| 178 |
}
|
| 179 |
|
| 180 |
pcre2_match_data_free(match_data);
|
| 181 |
|
| 182 |
return ret;
|
| 183 |
}
|
Parent Directory
| 
Revision Log