xref: /PHP-7.2/ext/standard/quot_print.c (revision 7a7ec01a)
1 /*
2    +----------------------------------------------------------------------+
3    | PHP Version 7                                                        |
4    +----------------------------------------------------------------------+
5    | Copyright (c) 1997-2018 The PHP Group                                |
6    +----------------------------------------------------------------------+
7    | This source file is subject to version 3.01 of the PHP license,      |
8    | that is bundled with this package in the file LICENSE, and is        |
9    | available through the world-wide-web at the following url:           |
10    | http://www.php.net/license/3_01.txt                                  |
11    | If you did not receive a copy of the PHP license and are unable to   |
12    | obtain it through the world-wide-web, please send a note to          |
13    | license@php.net so we can mail you a copy immediately.               |
14    +----------------------------------------------------------------------+
15    | Author: Kirill Maximov <kir@actimind.com>                            |
16    +----------------------------------------------------------------------+
17  */
18 
19 /* $Id$ */
20 
21 #include <stdlib.h>
22 
23 #ifdef HAVE_UNISTD_H
24 #include <unistd.h>
25 #endif
26 #include <string.h>
27 #include <errno.h>
28 
29 #include "php.h"
30 #include "quot_print.h"
31 
32 #include <stdio.h>
33 
34 /*
35 *  Converting HEX char to INT value
36 */
php_hex2int(int c)37 static char php_hex2int(int c) /* {{{ */
38 {
39 	if (isdigit(c)) {
40 		return c - '0';
41 	}
42 	else if (c >= 'A' && c <= 'F') {
43 		return c - 'A' + 10;
44 	}
45 	else if (c >= 'a' && c <= 'f') {
46 		return c - 'a' + 10;
47 	}
48 	else {
49 		return -1;
50 	}
51 }
52 /* }}} */
53 
php_quot_print_decode(const unsigned char * str,size_t length,int replace_us_by_ws)54 PHPAPI zend_string *php_quot_print_decode(const unsigned char *str, size_t length, int replace_us_by_ws) /* {{{ */
55 {
56 	register size_t i;
57 	register unsigned const char *p1;
58 	register unsigned char *p2;
59 	register unsigned int h_nbl, l_nbl;
60 
61 	size_t decoded_len, buf_size;
62 	zend_string *retval;
63 
64 	static unsigned int hexval_tbl[256] = {
65 		64, 64, 64, 64, 64, 64, 64, 64, 64, 32, 16, 64, 64, 16, 64, 64,
66 		64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64,
67 		32, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64,
68 		 0,  1,  2,  3,  4,  5,  6,  7,  8,  9, 64, 64, 64, 64, 64, 64,
69 		64, 10, 11, 12, 13, 14, 15, 64, 64, 64, 64, 64, 64, 64, 64, 64,
70 		64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64,
71 		64, 10, 11, 12, 13, 14, 15, 64, 64, 64, 64, 64, 64, 64, 64, 64,
72 		64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64,
73 		64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64,
74 		64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64,
75 		64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64,
76 		64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64,
77 		64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64,
78 		64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64,
79 		64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64,
80 		64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 64
81 	};
82 
83 	if (replace_us_by_ws) {
84 		replace_us_by_ws = '_';
85 	}
86 
87 	i = length, p1 = str; buf_size = length;
88 
89 	while (i > 1 && *p1 != '\0') {
90 		if (*p1 == '=') {
91 			buf_size -= 2;
92 			p1++;
93 			i--;
94 		}
95 		p1++;
96 		i--;
97 	}
98 
99 	retval = zend_string_alloc(buf_size, 0);
100 	i = length; p1 = str; p2 = (unsigned char*)ZSTR_VAL(retval);
101 	decoded_len = 0;
102 
103 	while (i > 0 && *p1 != '\0') {
104 		if (*p1 == '=') {
105 			i--, p1++;
106 			if (i == 0 || *p1 == '\0') {
107 				break;
108 			}
109 			h_nbl = hexval_tbl[*p1];
110 			if (h_nbl < 16) {
111 				/* next char should be a hexadecimal digit */
112 				if ((--i) == 0 || (l_nbl = hexval_tbl[*(++p1)]) >= 16) {
113 					efree(retval);
114 					return NULL;
115 				}
116 				*(p2++) = (h_nbl << 4) | l_nbl, decoded_len++;
117 				i--, p1++;
118 			} else if (h_nbl < 64) {
119 				/* soft line break */
120 				while (h_nbl == 32) {
121 					if (--i == 0 || (h_nbl = hexval_tbl[*(++p1)]) == 64) {
122 						efree(retval);
123 						return NULL;
124 					}
125 				}
126 				if (p1[0] == '\r' && i >= 2 && p1[1] == '\n') {
127 					i--, p1++;
128 				}
129 				i--, p1++;
130 			} else {
131 				efree(retval);
132 				return NULL;
133 			}
134 		} else {
135 			*(p2++) = (replace_us_by_ws == *p1 ? '\x20': *p1);
136 			i--, p1++, decoded_len++;
137 		}
138 	}
139 
140 	*p2 = '\0';
141 	ZSTR_LEN(retval) = decoded_len;
142 	return retval;
143 }
144 /* }}} */
145 
146 #define PHP_QPRINT_MAXL 75
147 
php_quot_print_encode(const unsigned char * str,size_t length)148 PHPAPI zend_string *php_quot_print_encode(const unsigned char *str, size_t length) /* {{{ */
149 {
150 	zend_ulong lp = 0;
151 	unsigned char c, *d;
152 	char *hex = "0123456789ABCDEF";
153 	zend_string *ret;
154 
155 	ret = zend_string_safe_alloc(3, (length + (((3 * length)/(PHP_QPRINT_MAXL-9)) + 1)), 0, 0);
156 	d = (unsigned char*)ZSTR_VAL(ret);
157 
158 	while (length--) {
159 		if (((c = *str++) == '\015') && (*str == '\012') && length > 0) {
160 			*d++ = '\015';
161 			*d++ = *str++;
162 			length--;
163 			lp = 0;
164 		} else {
165 			if (iscntrl (c) || (c == 0x7f) || (c & 0x80) || (c == '=') || ((c == ' ') && (*str == '\015'))) {
166 				if ((((lp+= 3) > PHP_QPRINT_MAXL) && (c <= 0x7f))
167             || ((c > 0x7f) && (c <= 0xdf) && ((lp + 3) > PHP_QPRINT_MAXL))
168             || ((c > 0xdf) && (c <= 0xef) && ((lp + 6) > PHP_QPRINT_MAXL))
169             || ((c > 0xef) && (c <= 0xf4) && ((lp + 9) > PHP_QPRINT_MAXL))) {
170 					*d++ = '=';
171 					*d++ = '\015';
172 					*d++ = '\012';
173 					lp = 3;
174 				}
175 				*d++ = '=';
176 				*d++ = hex[c >> 4];
177 				*d++ = hex[c & 0xf];
178 			} else {
179 				if ((++lp) > PHP_QPRINT_MAXL) {
180 					*d++ = '=';
181 					*d++ = '\015';
182 					*d++ = '\012';
183 					lp = 1;
184 				}
185 				*d++ = c;
186 			}
187 		}
188 	}
189 	*d = '\0';
190 	ret = zend_string_truncate(ret, d - (unsigned char*)ZSTR_VAL(ret), 0);
191 	return ret;
192 }
193 /* }}} */
194 
195 /*
196 *
197 * Decoding  Quoted-printable string.
198 *
199 */
200 /* {{{ proto string quoted_printable_decode(string str)
201    Convert a quoted-printable string to an 8 bit string */
PHP_FUNCTION(quoted_printable_decode)202 PHP_FUNCTION(quoted_printable_decode)
203 {
204 	zend_string *arg1;
205 	char *str_in;
206 	zend_string *str_out;
207 	size_t i = 0, j = 0, k;
208 
209 	ZEND_PARSE_PARAMETERS_START(1, 1)
210 		Z_PARAM_STR(arg1)
211 	ZEND_PARSE_PARAMETERS_END();
212 
213 	if (ZSTR_LEN(arg1) == 0) {
214 		/* shortcut */
215 		RETURN_EMPTY_STRING();
216 	}
217 
218 	str_in = ZSTR_VAL(arg1);
219 	str_out = zend_string_alloc(ZSTR_LEN(arg1), 0);
220 	while (str_in[i]) {
221 		switch (str_in[i]) {
222 		case '=':
223 			if (str_in[i + 1] && str_in[i + 2] &&
224 				isxdigit((int) str_in[i + 1]) &&
225 				isxdigit((int) str_in[i + 2]))
226 			{
227 				ZSTR_VAL(str_out)[j++] = (php_hex2int((int) str_in[i + 1]) << 4)
228 						+ php_hex2int((int) str_in[i + 2]);
229 				i += 3;
230 			} else  /* check for soft line break according to RFC 2045*/ {
231 				k = 1;
232 				while (str_in[i + k] && ((str_in[i + k] == 32) || (str_in[i + k] == 9))) {
233 					/* Possibly, skip spaces/tabs at the end of line */
234 					k++;
235 				}
236 				if (!str_in[i + k]) {
237 					/* End of line reached */
238 					i += k;
239 				}
240 				else if ((str_in[i + k] == 13) && (str_in[i + k + 1] == 10)) {
241 					/* CRLF */
242 					i += k + 2;
243 				}
244 				else if ((str_in[i + k] == 13) || (str_in[i + k] == 10)) {
245 					/* CR or LF */
246 					i += k + 1;
247 				}
248 				else {
249 					ZSTR_VAL(str_out)[j++] = str_in[i++];
250 				}
251 			}
252 			break;
253 		default:
254 			ZSTR_VAL(str_out)[j++] = str_in[i++];
255 		}
256 	}
257 	ZSTR_VAL(str_out)[j] = '\0';
258 	ZSTR_LEN(str_out) = j;
259 
260 	RETVAL_NEW_STR(str_out);
261 }
262 /* }}} */
263 
264 /* {{{ proto string quoted_printable_encode(string str) */
PHP_FUNCTION(quoted_printable_encode)265 PHP_FUNCTION(quoted_printable_encode)
266 {
267 	zend_string *str;
268 	zend_string *new_str;
269 
270 	ZEND_PARSE_PARAMETERS_START(1, 1)
271 		Z_PARAM_STR(str)
272 	ZEND_PARSE_PARAMETERS_END();
273 
274 	if (!ZSTR_LEN(str)) {
275 		RETURN_EMPTY_STRING();
276 	}
277 
278 	new_str = php_quot_print_encode((unsigned char *)ZSTR_VAL(str), ZSTR_LEN(str));
279 	RETURN_STR(new_str);
280 }
281 /* }}} */
282 
283 /*
284  * Local variables:
285  * tab-width: 4
286  * c-basic-offset: 4
287  * End:
288  * vim600: sw=4 ts=4 fdm=marker
289  * vim<600: sw=4 ts=4
290  */
291