xref: /PHP-5.5/ext/standard/quot_print.c (revision 73c1be26)
1 /*
2    +----------------------------------------------------------------------+
3    | PHP Version 5                                                        |
4    +----------------------------------------------------------------------+
5    | Copyright (c) 1997-2015 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,size_t * ret_length,int replace_us_by_ws)54 PHPAPI unsigned char *php_quot_print_decode(const unsigned char *str, size_t length, size_t *ret_length, int replace_us_by_ws) /* {{{ */
55 {
56 	register unsigned int 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 	unsigned char *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 = emalloc(buf_size + 1);
100 	i = length; p1 = str; p2 = 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 	*ret_length = 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,size_t * ret_length)148 PHPAPI unsigned char *php_quot_print_encode(const unsigned char *str, size_t length, size_t *ret_length) /* {{{ */
149 {
150 	unsigned long lp = 0;
151 	unsigned char c, *ret, *d;
152 	char *hex = "0123456789ABCDEF";
153 
154 	ret = safe_emalloc(3, length + (((3 * length)/(PHP_QPRINT_MAXL-9)) + 1), 1);
155 	d = ret;
156 
157 	while (length--) {
158 		if (((c = *str++) == '\015') && (*str == '\012') && length > 0) {
159 			*d++ = '\015';
160 			*d++ = *str++;
161 			length--;
162 			lp = 0;
163 		} else {
164 			if (iscntrl (c) || (c == 0x7f) || (c & 0x80) || (c == '=') || ((c == ' ') && (*str == '\015'))) {
165 				if ((((lp+= 3) > PHP_QPRINT_MAXL) && (c <= 0x7f))
166             || ((c > 0x7f) && (c <= 0xdf) && ((lp + 3) > PHP_QPRINT_MAXL))
167             || ((c > 0xdf) && (c <= 0xef) && ((lp + 6) > PHP_QPRINT_MAXL))
168             || ((c > 0xef) && (c <= 0xf4) && ((lp + 9) > PHP_QPRINT_MAXL))) {
169 					*d++ = '=';
170 					*d++ = '\015';
171 					*d++ = '\012';
172 					lp = 3;
173 				}
174 				*d++ = '=';
175 				*d++ = hex[c >> 4];
176 				*d++ = hex[c & 0xf];
177 			} else {
178 				if ((++lp) > PHP_QPRINT_MAXL) {
179 					*d++ = '=';
180 					*d++ = '\015';
181 					*d++ = '\012';
182 					lp = 1;
183 				}
184 				*d++ = c;
185 			}
186 		}
187 	}
188 	*d = '\0';
189 	*ret_length = d - ret;
190 
191 	ret = erealloc(ret, *ret_length + 1);
192 	return ret;
193 }
194 /* }}} */
195 
196 /*
197 *
198 * Decoding  Quoted-printable string.
199 *
200 */
201 /* {{{ proto string quoted_printable_decode(string str)
202    Convert a quoted-printable string to an 8 bit string */
PHP_FUNCTION(quoted_printable_decode)203 PHP_FUNCTION(quoted_printable_decode)
204 {
205 	char *arg1, *str_in, *str_out;
206 	int arg1_len, i = 0, j = 0, k;
207 
208 	if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "s", &arg1, &arg1_len) == FAILURE) {
209 		return;
210 	}
211 
212 	if (arg1_len == 0) {
213 		/* shortcut */
214 		RETURN_EMPTY_STRING();
215 	}
216 
217 	str_in = arg1;
218 	str_out = emalloc(arg1_len + 1);
219 	while (str_in[i]) {
220 		switch (str_in[i]) {
221 		case '=':
222 			if (str_in[i + 1] && str_in[i + 2] &&
223 				isxdigit((int) str_in[i + 1]) &&
224 				isxdigit((int) str_in[i + 2]))
225 			{
226 				str_out[j++] = (php_hex2int((int) str_in[i + 1]) << 4)
227 						+ php_hex2int((int) str_in[i + 2]);
228 				i += 3;
229 			} else  /* check for soft line break according to RFC 2045*/ {
230 				k = 1;
231 				while (str_in[i + k] && ((str_in[i + k] == 32) || (str_in[i + k] == 9))) {
232 					/* Possibly, skip spaces/tabs at the end of line */
233 					k++;
234 				}
235 				if (!str_in[i + k]) {
236 					/* End of line reached */
237 					i += k;
238 				}
239 				else if ((str_in[i + k] == 13) && (str_in[i + k + 1] == 10)) {
240 					/* CRLF */
241 					i += k + 2;
242 				}
243 				else if ((str_in[i + k] == 13) || (str_in[i + k] == 10)) {
244 					/* CR or LF */
245 					i += k + 1;
246 				}
247 				else {
248 					str_out[j++] = str_in[i++];
249 				}
250 			}
251 			break;
252 		default:
253 			str_out[j++] = str_in[i++];
254 		}
255 	}
256 	str_out[j] = '\0';
257 
258 	RETVAL_STRINGL(str_out, j, 0);
259 }
260 /* }}} */
261 
262 /* {{{ proto string quoted_printable_encode(string str) */
PHP_FUNCTION(quoted_printable_encode)263 PHP_FUNCTION(quoted_printable_encode)
264 {
265 	char *str, *new_str;
266 	int str_len;
267 	size_t new_str_len;
268 
269 	if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "s", &str, &str_len) != SUCCESS) {
270 		return;
271 	}
272 
273 	if (!str_len) {
274 		RETURN_EMPTY_STRING();
275 	}
276 
277 	new_str = (char *)php_quot_print_encode((unsigned char *)str, (size_t)str_len, &new_str_len);
278 	RETURN_STRINGL(new_str, new_str_len, 0);
279 }
280 /* }}} */
281 
282 /*
283  * Local variables:
284  * tab-width: 4
285  * c-basic-offset: 4
286  * End:
287  * vim600: sw=4 ts=4 fdm=marker
288  * vim<600: sw=4 ts=4
289  */
290