1 /* vi: set sw=4 ts=4: */
2 /*
3  * Utility routines.
4  *
5  * Copyright (C) Manuel Novoa III <mjn3@codepoet.org>
6  * and Vladimir Oleynik <dzo@simtreas.ru>
7  *
8  * Licensed under GPLv2 or later, see file LICENSE in this source tree.
9  */
10 #include "libbb.h"
11 
12 #define WANT_HEX_ESCAPES 1
13 
14 /* Usual "this only works for ascii compatible encodings" disclaimer. */
15 #undef _tolower
16 #define _tolower(X) ((X)|((char) 0x20))
17 
bb_process_escape_sequence(const char ** ptr)18 char FAST_FUNC bb_process_escape_sequence(const char **ptr)
19 {
20 	const char *q;
21 	unsigned num_digits;
22 	unsigned n;
23 	unsigned base;
24 
25 	num_digits = n = 0;
26 	base = 8;
27 	q = *ptr;
28 
29 	if (WANT_HEX_ESCAPES && *q == 'x') {
30 		++q;
31 		base = 16;
32 		++num_digits;
33 	}
34 
35 	/* bash requires leading 0 in octal escapes:
36 	 * \02 works, \2 does not (prints \ and 2).
37 	 * We treat \2 as a valid octal escape sequence. */
38 	do {
39 		unsigned r;
40 		unsigned d = (unsigned char)(*q) - '0';
41 #if WANT_HEX_ESCAPES
42 		if (d >= 10) {
43 			d = (unsigned char)_tolower(*q) - 'a';
44 			//d += 10;
45 			/* The above would map 'A'-'F' and 'a'-'f' to 10-15,
46 			 * however, some chars like '@' would map to 9 < base.
47 			 * Do not allow that, map invalid chars to N > base:
48 			 */
49 			if ((int)d >= 0)
50 				d += 10;
51 		}
52 #endif
53 		if (d >= base) {
54 			if (WANT_HEX_ESCAPES && base == 16) {
55 				--num_digits;
56 				if (num_digits == 0) {
57 					/* \x<bad_char>: return '\',
58 					 * leave ptr pointing to x */
59 					return '\\';
60 				}
61 			}
62 			break;
63 		}
64 
65 		r = n * base + d;
66 		if (r > UCHAR_MAX) {
67 			break;
68 		}
69 
70 		n = r;
71 		++q;
72 	} while (++num_digits < 3);
73 
74 	if (num_digits == 0) {
75 		/* Not octal or hex escape sequence.
76 		 * Is it one-letter one? */
77 
78 		/* bash builtin "echo -e '\ec'" interprets \e as ESC,
79 		 * but coreutils "/bin/echo -e '\ec'" does not.
80 		 * Manpages tend to support coreutils way.
81 		 * Update: coreutils added support for \e on 28 Oct 2009. */
82 		static const char charmap[] ALIGN1 = {
83 			'a',  'b', 'e', 'f',  'n',  'r',  't',  'v',  '\\', '\0',
84 			'\a', '\b', 27, '\f', '\n', '\r', '\t', '\v', '\\', '\\',
85 		};
86 		const char *p = charmap;
87 		do {
88 			if (*p == *q) {
89 				q++;
90 				break;
91 			}
92 		} while (*++p != '\0');
93 		/* p points to found escape char or NUL,
94 		 * advance it and find what it translates to.
95 		 * Note that \NUL and unrecognized sequence \z return '\'
96 		 * and leave ptr pointing to NUL or z. */
97 		n = p[sizeof(charmap) / 2];
98 	}
99 
100 	*ptr = q;
101 
102 	return (char) n;
103 }
104 
strcpy_and_process_escape_sequences(char * dst,const char * src)105 char* FAST_FUNC strcpy_and_process_escape_sequences(char *dst, const char *src)
106 {
107 	while (1) {
108 		char c, c1;
109 		c = c1 = *src++;
110 		if (c1 == '\\')
111 			c1 = bb_process_escape_sequence(&src);
112 		*dst = c1;
113 		if (c == '\0')
114 			return dst;
115 		dst++;
116 	}
117 }
118