blob: 346ecfa1e2c0c5de4bc299ee10c12d4c465ce685
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 | |
11 | #include "libbb.h" |
12 | |
13 | #define WANT_HEX_ESCAPES 1 |
14 | |
15 | /* Usual "this only works for ascii compatible encodings" disclaimer. */ |
16 | #undef _tolower |
17 | #define _tolower(X) ((X)|((char) 0x20)) |
18 | |
19 | char FAST_FUNC bb_process_escape_sequence(const char **ptr) |
20 | { |
21 | const char *q; |
22 | unsigned num_digits; |
23 | unsigned n; |
24 | unsigned base; |
25 | |
26 | num_digits = n = 0; |
27 | base = 8; |
28 | q = *ptr; |
29 | |
30 | if (WANT_HEX_ESCAPES && *q == 'x') { |
31 | ++q; |
32 | base = 16; |
33 | ++num_digits; |
34 | } |
35 | |
36 | /* bash requires leading 0 in octal escapes: |
37 | * \02 works, \2 does not (prints \ and 2). |
38 | * We treat \2 as a valid octal escape sequence. */ |
39 | do { |
40 | unsigned r; |
41 | #if !WANT_HEX_ESCAPES |
42 | unsigned d = (unsigned char)(*q) - '0'; |
43 | #else |
44 | unsigned d = (unsigned char)_tolower(*q) - '0'; |
45 | if (d >= 10) |
46 | d += ('0' - 'a' + 10); |
47 | #endif |
48 | if (d >= base) { |
49 | if (WANT_HEX_ESCAPES && base == 16) { |
50 | --num_digits; |
51 | if (num_digits == 0) { |
52 | /* \x<bad_char>: return '\', |
53 | * leave ptr pointing to x */ |
54 | return '\\'; |
55 | } |
56 | } |
57 | break; |
58 | } |
59 | |
60 | r = n * base + d; |
61 | if (r > UCHAR_MAX) { |
62 | break; |
63 | } |
64 | |
65 | n = r; |
66 | ++q; |
67 | } while (++num_digits < 3); |
68 | |
69 | if (num_digits == 0) { |
70 | /* Not octal or hex escape sequence. |
71 | * Is it one-letter one? */ |
72 | |
73 | /* bash builtin "echo -e '\ec'" interprets \e as ESC, |
74 | * but coreutils "/bin/echo -e '\ec'" does not. |
75 | * Manpages tend to support coreutils way. |
76 | * Update: coreutils added support for \e on 28 Oct 2009. */ |
77 | static const char charmap[] ALIGN1 = { |
78 | 'a', 'b', 'e', 'f', 'n', 'r', 't', 'v', '\\', '\0', |
79 | '\a', '\b', 27, '\f', '\n', '\r', '\t', '\v', '\\', '\\', |
80 | }; |
81 | const char *p = charmap; |
82 | do { |
83 | if (*p == *q) { |
84 | q++; |
85 | break; |
86 | } |
87 | } while (*++p != '\0'); |
88 | /* p points to found escape char or NUL, |
89 | * advance it and find what it translates to. |
90 | * Note that \NUL and unrecognized sequence \z return '\' |
91 | * and leave ptr pointing to NUL or z. */ |
92 | n = p[sizeof(charmap) / 2]; |
93 | } |
94 | |
95 | *ptr = q; |
96 | |
97 | return (char) n; |
98 | } |
99 | |
100 | char* FAST_FUNC strcpy_and_process_escape_sequences(char *dst, const char *src) |
101 | { |
102 | while (1) { |
103 | char c, c1; |
104 | c = c1 = *src++; |
105 | if (c1 == '\\') |
106 | c1 = bb_process_escape_sequence(&src); |
107 | *dst = c1; |
108 | if (c == '\0') |
109 | return dst; |
110 | dst++; |
111 | } |
112 | } |
113 |