1 | //===----------------------------------------------------------------------===// |
2 | // |
3 | // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions. |
4 | // See https://llvm.org/LICENSE.txt for license information. |
5 | // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception |
6 | // |
7 | //===----------------------------------------------------------------------===// |
8 | |
9 | #include <cstddef> // size_t |
10 | #include <cwchar> // mbstate_t |
11 | #include <limits.h> // MB_LEN_MAX |
12 | #include <string.h> // wmemcpy |
13 | |
14 | // Returns the number of wide characters found in the multi byte sequence `src` |
15 | // (of `src_size_bytes`), that fit in the buffer `dst` (of `max_dest_chars` |
16 | // elements size). The count returned excludes the null terminator. |
17 | // When `dst` is NULL, no characters are copied to `dst`. |
18 | // Returns (size_t) -1 when an invalid sequence is encountered. |
19 | // Leaves *`src` pointing to the next character to convert or NULL |
20 | // if a null character was converted from *`src`. |
21 | _LIBCPP_EXPORTED_FROM_ABI size_t mbsnrtowcs( |
22 | wchar_t* __restrict dst, |
23 | const char** __restrict src, |
24 | size_t src_size_bytes, |
25 | size_t max_dest_chars, |
26 | mbstate_t* __restrict ps) { |
27 | const size_t terminated_sequence = static_cast<size_t>(0); |
28 | const size_t invalid_sequence = static_cast<size_t>(-1); |
29 | const size_t incomplete_sequence = static_cast<size_t>(-2); |
30 | |
31 | size_t source_converted; |
32 | size_t dest_converted; |
33 | size_t result = 0; |
34 | |
35 | // If `dst` is null then `max_dest_chars` should be ignored according to the |
36 | // standard. Setting `max_dest_chars` to a large value has this effect. |
37 | if (dst == nullptr) |
38 | max_dest_chars = static_cast<size_t>(-1); |
39 | |
40 | for (dest_converted = source_converted = 0; |
41 | source_converted < src_size_bytes && (!dst || dest_converted < max_dest_chars); |
42 | ++dest_converted, source_converted += result) { |
43 | // Converts one multi byte character. |
44 | // If result (char_size) is greater than 0, it's the size in bytes of that character. |
45 | // If result (char_size) is zero, it indicates that the null character has been found. |
46 | // Otherwise, it's an error and errno may be set. |
47 | size_t source_remaining = src_size_bytes - source_converted; |
48 | size_t dest_remaining = max_dest_chars - dest_converted; |
49 | |
50 | if (dst == nullptr) { |
51 | result = mbrtowc(NULL, *src + source_converted, source_remaining, ps); |
52 | } else if (dest_remaining >= source_remaining) { |
53 | // dst has enough space to translate in-place. |
54 | result = mbrtowc(dst + dest_converted, *src + source_converted, source_remaining, ps); |
55 | } else { |
56 | /* |
57 | * dst may not have enough space, so use a temporary buffer. |
58 | * |
59 | * We need to save a copy of the conversion state |
60 | * here so we can restore it if the multibyte |
61 | * character is too long for the buffer. |
62 | */ |
63 | wchar_t buff[MB_LEN_MAX]; |
64 | mbstate_t mbstate_tmp; |
65 | |
66 | if (ps != nullptr) |
67 | mbstate_tmp = *ps; |
68 | result = mbrtowc(buff, *src + source_converted, source_remaining, ps); |
69 | |
70 | if (result > dest_remaining) { |
71 | // Multi-byte sequence for character won't fit. |
72 | if (ps != nullptr) |
73 | *ps = mbstate_tmp; |
74 | break; |
75 | } else { |
76 | // The buffer was used, so we need copy the translation to dst. |
77 | wmemcpy(dst, buff, result); |
78 | } |
79 | } |
80 | |
81 | // Don't do anything to change errno from here on. |
82 | if (result == invalid_sequence || result == terminated_sequence || result == incomplete_sequence) { |
83 | break; |
84 | } |
85 | } |
86 | |
87 | if (dst) { |
88 | if (result == terminated_sequence) |
89 | *src = NULL; |
90 | else |
91 | *src += source_converted; |
92 | } |
93 | if (result == invalid_sequence) |
94 | return invalid_sequence; |
95 | |
96 | return dest_converted; |
97 | } |
98 | |