OLD | NEW |
1 // Copyright (c) 2006-2008 The Chromium Authors. All rights reserved. | 1 // Copyright (c) 2006-2008 The Chromium Authors. All rights reserved. |
2 // Use of this source code is governed by a BSD-style license that can be | 2 // Use of this source code is governed by a BSD-style license that can be |
3 // found in the LICENSE file. | 3 // found in the LICENSE file. |
4 | 4 |
5 #include "base/sys_string_conversions.h" | 5 #include "base/sys_string_conversions.h" |
6 | 6 |
7 #include <wchar.h> | |
8 | |
9 #include "base/string_piece.h" | 7 #include "base/string_piece.h" |
10 #include "base/string_util.h" | 8 #include "base/string_util.h" |
11 | 9 |
12 namespace base { | 10 namespace base { |
13 | 11 |
14 std::string SysWideToUTF8(const std::wstring& wide) { | 12 std::string SysWideToUTF8(const std::wstring& wide) { |
15 // In theory this should be using the system-provided conversion rather | 13 // In theory this should be using the system-provided conversion rather |
16 // than our ICU, but this will do for now. | 14 // than our ICU, but this will do for now. |
17 return WideToUTF8(wide); | 15 return WideToUTF8(wide); |
18 } | 16 } |
19 std::wstring SysUTF8ToWide(const StringPiece& utf8) { | 17 std::wstring SysUTF8ToWide(const StringPiece& utf8) { |
20 // In theory this should be using the system-provided conversion rather | 18 // In theory this should be using the system-provided conversion rather |
21 // than our ICU, but this will do for now. | 19 // than our ICU, but this will do for now. |
22 std::wstring out; | 20 std::wstring out; |
23 UTF8ToWide(utf8.data(), utf8.size(), &out); | 21 UTF8ToWide(utf8.data(), utf8.size(), &out); |
24 return out; | 22 return out; |
25 } | 23 } |
26 | 24 |
27 std::string SysWideToNativeMB(const std::wstring& wide) { | 25 std::string SysWideToNativeMB(const std::wstring& wide) { |
28 mbstate_t ps; | 26 // TODO(evanm): we can't assume Linux is UTF-8. |
29 | 27 return SysWideToUTF8(wide); |
30 // Calculate the number of multi-byte characters. We walk through the string | |
31 // without writing the output, counting the number of multi-byte characters. | |
32 size_t num_out_chars = 0; | |
33 memset(&ps, 0, sizeof(ps)); | |
34 for (size_t i = 0; i < wide.size(); ++i) { | |
35 const wchar_t src = wide[i]; | |
36 // Use a temp buf since a output of NULL does not do what we want. | |
37 char buf[16]; | |
38 // We don't want wcrtomb to do it's funkiness for embedded NULLs. | |
39 size_t res = src ? wcrtomb(buf, src, &ps) : 0; | |
40 switch (res) { | |
41 // Handle any errors and return an empty string. | |
42 case -1: | |
43 return std::string(); | |
44 break; | |
45 case 0: | |
46 // We hit an embedded null byte, keep going. | |
47 ++num_out_chars; | |
48 default: | |
49 num_out_chars += res; | |
50 break; | |
51 } | |
52 } | |
53 | |
54 if (num_out_chars == 0) | |
55 return std::string(); | |
56 | |
57 std::string out; | |
58 out.resize(num_out_chars); | |
59 | |
60 // We walk the input string again, with |i| tracking the index of the | |
61 // wide input, and |j| tracking the multi-byte output. | |
62 memset(&ps, 0, sizeof(ps)); | |
63 for (size_t i = 0, j = 0; i < wide.size(); ++i) { | |
64 const wchar_t src = wide[i]; | |
65 // We don't want wcrtomb to do it's funkiness for embedded NULLs. | |
66 size_t res = src ? wcrtomb(&out[j], src, &ps) : 0; | |
67 switch (res) { | |
68 // Handle any errors and return an empty string. | |
69 case -1: | |
70 return std::string(); | |
71 break; | |
72 case 0: | |
73 // We hit an embedded null byte, keep going. | |
74 ++j; // Output is already 0. | |
75 default: | |
76 j += res; | |
77 break; | |
78 } | |
79 } | |
80 | |
81 return out; | |
82 } | 28 } |
83 | 29 |
84 std::wstring SysNativeMBToWide(const StringPiece& native_mb) { | 30 std::wstring SysNativeMBToWide(const StringPiece& native_mb) { |
85 mbstate_t ps; | 31 // TODO(evanm): we can't assume Linux is UTF-8. |
86 | 32 return SysUTF8ToWide(native_mb); |
87 // Calculate the number of wide characters. We walk through the string | |
88 // without writing the output, counting the number of wide characters. | |
89 size_t num_out_chars = 0; | |
90 memset(&ps, 0, sizeof(ps)); | |
91 for (size_t i = 0; i < native_mb.size(); ) { | |
92 const char* src = native_mb.data() + i; | |
93 size_t res = mbrtowc(NULL, src, native_mb.size() - i, &ps); | |
94 switch (res) { | |
95 // Handle any errors and return an empty string. | |
96 case -2: | |
97 case -1: | |
98 return std::wstring(); | |
99 break; | |
100 case 0: | |
101 // We hit an embedded null byte, keep going. | |
102 i += 1; // Fall through. | |
103 default: | |
104 i += res; | |
105 ++num_out_chars; | |
106 break; | |
107 } | |
108 } | |
109 | |
110 if (num_out_chars == 0) | |
111 return std::wstring(); | |
112 | |
113 std::wstring out; | |
114 out.resize(num_out_chars); | |
115 | |
116 memset(&ps, 0, sizeof(ps)); // Clear the shift state. | |
117 // We walk the input string again, with |i| tracking the index of the | |
118 // multi-byte input, and |j| tracking the wide output. | |
119 for (size_t i = 0, j = 0; i < native_mb.size(); ++j) { | |
120 const char* src = native_mb.data() + i; | |
121 wchar_t* dst = &out[j]; | |
122 size_t res = mbrtowc(dst, src, native_mb.size() - i, &ps); | |
123 switch (res) { | |
124 // Handle any errors and return an empty string. | |
125 case -2: | |
126 case -1: | |
127 return std::wstring(); | |
128 break; | |
129 case 0: | |
130 i += 1; // Skip null, fall through. | |
131 default: | |
132 i += res; | |
133 break; | |
134 } | |
135 } | |
136 | |
137 return out; | |
138 } | 33 } |
139 | 34 |
140 } // namespace base | 35 } // namespace base |
OLD | NEW |