OLD | NEW |
| (Empty) |
1 // Copyright (c) 2011 The Chromium Authors. All rights reserved. | |
2 // Use of this source code is governed by a BSD-style license that can be | |
3 // found in the LICENSE file. | |
4 | |
5 #include "ppapi/shared_impl/ppb_char_set_shared.h" | |
6 | |
7 #include "base/i18n/icu_string_conversions.h" | |
8 #include "ppapi/c/dev/ppb_memory_dev.h" | |
9 #include "ppapi/thunk/thunk.h" | |
10 #include "unicode/ucnv.h" | |
11 #include "unicode/ucnv_cb.h" | |
12 #include "unicode/ucnv_err.h" | |
13 #include "unicode/ustring.h" | |
14 | |
15 namespace ppapi { | |
16 | |
17 namespace { | |
18 | |
19 // Converts the given PP error handling behavior to the version in base, | |
20 // placing the result in |*result| and returning true on success. Returns false | |
21 // if the enum is invalid. | |
22 bool PPToBaseConversionError(PP_CharSet_ConversionError on_error, | |
23 base::OnStringConversionError::Type* result) { | |
24 switch (on_error) { | |
25 case PP_CHARSET_CONVERSIONERROR_FAIL: | |
26 *result = base::OnStringConversionError::FAIL; | |
27 return true; | |
28 case PP_CHARSET_CONVERSIONERROR_SKIP: | |
29 *result = base::OnStringConversionError::SKIP; | |
30 return true; | |
31 case PP_CHARSET_CONVERSIONERROR_SUBSTITUTE: | |
32 *result = base::OnStringConversionError::SUBSTITUTE; | |
33 return true; | |
34 default: | |
35 return false; | |
36 } | |
37 } | |
38 | |
39 } // namespace | |
40 | |
41 // static | |
42 // The "substitution" behavior of this function does not match the | |
43 // implementation in base, so we partially duplicate the code from | |
44 // icu_string_conversions.cc with the correct error handling setup required | |
45 // by the PPAPI interface. | |
46 char* PPB_CharSet_Shared::UTF16ToCharSet( | |
47 const uint16_t* utf16, | |
48 uint32_t utf16_len, | |
49 const char* output_char_set, | |
50 PP_CharSet_ConversionError on_error, | |
51 uint32_t* output_length) { | |
52 if (!utf16 || !output_char_set || !output_length) | |
53 return NULL; | |
54 | |
55 *output_length = 0; | |
56 | |
57 UErrorCode status = U_ZERO_ERROR; | |
58 UConverter* converter = ucnv_open(output_char_set, &status); | |
59 if (!U_SUCCESS(status)) | |
60 return NULL; | |
61 | |
62 int encoded_max_length = UCNV_GET_MAX_BYTES_FOR_STRING(utf16_len, | |
63 ucnv_getMaxCharSize(converter)); | |
64 | |
65 // Setup our error handler. | |
66 switch (on_error) { | |
67 case PP_CHARSET_CONVERSIONERROR_FAIL: | |
68 ucnv_setFromUCallBack(converter, UCNV_FROM_U_CALLBACK_STOP, 0, | |
69 NULL, NULL, &status); | |
70 break; | |
71 case PP_CHARSET_CONVERSIONERROR_SKIP: | |
72 ucnv_setFromUCallBack(converter, UCNV_FROM_U_CALLBACK_SKIP, 0, | |
73 NULL, NULL, &status); | |
74 break; | |
75 case PP_CHARSET_CONVERSIONERROR_SUBSTITUTE: { | |
76 // ICU sets the substitution char for some character sets (like latin1) | |
77 // to be the ASCII "substitution character" (26). We want to use '?' | |
78 // instead for backwards-compat with Windows behavior. | |
79 char subst_chars[32]; | |
80 int8_t subst_chars_len = 32; | |
81 ucnv_getSubstChars(converter, subst_chars, &subst_chars_len, &status); | |
82 if (subst_chars_len == 1 && subst_chars[0] == 26) { | |
83 // Override to the question mark character if possible. When using | |
84 // setSubstString, the input is a Unicode character. The function will | |
85 // try to convert it to the destination character set and fail if that | |
86 // can not be converted to the destination character set. | |
87 // | |
88 // We just ignore any failure. If the dest char set has no | |
89 // representation for '?', then we'll just stick to the ICU default | |
90 // substitution character. | |
91 UErrorCode subst_status = U_ZERO_ERROR; | |
92 UChar question_mark = '?'; | |
93 ucnv_setSubstString(converter, &question_mark, 1, &subst_status); | |
94 } | |
95 | |
96 ucnv_setFromUCallBack(converter, UCNV_FROM_U_CALLBACK_SUBSTITUTE, 0, | |
97 NULL, NULL, &status); | |
98 break; | |
99 } | |
100 default: | |
101 return NULL; | |
102 } | |
103 | |
104 // ucnv_fromUChars returns size not including terminating null. | |
105 char* encoded = static_cast<char*>( | |
106 thunk::GetPPB_Memory_Dev_Thunk()->MemAlloc(encoded_max_length + 1)); | |
107 int actual_size = ucnv_fromUChars(converter, encoded, | |
108 encoded_max_length, reinterpret_cast<const UChar*>(utf16), utf16_len, | |
109 &status); | |
110 ucnv_close(converter); | |
111 if (!U_SUCCESS(status)) { | |
112 thunk::GetPPB_Memory_Dev_Thunk()->MemFree(encoded); | |
113 return NULL; | |
114 } | |
115 encoded[actual_size] = 0; | |
116 *output_length = actual_size; | |
117 return encoded; | |
118 } | |
119 | |
120 // static | |
121 uint16_t* PPB_CharSet_Shared::CharSetToUTF16( | |
122 const char* input, | |
123 uint32_t input_len, | |
124 const char* input_char_set, | |
125 PP_CharSet_ConversionError on_error, | |
126 uint32_t* output_length) { | |
127 if (!input || !input_char_set || !output_length) | |
128 return NULL; | |
129 | |
130 *output_length = 0; | |
131 | |
132 base::OnStringConversionError::Type base_on_error; | |
133 if (!PPToBaseConversionError(on_error, &base_on_error)) | |
134 return NULL; // Invalid enum value. | |
135 | |
136 // We can convert this call to the implementation in base to avoid code | |
137 // duplication, although this does introduce an extra copy of the data. | |
138 string16 output; | |
139 if (!base::CodepageToUTF16(std::string(input, input_len), input_char_set, | |
140 base_on_error, &output)) | |
141 return NULL; | |
142 | |
143 uint16_t* ret_buf = static_cast<uint16_t*>( | |
144 thunk::GetPPB_Memory_Dev_Thunk()->MemAlloc( | |
145 (output.size() + 1) * sizeof(uint16_t))); | |
146 if (!ret_buf) | |
147 return NULL; | |
148 | |
149 *output_length = static_cast<uint32_t>(output.size()); | |
150 memcpy(ret_buf, output.c_str(), (output.size() + 1) * sizeof(uint16_t)); | |
151 return ret_buf; | |
152 } | |
153 | |
154 } // namespace ppapi | |
OLD | NEW |