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/char_set_impl.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* CharSetImpl::UTF16ToCharSet(const uint16_t* utf16, | |
47 uint32_t utf16_len, | |
48 const char* output_char_set, | |
49 PP_CharSet_ConversionError on_error, | |
50 uint32_t* output_length) { | |
51 if (!utf16 || !output_char_set || !output_length) | |
52 return NULL; | |
53 | |
54 *output_length = 0; | |
55 | |
56 UErrorCode status = U_ZERO_ERROR; | |
57 UConverter* converter = ucnv_open(output_char_set, &status); | |
58 if (!U_SUCCESS(status)) | |
59 return NULL; | |
60 | |
61 int encoded_max_length = UCNV_GET_MAX_BYTES_FOR_STRING(utf16_len, | |
62 ucnv_getMaxCharSize(converter)); | |
63 | |
64 // Setup our error handler. | |
65 switch (on_error) { | |
66 case PP_CHARSET_CONVERSIONERROR_FAIL: | |
67 ucnv_setFromUCallBack(converter, UCNV_FROM_U_CALLBACK_STOP, 0, | |
68 NULL, NULL, &status); | |
69 break; | |
70 case PP_CHARSET_CONVERSIONERROR_SKIP: | |
71 ucnv_setFromUCallBack(converter, UCNV_FROM_U_CALLBACK_SKIP, 0, | |
72 NULL, NULL, &status); | |
73 break; | |
74 case PP_CHARSET_CONVERSIONERROR_SUBSTITUTE: { | |
75 // ICU sets the substitution char for some character sets (like latin1) | |
76 // to be the ASCII "substitution character" (26). We want to use '?' | |
77 // instead for backwards-compat with Windows behavior. | |
78 char subst_chars[32]; | |
79 int8_t subst_chars_len = 32; | |
80 ucnv_getSubstChars(converter, subst_chars, &subst_chars_len, &status); | |
81 if (subst_chars_len == 1 && subst_chars[0] == 26) { | |
82 // Override to the question mark character if possible. When using | |
83 // setSubstString, the input is a Unicode character. The function will | |
84 // try to convert it to the destination character set and fail if that | |
85 // can not be converted to the destination character set. | |
86 // | |
87 // We just ignore any failure. If the dest char set has no | |
88 // representation for '?', then we'll just stick to the ICU default | |
89 // substitution character. | |
90 UErrorCode subst_status = U_ZERO_ERROR; | |
91 UChar question_mark = '?'; | |
92 ucnv_setSubstString(converter, &question_mark, 1, &subst_status); | |
93 } | |
94 | |
95 ucnv_setFromUCallBack(converter, UCNV_FROM_U_CALLBACK_SUBSTITUTE, 0, | |
96 NULL, NULL, &status); | |
97 break; | |
98 } | |
99 default: | |
100 return NULL; | |
101 } | |
102 | |
103 // ucnv_fromUChars returns size not including terminating null. | |
104 char* encoded = static_cast<char*>( | |
105 thunk::GetPPB_Memory_Dev_Thunk()->MemAlloc(encoded_max_length + 1)); | |
106 int actual_size = ucnv_fromUChars(converter, encoded, | |
107 encoded_max_length, reinterpret_cast<const UChar*>(utf16), utf16_len, | |
108 &status); | |
109 ucnv_close(converter); | |
110 if (!U_SUCCESS(status)) { | |
111 thunk::GetPPB_Memory_Dev_Thunk()->MemFree(encoded); | |
112 return NULL; | |
113 } | |
114 encoded[actual_size] = 0; | |
115 *output_length = actual_size; | |
116 return encoded; | |
117 } | |
118 | |
119 // static | |
120 uint16_t* CharSetImpl::CharSetToUTF16(const char* input, | |
121 uint32_t input_len, | |
122 const char* input_char_set, | |
123 PP_CharSet_ConversionError on_error, | |
124 uint32_t* output_length) { | |
125 if (!input || !input_char_set || !output_length) | |
126 return NULL; | |
127 | |
128 *output_length = 0; | |
129 | |
130 base::OnStringConversionError::Type base_on_error; | |
131 if (!PPToBaseConversionError(on_error, &base_on_error)) | |
132 return NULL; // Invalid enum value. | |
133 | |
134 // We can convert this call to the implementation in base to avoid code | |
135 // duplication, although this does introduce an extra copy of the data. | |
136 string16 output; | |
137 if (!base::CodepageToUTF16(std::string(input, input_len), input_char_set, | |
138 base_on_error, &output)) | |
139 return NULL; | |
140 | |
141 uint16_t* ret_buf = static_cast<uint16_t*>( | |
142 thunk::GetPPB_Memory_Dev_Thunk()->MemAlloc( | |
143 (output.size() + 1) * sizeof(uint16_t))); | |
144 if (!ret_buf) | |
145 return NULL; | |
146 | |
147 *output_length = static_cast<uint32_t>(output.size()); | |
148 memcpy(ret_buf, output.c_str(), (output.size() + 1) * sizeof(uint16_t)); | |
149 return ret_buf; | |
150 } | |
151 | |
152 } // namespace ppapi | |
OLD | NEW |