1 // Copyright (c) 2009 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 "base/strings/utf_string_conversion_utils.h"
6
7 #include "base/third_party/icu/icu_utf.h"
8
9 namespace base {
10
11 // ReadUnicodeCharacter --------------------------------------------------------
12
ReadUnicodeCharacter(const char * src,int32_t src_len,int32_t * char_index,uint32_t * code_point_out)13 bool ReadUnicodeCharacter(const char* src,
14 int32_t src_len,
15 int32_t* char_index,
16 uint32_t* code_point_out) {
17 // U8_NEXT expects to be able to use -1 to signal an error, so we must
18 // use a signed type for code_point. But this function returns false
19 // on error anyway, so code_point_out is unsigned.
20 int32_t code_point;
21 CBU8_NEXT(src, *char_index, src_len, code_point);
22 *code_point_out = static_cast<uint32_t>(code_point);
23
24 // The ICU macro above moves to the next char, we want to point to the last
25 // char consumed.
26 (*char_index)--;
27
28 // Validate the decoded value.
29 return IsValidCodepoint(code_point);
30 }
31
ReadUnicodeCharacter(const char16 * src,int32_t src_len,int32_t * char_index,uint32_t * code_point)32 bool ReadUnicodeCharacter(const char16* src,
33 int32_t src_len,
34 int32_t* char_index,
35 uint32_t* code_point) {
36 if (CBU16_IS_SURROGATE(src[*char_index])) {
37 if (!CBU16_IS_SURROGATE_LEAD(src[*char_index]) ||
38 *char_index + 1 >= src_len ||
39 !CBU16_IS_TRAIL(src[*char_index + 1])) {
40 // Invalid surrogate pair.
41 return false;
42 }
43
44 // Valid surrogate pair.
45 *code_point = CBU16_GET_SUPPLEMENTARY(src[*char_index],
46 src[*char_index + 1]);
47 (*char_index)++;
48 } else {
49 // Not a surrogate, just one 16-bit word.
50 *code_point = src[*char_index];
51 }
52
53 return IsValidCodepoint(*code_point);
54 }
55
56 #if defined(WCHAR_T_IS_UTF32)
ReadUnicodeCharacter(const wchar_t * src,int32_t src_len,int32_t * char_index,uint32_t * code_point)57 bool ReadUnicodeCharacter(const wchar_t* src,
58 int32_t src_len,
59 int32_t* char_index,
60 uint32_t* code_point) {
61 // Conversion is easy since the source is 32-bit.
62 *code_point = src[*char_index];
63
64 // Validate the value.
65 return IsValidCodepoint(*code_point);
66 }
67 #endif // defined(WCHAR_T_IS_UTF32)
68
69 // WriteUnicodeCharacter -------------------------------------------------------
70
WriteUnicodeCharacter(uint32_t code_point,std::string * output)71 size_t WriteUnicodeCharacter(uint32_t code_point, std::string* output) {
72 if (code_point <= 0x7f) {
73 // Fast path the common case of one byte.
74 output->push_back(static_cast<char>(code_point));
75 return 1;
76 }
77
78
79 // CBU8_APPEND_UNSAFE can append up to 4 bytes.
80 size_t char_offset = output->length();
81 size_t original_char_offset = char_offset;
82 output->resize(char_offset + CBU8_MAX_LENGTH);
83
84 CBU8_APPEND_UNSAFE(&(*output)[0], char_offset, code_point);
85
86 // CBU8_APPEND_UNSAFE will advance our pointer past the inserted character, so
87 // it will represent the new length of the string.
88 output->resize(char_offset);
89 return char_offset - original_char_offset;
90 }
91
WriteUnicodeCharacter(uint32_t code_point,string16 * output)92 size_t WriteUnicodeCharacter(uint32_t code_point, string16* output) {
93 if (CBU16_LENGTH(code_point) == 1) {
94 // Thie code point is in the Basic Multilingual Plane (BMP).
95 output->push_back(static_cast<char16>(code_point));
96 return 1;
97 }
98 // Non-BMP characters use a double-character encoding.
99 size_t char_offset = output->length();
100 output->resize(char_offset + CBU16_MAX_LENGTH);
101 CBU16_APPEND_UNSAFE(&(*output)[0], char_offset, code_point);
102 return CBU16_MAX_LENGTH;
103 }
104
105 // Generalized Unicode converter -----------------------------------------------
106
107 template<typename CHAR>
PrepareForUTF8Output(const CHAR * src,size_t src_len,std::string * output)108 void PrepareForUTF8Output(const CHAR* src,
109 size_t src_len,
110 std::string* output) {
111 output->clear();
112 if (src_len == 0)
113 return;
114 if (src[0] < 0x80) {
115 // Assume that the entire input will be ASCII.
116 output->reserve(src_len);
117 } else {
118 // Assume that the entire input is non-ASCII and will have 3 bytes per char.
119 output->reserve(src_len * 3);
120 }
121 }
122
123 // Instantiate versions we know callers will need.
124 template void PrepareForUTF8Output(const wchar_t*, size_t, std::string*);
125 template void PrepareForUTF8Output(const char16*, size_t, std::string*);
126
127 template<typename STRING>
PrepareForUTF16Or32Output(const char * src,size_t src_len,STRING * output)128 void PrepareForUTF16Or32Output(const char* src,
129 size_t src_len,
130 STRING* output) {
131 output->clear();
132 if (src_len == 0)
133 return;
134 if (static_cast<unsigned char>(src[0]) < 0x80) {
135 // Assume the input is all ASCII, which means 1:1 correspondence.
136 output->reserve(src_len);
137 } else {
138 // Otherwise assume that the UTF-8 sequences will have 2 bytes for each
139 // character.
140 output->reserve(src_len / 2);
141 }
142 }
143
144 // Instantiate versions we know callers will need.
145 template void PrepareForUTF16Or32Output(const char*, size_t, std::wstring*);
146 template void PrepareForUTF16Or32Output(const char*, size_t, string16*);
147
148 } // namespace base
149