OLD | NEW |
1 // Copyright (c) 2011 The Chromium Authors. All rights reserved. | 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 | 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 #ifndef BASE_I18N_CHAR_ITERATOR_H_ | 5 #ifndef BASE_I18N_CHAR_ITERATOR_H_ |
6 #define BASE_I18N_CHAR_ITERATOR_H_ | 6 #define BASE_I18N_CHAR_ITERATOR_H_ |
7 | 7 |
| 8 #include <stddef.h> |
| 9 #include <stdint.h> |
| 10 |
8 #include <string> | 11 #include <string> |
9 | 12 |
10 #include "base/basictypes.h" | |
11 #include "base/i18n/base_i18n_export.h" | 13 #include "base/i18n/base_i18n_export.h" |
| 14 #include "base/macros.h" |
12 #include "base/strings/string16.h" | 15 #include "base/strings/string16.h" |
| 16 #include "build/build_config.h" |
13 | 17 |
14 // The CharIterator classes iterate through the characters in UTF8 and | 18 // The CharIterator classes iterate through the characters in UTF8 and |
15 // UTF16 strings. Example usage: | 19 // UTF16 strings. Example usage: |
16 // | 20 // |
17 // UTF8CharIterator iter(&str); | 21 // UTF8CharIterator iter(&str); |
18 // while (!iter.end()) { | 22 // while (!iter.end()) { |
19 // VLOG(1) << iter.get(); | 23 // VLOG(1) << iter.get(); |
20 // iter.Advance(); | 24 // iter.Advance(); |
21 // } | 25 // } |
22 | 26 |
23 #if defined(OS_WIN) | 27 #if defined(OS_WIN) |
24 typedef unsigned char uint8_t; | 28 typedef unsigned char uint8_t; |
25 #endif | 29 #endif |
26 | 30 |
27 namespace base { | 31 namespace base { |
28 namespace i18n { | 32 namespace i18n { |
29 | 33 |
30 class BASE_I18N_EXPORT UTF8CharIterator { | 34 class BASE_I18N_EXPORT UTF8CharIterator { |
31 public: | 35 public: |
32 // Requires |str| to live as long as the UTF8CharIterator does. | 36 // Requires |str| to live as long as the UTF8CharIterator does. |
33 explicit UTF8CharIterator(const std::string* str); | 37 explicit UTF8CharIterator(const std::string* str); |
34 ~UTF8CharIterator(); | 38 ~UTF8CharIterator(); |
35 | 39 |
36 // Return the starting array index of the current character within the | 40 // Return the starting array index of the current character within the |
37 // string. | 41 // string. |
38 int32 array_pos() const { return array_pos_; } | 42 int32_t array_pos() const { return array_pos_; } |
39 | 43 |
40 // Return the logical index of the current character, independent of the | 44 // Return the logical index of the current character, independent of the |
41 // number of bytes each character takes. | 45 // number of bytes each character takes. |
42 int32 char_pos() const { return char_pos_; } | 46 int32_t char_pos() const { return char_pos_; } |
43 | 47 |
44 // Return the current char. | 48 // Return the current char. |
45 int32 get() const { return char_; } | 49 int32_t get() const { return char_; } |
46 | 50 |
47 // Returns true if we're at the end of the string. | 51 // Returns true if we're at the end of the string. |
48 bool end() const { return array_pos_ == len_; } | 52 bool end() const { return array_pos_ == len_; } |
49 | 53 |
50 // Advance to the next actual character. Returns false if we're at the | 54 // Advance to the next actual character. Returns false if we're at the |
51 // end of the string. | 55 // end of the string. |
52 bool Advance(); | 56 bool Advance(); |
53 | 57 |
54 private: | 58 private: |
55 // The string we're iterating over. | 59 // The string we're iterating over. |
56 const uint8_t* str_; | 60 const uint8_t* str_; |
57 | 61 |
58 // The length of the encoded string. | 62 // The length of the encoded string. |
59 int32 len_; | 63 int32_t len_; |
60 | 64 |
61 // Array index. | 65 // Array index. |
62 int32 array_pos_; | 66 int32_t array_pos_; |
63 | 67 |
64 // The next array index. | 68 // The next array index. |
65 int32 next_pos_; | 69 int32_t next_pos_; |
66 | 70 |
67 // Character index. | 71 // Character index. |
68 int32 char_pos_; | 72 int32_t char_pos_; |
69 | 73 |
70 // The current character. | 74 // The current character. |
71 int32 char_; | 75 int32_t char_; |
72 | 76 |
73 DISALLOW_COPY_AND_ASSIGN(UTF8CharIterator); | 77 DISALLOW_COPY_AND_ASSIGN(UTF8CharIterator); |
74 }; | 78 }; |
75 | 79 |
76 class BASE_I18N_EXPORT UTF16CharIterator { | 80 class BASE_I18N_EXPORT UTF16CharIterator { |
77 public: | 81 public: |
78 // Requires |str| to live as long as the UTF16CharIterator does. | 82 // Requires |str| to live as long as the UTF16CharIterator does. |
79 explicit UTF16CharIterator(const string16* str); | 83 explicit UTF16CharIterator(const string16* str); |
80 UTF16CharIterator(const char16* str, size_t str_len); | 84 UTF16CharIterator(const char16* str, size_t str_len); |
81 ~UTF16CharIterator(); | 85 ~UTF16CharIterator(); |
82 | 86 |
83 // Return the starting array index of the current character within the | 87 // Return the starting array index of the current character within the |
84 // string. | 88 // string. |
85 int32 array_pos() const { return array_pos_; } | 89 int32_t array_pos() const { return array_pos_; } |
86 | 90 |
87 // Return the logical index of the current character, independent of the | 91 // Return the logical index of the current character, independent of the |
88 // number of codewords each character takes. | 92 // number of codewords each character takes. |
89 int32 char_pos() const { return char_pos_; } | 93 int32_t char_pos() const { return char_pos_; } |
90 | 94 |
91 // Return the current char. | 95 // Return the current char. |
92 int32 get() const { return char_; } | 96 int32_t get() const { return char_; } |
93 | 97 |
94 // Returns true if we're at the end of the string. | 98 // Returns true if we're at the end of the string. |
95 bool end() const { return array_pos_ == len_; } | 99 bool end() const { return array_pos_ == len_; } |
96 | 100 |
97 // Advance to the next actual character. Returns false if we're at the | 101 // Advance to the next actual character. Returns false if we're at the |
98 // end of the string. | 102 // end of the string. |
99 bool Advance(); | 103 bool Advance(); |
100 | 104 |
101 private: | 105 private: |
102 // Fills in the current character we found and advances to the next | 106 // Fills in the current character we found and advances to the next |
103 // character, updating all flags as necessary. | 107 // character, updating all flags as necessary. |
104 void ReadChar(); | 108 void ReadChar(); |
105 | 109 |
106 // The string we're iterating over. | 110 // The string we're iterating over. |
107 const char16* str_; | 111 const char16* str_; |
108 | 112 |
109 // The length of the encoded string. | 113 // The length of the encoded string. |
110 int32 len_; | 114 int32_t len_; |
111 | 115 |
112 // Array index. | 116 // Array index. |
113 int32 array_pos_; | 117 int32_t array_pos_; |
114 | 118 |
115 // The next array index. | 119 // The next array index. |
116 int32 next_pos_; | 120 int32_t next_pos_; |
117 | 121 |
118 // Character index. | 122 // Character index. |
119 int32 char_pos_; | 123 int32_t char_pos_; |
120 | 124 |
121 // The current character. | 125 // The current character. |
122 int32 char_; | 126 int32_t char_; |
123 | 127 |
124 DISALLOW_COPY_AND_ASSIGN(UTF16CharIterator); | 128 DISALLOW_COPY_AND_ASSIGN(UTF16CharIterator); |
125 }; | 129 }; |
126 | 130 |
127 } // namespace i18n | 131 } // namespace i18n |
128 } // namespace base | 132 } // namespace base |
129 | 133 |
130 #endif // BASE_I18N_CHAR_ITERATOR_H_ | 134 #endif // BASE_I18N_CHAR_ITERATOR_H_ |
OLD | NEW |