|
1 /* |
|
2 ********************************************************************** |
|
3 * Copyright (C) 1999-2011, International Business Machines |
|
4 * Corporation and others. All Rights Reserved. |
|
5 ********************************************************************** |
|
6 * file name: ustr_imp.h |
|
7 * encoding: US-ASCII |
|
8 * tab size: 8 (not used) |
|
9 * indentation:4 |
|
10 * |
|
11 * created on: 2001jan30 |
|
12 * created by: Markus W. Scherer |
|
13 */ |
|
14 |
|
15 #ifndef __USTR_IMP_H__ |
|
16 #define __USTR_IMP_H__ |
|
17 |
|
18 #include "unicode/utypes.h" |
|
19 #include "unicode/uiter.h" |
|
20 #include "ucase.h" |
|
21 |
|
22 /** Simple declaration to avoid including unicode/ubrk.h. */ |
|
23 #ifndef UBRK_TYPEDEF_UBREAK_ITERATOR |
|
24 # define UBRK_TYPEDEF_UBREAK_ITERATOR |
|
25 typedef struct UBreakIterator UBreakIterator; |
|
26 #endif |
|
27 |
|
28 #ifndef U_COMPARE_IGNORE_CASE |
|
29 /* see also unorm.h */ |
|
30 /** |
|
31 * Option bit for unorm_compare: |
|
32 * Perform case-insensitive comparison. |
|
33 */ |
|
34 #define U_COMPARE_IGNORE_CASE 0x10000 |
|
35 #endif |
|
36 |
|
37 /** |
|
38 * Internal option for unorm_cmpEquivFold() for strncmp style. |
|
39 * If set, checks for both string length and terminating NUL. |
|
40 */ |
|
41 #define _STRNCMP_STYLE 0x1000 |
|
42 |
|
43 /** |
|
44 * Compare two strings in code point order or code unit order. |
|
45 * Works in strcmp style (both lengths -1), |
|
46 * strncmp style (lengths equal and >=0, flag TRUE), |
|
47 * and memcmp/UnicodeString style (at least one length >=0). |
|
48 */ |
|
49 U_CFUNC int32_t U_EXPORT2 |
|
50 uprv_strCompare(const UChar *s1, int32_t length1, |
|
51 const UChar *s2, int32_t length2, |
|
52 UBool strncmpStyle, UBool codePointOrder); |
|
53 |
|
54 /** |
|
55 * Internal API, used by u_strcasecmp() etc. |
|
56 * Compare strings case-insensitively, |
|
57 * in code point order or code unit order. |
|
58 */ |
|
59 U_CFUNC int32_t |
|
60 u_strcmpFold(const UChar *s1, int32_t length1, |
|
61 const UChar *s2, int32_t length2, |
|
62 uint32_t options, |
|
63 UErrorCode *pErrorCode); |
|
64 |
|
65 /** |
|
66 * Are the Unicode properties loaded? |
|
67 * This must be used before internal functions are called that do |
|
68 * not perform this check. |
|
69 * Generate a debug assertion failure if data is not loaded. |
|
70 */ |
|
71 U_CFUNC UBool |
|
72 uprv_haveProperties(UErrorCode *pErrorCode); |
|
73 |
|
74 /** |
|
75 * Load the Unicode property data. |
|
76 * Intended primarily for use from u_init(). |
|
77 * Has no effect if property data is already loaded. |
|
78 * NOT thread safe. |
|
79 */ |
|
80 /*U_CFUNC int8_t |
|
81 uprv_loadPropsData(UErrorCode *errorCode);*/ |
|
82 |
|
83 /* |
|
84 * Internal string casing functions implementing |
|
85 * ustring.h/ustrcase.c and UnicodeString case mapping functions. |
|
86 */ |
|
87 |
|
88 struct UCaseMap { |
|
89 const UCaseProps *csp; |
|
90 #if !UCONFIG_NO_BREAK_ITERATION |
|
91 UBreakIterator *iter; /* We adopt the iterator, so we own it. */ |
|
92 #endif |
|
93 char locale[32]; |
|
94 int32_t locCache; |
|
95 uint32_t options; |
|
96 }; |
|
97 |
|
98 #ifndef __UCASEMAP_H__ |
|
99 typedef struct UCaseMap UCaseMap; |
|
100 #endif |
|
101 |
|
102 #if UCONFIG_NO_BREAK_ITERATION |
|
103 # define UCASEMAP_INITIALIZER { NULL, { 0 }, 0, 0 } |
|
104 #else |
|
105 # define UCASEMAP_INITIALIZER { NULL, NULL, { 0 }, 0, 0 } |
|
106 #endif |
|
107 |
|
108 U_CFUNC void |
|
109 ustrcase_setTempCaseMapLocale(UCaseMap *csm, const char *locale); |
|
110 |
|
111 #ifndef U_STRING_CASE_MAPPER_DEFINED |
|
112 #define U_STRING_CASE_MAPPER_DEFINED |
|
113 |
|
114 /** |
|
115 * String case mapping function type, used by ustrcase_map(). |
|
116 * All error checking must be done. |
|
117 * The UCaseMap must be fully initialized, with locale and/or iter set as needed. |
|
118 * src and dest must not overlap. |
|
119 */ |
|
120 typedef int32_t U_CALLCONV |
|
121 UStringCaseMapper(const UCaseMap *csm, |
|
122 UChar *dest, int32_t destCapacity, |
|
123 const UChar *src, int32_t srcLength, |
|
124 UErrorCode *pErrorCode); |
|
125 |
|
126 #endif |
|
127 |
|
128 /** Implements UStringCaseMapper. */ |
|
129 U_CFUNC int32_t U_CALLCONV |
|
130 ustrcase_internalToLower(const UCaseMap *csm, |
|
131 UChar *dest, int32_t destCapacity, |
|
132 const UChar *src, int32_t srcLength, |
|
133 UErrorCode *pErrorCode); |
|
134 |
|
135 /** Implements UStringCaseMapper. */ |
|
136 U_CFUNC int32_t U_CALLCONV |
|
137 ustrcase_internalToUpper(const UCaseMap *csm, |
|
138 UChar *dest, int32_t destCapacity, |
|
139 const UChar *src, int32_t srcLength, |
|
140 UErrorCode *pErrorCode); |
|
141 |
|
142 #if !UCONFIG_NO_BREAK_ITERATION |
|
143 |
|
144 /** Implements UStringCaseMapper. */ |
|
145 U_CFUNC int32_t U_CALLCONV |
|
146 ustrcase_internalToTitle(const UCaseMap *csm, |
|
147 UChar *dest, int32_t destCapacity, |
|
148 const UChar *src, int32_t srcLength, |
|
149 UErrorCode *pErrorCode); |
|
150 |
|
151 #endif |
|
152 |
|
153 /** Implements UStringCaseMapper. */ |
|
154 U_CFUNC int32_t U_CALLCONV |
|
155 ustrcase_internalFold(const UCaseMap *csm, |
|
156 UChar *dest, int32_t destCapacity, |
|
157 const UChar *src, int32_t srcLength, |
|
158 UErrorCode *pErrorCode); |
|
159 |
|
160 /** |
|
161 * Implements argument checking and buffer handling |
|
162 * for string case mapping as a common function. |
|
163 */ |
|
164 U_CFUNC int32_t |
|
165 ustrcase_map(const UCaseMap *csm, |
|
166 UChar *dest, int32_t destCapacity, |
|
167 const UChar *src, int32_t srcLength, |
|
168 UStringCaseMapper *stringCaseMapper, |
|
169 UErrorCode *pErrorCode); |
|
170 |
|
171 /** |
|
172 * UTF-8 string case mapping function type, used by ucasemap_mapUTF8(). |
|
173 * UTF-8 version of UStringCaseMapper. |
|
174 * All error checking must be done. |
|
175 * The UCaseMap must be fully initialized, with locale and/or iter set as needed. |
|
176 * src and dest must not overlap. |
|
177 */ |
|
178 typedef int32_t U_CALLCONV |
|
179 UTF8CaseMapper(const UCaseMap *csm, |
|
180 uint8_t *dest, int32_t destCapacity, |
|
181 const uint8_t *src, int32_t srcLength, |
|
182 UErrorCode *pErrorCode); |
|
183 |
|
184 /** Implements UTF8CaseMapper. */ |
|
185 U_CFUNC int32_t U_CALLCONV |
|
186 ucasemap_internalUTF8ToTitle(const UCaseMap *csm, |
|
187 uint8_t *dest, int32_t destCapacity, |
|
188 const uint8_t *src, int32_t srcLength, |
|
189 UErrorCode *pErrorCode); |
|
190 |
|
191 /** |
|
192 * Implements argument checking and buffer handling |
|
193 * for UTF-8 string case mapping as a common function. |
|
194 */ |
|
195 U_CFUNC int32_t |
|
196 ucasemap_mapUTF8(const UCaseMap *csm, |
|
197 uint8_t *dest, int32_t destCapacity, |
|
198 const uint8_t *src, int32_t srcLength, |
|
199 UTF8CaseMapper *stringCaseMapper, |
|
200 UErrorCode *pErrorCode); |
|
201 |
|
202 U_CAPI int32_t U_EXPORT2 |
|
203 ustr_hashUCharsN(const UChar *str, int32_t length); |
|
204 |
|
205 U_CAPI int32_t U_EXPORT2 |
|
206 ustr_hashCharsN(const char *str, int32_t length); |
|
207 |
|
208 U_CAPI int32_t U_EXPORT2 |
|
209 ustr_hashICharsN(const char *str, int32_t length); |
|
210 |
|
211 /** |
|
212 * NUL-terminate a UChar * string if possible. |
|
213 * If length < destCapacity then NUL-terminate. |
|
214 * If length == destCapacity then do not terminate but set U_STRING_NOT_TERMINATED_WARNING. |
|
215 * If length > destCapacity then do not terminate but set U_BUFFER_OVERFLOW_ERROR. |
|
216 * |
|
217 * @param dest Destination buffer, can be NULL if destCapacity==0. |
|
218 * @param destCapacity Number of UChars available at dest. |
|
219 * @param length Number of UChars that were (to be) written to dest. |
|
220 * @param pErrorCode ICU error code. |
|
221 * @return length |
|
222 */ |
|
223 U_CAPI int32_t U_EXPORT2 |
|
224 u_terminateUChars(UChar *dest, int32_t destCapacity, int32_t length, UErrorCode *pErrorCode); |
|
225 |
|
226 /** |
|
227 * NUL-terminate a char * string if possible. |
|
228 * Same as u_terminateUChars() but for a different string type. |
|
229 */ |
|
230 U_CAPI int32_t U_EXPORT2 |
|
231 u_terminateChars(char *dest, int32_t destCapacity, int32_t length, UErrorCode *pErrorCode); |
|
232 |
|
233 /** |
|
234 * NUL-terminate a UChar32 * string if possible. |
|
235 * Same as u_terminateUChars() but for a different string type. |
|
236 */ |
|
237 U_CAPI int32_t U_EXPORT2 |
|
238 u_terminateUChar32s(UChar32 *dest, int32_t destCapacity, int32_t length, UErrorCode *pErrorCode); |
|
239 |
|
240 /** |
|
241 * NUL-terminate a wchar_t * string if possible. |
|
242 * Same as u_terminateUChars() but for a different string type. |
|
243 */ |
|
244 U_CAPI int32_t U_EXPORT2 |
|
245 u_terminateWChars(wchar_t *dest, int32_t destCapacity, int32_t length, UErrorCode *pErrorCode); |
|
246 |
|
247 #endif |