ucase.h revision b26ce3a7367e4ed2ee7ddddcdc3f3d3377a455c2 
1/*
2*******************************************************************************
3*
4*   Copyright (C) 2004-2010, International Business Machines
5*   Corporation and others.  All Rights Reserved.
6*
7*******************************************************************************
8*   file name:  ucase.h
9*   encoding:   US-ASCII
10*   tab size:   8 (not used)
11*   indentation:4
12*
13*   created on: 2004aug30
14*   created by: Markus W. Scherer
15*
16*   Low-level Unicode character/string case mapping code.
17*/
18
19#ifndef __UCASE_H__
20#define __UCASE_H__
21
22#include "unicode/utypes.h"
23#include "unicode/uset.h"
24#include "uset_imp.h"
25#include "udataswp.h"
26
27U_CDECL_BEGIN
28
29/* library API -------------------------------------------------------------- */
30
31struct UCaseProps;
32typedef struct UCaseProps UCaseProps;
33
34U_CAPI const UCaseProps * U_EXPORT2
35ucase_getSingleton(void);
36
37U_CFUNC void U_EXPORT2
38ucase_addPropertyStarts(const UCaseProps *csp, const USetAdder *sa, UErrorCode *pErrorCode);
39
40/**
41 * Requires non-NULL locale ID but otherwise does the equivalent of
42 * checking for language codes as if uloc_getLanguage() were called:
43 * Accepts both 2- and 3-letter codes and accepts case variants.
44 */
45U_CFUNC int32_t
46ucase_getCaseLocale(const char *locale, int32_t *locCache);
47
48/* Casing locale types for ucase_getCaseLocale */
49enum {
50    UCASE_LOC_UNKNOWN,
51    UCASE_LOC_ROOT,
52    UCASE_LOC_TURKISH,
53    UCASE_LOC_LITHUANIAN,
54    UCASE_LOC_DUTCH
55};
56
57/**
58 * Bit mask for getting just the options from a string compare options word
59 * that are relevant for case-insensitive string comparison.
60 * See uchar.h. Also include _STRNCMP_STYLE and U_COMPARE_CODE_POINT_ORDER.
61 * @internal
62 */
63#define _STRCASECMP_OPTIONS_MASK 0xffff
64
65/**
66 * Bit mask for getting just the options from a string compare options word
67 * that are relevant for case folding (of a single string or code point).
68 * See uchar.h.
69 * @internal
70 */
71#define _FOLD_CASE_OPTIONS_MASK 0xff
72
73/* single-code point functions */
74
75U_CAPI UChar32 U_EXPORT2
76ucase_tolower(const UCaseProps *csp, UChar32 c);
77
78U_CAPI UChar32 U_EXPORT2
79ucase_toupper(const UCaseProps *csp, UChar32 c);
80
81U_CAPI UChar32 U_EXPORT2
82ucase_totitle(const UCaseProps *csp, UChar32 c);
83
84U_CAPI UChar32 U_EXPORT2
85ucase_fold(const UCaseProps *csp, UChar32 c, uint32_t options);
86
87/**
88 * Adds all simple case mappings and the full case folding for c to sa,
89 * and also adds special case closure mappings.
90 * c itself is not added.
91 * For example, the mappings
92 * - for s include long s
93 * - for sharp s include ss
94 * - for k include the Kelvin sign
95 */
96U_CFUNC void U_EXPORT2
97ucase_addCaseClosure(const UCaseProps *csp, UChar32 c, const USetAdder *sa);
98
99/**
100 * Maps the string to single code points and adds the associated case closure
101 * mappings.
102 * The string is mapped to code points if it is their full case folding string.
103 * In other words, this performs a reverse full case folding and then
104 * adds the case closure items of the resulting code points.
105 * If the string is found and its closure applied, then
106 * the string itself is added as well as part of its code points' closure.
107 * It must be length>=0.
108 *
109 * @return TRUE if the string was found
110 */
111U_CFUNC UBool U_EXPORT2
112ucase_addStringCaseClosure(const UCaseProps *csp, const UChar *s, int32_t length, const USetAdder *sa);
113
114/** @return UCASE_NONE, UCASE_LOWER, UCASE_UPPER, UCASE_TITLE */
115U_CAPI int32_t U_EXPORT2
116ucase_getType(const UCaseProps *csp, UChar32 c);
117
118/** @return same as ucase_getType(), or <0 if c is case-ignorable */
119U_CAPI int32_t U_EXPORT2
120ucase_getTypeOrIgnorable(const UCaseProps *csp, UChar32 c);
121
122U_CAPI UBool U_EXPORT2
123ucase_isSoftDotted(const UCaseProps *csp, UChar32 c);
124
125U_CAPI UBool U_EXPORT2
126ucase_isCaseSensitive(const UCaseProps *csp, UChar32 c);
127
128/* string case mapping functions */
129
130/**
131 * Iterator function for string case mappings, which need to look at the
132 * context (surrounding text) of a given character for conditional mappings.
133 *
134 * The iterator only needs to go backward or forward away from the
135 * character in question. It does not use any indexes on this interface.
136 * It does not support random access or an arbitrary change of
137 * iteration direction.
138 *
139 * The code point being case-mapped itself is never returned by
140 * this iterator.
141 *
142 * @param context A pointer to the iterator's working data.
143 * @param dir If <0 then start iterating backward from the character;
144 *            if >0 then start iterating forward from the character;
145 *            if 0 then continue iterating in the current direction.
146 * @return Next code point, or <0 when the iteration is done.
147 */
148typedef UChar32 U_CALLCONV
149UCaseContextIterator(void *context, int8_t dir);
150
151/**
152 * Sample struct which may be used by some implementations of
153 * UCaseContextIterator.
154 */
155struct UCaseContext {
156    void *p;
157    int32_t start, index, limit;
158    int32_t cpStart, cpLimit;
159    int8_t dir;
160    int8_t b1, b2, b3;
161};
162typedef struct UCaseContext UCaseContext;
163
164enum {
165    /**
166     * For string case mappings, a single character (a code point) is mapped
167     * either to itself (in which case in-place mapping functions do nothing),
168     * or to another single code point, or to a string.
169     * Aside from the string contents, these are indicated with a single int32_t
170     * value as follows:
171     *
172     * Mapping to self: Negative values (~self instead of -self to support U+0000)
173     *
174     * Mapping to another code point: Positive values >UCASE_MAX_STRING_LENGTH
175     *
176     * Mapping to a string: The string length (0..UCASE_MAX_STRING_LENGTH) is
177     * returned. Note that the string result may indeed have zero length.
178     */
179    UCASE_MAX_STRING_LENGTH=0x1f
180};
181
182/**
183 * Get the full lowercase mapping for c.
184 *
185 * @param csp Case mapping properties.
186 * @param c Character to be mapped.
187 * @param iter Character iterator, used for context-sensitive mappings.
188 *             See UCaseContextIterator for details.
189 *             If iter==NULL then a context-independent result is returned.
190 * @param context Pointer to be passed into iter.
191 * @param pString If the mapping result is a string, then the pointer is
192 *                written to *pString.
193 * @param locale Locale ID for locale-dependent mappings.
194 * @param locCache Initialize to 0; may be used to cache the result of parsing
195 *                 the locale ID for subsequent calls.
196 *                 Can be NULL.
197 * @return Output code point or string length, see UCASE_MAX_STRING_LENGTH.
198 *
199 * @see UCaseContextIterator
200 * @see UCASE_MAX_STRING_LENGTH
201 * @internal
202 */
203U_CAPI int32_t U_EXPORT2
204ucase_toFullLower(const UCaseProps *csp, UChar32 c,
205                  UCaseContextIterator *iter, void *context,
206                  const UChar **pString,
207                  const char *locale, int32_t *locCache);
208
209U_CAPI int32_t U_EXPORT2
210ucase_toFullUpper(const UCaseProps *csp, UChar32 c,
211                  UCaseContextIterator *iter, void *context,
212                  const UChar **pString,
213                  const char *locale, int32_t *locCache);
214
215U_CAPI int32_t U_EXPORT2
216ucase_toFullTitle(const UCaseProps *csp, UChar32 c,
217                  UCaseContextIterator *iter, void *context,
218                  const UChar **pString,
219                  const char *locale, int32_t *locCache);
220
221U_CAPI int32_t U_EXPORT2
222ucase_toFullFolding(const UCaseProps *csp, UChar32 c,
223                    const UChar **pString,
224                    uint32_t options);
225
226U_CFUNC int32_t U_EXPORT2
227ucase_hasBinaryProperty(UChar32 c, UProperty which);
228
229
230U_CDECL_BEGIN
231
232/**
233 * @internal
234 */
235typedef int32_t U_CALLCONV
236UCaseMapFull(const UCaseProps *csp, UChar32 c,
237             UCaseContextIterator *iter, void *context,
238             const UChar **pString,
239             const char *locale, int32_t *locCache);
240
241U_CDECL_END
242
243/* file definitions --------------------------------------------------------- */
244
245#define UCASE_DATA_NAME "ucase"
246#define UCASE_DATA_TYPE "icu"
247
248/* format "cAsE" */
249#define UCASE_FMT_0 0x63
250#define UCASE_FMT_1 0x41
251#define UCASE_FMT_2 0x53
252#define UCASE_FMT_3 0x45
253
254/* indexes into indexes[] */
255enum {
256    UCASE_IX_INDEX_TOP,
257    UCASE_IX_LENGTH,
258    UCASE_IX_TRIE_SIZE,
259    UCASE_IX_EXC_LENGTH,
260    UCASE_IX_UNFOLD_LENGTH,
261
262    UCASE_IX_MAX_FULL_LENGTH=15,
263    UCASE_IX_TOP=16
264};
265
266/* definitions for 16-bit case properties word ------------------------------ */
267
268/* 2-bit constants for types of cased characters */
269#define UCASE_TYPE_MASK     3
270enum {
271    UCASE_NONE,
272    UCASE_LOWER,
273    UCASE_UPPER,
274    UCASE_TITLE
275};
276
277#define UCASE_GET_TYPE(props) ((props)&UCASE_TYPE_MASK)
278
279#define UCASE_SENSITIVE     4
280#define UCASE_EXCEPTION     8
281
282#define UCASE_DOT_MASK      0x30
283enum {
284    UCASE_NO_DOT=0,         /* normal characters with cc=0 */
285    UCASE_SOFT_DOTTED=0x10, /* soft-dotted characters with cc=0 */
286    UCASE_ABOVE=0x20,       /* "above" accents with cc=230 */
287    UCASE_OTHER_ACCENT=0x30 /* other accent character (0<cc!=230) */
288};
289
290/* no exception: bits 15..6 are a 10-bit signed case mapping delta */
291#define UCASE_DELTA_SHIFT   6
292#define UCASE_DELTA_MASK    0xffc0
293#define UCASE_MAX_DELTA     0x1ff
294#define UCASE_MIN_DELTA     (-UCASE_MAX_DELTA-1)
295
296#define UCASE_GET_DELTA(props) ((int16_t)(props)>>UCASE_DELTA_SHIFT)
297
298/* case-ignorable uses one of the delta bits, see gencase/store.c */
299#define UCASE_CASE_IGNORABLE 0x40
300
301/* exception: bits 15..4 are an unsigned 12-bit index into the exceptions array */
302#define UCASE_EXC_SHIFT     4
303#define UCASE_EXC_MASK      0xfff0
304#define UCASE_MAX_EXCEPTIONS 0x1000
305
306/* definitions for 16-bit main exceptions word ------------------------------ */
307
308/* first 8 bits indicate values in optional slots */
309enum {
310    UCASE_EXC_LOWER,
311    UCASE_EXC_FOLD,
312    UCASE_EXC_UPPER,
313    UCASE_EXC_TITLE,
314    UCASE_EXC_4,            /* reserved */
315    UCASE_EXC_5,            /* reserved */
316    UCASE_EXC_CLOSURE,
317    UCASE_EXC_FULL_MAPPINGS,
318    UCASE_EXC_ALL_SLOTS     /* one past the last slot */
319};
320
321/* each slot is 2 uint16_t instead of 1 */
322#define UCASE_EXC_DOUBLE_SLOTS      0x100
323
324/* reserved: exception bits 10..9 */
325
326#define UCASE_EXC_CASE_IGNORABLE        0x800
327
328/* UCASE_EXC_DOT_MASK=UCASE_DOT_MASK<<UCASE_EXC_DOT_SHIFT */
329#define UCASE_EXC_DOT_SHIFT     8
330
331/* normally stored in the main word, but pushed out for larger exception indexes */
332#define UCASE_EXC_DOT_MASK      0x3000
333enum {
334    UCASE_EXC_NO_DOT=0,
335    UCASE_EXC_SOFT_DOTTED=0x1000,
336    UCASE_EXC_ABOVE=0x2000,         /* "above" accents with cc=230 */
337    UCASE_EXC_OTHER_ACCENT=0x3000   /* other character (0<cc!=230) */
338};
339
340/* complex/conditional mappings */
341#define UCASE_EXC_CONDITIONAL_SPECIAL   0x4000
342#define UCASE_EXC_CONDITIONAL_FOLD      0x8000
343
344/* definitions for lengths word for full case mappings */
345#define UCASE_FULL_LOWER    0xf
346#define UCASE_FULL_FOLDING  0xf0
347#define UCASE_FULL_UPPER    0xf00
348#define UCASE_FULL_TITLE    0xf000
349
350/* maximum lengths */
351#define UCASE_FULL_MAPPINGS_MAX_LENGTH (4*0xf)
352#define UCASE_CLOSURE_MAX_LENGTH 0xf
353
354/* constants for reverse case folding ("unfold") data */
355enum {
356    UCASE_UNFOLD_ROWS,
357    UCASE_UNFOLD_ROW_WIDTH,
358    UCASE_UNFOLD_STRING_WIDTH
359};
360
361U_CDECL_END
362
363#endif
364