ICU 66.1 66.1
ucptrie.h
Go to the documentation of this file.
1// © 2017 and later: Unicode, Inc. and others.
2// License & terms of use: http://www.unicode.org/copyright.html
3
4// ucptrie.h (modified from utrie2.h)
5// created: 2017dec29 Markus W. Scherer
6
7#ifndef __UCPTRIE_H__
8#define __UCPTRIE_H__
9
10#include "unicode/utypes.h"
11
13#include "unicode/ucpmap.h"
14#include "unicode/utf8.h"
15
17
27#ifndef U_IN_DOXYGEN
29typedef union UCPTrieData {
31 const void *ptr0;
33 const uint16_t *ptr16;
35 const uint32_t *ptr32;
37 const uint8_t *ptr8;
38} UCPTrieData;
39#endif
40
58struct UCPTrie {
59#ifndef U_IN_DOXYGEN
61 const uint16_t *index;
63 UCPTrieData data;
64
66 int32_t indexLength;
68 int32_t dataLength;
70 UChar32 highStart;
72 uint16_t shifted12HighStart;
73
75 int8_t type; // UCPTrieType
77 int8_t valueWidth; // UCPTrieValueWidth
78
80 uint32_t reserved32;
82 uint16_t reserved16;
83
89 uint16_t index3NullOffset;
95 int32_t dataNullOffset;
97 uint32_t nullValue;
98
99#ifdef UCPTRIE_DEBUG
101 const char *name;
102#endif
103#endif
104};
105#ifndef U_IN_DOXYGEN
106typedef struct UCPTrie UCPTrie;
107#endif
108
136#ifndef U_IN_DOXYGEN
137typedef enum UCPTrieType UCPTrieType;
138#endif
139
173#ifndef U_IN_DOXYGEN
175#endif
176
203U_CAPI UCPTrie * U_EXPORT2
205 const void *data, int32_t length, int32_t *pActualLength,
206 UErrorCode *pErrorCode);
207
214U_CAPI void U_EXPORT2
216
226U_CAPI UCPTrieType U_EXPORT2
228
240
255U_CAPI uint32_t U_EXPORT2
257
294U_CAPI UChar32 U_EXPORT2
296 UCPMapRangeOption option, uint32_t surrogateValue,
297 UCPMapValueFilter *filter, const void *context, uint32_t *pValue);
298
314U_CAPI int32_t U_EXPORT2
315ucptrie_toBinary(const UCPTrie *trie, void *data, int32_t capacity, UErrorCode *pErrorCode);
316
325#define UCPTRIE_16(trie, i) ((trie)->data.ptr16[i])
326
335#define UCPTRIE_32(trie, i) ((trie)->data.ptr32[i])
336
345#define UCPTRIE_8(trie, i) ((trie)->data.ptr8[i])
346
357#define UCPTRIE_FAST_GET(trie, dataAccess, c) dataAccess(trie, _UCPTRIE_CP_INDEX(trie, 0xffff, c))
358
369#define UCPTRIE_SMALL_GET(trie, dataAccess, c) \
370 dataAccess(trie, _UCPTRIE_CP_INDEX(trie, UCPTRIE_SMALL_MAX, c))
371
385#define UCPTRIE_FAST_U16_NEXT(trie, dataAccess, src, limit, c, result) UPRV_BLOCK_MACRO_BEGIN { \
386 (c) = *(src)++; \
387 int32_t __index; \
388 if (!U16_IS_SURROGATE(c)) { \
389 __index = _UCPTRIE_FAST_INDEX(trie, c); \
390 } else { \
391 uint16_t __c2; \
392 if (U16_IS_SURROGATE_LEAD(c) && (src) != (limit) && U16_IS_TRAIL(__c2 = *(src))) { \
393 ++(src); \
394 (c) = U16_GET_SUPPLEMENTARY((c), __c2); \
395 __index = _UCPTRIE_SMALL_INDEX(trie, c); \
396 } else { \
397 __index = (trie)->dataLength - UCPTRIE_ERROR_VALUE_NEG_DATA_OFFSET; \
398 } \
399 } \
400 (result) = dataAccess(trie, __index); \
401} UPRV_BLOCK_MACRO_END
402
416#define UCPTRIE_FAST_U16_PREV(trie, dataAccess, start, src, c, result) UPRV_BLOCK_MACRO_BEGIN { \
417 (c) = *--(src); \
418 int32_t __index; \
419 if (!U16_IS_SURROGATE(c)) { \
420 __index = _UCPTRIE_FAST_INDEX(trie, c); \
421 } else { \
422 uint16_t __c2; \
423 if (U16_IS_SURROGATE_TRAIL(c) && (src) != (start) && U16_IS_LEAD(__c2 = *((src) - 1))) { \
424 --(src); \
425 (c) = U16_GET_SUPPLEMENTARY(__c2, (c)); \
426 __index = _UCPTRIE_SMALL_INDEX(trie, c); \
427 } else { \
428 __index = (trie)->dataLength - UCPTRIE_ERROR_VALUE_NEG_DATA_OFFSET; \
429 } \
430 } \
431 (result) = dataAccess(trie, __index); \
432} UPRV_BLOCK_MACRO_END
433
450#define UCPTRIE_FAST_U8_NEXT(trie, dataAccess, src, limit, result) UPRV_BLOCK_MACRO_BEGIN { \
451 int32_t __lead = (uint8_t)*(src)++; \
452 if (!U8_IS_SINGLE(__lead)) { \
453 uint8_t __t1, __t2, __t3; \
454 if ((src) != (limit) && \
455 (__lead >= 0xe0 ? \
456 __lead < 0xf0 ? /* U+0800..U+FFFF except surrogates */ \
457 U8_LEAD3_T1_BITS[__lead &= 0xf] & (1 << ((__t1 = *(src)) >> 5)) && \
458 ++(src) != (limit) && (__t2 = *(src) - 0x80) <= 0x3f && \
459 (__lead = ((int32_t)(trie)->index[(__lead << 6) + (__t1 & 0x3f)]) + __t2, 1) \
460 : /* U+10000..U+10FFFF */ \
461 (__lead -= 0xf0) <= 4 && \
462 U8_LEAD4_T1_BITS[(__t1 = *(src)) >> 4] & (1 << __lead) && \
463 (__lead = (__lead << 6) | (__t1 & 0x3f), ++(src) != (limit)) && \
464 (__t2 = *(src) - 0x80) <= 0x3f && \
465 ++(src) != (limit) && (__t3 = *(src) - 0x80) <= 0x3f && \
466 (__lead = __lead >= (trie)->shifted12HighStart ? \
467 (trie)->dataLength - UCPTRIE_HIGH_VALUE_NEG_DATA_OFFSET : \
468 ucptrie_internalSmallU8Index((trie), __lead, __t2, __t3), 1) \
469 : /* U+0080..U+07FF */ \
470 __lead >= 0xc2 && (__t1 = *(src) - 0x80) <= 0x3f && \
471 (__lead = (int32_t)(trie)->index[__lead & 0x1f] + __t1, 1))) { \
472 ++(src); \
473 } else { \
474 __lead = (trie)->dataLength - UCPTRIE_ERROR_VALUE_NEG_DATA_OFFSET; /* ill-formed*/ \
475 } \
476 } \
477 (result) = dataAccess(trie, __lead); \
478} UPRV_BLOCK_MACRO_END
479
496#define UCPTRIE_FAST_U8_PREV(trie, dataAccess, start, src, result) UPRV_BLOCK_MACRO_BEGIN { \
497 int32_t __index = (uint8_t)*--(src); \
498 if (!U8_IS_SINGLE(__index)) { \
499 __index = ucptrie_internalU8PrevIndex((trie), __index, (const uint8_t *)(start), \
500 (const uint8_t *)(src)); \
501 (src) -= __index & 7; \
502 __index >>= 3; \
503 } \
504 (result) = dataAccess(trie, __index); \
505} UPRV_BLOCK_MACRO_END
506
516#define UCPTRIE_ASCII_GET(trie, dataAccess, c) dataAccess(trie, c)
517
529#define UCPTRIE_FAST_BMP_GET(trie, dataAccess, c) dataAccess(trie, _UCPTRIE_FAST_INDEX(trie, c))
530
541#define UCPTRIE_FAST_SUPP_GET(trie, dataAccess, c) dataAccess(trie, _UCPTRIE_SMALL_INDEX(trie, c))
542
543/* Internal definitions ----------------------------------------------------- */
544
545#ifndef U_IN_DOXYGEN
546
552enum {
554 UCPTRIE_FAST_SHIFT = 6,
555
557 UCPTRIE_FAST_DATA_BLOCK_LENGTH = 1 << UCPTRIE_FAST_SHIFT,
558
560 UCPTRIE_FAST_DATA_MASK = UCPTRIE_FAST_DATA_BLOCK_LENGTH - 1,
561
563 UCPTRIE_SMALL_MAX = 0xfff,
564
570 UCPTRIE_ERROR_VALUE_NEG_DATA_OFFSET = 1,
576 UCPTRIE_HIGH_VALUE_NEG_DATA_OFFSET = 2
577};
578
579/* Internal functions and macros -------------------------------------------- */
580// Do not conditionalize with #ifndef U_HIDE_INTERNAL_API, needed for public API
581
583U_INTERNAL int32_t U_EXPORT2
584ucptrie_internalSmallIndex(const UCPTrie *trie, UChar32 c);
585
587U_INTERNAL int32_t U_EXPORT2
588ucptrie_internalSmallU8Index(const UCPTrie *trie, int32_t lt1, uint8_t t2, uint8_t t3);
589
595U_INTERNAL int32_t U_EXPORT2
596ucptrie_internalU8PrevIndex(const UCPTrie *trie, UChar32 c,
597 const uint8_t *start, const uint8_t *src);
598
600#define _UCPTRIE_FAST_INDEX(trie, c) \
601 ((int32_t)(trie)->index[(c) >> UCPTRIE_FAST_SHIFT] + ((c) & UCPTRIE_FAST_DATA_MASK))
602
604#define _UCPTRIE_SMALL_INDEX(trie, c) \
605 ((c) >= (trie)->highStart ? \
606 (trie)->dataLength - UCPTRIE_HIGH_VALUE_NEG_DATA_OFFSET : \
607 ucptrie_internalSmallIndex(trie, c))
608
614#define _UCPTRIE_CP_INDEX(trie, fastMax, c) \
615 ((uint32_t)(c) <= (uint32_t)(fastMax) ? \
616 _UCPTRIE_FAST_INDEX(trie, c) : \
617 (uint32_t)(c) <= 0x10ffff ? \
618 _UCPTRIE_SMALL_INDEX(trie, c) : \
619 (trie)->dataLength - UCPTRIE_ERROR_VALUE_NEG_DATA_OFFSET)
620
622
623#endif // U_IN_DOXYGEN
624
625#if U_SHOW_CPLUSPLUS_API
626
627U_NAMESPACE_BEGIN
628
639
640U_NAMESPACE_END
641
642#endif // U_SHOW_CPLUSPLUS_API
643
644#endif
"Smart pointer" class, closes a UCPTrie via ucptrie_close().
C++ API: "Smart pointers" for use with and in ICU4C C++ code.
#define U_DEFINE_LOCAL_OPEN_POINTER(LocalPointerClassName, Type, closeFunction)
"Smart pointer" definition macro, deletes objects via the closeFunction.
Definition: localpointer.h:562
Immutable Unicode code point trie structure.
Definition: ucptrie.h:58
This file defines an abstract map from Unicode code points to integer values.
uint32_t UCPMapValueFilter(const void *context, uint32_t value)
Callback function type: Modifies a map value.
Definition: ucpmap.h:114
UCPMapRangeOption
Selectors for how ucpmap_getRange() etc.
Definition: ucpmap.h:42
UCPTrieType
Selectors for the type of a UCPTrie.
Definition: ucptrie.h:118
@ UCPTRIE_TYPE_FAST
Fast/simple/larger BMP data structure.
Definition: ucptrie.h:129
@ UCPTRIE_TYPE_SMALL
Small/slower BMP data structure.
Definition: ucptrie.h:134
@ UCPTRIE_TYPE_ANY
For ucptrie_openFromBinary() to accept any type.
Definition: ucptrie.h:124
U_CAPI void ucptrie_close(UCPTrie *trie)
Closes a trie and releases associated memory.
U_CAPI UCPTrieValueWidth ucptrie_getValueWidth(const UCPTrie *trie)
Returns the number of bits in a trie data value.
U_CAPI UChar32 ucptrie_getRange(const UCPTrie *trie, UChar32 start, UCPMapRangeOption option, uint32_t surrogateValue, UCPMapValueFilter *filter, const void *context, uint32_t *pValue)
Returns the last code point such that all those from start to there have the same value.
U_CAPI UCPTrie * ucptrie_openFromBinary(UCPTrieType type, UCPTrieValueWidth valueWidth, const void *data, int32_t length, int32_t *pActualLength, UErrorCode *pErrorCode)
Opens a trie from its binary form, stored in 32-bit-aligned memory.
U_CAPI int32_t ucptrie_toBinary(const UCPTrie *trie, void *data, int32_t capacity, UErrorCode *pErrorCode)
Writes a memory-mappable form of the trie into 32-bit aligned memory.
U_CAPI UCPTrieType ucptrie_getType(const UCPTrie *trie)
Returns the trie type.
UCPTrieValueWidth
Selectors for the number of bits in a UCPTrie data value.
Definition: ucptrie.h:148
@ UCPTRIE_VALUE_BITS_ANY
For ucptrie_openFromBinary() to accept any data value width.
Definition: ucptrie.h:154
@ UCPTRIE_VALUE_BITS_8
The trie stores 8 bits per data value.
Definition: ucptrie.h:171
@ UCPTRIE_VALUE_BITS_16
The trie stores 16 bits per data value.
Definition: ucptrie.h:160
@ UCPTRIE_VALUE_BITS_32
The trie stores 32 bits per data value.
Definition: ucptrie.h:165
U_CAPI uint32_t ucptrie_get(const UCPTrie *trie, UChar32 c)
Returns the value for a code point as stored in the trie, with range checking.
int32_t UChar32
Define UChar32 as a type for single Unicode code points.
Definition: umachine.h:425
#define U_INTERNAL
This is used to declare a function as an internal ICU C API
Definition: umachine.h:119
#define U_CDECL_END
This is used to end a declaration of a library private ICU C API.
Definition: umachine.h:85
#define U_CAPI
This is used to declare a function as a public ICU C API.
Definition: umachine.h:109
#define U_CDECL_BEGIN
This is used to begin a declaration of a library private ICU C API.
Definition: umachine.h:84
C API: 8-bit Unicode handling macros.
Basic definitions for ICU, for both C and C++ APIs.
UErrorCode
Standard ICU4C error code type, a substitute for exceptions.
Definition: utypes.h:415