1 // Copyright (C) 2016 and later: Unicode, Inc. and others.
2 // License & terms of use: http://www.unicode.org/copyright.html
3 /*
4 *******************************************************************************
5 *
6 * Copyright (C) 2004-2014, International Business Machines
7 * Corporation and others. All Rights Reserved.
8 *
9 *******************************************************************************
10 * file name: ubidi_props.c
11 * encoding: US-ASCII
12 * tab size: 8 (not used)
13 * indentation:4
14 *
15 * created on: 2004dec30
16 * created by: Markus W. Scherer
17 *
18 * Low-level Unicode bidi/shaping properties access.
19 */
20
21 #include "unicode/utypes.h"
22 #include "unicode/uset.h"
23 #include "unicode/udata.h" /* UDataInfo */
24 #include "ucmndata.h" /* DataHeader */
25 #include "udatamem.h"
26 #include "uassert.h"
27 #include "cmemory.h"
28 #include "utrie2.h"
29 #include "ubidi_props.h"
30 #include "ucln_cmn.h"
31
32 struct UBiDiProps {
33 UDataMemory *mem;
34 const int32_t *indexes;
35 const uint32_t *mirrors;
36 const uint8_t *jgArray;
37 const uint8_t *jgArray2;
38
39 UTrie2 trie;
40 uint8_t formatVersion[4];
41 };
42
43 /* ubidi_props_data.h is machine-generated by genbidi --csource */
44 #define INCLUDED_FROM_UBIDI_PROPS_C
45 #include "ubidi_props_data.h"
46
47 /* UBiDiProps singleton ----------------------------------------------------- */
48
49 U_CFUNC const UBiDiProps *
ubidi_getSingleton()50 ubidi_getSingleton() {
51 return &ubidi_props_singleton;
52 }
53
54 /* set of property starts for UnicodeSet ------------------------------------ */
55
56 static UBool U_CALLCONV
_enumPropertyStartsRange(const void * context,UChar32 start,UChar32 end,uint32_t value)57 _enumPropertyStartsRange(const void *context, UChar32 start, UChar32 end, uint32_t value) {
58 /* add the start code point to the USet */
59 const USetAdder *sa=(const USetAdder *)context;
60 sa->add(sa->set, start);
61 return TRUE;
62 }
63
64 U_CFUNC void
ubidi_addPropertyStarts(const UBiDiProps * bdp,const USetAdder * sa,UErrorCode * pErrorCode)65 ubidi_addPropertyStarts(const UBiDiProps *bdp, const USetAdder *sa, UErrorCode *pErrorCode) {
66 int32_t i, length;
67 UChar32 c, start, limit;
68
69 const uint8_t *jgArray;
70 uint8_t prev, jg;
71
72 if(U_FAILURE(*pErrorCode)) {
73 return;
74 }
75
76 /* add the start code point of each same-value range of the trie */
77 utrie2_enum(&bdp->trie, NULL, _enumPropertyStartsRange, sa);
78
79 /* add the code points from the bidi mirroring table */
80 length=bdp->indexes[UBIDI_IX_MIRROR_LENGTH];
81 for(i=0; i<length; ++i) {
82 c=UBIDI_GET_MIRROR_CODE_POINT(bdp->mirrors[i]);
83 sa->addRange(sa->set, c, c+1);
84 }
85
86 /* add the code points from the Joining_Group array where the value changes */
87 start=bdp->indexes[UBIDI_IX_JG_START];
88 limit=bdp->indexes[UBIDI_IX_JG_LIMIT];
89 jgArray=bdp->jgArray;
90 for(;;) {
91 prev=0;
92 while(start<limit) {
93 jg=*jgArray++;
94 if(jg!=prev) {
95 sa->add(sa->set, start);
96 prev=jg;
97 }
98 ++start;
99 }
100 if(prev!=0) {
101 /* add the limit code point if the last value was not 0 (it is now start==limit) */
102 sa->add(sa->set, limit);
103 }
104 if(limit==bdp->indexes[UBIDI_IX_JG_LIMIT]) {
105 /* switch to the second Joining_Group range */
106 start=bdp->indexes[UBIDI_IX_JG_START2];
107 limit=bdp->indexes[UBIDI_IX_JG_LIMIT2];
108 jgArray=bdp->jgArray2;
109 } else {
110 break;
111 }
112 }
113
114 /* add code points with hardcoded properties, plus the ones following them */
115
116 /* (none right now) */
117 }
118
119 /* property access functions ------------------------------------------------ */
120
121 U_CFUNC int32_t
ubidi_getMaxValue(const UBiDiProps * bdp,UProperty which)122 ubidi_getMaxValue(const UBiDiProps *bdp, UProperty which) {
123 int32_t max;
124
125 if(bdp==NULL) {
126 return -1;
127 }
128
129 max=bdp->indexes[UBIDI_MAX_VALUES_INDEX];
130 switch(which) {
131 case UCHAR_BIDI_CLASS:
132 return (max&UBIDI_CLASS_MASK);
133 case UCHAR_JOINING_GROUP:
134 return (max&UBIDI_MAX_JG_MASK)>>UBIDI_MAX_JG_SHIFT;
135 case UCHAR_JOINING_TYPE:
136 return (max&UBIDI_JT_MASK)>>UBIDI_JT_SHIFT;
137 case UCHAR_BIDI_PAIRED_BRACKET_TYPE:
138 return (max&UBIDI_BPT_MASK)>>UBIDI_BPT_SHIFT;
139 default:
140 return -1; /* undefined */
141 }
142 }
143
144 U_CAPI UCharDirection
ubidi_getClass(const UBiDiProps * bdp,UChar32 c)145 ubidi_getClass(const UBiDiProps *bdp, UChar32 c) {
146 uint16_t props=UTRIE2_GET16(&bdp->trie, c);
147 return (UCharDirection)UBIDI_GET_CLASS(props);
148 }
149
150 U_CFUNC UBool
ubidi_isMirrored(const UBiDiProps * bdp,UChar32 c)151 ubidi_isMirrored(const UBiDiProps *bdp, UChar32 c) {
152 uint16_t props=UTRIE2_GET16(&bdp->trie, c);
153 return (UBool)UBIDI_GET_FLAG(props, UBIDI_IS_MIRRORED_SHIFT);
154 }
155
156 static UChar32
getMirror(const UBiDiProps * bdp,UChar32 c,uint16_t props)157 getMirror(const UBiDiProps *bdp, UChar32 c, uint16_t props) {
158 int32_t delta=UBIDI_GET_MIRROR_DELTA(props);
159 if(delta!=UBIDI_ESC_MIRROR_DELTA) {
160 return c+delta;
161 } else {
162 /* look for mirror code point in the mirrors[] table */
163 const uint32_t *mirrors;
164 uint32_t m;
165 int32_t i, length;
166 UChar32 c2;
167
168 mirrors=bdp->mirrors;
169 length=bdp->indexes[UBIDI_IX_MIRROR_LENGTH];
170
171 /* linear search */
172 for(i=0; i<length; ++i) {
173 m=mirrors[i];
174 c2=UBIDI_GET_MIRROR_CODE_POINT(m);
175 if(c==c2) {
176 /* found c, return its mirror code point using the index in m */
177 return UBIDI_GET_MIRROR_CODE_POINT(mirrors[UBIDI_GET_MIRROR_INDEX(m)]);
178 } else if(c<c2) {
179 break;
180 }
181 }
182
183 /* c not found, return it itself */
184 return c;
185 }
186 }
187
188 U_CFUNC UChar32
ubidi_getMirror(const UBiDiProps * bdp,UChar32 c)189 ubidi_getMirror(const UBiDiProps *bdp, UChar32 c) {
190 uint16_t props=UTRIE2_GET16(&bdp->trie, c);
191 return getMirror(bdp, c, props);
192 }
193
194 U_CFUNC UBool
ubidi_isBidiControl(const UBiDiProps * bdp,UChar32 c)195 ubidi_isBidiControl(const UBiDiProps *bdp, UChar32 c) {
196 uint16_t props=UTRIE2_GET16(&bdp->trie, c);
197 return (UBool)UBIDI_GET_FLAG(props, UBIDI_BIDI_CONTROL_SHIFT);
198 }
199
200 U_CFUNC UBool
ubidi_isJoinControl(const UBiDiProps * bdp,UChar32 c)201 ubidi_isJoinControl(const UBiDiProps *bdp, UChar32 c) {
202 uint16_t props=UTRIE2_GET16(&bdp->trie, c);
203 return (UBool)UBIDI_GET_FLAG(props, UBIDI_JOIN_CONTROL_SHIFT);
204 }
205
206 U_CFUNC UJoiningType
ubidi_getJoiningType(const UBiDiProps * bdp,UChar32 c)207 ubidi_getJoiningType(const UBiDiProps *bdp, UChar32 c) {
208 uint16_t props=UTRIE2_GET16(&bdp->trie, c);
209 return (UJoiningType)((props&UBIDI_JT_MASK)>>UBIDI_JT_SHIFT);
210 }
211
212 U_CFUNC UJoiningGroup
ubidi_getJoiningGroup(const UBiDiProps * bdp,UChar32 c)213 ubidi_getJoiningGroup(const UBiDiProps *bdp, UChar32 c) {
214 UChar32 start, limit;
215
216 start=bdp->indexes[UBIDI_IX_JG_START];
217 limit=bdp->indexes[UBIDI_IX_JG_LIMIT];
218 if(start<=c && c<limit) {
219 return (UJoiningGroup)bdp->jgArray[c-start];
220 }
221 start=bdp->indexes[UBIDI_IX_JG_START2];
222 limit=bdp->indexes[UBIDI_IX_JG_LIMIT2];
223 if(start<=c && c<limit) {
224 return (UJoiningGroup)bdp->jgArray2[c-start];
225 }
226 return U_JG_NO_JOINING_GROUP;
227 }
228
229 U_CFUNC UBidiPairedBracketType
ubidi_getPairedBracketType(const UBiDiProps * bdp,UChar32 c)230 ubidi_getPairedBracketType(const UBiDiProps *bdp, UChar32 c) {
231 uint16_t props=UTRIE2_GET16(&bdp->trie, c);
232 return (UBidiPairedBracketType)((props&UBIDI_BPT_MASK)>>UBIDI_BPT_SHIFT);
233 }
234
235 U_CFUNC UChar32
ubidi_getPairedBracket(const UBiDiProps * bdp,UChar32 c)236 ubidi_getPairedBracket(const UBiDiProps *bdp, UChar32 c) {
237 uint16_t props=UTRIE2_GET16(&bdp->trie, c);
238 if((props&UBIDI_BPT_MASK)==0) {
239 return c;
240 } else {
241 return getMirror(bdp, c, props);
242 }
243 }
244
245 /* public API (see uchar.h) ------------------------------------------------- */
246
247 U_CFUNC UCharDirection
u_charDirection(UChar32 c)248 u_charDirection(UChar32 c) {
249 return ubidi_getClass(&ubidi_props_singleton, c);
250 }
251
252 U_CFUNC UBool
u_isMirrored(UChar32 c)253 u_isMirrored(UChar32 c) {
254 return ubidi_isMirrored(&ubidi_props_singleton, c);
255 }
256
257 U_CFUNC UChar32
u_charMirror(UChar32 c)258 u_charMirror(UChar32 c) {
259 return ubidi_getMirror(&ubidi_props_singleton, c);
260 }
261
262 U_STABLE UChar32 U_EXPORT2
u_getBidiPairedBracket(UChar32 c)263 u_getBidiPairedBracket(UChar32 c) {
264 return ubidi_getPairedBracket(&ubidi_props_singleton, c);
265 }
266