/src/icu/source/common/ubidi_props.cpp
Line  | Count  | Source (jump to first uncovered line)  | 
1  |  | // © 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:   UTF-8  | 
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  |  | /* set of property starts for UnicodeSet ------------------------------------ */  | 
48  |  |  | 
49  |  | static UBool U_CALLCONV  | 
50  | 0  | _enumPropertyStartsRange(const void *context, UChar32 start, UChar32 end, uint32_t value) { | 
51  | 0  |     (void)end;  | 
52  | 0  |     (void)value;  | 
53  |  |     /* add the start code point to the USet */  | 
54  | 0  |     const USetAdder *sa=(const USetAdder *)context;  | 
55  | 0  |     sa->add(sa->set, start);  | 
56  | 0  |     return TRUE;  | 
57  | 0  | }  | 
58  |  |  | 
59  |  | U_CFUNC void  | 
60  | 0  | ubidi_addPropertyStarts(const USetAdder *sa, UErrorCode *pErrorCode) { | 
61  | 0  |     int32_t i, length;  | 
62  | 0  |     UChar32 c, start, limit;  | 
63  |  | 
  | 
64  | 0  |     const uint8_t *jgArray;  | 
65  | 0  |     uint8_t prev, jg;  | 
66  |  | 
  | 
67  | 0  |     if(U_FAILURE(*pErrorCode)) { | 
68  | 0  |         return;  | 
69  | 0  |     }  | 
70  |  |  | 
71  |  |     /* add the start code point of each same-value range of the trie */  | 
72  | 0  |     utrie2_enum(&ubidi_props_singleton.trie, NULL, _enumPropertyStartsRange, sa);  | 
73  |  |  | 
74  |  |     /* add the code points from the bidi mirroring table */  | 
75  | 0  |     length=ubidi_props_singleton.indexes[UBIDI_IX_MIRROR_LENGTH];  | 
76  | 0  |     for(i=0; i<length; ++i) { | 
77  | 0  |         c=UBIDI_GET_MIRROR_CODE_POINT(ubidi_props_singleton.mirrors[i]);  | 
78  | 0  |         sa->addRange(sa->set, c, c+1);  | 
79  | 0  |     }  | 
80  |  |  | 
81  |  |     /* add the code points from the Joining_Group array where the value changes */  | 
82  | 0  |     start=ubidi_props_singleton.indexes[UBIDI_IX_JG_START];  | 
83  | 0  |     limit=ubidi_props_singleton.indexes[UBIDI_IX_JG_LIMIT];  | 
84  | 0  |     jgArray=ubidi_props_singleton.jgArray;  | 
85  | 0  |     for(;;) { | 
86  | 0  |         prev=0;  | 
87  | 0  |         while(start<limit) { | 
88  | 0  |             jg=*jgArray++;  | 
89  | 0  |             if(jg!=prev) { | 
90  | 0  |                 sa->add(sa->set, start);  | 
91  | 0  |                 prev=jg;  | 
92  | 0  |             }  | 
93  | 0  |             ++start;  | 
94  | 0  |         }  | 
95  | 0  |         if(prev!=0) { | 
96  |  |             /* add the limit code point if the last value was not 0 (it is now start==limit) */  | 
97  | 0  |             sa->add(sa->set, limit);  | 
98  | 0  |         }  | 
99  | 0  |         if(limit==ubidi_props_singleton.indexes[UBIDI_IX_JG_LIMIT]) { | 
100  |  |             /* switch to the second Joining_Group range */  | 
101  | 0  |             start=ubidi_props_singleton.indexes[UBIDI_IX_JG_START2];  | 
102  | 0  |             limit=ubidi_props_singleton.indexes[UBIDI_IX_JG_LIMIT2];  | 
103  | 0  |             jgArray=ubidi_props_singleton.jgArray2;  | 
104  | 0  |         } else { | 
105  | 0  |             break;  | 
106  | 0  |         }  | 
107  | 0  |     }  | 
108  |  |  | 
109  |  |     /* add code points with hardcoded properties, plus the ones following them */  | 
110  |  |  | 
111  |  |     /* (none right now) */  | 
112  | 0  | }  | 
113  |  |  | 
114  |  | /* property access functions ------------------------------------------------ */  | 
115  |  |  | 
116  |  | U_CFUNC int32_t  | 
117  | 0  | ubidi_getMaxValue(UProperty which) { | 
118  | 0  |     int32_t max=ubidi_props_singleton.indexes[UBIDI_MAX_VALUES_INDEX];  | 
119  | 0  |     switch(which) { | 
120  | 0  |     case UCHAR_BIDI_CLASS:  | 
121  | 0  |         return (max&UBIDI_CLASS_MASK);  | 
122  | 0  |     case UCHAR_JOINING_GROUP:  | 
123  | 0  |         return (max&UBIDI_MAX_JG_MASK)>>UBIDI_MAX_JG_SHIFT;  | 
124  | 0  |     case UCHAR_JOINING_TYPE:  | 
125  | 0  |         return (max&UBIDI_JT_MASK)>>UBIDI_JT_SHIFT;  | 
126  | 0  |     case UCHAR_BIDI_PAIRED_BRACKET_TYPE:  | 
127  | 0  |         return (max&UBIDI_BPT_MASK)>>UBIDI_BPT_SHIFT;  | 
128  | 0  |     default:  | 
129  | 0  |         return -1; /* undefined */  | 
130  | 0  |     }  | 
131  | 0  | }  | 
132  |  |  | 
133  |  | U_CAPI UCharDirection  | 
134  | 0  | ubidi_getClass(UChar32 c) { | 
135  | 0  |     uint16_t props=UTRIE2_GET16(&ubidi_props_singleton.trie, c);  | 
136  | 0  |     return (UCharDirection)UBIDI_GET_CLASS(props);  | 
137  | 0  | }  | 
138  |  |  | 
139  |  | U_CFUNC UBool  | 
140  | 0  | ubidi_isMirrored(UChar32 c) { | 
141  | 0  |     uint16_t props=UTRIE2_GET16(&ubidi_props_singleton.trie, c);  | 
142  | 0  |     return (UBool)UBIDI_GET_FLAG(props, UBIDI_IS_MIRRORED_SHIFT);  | 
143  | 0  | }  | 
144  |  |  | 
145  |  | static UChar32  | 
146  | 0  | getMirror(UChar32 c, uint16_t props) { | 
147  | 0  |     int32_t delta=UBIDI_GET_MIRROR_DELTA(props);  | 
148  | 0  |     if(delta!=UBIDI_ESC_MIRROR_DELTA) { | 
149  | 0  |         return c+delta;  | 
150  | 0  |     } else { | 
151  |  |         /* look for mirror code point in the mirrors[] table */  | 
152  | 0  |         const uint32_t *mirrors;  | 
153  | 0  |         uint32_t m;  | 
154  | 0  |         int32_t i, length;  | 
155  | 0  |         UChar32 c2;  | 
156  |  | 
  | 
157  | 0  |         mirrors=ubidi_props_singleton.mirrors;  | 
158  | 0  |         length=ubidi_props_singleton.indexes[UBIDI_IX_MIRROR_LENGTH];  | 
159  |  |  | 
160  |  |         /* linear search */  | 
161  | 0  |         for(i=0; i<length; ++i) { | 
162  | 0  |             m=mirrors[i];  | 
163  | 0  |             c2=UBIDI_GET_MIRROR_CODE_POINT(m);  | 
164  | 0  |             if(c==c2) { | 
165  |  |                 /* found c, return its mirror code point using the index in m */  | 
166  | 0  |                 return UBIDI_GET_MIRROR_CODE_POINT(mirrors[UBIDI_GET_MIRROR_INDEX(m)]);  | 
167  | 0  |             } else if(c<c2) { | 
168  | 0  |                 break;  | 
169  | 0  |             }  | 
170  | 0  |         }  | 
171  |  |  | 
172  |  |         /* c not found, return it itself */  | 
173  | 0  |         return c;  | 
174  | 0  |     }  | 
175  | 0  | }  | 
176  |  |  | 
177  |  | U_CFUNC UChar32  | 
178  | 0  | ubidi_getMirror(UChar32 c) { | 
179  | 0  |     uint16_t props=UTRIE2_GET16(&ubidi_props_singleton.trie, c);  | 
180  | 0  |     return getMirror(c, props);  | 
181  | 0  | }  | 
182  |  |  | 
183  |  | U_CFUNC UBool  | 
184  | 0  | ubidi_isBidiControl(UChar32 c) { | 
185  | 0  |     uint16_t props=UTRIE2_GET16(&ubidi_props_singleton.trie, c);  | 
186  | 0  |     return (UBool)UBIDI_GET_FLAG(props, UBIDI_BIDI_CONTROL_SHIFT);  | 
187  | 0  | }  | 
188  |  |  | 
189  |  | U_CFUNC UBool  | 
190  | 0  | ubidi_isJoinControl(UChar32 c) { | 
191  | 0  |     uint16_t props=UTRIE2_GET16(&ubidi_props_singleton.trie, c);  | 
192  | 0  |     return (UBool)UBIDI_GET_FLAG(props, UBIDI_JOIN_CONTROL_SHIFT);  | 
193  | 0  | }  | 
194  |  |  | 
195  |  | U_CFUNC UJoiningType  | 
196  | 0  | ubidi_getJoiningType(UChar32 c) { | 
197  | 0  |     uint16_t props=UTRIE2_GET16(&ubidi_props_singleton.trie, c);  | 
198  | 0  |     return (UJoiningType)((props&UBIDI_JT_MASK)>>UBIDI_JT_SHIFT);  | 
199  | 0  | }  | 
200  |  |  | 
201  |  | U_CFUNC UJoiningGroup  | 
202  | 0  | ubidi_getJoiningGroup(UChar32 c) { | 
203  | 0  |     UChar32 start, limit;  | 
204  |  | 
  | 
205  | 0  |     start=ubidi_props_singleton.indexes[UBIDI_IX_JG_START];  | 
206  | 0  |     limit=ubidi_props_singleton.indexes[UBIDI_IX_JG_LIMIT];  | 
207  | 0  |     if(start<=c && c<limit) { | 
208  | 0  |         return (UJoiningGroup)ubidi_props_singleton.jgArray[c-start];  | 
209  | 0  |     }  | 
210  | 0  |     start=ubidi_props_singleton.indexes[UBIDI_IX_JG_START2];  | 
211  | 0  |     limit=ubidi_props_singleton.indexes[UBIDI_IX_JG_LIMIT2];  | 
212  | 0  |     if(start<=c && c<limit) { | 
213  | 0  |         return (UJoiningGroup)ubidi_props_singleton.jgArray2[c-start];  | 
214  | 0  |     }  | 
215  | 0  |     return U_JG_NO_JOINING_GROUP;  | 
216  | 0  | }  | 
217  |  |  | 
218  |  | U_CFUNC UBidiPairedBracketType  | 
219  | 0  | ubidi_getPairedBracketType(UChar32 c) { | 
220  | 0  |     uint16_t props=UTRIE2_GET16(&ubidi_props_singleton.trie, c);  | 
221  | 0  |     return (UBidiPairedBracketType)((props&UBIDI_BPT_MASK)>>UBIDI_BPT_SHIFT);  | 
222  | 0  | }  | 
223  |  |  | 
224  |  | U_CFUNC UChar32  | 
225  | 0  | ubidi_getPairedBracket(UChar32 c) { | 
226  | 0  |     uint16_t props=UTRIE2_GET16(&ubidi_props_singleton.trie, c);  | 
227  | 0  |     if((props&UBIDI_BPT_MASK)==0) { | 
228  | 0  |         return c;  | 
229  | 0  |     } else { | 
230  | 0  |         return getMirror(c, props);  | 
231  | 0  |     }  | 
232  | 0  | }  | 
233  |  |  | 
234  |  | /* public API (see uchar.h) ------------------------------------------------- */  | 
235  |  |  | 
236  |  | U_CFUNC UCharDirection  | 
237  | 0  | u_charDirection(UChar32 c) { | 
238  | 0  |     return ubidi_getClass(c);  | 
239  | 0  | }  | 
240  |  |  | 
241  |  | U_CFUNC UBool  | 
242  | 0  | u_isMirrored(UChar32 c) { | 
243  | 0  |     return ubidi_isMirrored(c);  | 
244  | 0  | }  | 
245  |  |  | 
246  |  | U_CFUNC UChar32  | 
247  | 0  | u_charMirror(UChar32 c) { | 
248  | 0  |     return ubidi_getMirror(c);  | 
249  | 0  | }  | 
250  |  |  | 
251  |  | U_CAPI UChar32 U_EXPORT2  | 
252  | 0  | u_getBidiPairedBracket(UChar32 c) { | 
253  | 0  |     return ubidi_getPairedBracket(c);  | 
254  | 0  | }  |