1 // © 2016 and later: Unicode, Inc. and others.
2 // License & terms of use: http://www.unicode.org/copyright.html
3 /*
4 ********************************************************************************
5 *   Copyright (C) 1996-2016, International Business Machines
6 *   Corporation and others.  All Rights Reserved.
7 ********************************************************************************
8 *
9 * File UCHAR.C
10 *
11 * Modification History:
12 *
13 *   Date        Name        Description
14 *   04/02/97    aliu        Creation.
15 *   4/15/99     Madhu       Updated all the function definitions for C Implementation
16 *   5/20/99     Madhu       Added the function u_getVersion()
17 *   8/19/1999   srl         Upgraded scripts to Unicode3.0
18 *   11/11/1999  weiv        added u_isalnum(), cleaned comments
19 *   01/11/2000  helena      Renamed u_getVersion to u_getUnicodeVersion.
20 *   06/20/2000  helena      OS/400 port changes; mostly typecast.
21 ******************************************************************************
22 */
23 
24 #include "unicode/utypes.h"
25 #include "unicode/uchar.h"
26 #include "unicode/uscript.h"
27 #include "unicode/udata.h"
28 #include "uassert.h"
29 #include "cmemory.h"
30 #include "ucln_cmn.h"
31 #include "utrie2.h"
32 #include "udataswp.h"
33 #include "uprops.h"
34 #include "ustr_imp.h"
35 
36 /* uchar_props_data.h is machine-generated by genprops --csource */
37 #define INCLUDED_FROM_UCHAR_C
38 #include "uchar_props_data.h"
39 
40 /* constants and macros for access to the data ------------------------------ */
41 
42 /* getting a uint32_t properties word from the data */
43 #define GET_PROPS(c, result) ((result)=UTRIE2_GET16(&propsTrie, c))
44 
45 /* API functions ------------------------------------------------------------ */
46 
47 /* Gets the Unicode character's general category.*/
48 U_CAPI int8_t U_EXPORT2
u_charType(UChar32 c)49 u_charType(UChar32 c) {
50     uint32_t props;
51     GET_PROPS(c, props);
52     return (int8_t)GET_CATEGORY(props);
53 }
54 
55 /* Enumerate all code points with their general categories. */
56 struct _EnumTypeCallback {
57     UCharEnumTypeRange *enumRange;
58     const void *context;
59 };
60 
61 static uint32_t U_CALLCONV
_enumTypeValue(const void * context,uint32_t value)62 _enumTypeValue(const void *context, uint32_t value) {
63     (void)context;
64     return GET_CATEGORY(value);
65 }
66 
67 static UBool U_CALLCONV
_enumTypeRange(const void * context,UChar32 start,UChar32 end,uint32_t value)68 _enumTypeRange(const void *context, UChar32 start, UChar32 end, uint32_t value) {
69     /* just cast the value to UCharCategory */
70     return ((struct _EnumTypeCallback *)context)->
71         enumRange(((struct _EnumTypeCallback *)context)->context,
72                   start, end+1, (UCharCategory)value);
73 }
74 
75 U_CAPI void U_EXPORT2
u_enumCharTypes(UCharEnumTypeRange * enumRange,const void * context)76 u_enumCharTypes(UCharEnumTypeRange *enumRange, const void *context) {
77     struct _EnumTypeCallback callback;
78 
79     if(enumRange==NULL) {
80         return;
81     }
82 
83     callback.enumRange=enumRange;
84     callback.context=context;
85     utrie2_enum(&propsTrie, _enumTypeValue, _enumTypeRange, &callback);
86 }
87 
88 /* Checks if ch is a lower case letter.*/
89 U_CAPI UBool U_EXPORT2
u_islower(UChar32 c)90 u_islower(UChar32 c) {
91     uint32_t props;
92     GET_PROPS(c, props);
93     return (UBool)(GET_CATEGORY(props)==U_LOWERCASE_LETTER);
94 }
95 
96 /* Checks if ch is an upper case letter.*/
97 U_CAPI UBool U_EXPORT2
u_isupper(UChar32 c)98 u_isupper(UChar32 c) {
99     uint32_t props;
100     GET_PROPS(c, props);
101     return (UBool)(GET_CATEGORY(props)==U_UPPERCASE_LETTER);
102 }
103 
104 /* Checks if ch is a title case letter; usually upper case letters.*/
105 U_CAPI UBool U_EXPORT2
u_istitle(UChar32 c)106 u_istitle(UChar32 c) {
107     uint32_t props;
108     GET_PROPS(c, props);
109     return (UBool)(GET_CATEGORY(props)==U_TITLECASE_LETTER);
110 }
111 
112 /* Checks if ch is a decimal digit. */
113 U_CAPI UBool U_EXPORT2
u_isdigit(UChar32 c)114 u_isdigit(UChar32 c) {
115     uint32_t props;
116     GET_PROPS(c, props);
117     return (UBool)(GET_CATEGORY(props)==U_DECIMAL_DIGIT_NUMBER);
118 }
119 
120 U_CAPI UBool U_EXPORT2
u_isxdigit(UChar32 c)121 u_isxdigit(UChar32 c) {
122     uint32_t props;
123 
124     /* check ASCII and Fullwidth ASCII a-fA-F */
125     if(
126         (c<=0x66 && c>=0x41 && (c<=0x46 || c>=0x61)) ||
127         (c>=0xff21 && c<=0xff46 && (c<=0xff26 || c>=0xff41))
128     ) {
129         return TRUE;
130     }
131 
132     GET_PROPS(c, props);
133     return (UBool)(GET_CATEGORY(props)==U_DECIMAL_DIGIT_NUMBER);
134 }
135 
136 /* Checks if the Unicode character is a letter.*/
137 U_CAPI UBool U_EXPORT2
u_isalpha(UChar32 c)138 u_isalpha(UChar32 c) {
139     uint32_t props;
140     GET_PROPS(c, props);
141     return (UBool)((CAT_MASK(props)&U_GC_L_MASK)!=0);
142 }
143 
144 U_CAPI UBool U_EXPORT2
u_isUAlphabetic(UChar32 c)145 u_isUAlphabetic(UChar32 c) {
146     return (u_getUnicodeProperties(c, 1)&U_MASK(UPROPS_ALPHABETIC))!=0;
147 }
148 
149 /* Checks if c is a letter or a decimal digit */
150 U_CAPI UBool U_EXPORT2
u_isalnum(UChar32 c)151 u_isalnum(UChar32 c) {
152     uint32_t props;
153     GET_PROPS(c, props);
154     return (UBool)((CAT_MASK(props)&(U_GC_L_MASK|U_GC_ND_MASK))!=0);
155 }
156 
157 /**
158  * Checks if c is alphabetic, or a decimal digit; implements UCHAR_POSIX_ALNUM.
159  * @internal
160  */
161 U_CFUNC UBool
u_isalnumPOSIX(UChar32 c)162 u_isalnumPOSIX(UChar32 c) {
163     return (UBool)(u_isUAlphabetic(c) || u_isdigit(c));
164 }
165 
166 /* Checks if ch is a unicode character with assigned character type.*/
167 U_CAPI UBool U_EXPORT2
u_isdefined(UChar32 c)168 u_isdefined(UChar32 c) {
169     uint32_t props;
170     GET_PROPS(c, props);
171     return (UBool)(GET_CATEGORY(props)!=0);
172 }
173 
174 /* Checks if the Unicode character is a base form character that can take a diacritic.*/
175 U_CAPI UBool U_EXPORT2
u_isbase(UChar32 c)176 u_isbase(UChar32 c) {
177     uint32_t props;
178     GET_PROPS(c, props);
179     return (UBool)((CAT_MASK(props)&(U_GC_L_MASK|U_GC_N_MASK|U_GC_MC_MASK|U_GC_ME_MASK))!=0);
180 }
181 
182 /* Checks if the Unicode character is a control character.*/
183 U_CAPI UBool U_EXPORT2
u_iscntrl(UChar32 c)184 u_iscntrl(UChar32 c) {
185     uint32_t props;
186     GET_PROPS(c, props);
187     return (UBool)((CAT_MASK(props)&(U_GC_CC_MASK|U_GC_CF_MASK|U_GC_ZL_MASK|U_GC_ZP_MASK))!=0);
188 }
189 
190 U_CAPI UBool U_EXPORT2
u_isISOControl(UChar32 c)191 u_isISOControl(UChar32 c) {
192     return (uint32_t)c<=0x9f && (c<=0x1f || c>=0x7f);
193 }
194 
195 /* Some control characters that are used as space. */
196 #define IS_THAT_CONTROL_SPACE(c) \
197     (c<=0x9f && ((c>=TAB && c<=CR) || (c>=0x1c && c <=0x1f) || c==0x85))
198 
199 /* Java has decided that U+0085 New Line is not whitespace any more. */
200 #define IS_THAT_ASCII_CONTROL_SPACE(c) \
201     (c<=0x1f && c>=TAB && (c<=CR || c>=0x1c))
202 
203 /* Checks if the Unicode character is a space character.*/
204 U_CAPI UBool U_EXPORT2
u_isspace(UChar32 c)205 u_isspace(UChar32 c) {
206     uint32_t props;
207     GET_PROPS(c, props);
208     return (UBool)((CAT_MASK(props)&U_GC_Z_MASK)!=0 || IS_THAT_CONTROL_SPACE(c));
209 }
210 
211 U_CAPI UBool U_EXPORT2
u_isJavaSpaceChar(UChar32 c)212 u_isJavaSpaceChar(UChar32 c) {
213     uint32_t props;
214     GET_PROPS(c, props);
215     return (UBool)((CAT_MASK(props)&U_GC_Z_MASK)!=0);
216 }
217 
218 /* Checks if the Unicode character is a whitespace character.*/
219 U_CAPI UBool U_EXPORT2
u_isWhitespace(UChar32 c)220 u_isWhitespace(UChar32 c) {
221     uint32_t props;
222     GET_PROPS(c, props);
223     return (UBool)(
224                 ((CAT_MASK(props)&U_GC_Z_MASK)!=0 &&
225                     c!=NBSP && c!=FIGURESP && c!=NNBSP) || /* exclude no-break spaces */
226                 IS_THAT_ASCII_CONTROL_SPACE(c)
227            );
228 }
229 
230 U_CAPI UBool U_EXPORT2
u_isblank(UChar32 c)231 u_isblank(UChar32 c) {
232     if((uint32_t)c<=0x9f) {
233         return c==9 || c==0x20; /* TAB or SPACE */
234     } else {
235         /* Zs */
236         uint32_t props;
237         GET_PROPS(c, props);
238         return (UBool)(GET_CATEGORY(props)==U_SPACE_SEPARATOR);
239     }
240 }
241 
242 U_CAPI UBool U_EXPORT2
u_isUWhiteSpace(UChar32 c)243 u_isUWhiteSpace(UChar32 c) {
244     return (u_getUnicodeProperties(c, 1)&U_MASK(UPROPS_WHITE_SPACE))!=0;
245 }
246 
247 /* Checks if the Unicode character is printable.*/
248 U_CAPI UBool U_EXPORT2
u_isprint(UChar32 c)249 u_isprint(UChar32 c) {
250     uint32_t props;
251     GET_PROPS(c, props);
252     /* comparing ==0 returns FALSE for the categories mentioned */
253     return (UBool)((CAT_MASK(props)&U_GC_C_MASK)==0);
254 }
255 
256 /**
257  * Checks if c is in \p{graph}\p{blank} - \p{cntrl}.
258  * Implements UCHAR_POSIX_PRINT.
259  * @internal
260  */
261 U_CFUNC UBool
u_isprintPOSIX(UChar32 c)262 u_isprintPOSIX(UChar32 c) {
263     uint32_t props;
264     GET_PROPS(c, props);
265     /*
266      * The only cntrl character in graph+blank is TAB (in blank).
267      * Here we implement (blank-TAB)=Zs instead of calling u_isblank().
268      */
269     return (UBool)((GET_CATEGORY(props)==U_SPACE_SEPARATOR) || u_isgraphPOSIX(c));
270 }
271 
272 U_CAPI UBool U_EXPORT2
u_isgraph(UChar32 c)273 u_isgraph(UChar32 c) {
274     uint32_t props;
275     GET_PROPS(c, props);
276     /* comparing ==0 returns FALSE for the categories mentioned */
277     return (UBool)((CAT_MASK(props)&
278                     (U_GC_CC_MASK|U_GC_CF_MASK|U_GC_CS_MASK|U_GC_CN_MASK|U_GC_Z_MASK))
279                    ==0);
280 }
281 
282 /**
283  * Checks if c is in
284  * [^\p{space}\p{gc=Control}\p{gc=Surrogate}\p{gc=Unassigned}]
285  * with space=\p{Whitespace} and Control=Cc.
286  * Implements UCHAR_POSIX_GRAPH.
287  * @internal
288  */
289 U_CFUNC UBool
u_isgraphPOSIX(UChar32 c)290 u_isgraphPOSIX(UChar32 c) {
291     uint32_t props;
292     GET_PROPS(c, props);
293     /* \p{space}\p{gc=Control} == \p{gc=Z}\p{Control} */
294     /* comparing ==0 returns FALSE for the categories mentioned */
295     return (UBool)((CAT_MASK(props)&
296                     (U_GC_CC_MASK|U_GC_CS_MASK|U_GC_CN_MASK|U_GC_Z_MASK))
297                    ==0);
298 }
299 
300 U_CAPI UBool U_EXPORT2
u_ispunct(UChar32 c)301 u_ispunct(UChar32 c) {
302     uint32_t props;
303     GET_PROPS(c, props);
304     return (UBool)((CAT_MASK(props)&U_GC_P_MASK)!=0);
305 }
306 
307 /* Checks if the Unicode character can start a Unicode identifier.*/
308 U_CAPI UBool U_EXPORT2
u_isIDStart(UChar32 c)309 u_isIDStart(UChar32 c) {
310     /* same as u_isalpha() */
311     uint32_t props;
312     GET_PROPS(c, props);
313     return (UBool)((CAT_MASK(props)&(U_GC_L_MASK|U_GC_NL_MASK))!=0);
314 }
315 
316 /* Checks if the Unicode character can be a Unicode identifier part other than starting the
317  identifier.*/
318 U_CAPI UBool U_EXPORT2
u_isIDPart(UChar32 c)319 u_isIDPart(UChar32 c) {
320     uint32_t props;
321     GET_PROPS(c, props);
322     return (UBool)(
323            (CAT_MASK(props)&
324             (U_GC_ND_MASK|U_GC_NL_MASK|
325              U_GC_L_MASK|
326              U_GC_PC_MASK|U_GC_MC_MASK|U_GC_MN_MASK)
327            )!=0 ||
328            u_isIDIgnorable(c));
329 }
330 
331 /*Checks if the Unicode character can be ignorable in a Java or Unicode identifier.*/
332 U_CAPI UBool U_EXPORT2
u_isIDIgnorable(UChar32 c)333 u_isIDIgnorable(UChar32 c) {
334     if(c<=0x9f) {
335         return u_isISOControl(c) && !IS_THAT_ASCII_CONTROL_SPACE(c);
336     } else {
337         uint32_t props;
338         GET_PROPS(c, props);
339         return (UBool)(GET_CATEGORY(props)==U_FORMAT_CHAR);
340     }
341 }
342 
343 /*Checks if the Unicode character can start a Java identifier.*/
344 U_CAPI UBool U_EXPORT2
u_isJavaIDStart(UChar32 c)345 u_isJavaIDStart(UChar32 c) {
346     uint32_t props;
347     GET_PROPS(c, props);
348     return (UBool)((CAT_MASK(props)&(U_GC_L_MASK|U_GC_SC_MASK|U_GC_PC_MASK))!=0);
349 }
350 
351 /*Checks if the Unicode character can be a Java identifier part other than starting the
352  * identifier.
353  */
354 U_CAPI UBool U_EXPORT2
u_isJavaIDPart(UChar32 c)355 u_isJavaIDPart(UChar32 c) {
356     uint32_t props;
357     GET_PROPS(c, props);
358     return (UBool)(
359            (CAT_MASK(props)&
360             (U_GC_ND_MASK|U_GC_NL_MASK|
361              U_GC_L_MASK|
362              U_GC_SC_MASK|U_GC_PC_MASK|
363              U_GC_MC_MASK|U_GC_MN_MASK)
364            )!=0 ||
365            u_isIDIgnorable(c));
366 }
367 
368 U_CAPI int32_t U_EXPORT2
u_charDigitValue(UChar32 c)369 u_charDigitValue(UChar32 c) {
370     uint32_t props;
371     int32_t value;
372     GET_PROPS(c, props);
373     value=(int32_t)GET_NUMERIC_TYPE_VALUE(props)-UPROPS_NTV_DECIMAL_START;
374     if(value<=9) {
375         return value;
376     } else {
377         return -1;
378     }
379 }
380 
381 U_CAPI double U_EXPORT2
u_getNumericValue(UChar32 c)382 u_getNumericValue(UChar32 c) {
383     uint32_t props;
384     int32_t ntv;
385     GET_PROPS(c, props);
386     ntv=(int32_t)GET_NUMERIC_TYPE_VALUE(props);
387 
388     if(ntv==UPROPS_NTV_NONE) {
389         return U_NO_NUMERIC_VALUE;
390     } else if(ntv<UPROPS_NTV_DIGIT_START) {
391         /* decimal digit */
392         return ntv-UPROPS_NTV_DECIMAL_START;
393     } else if(ntv<UPROPS_NTV_NUMERIC_START) {
394         /* other digit */
395         return ntv-UPROPS_NTV_DIGIT_START;
396     } else if(ntv<UPROPS_NTV_FRACTION_START) {
397         /* small integer */
398         return ntv-UPROPS_NTV_NUMERIC_START;
399     } else if(ntv<UPROPS_NTV_LARGE_START) {
400         /* fraction */
401         int32_t numerator=(ntv>>4)-12;
402         int32_t denominator=(ntv&0xf)+1;
403         return (double)numerator/denominator;
404     } else if(ntv<UPROPS_NTV_BASE60_START) {
405         /* large, single-significant-digit integer */
406         double numValue;
407         int32_t mant=(ntv>>5)-14;
408         int32_t exp=(ntv&0x1f)+2;
409         numValue=mant;
410 
411         /* multiply by 10^exp without math.h */
412         while(exp>=4) {
413             numValue*=10000.;
414             exp-=4;
415         }
416         switch(exp) {
417         case 3:
418             numValue*=1000.;
419             break;
420         case 2:
421             numValue*=100.;
422             break;
423         case 1:
424             numValue*=10.;
425             break;
426         case 0:
427         default:
428             break;
429         }
430 
431         return numValue;
432     } else if(ntv<UPROPS_NTV_FRACTION20_START) {
433         /* sexagesimal (base 60) integer */
434         int32_t numValue=(ntv>>2)-0xbf;
435         int32_t exp=(ntv&3)+1;
436 
437         switch(exp) {
438         case 4:
439             numValue*=60*60*60*60;
440             break;
441         case 3:
442             numValue*=60*60*60;
443             break;
444         case 2:
445             numValue*=60*60;
446             break;
447         case 1:
448             numValue*=60;
449             break;
450         case 0:
451         default:
452             break;
453         }
454 
455         return numValue;
456     } else if(ntv<UPROPS_NTV_FRACTION32_START) {
457         // fraction-20 e.g. 3/80
458         int32_t frac20=ntv-UPROPS_NTV_FRACTION20_START;  // 0..0x17
459         int32_t numerator=2*(frac20&3)+1;
460         int32_t denominator=20<<(frac20>>2);
461         return (double)numerator/denominator;
462     } else if(ntv<UPROPS_NTV_RESERVED_START) {
463         // fraction-32 e.g. 3/64
464         int32_t frac32=ntv-UPROPS_NTV_FRACTION32_START;  // 0..15
465         int32_t numerator=2*(frac32&3)+1;
466         int32_t denominator=32<<(frac32>>2);
467         return (double)numerator/denominator;
468     } else {
469         /* reserved */
470         return U_NO_NUMERIC_VALUE;
471     }
472 }
473 
474 U_CAPI int32_t U_EXPORT2
u_digit(UChar32 ch,int8_t radix)475 u_digit(UChar32 ch, int8_t radix) {
476     int8_t value;
477     if((uint8_t)(radix-2)<=(36-2)) {
478         value=(int8_t)u_charDigitValue(ch);
479         if(value<0) {
480             /* ch is not a decimal digit, try latin letters */
481             if(ch>=0x61 && ch<=0x7A) {
482                 value=(int8_t)(ch-0x57);  /* ch - 'a' + 10 */
483             } else if(ch>=0x41 && ch<=0x5A) {
484                 value=(int8_t)(ch-0x37);  /* ch - 'A' + 10 */
485             } else if(ch>=0xFF41 && ch<=0xFF5A) {
486                 value=(int8_t)(ch-0xFF37);  /* fullwidth ASCII a-z */
487             } else if(ch>=0xFF21 && ch<=0xFF3A) {
488                 value=(int8_t)(ch-0xFF17);  /* fullwidth ASCII A-Z */
489             }
490         }
491     } else {
492         value=-1;   /* invalid radix */
493     }
494     return (int8_t)((value<radix) ? value : -1);
495 }
496 
497 U_CAPI UChar32 U_EXPORT2
u_forDigit(int32_t digit,int8_t radix)498 u_forDigit(int32_t digit, int8_t radix) {
499     if((uint8_t)(radix-2)>(36-2) || (uint32_t)digit>=(uint32_t)radix) {
500         return 0;
501     } else if(digit<10) {
502         return (UChar32)(0x30+digit);
503     } else {
504         return (UChar32)((0x61-10)+digit);
505     }
506 }
507 
508 /* miscellaneous, and support for uprops.cpp -------------------------------- */
509 
510 U_CAPI void U_EXPORT2
u_getUnicodeVersion(UVersionInfo versionArray)511 u_getUnicodeVersion(UVersionInfo versionArray) {
512     if(versionArray!=NULL) {
513         uprv_memcpy(versionArray, dataVersion, U_MAX_VERSION_LENGTH);
514     }
515 }
516 
517 U_CFUNC uint32_t
u_getMainProperties(UChar32 c)518 u_getMainProperties(UChar32 c) {
519     uint32_t props;
520     GET_PROPS(c, props);
521     return props;
522 }
523 
524 U_CFUNC uint32_t
u_getUnicodeProperties(UChar32 c,int32_t column)525 u_getUnicodeProperties(UChar32 c, int32_t column) {
526     U_ASSERT(column>=0);
527     if(column>=propsVectorsColumns) {
528         return 0;
529     } else {
530         uint16_t vecIndex=UTRIE2_GET16(&propsVectorsTrie, c);
531         return propsVectors[vecIndex+column];
532     }
533 }
534 
535 U_CFUNC int32_t
uprv_getMaxValues(int32_t column)536 uprv_getMaxValues(int32_t column) {
537     switch(column) {
538     case 0:
539         return indexes[UPROPS_MAX_VALUES_INDEX];
540     case 2:
541         return indexes[UPROPS_MAX_VALUES_2_INDEX];
542     default:
543         return 0;
544     }
545 }
546 
547 U_CAPI void U_EXPORT2
u_charAge(UChar32 c,UVersionInfo versionArray)548 u_charAge(UChar32 c, UVersionInfo versionArray) {
549     if(versionArray!=NULL) {
550         uint32_t version=u_getUnicodeProperties(c, 0)>>UPROPS_AGE_SHIFT;
551         versionArray[0]=(uint8_t)(version>>4);
552         versionArray[1]=(uint8_t)(version&0xf);
553         versionArray[2]=versionArray[3]=0;
554     }
555 }
556 
557 U_CAPI UScriptCode U_EXPORT2
uscript_getScript(UChar32 c,UErrorCode * pErrorCode)558 uscript_getScript(UChar32 c, UErrorCode *pErrorCode) {
559     if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
560         return USCRIPT_INVALID_CODE;
561     }
562     if((uint32_t)c>0x10ffff) {
563         *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
564         return USCRIPT_INVALID_CODE;
565     }
566     uint32_t scriptX=u_getUnicodeProperties(c, 0)&UPROPS_SCRIPT_X_MASK;
567     uint32_t codeOrIndex=uprops_mergeScriptCodeOrIndex(scriptX);
568     if(scriptX<UPROPS_SCRIPT_X_WITH_COMMON) {
569         return (UScriptCode)codeOrIndex;
570     } else if(scriptX<UPROPS_SCRIPT_X_WITH_INHERITED) {
571         return USCRIPT_COMMON;
572     } else if(scriptX<UPROPS_SCRIPT_X_WITH_OTHER) {
573         return USCRIPT_INHERITED;
574     } else {
575         return (UScriptCode)scriptExtensions[codeOrIndex];
576     }
577 }
578 
579 U_CAPI UBool U_EXPORT2
uscript_hasScript(UChar32 c,UScriptCode sc)580 uscript_hasScript(UChar32 c, UScriptCode sc) {
581     uint32_t scriptX=u_getUnicodeProperties(c, 0)&UPROPS_SCRIPT_X_MASK;
582     uint32_t codeOrIndex=uprops_mergeScriptCodeOrIndex(scriptX);
583     if(scriptX<UPROPS_SCRIPT_X_WITH_COMMON) {
584         return sc==(UScriptCode)codeOrIndex;
585     }
586 
587     const uint16_t *scx=scriptExtensions+codeOrIndex;
588     if(scriptX>=UPROPS_SCRIPT_X_WITH_OTHER) {
589         scx=scriptExtensions+scx[1];
590     }
591     uint32_t sc32=sc;
592     if(sc32>0x7fff) {
593         /* Guard against bogus input that would make us go past the Script_Extensions terminator. */
594         return FALSE;
595     }
596     while(sc32>*scx) {
597         ++scx;
598     }
599     return sc32==(*scx&0x7fff);
600 }
601 
602 U_CAPI int32_t U_EXPORT2
uscript_getScriptExtensions(UChar32 c,UScriptCode * scripts,int32_t capacity,UErrorCode * pErrorCode)603 uscript_getScriptExtensions(UChar32 c,
604                             UScriptCode *scripts, int32_t capacity,
605                             UErrorCode *pErrorCode) {
606     if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
607         return 0;
608     }
609     if(capacity<0 || (capacity>0 && scripts==NULL)) {
610         *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
611         return 0;
612     }
613     uint32_t scriptX=u_getUnicodeProperties(c, 0)&UPROPS_SCRIPT_X_MASK;
614     uint32_t codeOrIndex=uprops_mergeScriptCodeOrIndex(scriptX);
615     if(scriptX<UPROPS_SCRIPT_X_WITH_COMMON) {
616         if(capacity==0) {
617             *pErrorCode=U_BUFFER_OVERFLOW_ERROR;
618         } else {
619             scripts[0]=(UScriptCode)codeOrIndex;
620         }
621         return 1;
622     }
623 
624     const uint16_t *scx=scriptExtensions+codeOrIndex;
625     if(scriptX>=UPROPS_SCRIPT_X_WITH_OTHER) {
626         scx=scriptExtensions+scx[1];
627     }
628     int32_t length=0;
629     uint16_t sx;
630     do {
631         sx=*scx++;
632         if(length<capacity) {
633             scripts[length]=(UScriptCode)(sx&0x7fff);
634         }
635         ++length;
636     } while(sx<0x8000);
637     if(length>capacity) {
638         *pErrorCode=U_BUFFER_OVERFLOW_ERROR;
639     }
640     return length;
641 }
642 
643 U_CAPI UBlockCode U_EXPORT2
ublock_getCode(UChar32 c)644 ublock_getCode(UChar32 c) {
645     return (UBlockCode)((u_getUnicodeProperties(c, 0)&UPROPS_BLOCK_MASK)>>UPROPS_BLOCK_SHIFT);
646 }
647 
648 /* property starts for UnicodeSet ------------------------------------------- */
649 
650 static UBool U_CALLCONV
_enumPropertyStartsRange(const void * context,UChar32 start,UChar32 end,uint32_t value)651 _enumPropertyStartsRange(const void *context, UChar32 start, UChar32 end, uint32_t value) {
652     /* add the start code point to the USet */
653     const USetAdder *sa=(const USetAdder *)context;
654     sa->add(sa->set, start);
655     (void)end;
656     (void)value;
657     return TRUE;
658 }
659 
660 #define USET_ADD_CP_AND_NEXT(sa, cp) sa->add(sa->set, cp); sa->add(sa->set, cp+1)
661 
662 U_CFUNC void U_EXPORT2
uchar_addPropertyStarts(const USetAdder * sa,UErrorCode * pErrorCode)663 uchar_addPropertyStarts(const USetAdder *sa, UErrorCode *pErrorCode) {
664     if(U_FAILURE(*pErrorCode)) {
665         return;
666     }
667 
668     /* add the start code point of each same-value range of the main trie */
669     utrie2_enum(&propsTrie, NULL, _enumPropertyStartsRange, sa);
670 
671     /* add code points with hardcoded properties, plus the ones following them */
672 
673     /* add for u_isblank() */
674     USET_ADD_CP_AND_NEXT(sa, TAB);
675 
676     /* add for IS_THAT_CONTROL_SPACE() */
677     sa->add(sa->set, CR+1); /* range TAB..CR */
678     sa->add(sa->set, 0x1c);
679     sa->add(sa->set, 0x1f+1);
680     USET_ADD_CP_AND_NEXT(sa, 0x85);  // NEXT LINE (NEL)
681 
682     /* add for u_isIDIgnorable() what was not added above */
683     sa->add(sa->set, 0x7f); /* range DEL..NBSP-1, NBSP added below */
684     sa->add(sa->set, HAIRSP);
685     sa->add(sa->set, RLM+1);
686     sa->add(sa->set, 0x206a);  // INHIBIT SYMMETRIC SWAPPING
687     sa->add(sa->set, 0x206f+1);  // NOMINAL DIGIT SHAPES
688     USET_ADD_CP_AND_NEXT(sa, ZWNBSP);
689 
690     /* add no-break spaces for u_isWhitespace() what was not added above */
691     USET_ADD_CP_AND_NEXT(sa, NBSP);
692     USET_ADD_CP_AND_NEXT(sa, FIGURESP);
693     USET_ADD_CP_AND_NEXT(sa, NNBSP);
694 
695     /* add for u_digit() */
696     sa->add(sa->set, u'a');
697     sa->add(sa->set, u'z'+1);
698     sa->add(sa->set, u'A');
699     sa->add(sa->set, u'Z'+1);
700     // fullwidth
701     sa->add(sa->set, u'ａ');
702     sa->add(sa->set, u'ｚ'+1);
703     sa->add(sa->set, u'Ａ');
704     sa->add(sa->set, u'Ｚ'+1);
705 
706     /* add for u_isxdigit() */
707     sa->add(sa->set, u'f'+1);
708     sa->add(sa->set, u'F'+1);
709     // fullwidth
710     sa->add(sa->set, u'ｆ'+1);
711     sa->add(sa->set, u'Ｆ'+1);
712 
713     /* add for UCHAR_DEFAULT_IGNORABLE_CODE_POINT what was not added above */
714     sa->add(sa->set, 0x2060); /* range 2060..206f */
715     sa->add(sa->set, 0xfff0);
716     sa->add(sa->set, 0xfffb+1);
717     sa->add(sa->set, 0xe0000);
718     sa->add(sa->set, 0xe0fff+1);
719 
720     /* add for UCHAR_GRAPHEME_BASE and others */
721     USET_ADD_CP_AND_NEXT(sa, CGJ);
722 }
723 
724 U_CFUNC void U_EXPORT2
upropsvec_addPropertyStarts(const USetAdder * sa,UErrorCode * pErrorCode)725 upropsvec_addPropertyStarts(const USetAdder *sa, UErrorCode *pErrorCode) {
726     if(U_FAILURE(*pErrorCode)) {
727         return;
728     }
729 
730     /* add the start code point of each same-value range of the properties vectors trie */
731     utrie2_enum(&propsVectorsTrie, NULL, _enumPropertyStartsRange, sa);
732 }
733