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 U_CFUNC UBool
uprv_haveProperties(UErrorCode * pErrorCode)46 uprv_haveProperties(UErrorCode *pErrorCode) {
47     if(U_FAILURE(*pErrorCode)) {
48         return FALSE;
49     }
50     return TRUE;
51 }
52 
53 /* API functions ------------------------------------------------------------ */
54 
55 /* Gets the Unicode character's general category.*/
56 U_CAPI int8_t U_EXPORT2
u_charType(UChar32 c)57 u_charType(UChar32 c) {
58     uint32_t props;
59     GET_PROPS(c, props);
60     return (int8_t)GET_CATEGORY(props);
61 }
62 
63 /* Enumerate all code points with their general categories. */
64 struct _EnumTypeCallback {
65     UCharEnumTypeRange *enumRange;
66     const void *context;
67 };
68 
69 static uint32_t U_CALLCONV
_enumTypeValue(const void * context,uint32_t value)70 _enumTypeValue(const void *context, uint32_t value) {
71     (void)context;
72     return GET_CATEGORY(value);
73 }
74 
75 static UBool U_CALLCONV
_enumTypeRange(const void * context,UChar32 start,UChar32 end,uint32_t value)76 _enumTypeRange(const void *context, UChar32 start, UChar32 end, uint32_t value) {
77     /* just cast the value to UCharCategory */
78     return ((struct _EnumTypeCallback *)context)->
79         enumRange(((struct _EnumTypeCallback *)context)->context,
80                   start, end+1, (UCharCategory)value);
81 }
82 
83 U_CAPI void U_EXPORT2
u_enumCharTypes(UCharEnumTypeRange * enumRange,const void * context)84 u_enumCharTypes(UCharEnumTypeRange *enumRange, const void *context) {
85     struct _EnumTypeCallback callback;
86 
87     if(enumRange==NULL) {
88         return;
89     }
90 
91     callback.enumRange=enumRange;
92     callback.context=context;
93     utrie2_enum(&propsTrie, _enumTypeValue, _enumTypeRange, &callback);
94 }
95 
96 /* Checks if ch is a lower case letter.*/
97 U_CAPI UBool U_EXPORT2
u_islower(UChar32 c)98 u_islower(UChar32 c) {
99     uint32_t props;
100     GET_PROPS(c, props);
101     return (UBool)(GET_CATEGORY(props)==U_LOWERCASE_LETTER);
102 }
103 
104 /* Checks if ch is an upper case letter.*/
105 U_CAPI UBool U_EXPORT2
u_isupper(UChar32 c)106 u_isupper(UChar32 c) {
107     uint32_t props;
108     GET_PROPS(c, props);
109     return (UBool)(GET_CATEGORY(props)==U_UPPERCASE_LETTER);
110 }
111 
112 /* Checks if ch is a title case letter; usually upper case letters.*/
113 U_CAPI UBool U_EXPORT2
u_istitle(UChar32 c)114 u_istitle(UChar32 c) {
115     uint32_t props;
116     GET_PROPS(c, props);
117     return (UBool)(GET_CATEGORY(props)==U_TITLECASE_LETTER);
118 }
119 
120 /* Checks if ch is a decimal digit. */
121 U_CAPI UBool U_EXPORT2
u_isdigit(UChar32 c)122 u_isdigit(UChar32 c) {
123     uint32_t props;
124     GET_PROPS(c, props);
125     return (UBool)(GET_CATEGORY(props)==U_DECIMAL_DIGIT_NUMBER);
126 }
127 
128 U_CAPI UBool U_EXPORT2
u_isxdigit(UChar32 c)129 u_isxdigit(UChar32 c) {
130     uint32_t props;
131 
132     /* check ASCII and Fullwidth ASCII a-fA-F */
133     if(
134         (c<=0x66 && c>=0x41 && (c<=0x46 || c>=0x61)) ||
135         (c>=0xff21 && c<=0xff46 && (c<=0xff26 || c>=0xff41))
136     ) {
137         return TRUE;
138     }
139 
140     GET_PROPS(c, props);
141     return (UBool)(GET_CATEGORY(props)==U_DECIMAL_DIGIT_NUMBER);
142 }
143 
144 /* Checks if the Unicode character is a letter.*/
145 U_CAPI UBool U_EXPORT2
u_isalpha(UChar32 c)146 u_isalpha(UChar32 c) {
147     uint32_t props;
148     GET_PROPS(c, props);
149     return (UBool)((CAT_MASK(props)&U_GC_L_MASK)!=0);
150 }
151 
152 U_CAPI UBool U_EXPORT2
u_isUAlphabetic(UChar32 c)153 u_isUAlphabetic(UChar32 c) {
154     return (u_getUnicodeProperties(c, 1)&U_MASK(UPROPS_ALPHABETIC))!=0;
155 }
156 
157 /* Checks if c is a letter or a decimal digit */
158 U_CAPI UBool U_EXPORT2
u_isalnum(UChar32 c)159 u_isalnum(UChar32 c) {
160     uint32_t props;
161     GET_PROPS(c, props);
162     return (UBool)((CAT_MASK(props)&(U_GC_L_MASK|U_GC_ND_MASK))!=0);
163 }
164 
165 /**
166  * Checks if c is alphabetic, or a decimal digit; implements UCHAR_POSIX_ALNUM.
167  * @internal
168  */
169 U_CFUNC UBool
u_isalnumPOSIX(UChar32 c)170 u_isalnumPOSIX(UChar32 c) {
171     return (UBool)(u_isUAlphabetic(c) || u_isdigit(c));
172 }
173 
174 /* Checks if ch is a unicode character with assigned character type.*/
175 U_CAPI UBool U_EXPORT2
u_isdefined(UChar32 c)176 u_isdefined(UChar32 c) {
177     uint32_t props;
178     GET_PROPS(c, props);
179     return (UBool)(GET_CATEGORY(props)!=0);
180 }
181 
182 /* Checks if the Unicode character is a base form character that can take a diacritic.*/
183 U_CAPI UBool U_EXPORT2
u_isbase(UChar32 c)184 u_isbase(UChar32 c) {
185     uint32_t props;
186     GET_PROPS(c, props);
187     return (UBool)((CAT_MASK(props)&(U_GC_L_MASK|U_GC_N_MASK|U_GC_MC_MASK|U_GC_ME_MASK))!=0);
188 }
189 
190 /* Checks if the Unicode character is a control character.*/
191 U_CAPI UBool U_EXPORT2
u_iscntrl(UChar32 c)192 u_iscntrl(UChar32 c) {
193     uint32_t props;
194     GET_PROPS(c, props);
195     return (UBool)((CAT_MASK(props)&(U_GC_CC_MASK|U_GC_CF_MASK|U_GC_ZL_MASK|U_GC_ZP_MASK))!=0);
196 }
197 
198 U_CAPI UBool U_EXPORT2
u_isISOControl(UChar32 c)199 u_isISOControl(UChar32 c) {
200     return (uint32_t)c<=0x9f && (c<=0x1f || c>=0x7f);
201 }
202 
203 /* Some control characters that are used as space. */
204 #define IS_THAT_CONTROL_SPACE(c) \
205     (c<=0x9f && ((c>=TAB && c<=CR) || (c>=0x1c && c <=0x1f) || c==NL))
206 
207 /* Java has decided that U+0085 New Line is not whitespace any more. */
208 #define IS_THAT_ASCII_CONTROL_SPACE(c) \
209     (c<=0x1f && c>=TAB && (c<=CR || c>=0x1c))
210 
211 /* Checks if the Unicode character is a space character.*/
212 U_CAPI UBool U_EXPORT2
u_isspace(UChar32 c)213 u_isspace(UChar32 c) {
214     uint32_t props;
215     GET_PROPS(c, props);
216     return (UBool)((CAT_MASK(props)&U_GC_Z_MASK)!=0 || IS_THAT_CONTROL_SPACE(c));
217 }
218 
219 U_CAPI UBool U_EXPORT2
u_isJavaSpaceChar(UChar32 c)220 u_isJavaSpaceChar(UChar32 c) {
221     uint32_t props;
222     GET_PROPS(c, props);
223     return (UBool)((CAT_MASK(props)&U_GC_Z_MASK)!=0);
224 }
225 
226 /* Checks if the Unicode character is a whitespace character.*/
227 U_CAPI UBool U_EXPORT2
u_isWhitespace(UChar32 c)228 u_isWhitespace(UChar32 c) {
229     uint32_t props;
230     GET_PROPS(c, props);
231     return (UBool)(
232                 ((CAT_MASK(props)&U_GC_Z_MASK)!=0 &&
233                     c!=NBSP && c!=FIGURESP && c!=NNBSP) || /* exclude no-break spaces */
234                 IS_THAT_ASCII_CONTROL_SPACE(c)
235            );
236 }
237 
238 U_CAPI UBool U_EXPORT2
u_isblank(UChar32 c)239 u_isblank(UChar32 c) {
240     if((uint32_t)c<=0x9f) {
241         return c==9 || c==0x20; /* TAB or SPACE */
242     } else {
243         /* Zs */
244         uint32_t props;
245         GET_PROPS(c, props);
246         return (UBool)(GET_CATEGORY(props)==U_SPACE_SEPARATOR);
247     }
248 }
249 
250 U_CAPI UBool U_EXPORT2
u_isUWhiteSpace(UChar32 c)251 u_isUWhiteSpace(UChar32 c) {
252     return (u_getUnicodeProperties(c, 1)&U_MASK(UPROPS_WHITE_SPACE))!=0;
253 }
254 
255 /* Checks if the Unicode character is printable.*/
256 U_CAPI UBool U_EXPORT2
u_isprint(UChar32 c)257 u_isprint(UChar32 c) {
258     uint32_t props;
259     GET_PROPS(c, props);
260     /* comparing ==0 returns FALSE for the categories mentioned */
261     return (UBool)((CAT_MASK(props)&U_GC_C_MASK)==0);
262 }
263 
264 /**
265  * Checks if c is in \p{graph}\p{blank} - \p{cntrl}.
266  * Implements UCHAR_POSIX_PRINT.
267  * @internal
268  */
269 U_CFUNC UBool
u_isprintPOSIX(UChar32 c)270 u_isprintPOSIX(UChar32 c) {
271     uint32_t props;
272     GET_PROPS(c, props);
273     /*
274      * The only cntrl character in graph+blank is TAB (in blank).
275      * Here we implement (blank-TAB)=Zs instead of calling u_isblank().
276      */
277     return (UBool)((GET_CATEGORY(props)==U_SPACE_SEPARATOR) || u_isgraphPOSIX(c));
278 }
279 
280 U_CAPI UBool U_EXPORT2
u_isgraph(UChar32 c)281 u_isgraph(UChar32 c) {
282     uint32_t props;
283     GET_PROPS(c, props);
284     /* comparing ==0 returns FALSE for the categories mentioned */
285     return (UBool)((CAT_MASK(props)&
286                     (U_GC_CC_MASK|U_GC_CF_MASK|U_GC_CS_MASK|U_GC_CN_MASK|U_GC_Z_MASK))
287                    ==0);
288 }
289 
290 /**
291  * Checks if c is in
292  * [^\p{space}\p{gc=Control}\p{gc=Surrogate}\p{gc=Unassigned}]
293  * with space=\p{Whitespace} and Control=Cc.
294  * Implements UCHAR_POSIX_GRAPH.
295  * @internal
296  */
297 U_CFUNC UBool
u_isgraphPOSIX(UChar32 c)298 u_isgraphPOSIX(UChar32 c) {
299     uint32_t props;
300     GET_PROPS(c, props);
301     /* \p{space}\p{gc=Control} == \p{gc=Z}\p{Control} */
302     /* comparing ==0 returns FALSE for the categories mentioned */
303     return (UBool)((CAT_MASK(props)&
304                     (U_GC_CC_MASK|U_GC_CS_MASK|U_GC_CN_MASK|U_GC_Z_MASK))
305                    ==0);
306 }
307 
308 U_CAPI UBool U_EXPORT2
u_ispunct(UChar32 c)309 u_ispunct(UChar32 c) {
310     uint32_t props;
311     GET_PROPS(c, props);
312     return (UBool)((CAT_MASK(props)&U_GC_P_MASK)!=0);
313 }
314 
315 /* Checks if the Unicode character can start a Unicode identifier.*/
316 U_CAPI UBool U_EXPORT2
u_isIDStart(UChar32 c)317 u_isIDStart(UChar32 c) {
318     /* same as u_isalpha() */
319     uint32_t props;
320     GET_PROPS(c, props);
321     return (UBool)((CAT_MASK(props)&(U_GC_L_MASK|U_GC_NL_MASK))!=0);
322 }
323 
324 /* Checks if the Unicode character can be a Unicode identifier part other than starting the
325  identifier.*/
326 U_CAPI UBool U_EXPORT2
u_isIDPart(UChar32 c)327 u_isIDPart(UChar32 c) {
328     uint32_t props;
329     GET_PROPS(c, props);
330     return (UBool)(
331            (CAT_MASK(props)&
332             (U_GC_ND_MASK|U_GC_NL_MASK|
333              U_GC_L_MASK|
334              U_GC_PC_MASK|U_GC_MC_MASK|U_GC_MN_MASK)
335            )!=0 ||
336            u_isIDIgnorable(c));
337 }
338 
339 /*Checks if the Unicode character can be ignorable in a Java or Unicode identifier.*/
340 U_CAPI UBool U_EXPORT2
u_isIDIgnorable(UChar32 c)341 u_isIDIgnorable(UChar32 c) {
342     if(c<=0x9f) {
343         return u_isISOControl(c) && !IS_THAT_ASCII_CONTROL_SPACE(c);
344     } else {
345         uint32_t props;
346         GET_PROPS(c, props);
347         return (UBool)(GET_CATEGORY(props)==U_FORMAT_CHAR);
348     }
349 }
350 
351 /*Checks if the Unicode character can start a Java identifier.*/
352 U_CAPI UBool U_EXPORT2
u_isJavaIDStart(UChar32 c)353 u_isJavaIDStart(UChar32 c) {
354     uint32_t props;
355     GET_PROPS(c, props);
356     return (UBool)((CAT_MASK(props)&(U_GC_L_MASK|U_GC_SC_MASK|U_GC_PC_MASK))!=0);
357 }
358 
359 /*Checks if the Unicode character can be a Java identifier part other than starting the
360  * identifier.
361  */
362 U_CAPI UBool U_EXPORT2
u_isJavaIDPart(UChar32 c)363 u_isJavaIDPart(UChar32 c) {
364     uint32_t props;
365     GET_PROPS(c, props);
366     return (UBool)(
367            (CAT_MASK(props)&
368             (U_GC_ND_MASK|U_GC_NL_MASK|
369              U_GC_L_MASK|
370              U_GC_SC_MASK|U_GC_PC_MASK|
371              U_GC_MC_MASK|U_GC_MN_MASK)
372            )!=0 ||
373            u_isIDIgnorable(c));
374 }
375 
376 U_CAPI int32_t U_EXPORT2
u_charDigitValue(UChar32 c)377 u_charDigitValue(UChar32 c) {
378     uint32_t props;
379     int32_t value;
380     GET_PROPS(c, props);
381     value=(int32_t)GET_NUMERIC_TYPE_VALUE(props)-UPROPS_NTV_DECIMAL_START;
382     if(value<=9) {
383         return value;
384     } else {
385         return -1;
386     }
387 }
388 
389 U_CAPI double U_EXPORT2
u_getNumericValue(UChar32 c)390 u_getNumericValue(UChar32 c) {
391     uint32_t props;
392     int32_t ntv;
393     GET_PROPS(c, props);
394     ntv=(int32_t)GET_NUMERIC_TYPE_VALUE(props);
395 
396     if(ntv==UPROPS_NTV_NONE) {
397         return U_NO_NUMERIC_VALUE;
398     } else if(ntv<UPROPS_NTV_DIGIT_START) {
399         /* decimal digit */
400         return ntv-UPROPS_NTV_DECIMAL_START;
401     } else if(ntv<UPROPS_NTV_NUMERIC_START) {
402         /* other digit */
403         return ntv-UPROPS_NTV_DIGIT_START;
404     } else if(ntv<UPROPS_NTV_FRACTION_START) {
405         /* small integer */
406         return ntv-UPROPS_NTV_NUMERIC_START;
407     } else if(ntv<UPROPS_NTV_LARGE_START) {
408         /* fraction */
409         int32_t numerator=(ntv>>4)-12;
410         int32_t denominator=(ntv&0xf)+1;
411         return (double)numerator/denominator;
412     } else if(ntv<UPROPS_NTV_BASE60_START) {
413         /* large, single-significant-digit integer */
414         double numValue;
415         int32_t mant=(ntv>>5)-14;
416         int32_t exp=(ntv&0x1f)+2;
417         numValue=mant;
418 
419         /* multiply by 10^exp without math.h */
420         while(exp>=4) {
421             numValue*=10000.;
422             exp-=4;
423         }
424         switch(exp) {
425         case 3:
426             numValue*=1000.;
427             break;
428         case 2:
429             numValue*=100.;
430             break;
431         case 1:
432             numValue*=10.;
433             break;
434         case 0:
435         default:
436             break;
437         }
438 
439         return numValue;
440     } else if(ntv<UPROPS_NTV_FRACTION20_START) {
441         /* sexagesimal (base 60) integer */
442         int32_t numValue=(ntv>>2)-0xbf;
443         int32_t exp=(ntv&3)+1;
444 
445         switch(exp) {
446         case 4:
447             numValue*=60*60*60*60;
448             break;
449         case 3:
450             numValue*=60*60*60;
451             break;
452         case 2:
453             numValue*=60*60;
454             break;
455         case 1:
456             numValue*=60;
457             break;
458         case 0:
459         default:
460             break;
461         }
462 
463         return numValue;
464     } else if(ntv<UPROPS_NTV_RESERVED_START) {
465         // fraction-20 e.g. 3/80
466         int32_t frac20=ntv-UPROPS_NTV_FRACTION20_START;  // 0..0x17
467         int32_t numerator=2*(frac20&3)+1;
468         int32_t denominator=20<<(frac20>>2);
469         return (double)numerator/denominator;
470     } else {
471         /* reserved */
472         return U_NO_NUMERIC_VALUE;
473     }
474 }
475 
476 U_CAPI int32_t U_EXPORT2
u_digit(UChar32 ch,int8_t radix)477 u_digit(UChar32 ch, int8_t radix) {
478     int8_t value;
479     if((uint8_t)(radix-2)<=(36-2)) {
480         value=(int8_t)u_charDigitValue(ch);
481         if(value<0) {
482             /* ch is not a decimal digit, try latin letters */
483             if(ch>=0x61 && ch<=0x7A) {
484                 value=(int8_t)(ch-0x57);  /* ch - 'a' + 10 */
485             } else if(ch>=0x41 && ch<=0x5A) {
486                 value=(int8_t)(ch-0x37);  /* ch - 'A' + 10 */
487             } else if(ch>=0xFF41 && ch<=0xFF5A) {
488                 value=(int8_t)(ch-0xFF37);  /* fullwidth ASCII a-z */
489             } else if(ch>=0xFF21 && ch<=0xFF3A) {
490                 value=(int8_t)(ch-0xFF17);  /* fullwidth ASCII A-Z */
491             }
492         }
493     } else {
494         value=-1;   /* invalid radix */
495     }
496     return (int8_t)((value<radix) ? value : -1);
497 }
498 
499 U_CAPI UChar32 U_EXPORT2
u_forDigit(int32_t digit,int8_t radix)500 u_forDigit(int32_t digit, int8_t radix) {
501     if((uint8_t)(radix-2)>(36-2) || (uint32_t)digit>=(uint32_t)radix) {
502         return 0;
503     } else if(digit<10) {
504         return (UChar32)(0x30+digit);
505     } else {
506         return (UChar32)((0x61-10)+digit);
507     }
508 }
509 
510 /* miscellaneous, and support for uprops.cpp -------------------------------- */
511 
512 U_CAPI void U_EXPORT2
u_getUnicodeVersion(UVersionInfo versionArray)513 u_getUnicodeVersion(UVersionInfo versionArray) {
514     if(versionArray!=NULL) {
515         uprv_memcpy(versionArray, dataVersion, U_MAX_VERSION_LENGTH);
516     }
517 }
518 
519 U_CFUNC uint32_t
u_getMainProperties(UChar32 c)520 u_getMainProperties(UChar32 c) {
521     uint32_t props;
522     GET_PROPS(c, props);
523     return props;
524 }
525 
526 U_CFUNC uint32_t
u_getUnicodeProperties(UChar32 c,int32_t column)527 u_getUnicodeProperties(UChar32 c, int32_t column) {
528     U_ASSERT(column>=0);
529     if(column>=propsVectorsColumns) {
530         return 0;
531     } else {
532         uint16_t vecIndex=UTRIE2_GET16(&propsVectorsTrie, c);
533         return propsVectors[vecIndex+column];
534     }
535 }
536 
537 U_CFUNC int32_t
uprv_getMaxValues(int32_t column)538 uprv_getMaxValues(int32_t column) {
539     switch(column) {
540     case 0:
541         return indexes[UPROPS_MAX_VALUES_INDEX];
542     case 2:
543         return indexes[UPROPS_MAX_VALUES_2_INDEX];
544     default:
545         return 0;
546     }
547 }
548 
549 U_CAPI void U_EXPORT2
u_charAge(UChar32 c,UVersionInfo versionArray)550 u_charAge(UChar32 c, UVersionInfo versionArray) {
551     if(versionArray!=NULL) {
552         uint32_t version=u_getUnicodeProperties(c, 0)>>UPROPS_AGE_SHIFT;
553         versionArray[0]=(uint8_t)(version>>4);
554         versionArray[1]=(uint8_t)(version&0xf);
555         versionArray[2]=versionArray[3]=0;
556     }
557 }
558 
559 U_CAPI UScriptCode U_EXPORT2
uscript_getScript(UChar32 c,UErrorCode * pErrorCode)560 uscript_getScript(UChar32 c, UErrorCode *pErrorCode) {
561     uint32_t scriptX;
562     if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
563         return USCRIPT_INVALID_CODE;
564     }
565     if((uint32_t)c>0x10ffff) {
566         *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
567         return USCRIPT_INVALID_CODE;
568     }
569     scriptX=u_getUnicodeProperties(c, 0)&UPROPS_SCRIPT_X_MASK;
570     if(scriptX<UPROPS_SCRIPT_X_WITH_COMMON) {
571         return (UScriptCode)scriptX;
572     } else if(scriptX<UPROPS_SCRIPT_X_WITH_INHERITED) {
573         return USCRIPT_COMMON;
574     } else if(scriptX<UPROPS_SCRIPT_X_WITH_OTHER) {
575         return USCRIPT_INHERITED;
576     } else {
577         return (UScriptCode)scriptExtensions[scriptX&UPROPS_SCRIPT_MASK];
578     }
579 }
580 
581 U_CAPI UBool U_EXPORT2
uscript_hasScript(UChar32 c,UScriptCode sc)582 uscript_hasScript(UChar32 c, UScriptCode sc) {
583     const uint16_t *scx;
584     uint32_t scriptX=u_getUnicodeProperties(c, 0)&UPROPS_SCRIPT_X_MASK;
585     if(scriptX<UPROPS_SCRIPT_X_WITH_COMMON) {
586         return sc==(UScriptCode)scriptX;
587     }
588 
589     scx=scriptExtensions+(scriptX&UPROPS_SCRIPT_MASK);
590     if(scriptX>=UPROPS_SCRIPT_X_WITH_OTHER) {
591         scx=scriptExtensions+scx[1];
592     }
593     if(sc>=USCRIPT_CODE_LIMIT) {
594         /* Guard against bogus input that would make us go past the Script_Extensions terminator. */
595         return FALSE;
596     }
597     while(sc>*scx) {
598         ++scx;
599     }
600     return sc==(*scx&0x7fff);
601 }
602 
603 U_CAPI int32_t U_EXPORT2
uscript_getScriptExtensions(UChar32 c,UScriptCode * scripts,int32_t capacity,UErrorCode * pErrorCode)604 uscript_getScriptExtensions(UChar32 c,
605                             UScriptCode *scripts, int32_t capacity,
606                             UErrorCode *pErrorCode) {
607     uint32_t scriptX;
608     int32_t length;
609     const uint16_t *scx;
610     uint16_t sx;
611     if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
612         return 0;
613     }
614     if(capacity<0 || (capacity>0 && scripts==NULL)) {
615         *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
616         return 0;
617     }
618     scriptX=u_getUnicodeProperties(c, 0)&UPROPS_SCRIPT_X_MASK;
619     if(scriptX<UPROPS_SCRIPT_X_WITH_COMMON) {
620         if(capacity==0) {
621             *pErrorCode=U_BUFFER_OVERFLOW_ERROR;
622         } else {
623             scripts[0]=(UScriptCode)scriptX;
624         }
625         return 1;
626     }
627 
628     scx=scriptExtensions+(scriptX&UPROPS_SCRIPT_MASK);
629     if(scriptX>=UPROPS_SCRIPT_X_WITH_OTHER) {
630         scx=scriptExtensions+scx[1];
631     }
632     length=0;
633     do {
634         sx=*scx++;
635         if(length<capacity) {
636             scripts[length]=(UScriptCode)(sx&0x7fff);
637         }
638         ++length;
639     } while(sx<0x8000);
640     if(length>capacity) {
641         *pErrorCode=U_BUFFER_OVERFLOW_ERROR;
642     }
643     return length;
644 }
645 
646 U_CAPI UBlockCode U_EXPORT2
ublock_getCode(UChar32 c)647 ublock_getCode(UChar32 c) {
648     return (UBlockCode)((u_getUnicodeProperties(c, 0)&UPROPS_BLOCK_MASK)>>UPROPS_BLOCK_SHIFT);
649 }
650 
651 /* property starts for UnicodeSet ------------------------------------------- */
652 
653 static UBool U_CALLCONV
_enumPropertyStartsRange(const void * context,UChar32 start,UChar32 end,uint32_t value)654 _enumPropertyStartsRange(const void *context, UChar32 start, UChar32 end, uint32_t value) {
655     /* add the start code point to the USet */
656     const USetAdder *sa=(const USetAdder *)context;
657     sa->add(sa->set, start);
658     (void)end;
659     (void)value;
660     return TRUE;
661 }
662 
663 #define USET_ADD_CP_AND_NEXT(sa, cp) sa->add(sa->set, cp); sa->add(sa->set, cp+1)
664 
665 U_CFUNC void U_EXPORT2
uchar_addPropertyStarts(const USetAdder * sa,UErrorCode * pErrorCode)666 uchar_addPropertyStarts(const USetAdder *sa, UErrorCode *pErrorCode) {
667     if(U_FAILURE(*pErrorCode)) {
668         return;
669     }
670 
671     /* add the start code point of each same-value range of the main trie */
672     utrie2_enum(&propsTrie, NULL, _enumPropertyStartsRange, sa);
673 
674     /* add code points with hardcoded properties, plus the ones following them */
675 
676     /* add for u_isblank() */
677     USET_ADD_CP_AND_NEXT(sa, TAB);
678 
679     /* add for IS_THAT_CONTROL_SPACE() */
680     sa->add(sa->set, CR+1); /* range TAB..CR */
681     sa->add(sa->set, 0x1c);
682     sa->add(sa->set, 0x1f+1);
683     USET_ADD_CP_AND_NEXT(sa, NL);
684 
685     /* add for u_isIDIgnorable() what was not added above */
686     sa->add(sa->set, DEL); /* range DEL..NBSP-1, NBSP added below */
687     sa->add(sa->set, HAIRSP);
688     sa->add(sa->set, RLM+1);
689     sa->add(sa->set, INHSWAP);
690     sa->add(sa->set, NOMDIG+1);
691     USET_ADD_CP_AND_NEXT(sa, ZWNBSP);
692 
693     /* add no-break spaces for u_isWhitespace() what was not added above */
694     USET_ADD_CP_AND_NEXT(sa, NBSP);
695     USET_ADD_CP_AND_NEXT(sa, FIGURESP);
696     USET_ADD_CP_AND_NEXT(sa, NNBSP);
697 
698     /* add for u_digit() */
699     sa->add(sa->set, U_a);
700     sa->add(sa->set, U_z+1);
701     sa->add(sa->set, U_A);
702     sa->add(sa->set, U_Z+1);
703     sa->add(sa->set, U_FW_a);
704     sa->add(sa->set, U_FW_z+1);
705     sa->add(sa->set, U_FW_A);
706     sa->add(sa->set, U_FW_Z+1);
707 
708     /* add for u_isxdigit() */
709     sa->add(sa->set, U_f+1);
710     sa->add(sa->set, U_F+1);
711     sa->add(sa->set, U_FW_f+1);
712     sa->add(sa->set, U_FW_F+1);
713 
714     /* add for UCHAR_DEFAULT_IGNORABLE_CODE_POINT what was not added above */
715     sa->add(sa->set, WJ); /* range WJ..NOMDIG */
716     sa->add(sa->set, 0xfff0);
717     sa->add(sa->set, 0xfffb+1);
718     sa->add(sa->set, 0xe0000);
719     sa->add(sa->set, 0xe0fff+1);
720 
721     /* add for UCHAR_GRAPHEME_BASE and others */
722     USET_ADD_CP_AND_NEXT(sa, CGJ);
723 }
724 
725 U_CFUNC void U_EXPORT2
upropsvec_addPropertyStarts(const USetAdder * sa,UErrorCode * pErrorCode)726 upropsvec_addPropertyStarts(const USetAdder *sa, UErrorCode *pErrorCode) {
727     if(U_FAILURE(*pErrorCode)) {
728         return;
729     }
730 
731     /* add the start code point of each same-value range of the properties vectors trie */
732     utrie2_enum(&propsVectorsTrie, NULL, _enumPropertyStartsRange, sa);
733 }
734