static void
makeUnfoldData() {
static const UChar
iDot[2]= { 0x69, 0x307 };
UChar *p, *q;
int32_t i, j, k;
UErrorCode errorCode;
/*
* add a case folding that we missed because it's conditional:
* 0130; F; 0069 0307; # LATIN CAPITAL LETTER I WITH DOT ABOVE
*/
addUnfolding(0x130, iDot, 2);
/* sort the data */
errorCode=U_ZERO_ERROR;
uprv_sortArray(unfold+UGENCASE_UNFOLD_WIDTH, unfoldRows, UGENCASE_UNFOLD_WIDTH*2,
compareUnfold, NULL, FALSE, &errorCode);
/* make unique-string rows by merging adjacent ones' code point columns */
/* make p point to row i-1 */
p=(UChar *)unfold+UGENCASE_UNFOLD_WIDTH;
for(i=1; i<unfoldRows;) {
if(0==u_memcmp(p, p+UGENCASE_UNFOLD_WIDTH, UGENCASE_UNFOLD_STRING_WIDTH)) {
/* concatenate code point columns */
q=p+UGENCASE_UNFOLD_STRING_WIDTH;
for(j=1; j<UGENCASE_UNFOLD_CP_WIDTH && q[j]!=0; ++j) {}
for(k=0; k<UGENCASE_UNFOLD_CP_WIDTH && q[UGENCASE_UNFOLD_WIDTH+k]!=0; ++j, ++k) {
q[j]=q[UGENCASE_UNFOLD_WIDTH+k];
}
if(j>UGENCASE_UNFOLD_CP_WIDTH) {
fprintf(stderr, "gencase error: too many code points in unfold[]: %ld>%d=UGENCASE_UNFOLD_CP_WIDTH\n",
(long)j, UGENCASE_UNFOLD_CP_WIDTH);
exit(U_BUFFER_OVERFLOW_ERROR);
}
/* move following rows up one */
--unfoldRows;
unfoldTop-=UGENCASE_UNFOLD_WIDTH;
u_memmove(p+UGENCASE_UNFOLD_WIDTH, p+UGENCASE_UNFOLD_WIDTH*2, (unfoldRows-i)*UGENCASE_UNFOLD_WIDTH);
} else {
p+=UGENCASE_UNFOLD_WIDTH;
++i;
}
}
unfold[UCASE_UNFOLD_ROWS]=(UChar)unfoldRows;
if(beVerbose) {
puts("unfold data:");
p=(UChar *)unfold;
for(i=0; i<unfoldRows; ++i) {
p+=UGENCASE_UNFOLD_WIDTH;
printf("[%2d] %04x %04x %04x <- %04x %04x\n",
(int)i, p[0], p[1], p[2], p[3], p[4]);
}
}
}
void
CasePropsBuilder::setProps(const UniProps &props, const UnicodeSet &newValues,
UErrorCode &errorCode) {
if(U_FAILURE(errorCode) || newValues.containsNone(relevantProps)) { return; }
UChar32 start=props.start;
UChar32 end=props.end;
/* default: map to self */
int32_t delta=0;
uint32_t type;
if(props.binProps[UCHAR_LOWERCASE]) {
type=UCASE_LOWER;
} else if(props.binProps[UCHAR_UPPERCASE]) {
type=UCASE_UPPER;
} else if(props.getIntProp(UCHAR_GENERAL_CATEGORY)==U_TITLECASE_LETTER) {
type=UCASE_TITLE;
} else {
type=UCASE_NONE;
}
uint32_t value=type;
UBool hasMapping=FALSE;
if(props.suc>=0) {
/* uppercase mapping as delta if the character is lowercase */
hasMapping=TRUE;
if(type==UCASE_LOWER) {
delta=props.suc-start;
} else {
value|=UCASE_EXCEPTION;
}
}
if(props.slc>=0) {
/* lowercase mapping as delta if the character is uppercase or titlecase */
hasMapping=TRUE;
if(type>=UCASE_UPPER) {
delta=props.slc-start;
} else {
value|=UCASE_EXCEPTION;
}
}
if(props.stc>=0) {
hasMapping=TRUE;
}
if(props.suc!=props.stc) {
value|=UCASE_EXCEPTION;
}
if(!props.lc.isEmpty() || !props.uc.isEmpty() || !props.tc.isEmpty() ||
newValues.contains(PPUCD_CONDITIONAL_CASE_MAPPINGS)
) {
hasMapping=TRUE;
value|=UCASE_EXCEPTION;
}
if( (props.scf>=0 && props.scf!=props.slc) ||
(!props.cf.isEmpty() && props.cf!=UnicodeString(props.scf)) ||
newValues.contains(PPUCD_TURKIC_CASE_FOLDING)
) {
hasMapping=TRUE;
value|=UCASE_EXCEPTION;
}
// Simple case folding falls back to simple lowercasing.
// If there is no case folding but there is a lowercase mapping,
// then add a case folding mapping to the code point.
// For example: Cherokee uppercase syllables since Unicode 8.
// (Full case folding falls back to simple case folding,
// not to full lowercasing, so we need not also handle it specially
// for such cases.)
UChar32 scf=props.scf;
if(scf<0 && props.slc>=0) {
scf=start;
hasMapping=TRUE;
value|=UCASE_EXCEPTION;
}
if(delta<UCASE_MIN_DELTA || UCASE_MAX_DELTA<delta) {
value|=UCASE_EXCEPTION;
}
if(props.binProps[UCHAR_SOFT_DOTTED]) {
value|=UCASE_SOFT_DOTTED;
}
int32_t cc=props.getIntProp(UCHAR_CANONICAL_COMBINING_CLASS);
if(cc!=0) {
if(props.binProps[UCHAR_SOFT_DOTTED]) {
fprintf(stderr, "genprops error: a soft-dotted character has ccc!=0\n");
errorCode=U_ILLEGAL_ARGUMENT_ERROR;
return;
}
if(cc==230) {
value|=UCASE_ABOVE;
} else {
value|=UCASE_OTHER_ACCENT;
}
}
if(props.binProps[UCHAR_CASE_IGNORABLE]) {
value|=UCASE_IGNORABLE;
}
if((hasMapping || (value&UCASE_EXCEPTION)) && start!=end) {
fprintf(stderr,
"genprops error: range %04lX..%04lX has case mappings "
"or reasons for data structure exceptions\n",
(long)start, (long)end);
errorCode=U_ILLEGAL_ARGUMENT_ERROR;
return;
}
/* handle exceptions */
if(value&UCASE_EXCEPTION) {
/* simply store exceptions for later processing and encoding */
if(excPropsCount==MAX_EXC_COUNT) {
fprintf(stderr, "genprops error: casepropsbuilder: too many exceptions\n");
errorCode=U_INDEX_OUTOFBOUNDS_ERROR;
return;
}
ExcProps *newExcProps=new ExcProps(props);
if(newExcProps==NULL) {
fprintf(stderr,
"genprops error: casepropsbuilder out of memory allocating "
"exceptions properties\n");
errorCode=U_MEMORY_ALLOCATION_ERROR;
return;
}
newExcProps->props.scf=scf;
newExcProps->hasConditionalCaseMappings=newValues.contains(PPUCD_CONDITIONAL_CASE_MAPPINGS);
newExcProps->hasTurkicCaseFolding=newValues.contains(PPUCD_TURKIC_CASE_FOLDING);
value|=(uint32_t)excPropsCount<<UGENCASE_EXC_SHIFT;
excProps[excPropsCount++]=newExcProps;
} else {
/* store the simple case mapping delta */
value|=((uint32_t)delta<<UCASE_DELTA_SHIFT)&UCASE_DELTA_MASK;
}
utrie2_setRange32(pTrie, start, end, value, TRUE, &errorCode);
if(U_FAILURE(errorCode)) {
fprintf(stderr, "genprops error: unable to set case mapping values: %s\n",
u_errorName(errorCode));
return;
}
if(hasMapping) {
/* update the case-sensitive set */
caseSensitive.add(start);
if(scf>=0) { caseSensitive.add(scf); }
if(props.slc>=0) { caseSensitive.add(props.slc); }
if(props.suc>=0) { caseSensitive.add(props.suc); }
if(props.stc>=0) { caseSensitive.add(props.stc); }
caseSensitive.addAll(props.cf);
caseSensitive.addAll(props.lc);
caseSensitive.addAll(props.uc);
caseSensitive.addAll(props.tc);
/* update maxFullLength */
if(props.cf.length()>maxFullLength) { maxFullLength=props.cf.length(); }
if(props.lc.length()>maxFullLength) { maxFullLength=props.lc.length(); }
if(props.uc.length()>maxFullLength) { maxFullLength=props.uc.length(); }
if(props.tc.length()>maxFullLength) { maxFullLength=props.tc.length(); }
}
/* add the multi-character case folding to the "unfold" data */
if(props.cf.hasMoreChar32Than(0, 0x7fffffff, 1)) {
addUnfolding(start, props.cf, errorCode);
}
}
extern void
setProps(Props *p) {
UErrorCode errorCode;
uint32_t value, oldValue;
int32_t delta;
UBool isCaseIgnorable;
/* get the non-UnicodeData.txt properties */
value=oldValue=upvec_getValue(pv, p->code, 0);
/* default: map to self */
delta=0;
if(p->gc==U_TITLECASE_LETTER) {
/* the Titlecase property is read late, from UnicodeData.txt */
value|=UCASE_TITLE;
}
if(p->upperCase!=0) {
/* uppercase mapping as delta if the character is lowercase */
if((value&UCASE_TYPE_MASK)==UCASE_LOWER) {
delta=p->upperCase-p->code;
} else {
value|=UCASE_EXCEPTION;
}
}
if(p->lowerCase!=0) {
/* lowercase mapping as delta if the character is uppercase or titlecase */
if((value&UCASE_TYPE_MASK)>=UCASE_UPPER) {
delta=p->lowerCase-p->code;
} else {
value|=UCASE_EXCEPTION;
}
}
if(p->upperCase!=p->titleCase) {
value|=UCASE_EXCEPTION;
}
if(p->closure[0]!=0) {
value|=UCASE_EXCEPTION;
}
if(p->specialCasing!=NULL) {
value|=UCASE_EXCEPTION;
}
if(p->caseFolding!=NULL) {
value|=UCASE_EXCEPTION;
}
if(delta<UCASE_MIN_DELTA || UCASE_MAX_DELTA<delta) {
value|=UCASE_EXCEPTION;
}
if(p->cc!=0) {
if(value&UCASE_DOT_MASK) {
fprintf(stderr, "gencase: a soft-dotted character has cc!=0\n");
exit(U_INTERNAL_PROGRAM_ERROR);
}
if(p->cc==230) {
value|=UCASE_ABOVE;
} else {
value|=UCASE_OTHER_ACCENT;
}
}
/* encode case-ignorable as delta==1 on uncased characters */
isCaseIgnorable=FALSE;
if((value&UCASE_TYPE_MASK)==UCASE_NONE) {
if(ucdVersion>=UNI_4_1) {
/*
* Unicode 4.1 and up: (D47a) Word_Break=MidLetter or Mn, Me, Cf, Lm, Sk
* Unicode 5.1 and up: Word_Break=(MidLetter or MidNumLet) or Mn, Me, Cf, Lm, Sk
* The UGENCASE_IS_MID_LETTER_SHIFT bit is set for both WB=MidLetter and WB=MidNumLet.
*/
if(
(U_MASK(p->gc)&(U_GC_MN_MASK|U_GC_ME_MASK|U_GC_CF_MASK|U_GC_LM_MASK|U_GC_SK_MASK))!=0 ||
(upvec_getValue(pv, p->code, 1)&U_MASK(UGENCASE_IS_MID_LETTER_SHIFT))!=0
) {
isCaseIgnorable=TRUE;
}
} else {
/* before Unicode 4.1: Mn, Me, Cf, Lm, Sk or 0027 or 00AD or 2019 */
if(
(U_MASK(p->gc)&(U_GC_MN_MASK|U_GC_ME_MASK|U_GC_CF_MASK|U_GC_LM_MASK|U_GC_SK_MASK))!=0 ||
p->code==0x27 || p->code==0xad || p->code==0x2019
) {
isCaseIgnorable=TRUE;
}
}
}
if(isCaseIgnorable && p->code!=0x307) {
/*
* We use one of the delta/exception bits, which works because we only
* store the case-ignorable flag for uncased characters.
* There is no delta for uncased characters (see checks above).
* If there is an exception for an uncased, case-ignorable character
* (although there should not be any case mappings if it's uncased)
* then we have a problem.
* There is one character which is case-ignorable but has an exception:
* U+0307 is uncased, Mn, has conditional special casing and
* is therefore handled in code instead.
*/
if(value&UCASE_EXCEPTION) {
fprintf(stderr, "gencase error: unable to encode case-ignorable for U+%04lx with exceptions\n",
(unsigned long)p->code);
exit(U_INTERNAL_PROGRAM_ERROR);
}
delta=1;
}
/* handle exceptions */
if(value&UCASE_EXCEPTION) {
/* simply store exceptions for later processing and encoding */
value|=(uint32_t)exceptionsCount<<UGENCASE_EXC_SHIFT;
uprv_memcpy(excProps+exceptionsCount, p, sizeof(*p));
if(++exceptionsCount==MAX_EXC_COUNT) {
fprintf(stderr, "gencase: too many exceptions\n");
exit(U_INDEX_OUTOFBOUNDS_ERROR);
}
} else {
/* store the simple case mapping delta */
value|=((uint32_t)delta<<UCASE_DELTA_SHIFT)&UCASE_DELTA_MASK;
}
errorCode=U_ZERO_ERROR;
if(value!=oldValue) {
upvec_setValue(pv, p->code, p->code, 0, value, 0xffffffff, &errorCode);
if(U_FAILURE(errorCode)) {
fprintf(stderr, "gencase error: unable to set case mapping values, code: %s\n",
u_errorName(errorCode));
exit(errorCode);
}
}
/* add the multi-character case folding to the "unfold" data */
if(p->caseFolding!=NULL) {
int32_t length=p->caseFolding->full[0];
if(length>1 && u_strHasMoreChar32Than(p->caseFolding->full+1, length, 1)) {
addUnfolding(p->code, p->caseFolding->full+1, length);
}
}
}