+// Copyright (C) 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
/*
*******************************************************************************
*
if(result<0) {
/* (not) original code point */
c=~result;
- length=-1;
+ length=U16_LENGTH(c);
} else if(result<=UCASE_MAX_STRING_LENGTH) {
c=U_SENTINEL;
length=result;
} else {
c=result;
- length=-1;
+ length=U16_LENGTH(c);
+ }
+ if(length>(INT32_MAX-destIndex)) {
+ return -1; // integer overflow
}
if(destIndex<destCapacity) {
/* append the result */
- if(length<0) {
+ if(c>=0) {
/* code point */
UBool isError=FALSE;
U16_APPEND(dest, destIndex, destCapacity, c, isError);
if(isError) {
/* overflow, nothing written */
- destIndex+=U16_LENGTH(c);
+ destIndex+=length;
}
} else {
/* string */
}
} else {
/* preflight */
- if(length<0) {
- destIndex+=U16_LENGTH(c);
- } else {
- destIndex+=length;
+ destIndex+=length;
+ }
+ return destIndex;
+}
+
+static inline int32_t
+appendUChar(UChar *dest, int32_t destIndex, int32_t destCapacity, UChar c) {
+ if(destIndex<destCapacity) {
+ dest[destIndex]=c;
+ } else if(destIndex==INT32_MAX) {
+ return -1; // integer overflow
+ }
+ return destIndex+1;
+}
+
+static inline int32_t
+appendString(UChar *dest, int32_t destIndex, int32_t destCapacity,
+ const UChar *s, int32_t length) {
+ if(length>0) {
+ if(length>(INT32_MAX-destIndex)) {
+ return -1; // integer overflow
}
+ if((destIndex+length)<=destCapacity) {
+ u_memcpy(dest+destIndex, s, length);
+ }
+ destIndex+=length;
}
return destIndex;
}
dest[destIndex++]=(UChar)c2;
} else {
destIndex=appendResult(dest, destIndex, destCapacity, c, s);
+ if(destIndex<0) {
+ *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR;
+ return 0;
+ }
}
}
UErrorCode *pErrorCode) {
const UChar *s;
UChar32 c;
- int32_t prev, titleStart, titleLimit, idx, destIndex, length;
+ int32_t prev, titleStart, titleLimit, idx, destIndex;
UBool isFirstIndex;
if(U_FAILURE(*pErrorCode)) {
break; /* cased letter at [titleStart..titleLimit[ */
}
}
- length=titleStart-prev;
- if(length>0) {
- if((destIndex+length)<=destCapacity) {
- uprv_memcpy(dest+destIndex, src+prev, length*U_SIZEOF_UCHAR);
- }
- destIndex+=length;
+ destIndex=appendString(dest, destIndex, destCapacity, src+prev, titleStart-prev);
+ if(destIndex<0) {
+ *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR;
+ return 0;
}
}
csc.cpLimit=titleLimit;
c=ucase_toFullTitle(csm->csp, c, utf16_caseContextIterator, &csc, &s, csm->locale, &locCache);
destIndex=appendResult(dest, destIndex, destCapacity, c, s);
+ if(destIndex<0) {
+ *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR;
+ return 0;
+ }
/* Special case Dutch IJ titlecasing */
- if ( titleStart+1 < idx &&
- ucase_getCaseLocale(csm->locale,&locCache) == UCASE_LOC_DUTCH &&
- ( src[titleStart] == (UChar32) 0x0049 || src[titleStart] == (UChar32) 0x0069 ) &&
- ( src[titleStart+1] == (UChar32) 0x004A || src[titleStart+1] == (UChar32) 0x006A )) {
- c=(UChar32) 0x004A;
- destIndex=appendResult(dest, destIndex, destCapacity, c, s);
- titleLimit++;
+ if (titleStart+1 < idx &&
+ ucase_getCaseLocale(csm->locale,&locCache) == UCASE_LOC_DUTCH &&
+ (src[titleStart] == 0x0049 || src[titleStart] == 0x0069) &&
+ (src[titleStart+1] == 0x004A || src[titleStart+1] == 0x006A)) {
+ destIndex=appendUChar(dest, destIndex, destCapacity, 0x004A);
+ if(destIndex<0) {
+ *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR;
+ return 0;
+ }
+ titleLimit++;
}
/* lowercase [titleLimit..index[ */
src, &csc,
titleLimit, idx,
pErrorCode);
+ if(U_FAILURE(*pErrorCode)) {
+ return destIndex;
+ }
} else {
/* Optionally just copy the rest of the word unchanged. */
- length=idx-titleLimit;
- if((destIndex+length)<=destCapacity) {
- uprv_memcpy(dest+destIndex, src+titleLimit, length*U_SIZEOF_UCHAR);
+ destIndex=appendString(dest, destIndex, destCapacity, src+titleLimit, idx-titleLimit);
+ if(destIndex<0) {
+ *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR;
+ return 0;
}
- destIndex+=length;
}
}
}
#endif // !UCONFIG_NO_BREAK_ITERATION
+U_NAMESPACE_BEGIN
+namespace GreekUpper {
+
+// Data generated by prototype code, see
+// http://site.icu-project.org/design/case/greek-upper
+// TODO: Move this data into ucase.icu.
+static const uint16_t data0370[] = {
+ // U+0370..03FF
+ 0x0370,
+ 0x0370,
+ 0x0372,
+ 0x0372,
+ 0,
+ 0,
+ 0x0376,
+ 0x0376,
+ 0,
+ 0,
+ 0x037A,
+ 0x03FD,
+ 0x03FE,
+ 0x03FF,
+ 0,
+ 0x037F,
+ 0,
+ 0,
+ 0,
+ 0,
+ 0,
+ 0,
+ 0x0391 | HAS_VOWEL | HAS_ACCENT,
+ 0,
+ 0x0395 | HAS_VOWEL | HAS_ACCENT,
+ 0x0397 | HAS_VOWEL | HAS_ACCENT,
+ 0x0399 | HAS_VOWEL | HAS_ACCENT,
+ 0,
+ 0x039F | HAS_VOWEL | HAS_ACCENT,
+ 0,
+ 0x03A5 | HAS_VOWEL | HAS_ACCENT,
+ 0x03A9 | HAS_VOWEL | HAS_ACCENT,
+ 0x0399 | HAS_VOWEL | HAS_ACCENT | HAS_DIALYTIKA,
+ 0x0391 | HAS_VOWEL,
+ 0x0392,
+ 0x0393,
+ 0x0394,
+ 0x0395 | HAS_VOWEL,
+ 0x0396,
+ 0x0397 | HAS_VOWEL,
+ 0x0398,
+ 0x0399 | HAS_VOWEL,
+ 0x039A,
+ 0x039B,
+ 0x039C,
+ 0x039D,
+ 0x039E,
+ 0x039F | HAS_VOWEL,
+ 0x03A0,
+ 0x03A1,
+ 0,
+ 0x03A3,
+ 0x03A4,
+ 0x03A5 | HAS_VOWEL,
+ 0x03A6,
+ 0x03A7,
+ 0x03A8,
+ 0x03A9 | HAS_VOWEL,
+ 0x0399 | HAS_VOWEL | HAS_DIALYTIKA,
+ 0x03A5 | HAS_VOWEL | HAS_DIALYTIKA,
+ 0x0391 | HAS_VOWEL | HAS_ACCENT,
+ 0x0395 | HAS_VOWEL | HAS_ACCENT,
+ 0x0397 | HAS_VOWEL | HAS_ACCENT,
+ 0x0399 | HAS_VOWEL | HAS_ACCENT,
+ 0x03A5 | HAS_VOWEL | HAS_ACCENT | HAS_DIALYTIKA,
+ 0x0391 | HAS_VOWEL,
+ 0x0392,
+ 0x0393,
+ 0x0394,
+ 0x0395 | HAS_VOWEL,
+ 0x0396,
+ 0x0397 | HAS_VOWEL,
+ 0x0398,
+ 0x0399 | HAS_VOWEL,
+ 0x039A,
+ 0x039B,
+ 0x039C,
+ 0x039D,
+ 0x039E,
+ 0x039F | HAS_VOWEL,
+ 0x03A0,
+ 0x03A1,
+ 0x03A3,
+ 0x03A3,
+ 0x03A4,
+ 0x03A5 | HAS_VOWEL,
+ 0x03A6,
+ 0x03A7,
+ 0x03A8,
+ 0x03A9 | HAS_VOWEL,
+ 0x0399 | HAS_VOWEL | HAS_DIALYTIKA,
+ 0x03A5 | HAS_VOWEL | HAS_DIALYTIKA,
+ 0x039F | HAS_VOWEL | HAS_ACCENT,
+ 0x03A5 | HAS_VOWEL | HAS_ACCENT,
+ 0x03A9 | HAS_VOWEL | HAS_ACCENT,
+ 0x03CF,
+ 0x0392,
+ 0x0398,
+ 0x03D2,
+ 0x03D2 | HAS_ACCENT,
+ 0x03D2 | HAS_DIALYTIKA,
+ 0x03A6,
+ 0x03A0,
+ 0x03CF,
+ 0x03D8,
+ 0x03D8,
+ 0x03DA,
+ 0x03DA,
+ 0x03DC,
+ 0x03DC,
+ 0x03DE,
+ 0x03DE,
+ 0x03E0,
+ 0x03E0,
+ 0,
+ 0,
+ 0,
+ 0,
+ 0,
+ 0,
+ 0,
+ 0,
+ 0,
+ 0,
+ 0,
+ 0,
+ 0,
+ 0,
+ 0x039A,
+ 0x03A1,
+ 0x03F9,
+ 0x037F,
+ 0x03F4,
+ 0x0395 | HAS_VOWEL,
+ 0,
+ 0x03F7,
+ 0x03F7,
+ 0x03F9,
+ 0x03FA,
+ 0x03FA,
+ 0x03FC,
+ 0x03FD,
+ 0x03FE,
+ 0x03FF,
+};
+
+static const uint16_t data1F00[] = {
+ // U+1F00..1FFF
+ 0x0391 | HAS_VOWEL,
+ 0x0391 | HAS_VOWEL,
+ 0x0391 | HAS_VOWEL | HAS_ACCENT,
+ 0x0391 | HAS_VOWEL | HAS_ACCENT,
+ 0x0391 | HAS_VOWEL | HAS_ACCENT,
+ 0x0391 | HAS_VOWEL | HAS_ACCENT,
+ 0x0391 | HAS_VOWEL | HAS_ACCENT,
+ 0x0391 | HAS_VOWEL | HAS_ACCENT,
+ 0x0391 | HAS_VOWEL,
+ 0x0391 | HAS_VOWEL,
+ 0x0391 | HAS_VOWEL | HAS_ACCENT,
+ 0x0391 | HAS_VOWEL | HAS_ACCENT,
+ 0x0391 | HAS_VOWEL | HAS_ACCENT,
+ 0x0391 | HAS_VOWEL | HAS_ACCENT,
+ 0x0391 | HAS_VOWEL | HAS_ACCENT,
+ 0x0391 | HAS_VOWEL | HAS_ACCENT,
+ 0x0395 | HAS_VOWEL,
+ 0x0395 | HAS_VOWEL,
+ 0x0395 | HAS_VOWEL | HAS_ACCENT,
+ 0x0395 | HAS_VOWEL | HAS_ACCENT,
+ 0x0395 | HAS_VOWEL | HAS_ACCENT,
+ 0x0395 | HAS_VOWEL | HAS_ACCENT,
+ 0,
+ 0,
+ 0x0395 | HAS_VOWEL,
+ 0x0395 | HAS_VOWEL,
+ 0x0395 | HAS_VOWEL | HAS_ACCENT,
+ 0x0395 | HAS_VOWEL | HAS_ACCENT,
+ 0x0395 | HAS_VOWEL | HAS_ACCENT,
+ 0x0395 | HAS_VOWEL | HAS_ACCENT,
+ 0,
+ 0,
+ 0x0397 | HAS_VOWEL,
+ 0x0397 | HAS_VOWEL,
+ 0x0397 | HAS_VOWEL | HAS_ACCENT,
+ 0x0397 | HAS_VOWEL | HAS_ACCENT,
+ 0x0397 | HAS_VOWEL | HAS_ACCENT,
+ 0x0397 | HAS_VOWEL | HAS_ACCENT,
+ 0x0397 | HAS_VOWEL | HAS_ACCENT,
+ 0x0397 | HAS_VOWEL | HAS_ACCENT,
+ 0x0397 | HAS_VOWEL,
+ 0x0397 | HAS_VOWEL,
+ 0x0397 | HAS_VOWEL | HAS_ACCENT,
+ 0x0397 | HAS_VOWEL | HAS_ACCENT,
+ 0x0397 | HAS_VOWEL | HAS_ACCENT,
+ 0x0397 | HAS_VOWEL | HAS_ACCENT,
+ 0x0397 | HAS_VOWEL | HAS_ACCENT,
+ 0x0397 | HAS_VOWEL | HAS_ACCENT,
+ 0x0399 | HAS_VOWEL,
+ 0x0399 | HAS_VOWEL,
+ 0x0399 | HAS_VOWEL | HAS_ACCENT,
+ 0x0399 | HAS_VOWEL | HAS_ACCENT,
+ 0x0399 | HAS_VOWEL | HAS_ACCENT,
+ 0x0399 | HAS_VOWEL | HAS_ACCENT,
+ 0x0399 | HAS_VOWEL | HAS_ACCENT,
+ 0x0399 | HAS_VOWEL | HAS_ACCENT,
+ 0x0399 | HAS_VOWEL,
+ 0x0399 | HAS_VOWEL,
+ 0x0399 | HAS_VOWEL | HAS_ACCENT,
+ 0x0399 | HAS_VOWEL | HAS_ACCENT,
+ 0x0399 | HAS_VOWEL | HAS_ACCENT,
+ 0x0399 | HAS_VOWEL | HAS_ACCENT,
+ 0x0399 | HAS_VOWEL | HAS_ACCENT,
+ 0x0399 | HAS_VOWEL | HAS_ACCENT,
+ 0x039F | HAS_VOWEL,
+ 0x039F | HAS_VOWEL,
+ 0x039F | HAS_VOWEL | HAS_ACCENT,
+ 0x039F | HAS_VOWEL | HAS_ACCENT,
+ 0x039F | HAS_VOWEL | HAS_ACCENT,
+ 0x039F | HAS_VOWEL | HAS_ACCENT,
+ 0,
+ 0,
+ 0x039F | HAS_VOWEL,
+ 0x039F | HAS_VOWEL,
+ 0x039F | HAS_VOWEL | HAS_ACCENT,
+ 0x039F | HAS_VOWEL | HAS_ACCENT,
+ 0x039F | HAS_VOWEL | HAS_ACCENT,
+ 0x039F | HAS_VOWEL | HAS_ACCENT,
+ 0,
+ 0,
+ 0x03A5 | HAS_VOWEL,
+ 0x03A5 | HAS_VOWEL,
+ 0x03A5 | HAS_VOWEL | HAS_ACCENT,
+ 0x03A5 | HAS_VOWEL | HAS_ACCENT,
+ 0x03A5 | HAS_VOWEL | HAS_ACCENT,
+ 0x03A5 | HAS_VOWEL | HAS_ACCENT,
+ 0x03A5 | HAS_VOWEL | HAS_ACCENT,
+ 0x03A5 | HAS_VOWEL | HAS_ACCENT,
+ 0,
+ 0x03A5 | HAS_VOWEL,
+ 0,
+ 0x03A5 | HAS_VOWEL | HAS_ACCENT,
+ 0,
+ 0x03A5 | HAS_VOWEL | HAS_ACCENT,
+ 0,
+ 0x03A5 | HAS_VOWEL | HAS_ACCENT,
+ 0x03A9 | HAS_VOWEL,
+ 0x03A9 | HAS_VOWEL,
+ 0x03A9 | HAS_VOWEL | HAS_ACCENT,
+ 0x03A9 | HAS_VOWEL | HAS_ACCENT,
+ 0x03A9 | HAS_VOWEL | HAS_ACCENT,
+ 0x03A9 | HAS_VOWEL | HAS_ACCENT,
+ 0x03A9 | HAS_VOWEL | HAS_ACCENT,
+ 0x03A9 | HAS_VOWEL | HAS_ACCENT,
+ 0x03A9 | HAS_VOWEL,
+ 0x03A9 | HAS_VOWEL,
+ 0x03A9 | HAS_VOWEL | HAS_ACCENT,
+ 0x03A9 | HAS_VOWEL | HAS_ACCENT,
+ 0x03A9 | HAS_VOWEL | HAS_ACCENT,
+ 0x03A9 | HAS_VOWEL | HAS_ACCENT,
+ 0x03A9 | HAS_VOWEL | HAS_ACCENT,
+ 0x03A9 | HAS_VOWEL | HAS_ACCENT,
+ 0x0391 | HAS_VOWEL | HAS_ACCENT,
+ 0x0391 | HAS_VOWEL | HAS_ACCENT,
+ 0x0395 | HAS_VOWEL | HAS_ACCENT,
+ 0x0395 | HAS_VOWEL | HAS_ACCENT,
+ 0x0397 | HAS_VOWEL | HAS_ACCENT,
+ 0x0397 | HAS_VOWEL | HAS_ACCENT,
+ 0x0399 | HAS_VOWEL | HAS_ACCENT,
+ 0x0399 | HAS_VOWEL | HAS_ACCENT,
+ 0x039F | HAS_VOWEL | HAS_ACCENT,
+ 0x039F | HAS_VOWEL | HAS_ACCENT,
+ 0x03A5 | HAS_VOWEL | HAS_ACCENT,
+ 0x03A5 | HAS_VOWEL | HAS_ACCENT,
+ 0x03A9 | HAS_VOWEL | HAS_ACCENT,
+ 0x03A9 | HAS_VOWEL | HAS_ACCENT,
+ 0,
+ 0,
+ 0x0391 | HAS_VOWEL | HAS_YPOGEGRAMMENI,
+ 0x0391 | HAS_VOWEL | HAS_YPOGEGRAMMENI,
+ 0x0391 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
+ 0x0391 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
+ 0x0391 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
+ 0x0391 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
+ 0x0391 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
+ 0x0391 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
+ 0x0391 | HAS_VOWEL | HAS_YPOGEGRAMMENI,
+ 0x0391 | HAS_VOWEL | HAS_YPOGEGRAMMENI,
+ 0x0391 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
+ 0x0391 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
+ 0x0391 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
+ 0x0391 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
+ 0x0391 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
+ 0x0391 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
+ 0x0397 | HAS_VOWEL | HAS_YPOGEGRAMMENI,
+ 0x0397 | HAS_VOWEL | HAS_YPOGEGRAMMENI,
+ 0x0397 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
+ 0x0397 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
+ 0x0397 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
+ 0x0397 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
+ 0x0397 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
+ 0x0397 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
+ 0x0397 | HAS_VOWEL | HAS_YPOGEGRAMMENI,
+ 0x0397 | HAS_VOWEL | HAS_YPOGEGRAMMENI,
+ 0x0397 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
+ 0x0397 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
+ 0x0397 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
+ 0x0397 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
+ 0x0397 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
+ 0x0397 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
+ 0x03A9 | HAS_VOWEL | HAS_YPOGEGRAMMENI,
+ 0x03A9 | HAS_VOWEL | HAS_YPOGEGRAMMENI,
+ 0x03A9 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
+ 0x03A9 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
+ 0x03A9 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
+ 0x03A9 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
+ 0x03A9 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
+ 0x03A9 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
+ 0x03A9 | HAS_VOWEL | HAS_YPOGEGRAMMENI,
+ 0x03A9 | HAS_VOWEL | HAS_YPOGEGRAMMENI,
+ 0x03A9 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
+ 0x03A9 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
+ 0x03A9 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
+ 0x03A9 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
+ 0x03A9 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
+ 0x03A9 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
+ 0x0391 | HAS_VOWEL,
+ 0x0391 | HAS_VOWEL,
+ 0x0391 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
+ 0x0391 | HAS_VOWEL | HAS_YPOGEGRAMMENI,
+ 0x0391 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
+ 0,
+ 0x0391 | HAS_VOWEL | HAS_ACCENT,
+ 0x0391 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
+ 0x0391 | HAS_VOWEL,
+ 0x0391 | HAS_VOWEL,
+ 0x0391 | HAS_VOWEL | HAS_ACCENT,
+ 0x0391 | HAS_VOWEL | HAS_ACCENT,
+ 0x0391 | HAS_VOWEL | HAS_YPOGEGRAMMENI,
+ 0,
+ 0x0399 | HAS_VOWEL,
+ 0,
+ 0,
+ 0,
+ 0x0397 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
+ 0x0397 | HAS_VOWEL | HAS_YPOGEGRAMMENI,
+ 0x0397 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
+ 0,
+ 0x0397 | HAS_VOWEL | HAS_ACCENT,
+ 0x0397 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
+ 0x0395 | HAS_VOWEL | HAS_ACCENT,
+ 0x0395 | HAS_VOWEL | HAS_ACCENT,
+ 0x0397 | HAS_VOWEL | HAS_ACCENT,
+ 0x0397 | HAS_VOWEL | HAS_ACCENT,
+ 0x0397 | HAS_VOWEL | HAS_YPOGEGRAMMENI,
+ 0,
+ 0,
+ 0,
+ 0x0399 | HAS_VOWEL,
+ 0x0399 | HAS_VOWEL,
+ 0x0399 | HAS_VOWEL | HAS_ACCENT | HAS_DIALYTIKA,
+ 0x0399 | HAS_VOWEL | HAS_ACCENT | HAS_DIALYTIKA,
+ 0,
+ 0,
+ 0x0399 | HAS_VOWEL | HAS_ACCENT,
+ 0x0399 | HAS_VOWEL | HAS_ACCENT | HAS_DIALYTIKA,
+ 0x0399 | HAS_VOWEL,
+ 0x0399 | HAS_VOWEL,
+ 0x0399 | HAS_VOWEL | HAS_ACCENT,
+ 0x0399 | HAS_VOWEL | HAS_ACCENT,
+ 0,
+ 0,
+ 0,
+ 0,
+ 0x03A5 | HAS_VOWEL,
+ 0x03A5 | HAS_VOWEL,
+ 0x03A5 | HAS_VOWEL | HAS_ACCENT | HAS_DIALYTIKA,
+ 0x03A5 | HAS_VOWEL | HAS_ACCENT | HAS_DIALYTIKA,
+ 0x03A1,
+ 0x03A1,
+ 0x03A5 | HAS_VOWEL | HAS_ACCENT,
+ 0x03A5 | HAS_VOWEL | HAS_ACCENT | HAS_DIALYTIKA,
+ 0x03A5 | HAS_VOWEL,
+ 0x03A5 | HAS_VOWEL,
+ 0x03A5 | HAS_VOWEL | HAS_ACCENT,
+ 0x03A5 | HAS_VOWEL | HAS_ACCENT,
+ 0x03A1,
+ 0,
+ 0,
+ 0,
+ 0,
+ 0,
+ 0x03A9 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
+ 0x03A9 | HAS_VOWEL | HAS_YPOGEGRAMMENI,
+ 0x03A9 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
+ 0,
+ 0x03A9 | HAS_VOWEL | HAS_ACCENT,
+ 0x03A9 | HAS_VOWEL | HAS_YPOGEGRAMMENI | HAS_ACCENT,
+ 0x039F | HAS_VOWEL | HAS_ACCENT,
+ 0x039F | HAS_VOWEL | HAS_ACCENT,
+ 0x03A9 | HAS_VOWEL | HAS_ACCENT,
+ 0x03A9 | HAS_VOWEL | HAS_ACCENT,
+ 0x03A9 | HAS_VOWEL | HAS_YPOGEGRAMMENI,
+ 0,
+ 0,
+ 0,
+};
+
+// U+2126 Ohm sign
+static const uint16_t data2126 = 0x03A9 | HAS_VOWEL;
+
+uint32_t getLetterData(UChar32 c) {
+ if (c < 0x370 || 0x2126 < c || (0x3ff < c && c < 0x1f00)) {
+ return 0;
+ } else if (c <= 0x3ff) {
+ return data0370[c - 0x370];
+ } else if (c <= 0x1fff) {
+ return data1F00[c - 0x1f00];
+ } else if (c == 0x2126) {
+ return data2126;
+ } else {
+ return 0;
+ }
+}
+
+uint32_t getDiacriticData(UChar32 c) {
+ switch (c) {
+ case 0x0300: // varia
+ case 0x0301: // tonos = oxia
+ case 0x0342: // perispomeni
+ case 0x0302: // circumflex can look like perispomeni
+ case 0x0303: // tilde can look like perispomeni
+ case 0x0311: // inverted breve can look like perispomeni
+ return HAS_ACCENT;
+ case 0x0308: // dialytika = diaeresis
+ return HAS_COMBINING_DIALYTIKA;
+ case 0x0344: // dialytika tonos
+ return HAS_COMBINING_DIALYTIKA | HAS_ACCENT;
+ case 0x0345: // ypogegrammeni = iota subscript
+ return HAS_YPOGEGRAMMENI;
+ case 0x0304: // macron
+ case 0x0306: // breve
+ case 0x0313: // comma above
+ case 0x0314: // reversed comma above
+ case 0x0343: // koronis
+ return HAS_OTHER_GREEK_DIACRITIC;
+ default:
+ return 0;
+ }
+}
+
+UBool isFollowedByCasedLetter(const UCaseProps *csp, const UChar *s, int32_t i, int32_t length) {
+ while (i < length) {
+ UChar32 c;
+ U16_NEXT(s, i, length, c);
+ int32_t type = ucase_getTypeOrIgnorable(csp, c);
+ if ((type & UCASE_IGNORABLE) != 0) {
+ // Case-ignorable, continue with the loop.
+ } else if (type != UCASE_NONE) {
+ return TRUE; // Followed by cased letter.
+ } else {
+ return FALSE; // Uncased and not case-ignorable.
+ }
+ }
+ return FALSE; // Not followed by cased letter.
+}
+
+/**
+ * Greek string uppercasing with a state machine.
+ * Probably simpler than a stateless function that has to figure out complex context-before
+ * for each character.
+ * TODO: Try to re-consolidate one way or another with the non-Greek function.
+ */
+int32_t toUpper(const UCaseMap *csm,
+ UChar *dest, int32_t destCapacity,
+ const UChar *src, int32_t srcLength,
+ UErrorCode *pErrorCode) {
+ int32_t locCache = UCASE_LOC_GREEK;
+ int32_t destIndex=0;
+ uint32_t state = 0;
+ for (int32_t i = 0; i < srcLength;) {
+ int32_t nextIndex = i;
+ UChar32 c;
+ U16_NEXT(src, nextIndex, srcLength, c);
+ uint32_t nextState = 0;
+ int32_t type = ucase_getTypeOrIgnorable(csm->csp, c);
+ if ((type & UCASE_IGNORABLE) != 0) {
+ // c is case-ignorable
+ nextState |= (state & AFTER_CASED);
+ } else if (type != UCASE_NONE) {
+ // c is cased
+ nextState |= AFTER_CASED;
+ }
+ uint32_t data = getLetterData(c);
+ if (data > 0) {
+ uint32_t upper = data & UPPER_MASK;
+ // Add a dialytika to this iota or ypsilon vowel
+ // if we removed a tonos from the previous vowel,
+ // and that previous vowel did not also have (or gain) a dialytika.
+ // Adding one only to the final vowel in a longer sequence
+ // (which does not occur in normal writing) would require lookahead.
+ // Set the same flag as for preserving an existing dialytika.
+ if ((data & HAS_VOWEL) != 0 && (state & AFTER_VOWEL_WITH_ACCENT) != 0 &&
+ (upper == 0x399 || upper == 0x3A5)) {
+ data |= HAS_DIALYTIKA;
+ }
+ int32_t numYpogegrammeni = 0; // Map each one to a trailing, spacing, capital iota.
+ if ((data & HAS_YPOGEGRAMMENI) != 0) {
+ numYpogegrammeni = 1;
+ }
+ // Skip combining diacritics after this Greek letter.
+ while (nextIndex < srcLength) {
+ uint32_t diacriticData = getDiacriticData(src[nextIndex]);
+ if (diacriticData != 0) {
+ data |= diacriticData;
+ if ((diacriticData & HAS_YPOGEGRAMMENI) != 0) {
+ ++numYpogegrammeni;
+ }
+ ++nextIndex;
+ } else {
+ break; // not a Greek diacritic
+ }
+ }
+ if ((data & HAS_VOWEL_AND_ACCENT_AND_DIALYTIKA) == HAS_VOWEL_AND_ACCENT) {
+ nextState |= AFTER_VOWEL_WITH_ACCENT;
+ }
+ // Map according to Greek rules.
+ UBool addTonos = FALSE;
+ if (upper == 0x397 &&
+ (data & HAS_ACCENT) != 0 &&
+ numYpogegrammeni == 0 &&
+ (state & AFTER_CASED) == 0 &&
+ !isFollowedByCasedLetter(csm->csp, src, nextIndex, srcLength)) {
+ // Keep disjunctive "or" with (only) a tonos.
+ // We use the same "word boundary" conditions as for the Final_Sigma test.
+ if (i == nextIndex) {
+ upper = 0x389; // Preserve the precomposed form.
+ } else {
+ addTonos = TRUE;
+ }
+ } else if ((data & HAS_DIALYTIKA) != 0) {
+ // Preserve a vowel with dialytika in precomposed form if it exists.
+ if (upper == 0x399) {
+ upper = 0x3AA;
+ data &= ~HAS_EITHER_DIALYTIKA;
+ } else if (upper == 0x3A5) {
+ upper = 0x3AB;
+ data &= ~HAS_EITHER_DIALYTIKA;
+ }
+ }
+ destIndex=appendUChar(dest, destIndex, destCapacity, (UChar)upper);
+ if (destIndex >= 0 && (data & HAS_EITHER_DIALYTIKA) != 0) {
+ destIndex=appendUChar(dest, destIndex, destCapacity, 0x308); // restore or add a dialytika
+ }
+ if (destIndex >= 0 && addTonos) {
+ destIndex=appendUChar(dest, destIndex, destCapacity, 0x301);
+ }
+ while (destIndex >= 0 && numYpogegrammeni > 0) {
+ destIndex=appendUChar(dest, destIndex, destCapacity, 0x399);
+ --numYpogegrammeni;
+ }
+ if(destIndex<0) {
+ *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR;
+ return 0;
+ }
+ } else {
+ const UChar *s;
+ UChar32 c2 = 0;
+ c=ucase_toFullUpper(csm->csp, c, NULL, NULL, &s, csm->locale, &locCache);
+ if((destIndex<destCapacity) && (c<0 ? (c2=~c)<=0xffff : UCASE_MAX_STRING_LENGTH<c && (c2=c)<=0xffff)) {
+ /* fast path version of appendResult() for BMP results */
+ dest[destIndex++]=(UChar)c2;
+ } else {
+ destIndex=appendResult(dest, destIndex, destCapacity, c, s);
+ if(destIndex<0) {
+ *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR;
+ return 0;
+ }
+ }
+ }
+ i = nextIndex;
+ state = nextState;
+ }
+
+ if(destIndex>destCapacity) {
+ *pErrorCode=U_BUFFER_OVERFLOW_ERROR;
+ }
+ return destIndex;
+}
+
+} // namespace GreekUpper
+U_NAMESPACE_END
+
/* functions available in the common library (for unistr_case.cpp) */
U_CFUNC int32_t U_CALLCONV
UChar *dest, int32_t destCapacity,
const UChar *src, int32_t srcLength,
UErrorCode *pErrorCode) {
+ int32_t locCache = csm->locCache;
+ if (ucase_getCaseLocale(csm->locale, &locCache) == UCASE_LOC_GREEK) {
+ return GreekUpper::toUpper(csm, dest, destCapacity, src, srcLength, pErrorCode);
+ }
UCaseContext csc=UCASECONTEXT_INITIALIZER;
csc.p=(void *)src;
csc.limit=srcLength;
dest[destIndex++]=(UChar)c2;
} else {
destIndex=appendResult(dest, destIndex, destCapacity, c, s);
+ if(destIndex<0) {
+ *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR;
+ return 0;
+ }
}
}
if(destLength>0) {
int32_t copyLength= destLength<=destCapacity ? destLength : destCapacity;
if(copyLength>0) {
- uprv_memmove(dest, temp, copyLength*U_SIZEOF_UCHAR);
+ u_memmove(dest, temp, copyLength);
}
}
if(temp!=buffer) {
projects / platform / upstream / icu.git / blobdiff