AuroraMiddleware/scuffed-code
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

ICU-2021 replace Turkic-casefolding prototype with explanation

Browse Source 
X-SVN-Rev: 11278
This commit is contained in:
Markus Scherer 2003-03-10 00:59:10 +00:00
parent a32165212d
commit 25d1629bb7
1 changed files with 12 additions and 74 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

86
icu4c/source/common/uchar.c
View File

@ -1924,53 +1924,20 @@ u_foldCase(UChar32 c, uint32_t options) {
return c; /* no mapping - return c itself */
}
#if 0
/* ### TODO Turkic-i case folding prototype, j2021 */
enum {
FOLD_T_LENGTH=3
};
/*
* Turkic full case foldings.
* First UChar is the source, second the default mapping,
* then the Turkic mapping.
* Issue for canonical caseless match (UAX #21):
* Turkic casefolding (using "T" mappings in CaseFolding.txt) does not preserve
* canonical equivalence, unlike default-option casefolding.
* For example, I-grave and I + grave fold to strings that are not canonically
* equivalent.
* For more details, see the comment in unorm_compare() in unorm.cpp
* and the intermediate prototype changes for Jitterbug 2021.
* (For example, revision 1.104 of uchar.c and 1.4 of CaseFolding.txt.)
*
* This did not get fixed because it appears that it is not possible to fix
* it for uppercase and lowercase characters (I-grave vs. i-grave)
* together in a way that they still fold to common result strings.
*/
static const UChar fold_T[][2+FOLD_T_LENGTH]={
{ 0x0049, 0x0069, 0x131 },
{ 0x0069, 0x0069, 0x131, 0x307 },
{ 0x00cc, 0x00ec, 0x131, 0x300 },
{ 0x00cd, 0x00ed, 0x131, 0x301 },
{ 0x00ce, 0x00ee, 0x131, 0x302 },
{ 0x00cf, 0x00ef, 0x131, 0x308 },
{ 0x00ec, 0x00ec, 0x131, 0x300 },
{ 0x00ed, 0x00ed, 0x131, 0x301 },
{ 0x00ee, 0x00ee, 0x131, 0x302 },
{ 0x00ef, 0x00ef, 0x131, 0x308 },
{ 0x0128, 0x0129, 0x131, 0x303 },
{ 0x0129, 0x0129, 0x131, 0x303 },
{ 0x012a, 0x012b, 0x131, 0x304 },
{ 0x012b, 0x012b, 0x131, 0x304 },
{ 0x012c, 0x012d, 0x131, 0x306 },
{ 0x012d, 0x012d, 0x131, 0x306 },
{ 0x012e, 0x012f, 0x131, 0x328 },
{ 0x012f, 0x012f, 0x131, 0x328 },
{ 0x0130, 0, 0x131, 0x307 }, /* normal mapping is 0069 0307 */
{ 0x01cf, 0x01d0, 0x131, 0x30c },
{ 0x01d0, 0x01d0, 0x131, 0x30c },
{ 0x0208, 0x0209, 0x131, 0x30f },
{ 0x0209, 0x0209, 0x131, 0x30f },
{ 0x020a, 0x020b, 0x131, 0x311 },
{ 0x020b, 0x020b, 0x131, 0x311 },
{ 0x1e2c, 0x1e2d, 0x131, 0x330 },
{ 0x1e2d, 0x1e2d, 0x131, 0x330 },
{ 0x1e2e, 0x1e2f, 0x131, 0x308, 0x301 },
{ 0x1e2f, 0x1e2f, 0x131, 0x308, 0x301 },
{ 0x1ec8, 0x1ec9, 0x131, 0x309 },
{ 0x1ec9, 0x1ec9, 0x131, 0x309 },
{ 0x1eca, 0x1ecb, 0x131, 0x323 },
{ 0x1ecb, 0x1ecb, 0x131, 0x323 }
};
#endif
/* internal, see ustr_imp.h */
U_CAPI int32_t U_EXPORT2
@ -2026,37 +1993,9 @@ u_internalFoldCase(UChar32 c,
dest[1]=0x307;
}
return 2;
#if 0
/* ### TODO Turkic-i case folding prototype, j2021 */
} else if(c<=fold_T[LENGTHOF(fold_T)-1][0]) {
for(i=0; i<LENGTHOF(fold_T) && c>=fold_T[i][0]; ++i) {
if(c==fold_T[i][0]) {
result=fold_T[i][1];
break;
}
}
#endif
}
} else {
/* Turkic mappings */
#if 0
/* ### TODO Turkic-i case folding prototype, j2021 */
if(c<=fold_T[LENGTHOF(fold_T)-1][0]) {
for(i=0; i<LENGTHOF(fold_T) && c>=fold_T[i][0]; ++i) {
if(c==fold_T[i][0]) {
const UChar *p=&(fold_T[i][2]);
length=0;
while(length<FOLD_T_LENGTH && *p!=0) {
if(length<destCapacity) {
dest[length]=*p++;
}
++length;
}
return length;
}
}
}
#else
if(c==0x49) {
/* 0049; T; 0131; # LATIN CAPITAL LETTER I */
result=0x131;
@ -2064,7 +2003,6 @@ u_internalFoldCase(UChar32 c,
/* 0130; T; 0069; # LATIN CAPITAL LETTER I WITH DOT ABOVE */
result=0x69;
}
#endif
}
/* return c itself because there is no special mapping for it */
/* goto single; */