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

ICU-1274 reenable [and rename] arabic shaping option for tashkeel characters

Browse Source 
X-SVN-Rev: 6229
This commit is contained in:
Markus Scherer 2001-10-13 00:06:29 +00:00
parent 0ed91898b2
commit 94b2c62a59
3 changed files with 14 additions and 53 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

9
icu4c/source/common/unicode/ushape.h
View File

@ -150,8 +150,13 @@ u_shapeArabic(const UChar *source, int32_t sourceLength,
/** Letter shaping option: replace "shaped" letter characters by abstract ones. */
#define U_SHAPE_LETTERS_UNSHAPE 0x10
/** Not a valid option value. */
#define U_SHAPE_LETTERS_RESERVED 0x18
/**
* Letter shaping option: replace abstract letter characters by "shaped" ones.
* The only difference with U_SHAPE_LETTERS_SHAPE is that Tashkeel letters
* are always "shaped" into the isolated form instead of the medial form
* (selecting code points from the Arabic Presentation Forms-B block).
*/
#define U_SHAPE_LETTERS_SHAPE_TASHKEEL_ISOLATED 0x18
/** Bit mask for letter shaping options. */
#define U_SHAPE_LETTERS_MASK 0x18

19
icu4c/source/common/ushape.c
View File

@ -45,22 +45,6 @@
* + IBM Egypt has proposed to encode the tail in Unicode among Arabic Presentation Forms
*/
/*
* ### TODO
*
* The option U_SHAPE_LETTERS_EXCEPT_TASHKEEL was removed from ushape.h for
* ICU 1.8 because its semantics were not clear:
* It is not clear if the transformation of legacy codepage data might
* need an option to
* - ignore tashkeel characters, which would always leave them as is, or
* - always shape to a certain form them regardless of other letter shaping options
*/
/**
* Letter shaping option: replace abstract letter characters by "shaped" ones,
* but do not process Tashkeel characters.
*/
#define U_SHAPE_LETTERS_EXCEPT_TASHKEEL 0x18
/* definitions for Arabic letter shaping ------------------------------------ */
#define IRRELEVANT 4
@ -951,7 +935,6 @@ u_shapeArabic(const UChar *source, int32_t sourceLength,
if( source==NULL || sourceLength<-1 ||
(dest==NULL && destSize!=0) || destSize<0 ||
options>=U_SHAPE_DIGIT_TYPE_RESERVED ||
(options&U_SHAPE_LETTERS_MASK)==U_SHAPE_LETTERS_RESERVED ||
(options&U_SHAPE_DIGITS_MASK)>=U_SHAPE_DIGITS_RESERVED
) {
*pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
@ -1016,7 +999,7 @@ u_shapeArabic(const UChar *source, int32_t sourceLength,
/* Call the shaping function with tashkeel flag == 1 */
outputSize = shapeUnicode(tempbuffer,sourceLength,destSize,options,pErrorCode,1);
break;
case U_SHAPE_LETTERS_EXCEPT_TASHKEEL :
case U_SHAPE_LETTERS_SHAPE_TASHKEEL_ISOLATED :
/* Call the shaping function with tashkeel flag == 0 */
outputSize = shapeUnicode(tempbuffer,sourceLength,destSize,options,pErrorCode,0);
break;

39
icu4c/source/test/cintltst/cbiditst.c
View File

@ -741,15 +741,6 @@ doArabicShapingTest() {
log_err("failure in u_shapeArabic(destSize=-1), returned %s instead of U_ILLEGAL_ARGUMENT_ERROR\n", u_errorName(errorCode));
}
errorCode=U_ZERO_ERROR;
length=u_shapeArabic(source, LENGTHOF(source),
dest, LENGTHOF(dest),
U_SHAPE_LETTERS_RESERVED|U_SHAPE_DIGITS_EN2AN|U_SHAPE_DIGIT_TYPE_AN,
&errorCode);
if(errorCode!=U_ILLEGAL_ARGUMENT_ERROR) {
log_err("failure in u_shapeArabic(U_SHAPE_LETTERS_RESERVED), returned %s instead of U_ILLEGAL_ARGUMENT_ERROR\n", u_errorName(errorCode));
}
errorCode=U_ZERO_ERROR;
length=u_shapeArabic(source, LENGTHOF(source),
dest, LENGTHOF(dest),
@ -878,21 +869,13 @@ doLamAlefSpecialVLTRArabicShapingTest() {
log_err("failure in u_shapeArabic(LAMALEF shape_grow_shrink)\n");
}
#if 0
/*
* ### TODO
*
* The option U_SHAPE_LETTERS_EXCEPT_TASHKEEL was removed from ushape.h for
* ICU 1.8 because its semantics were not clear.
* See ushape.c.
*/
/* ==================== SHAPE_EXCEPT_TASHKEEL ==================== */
/* ==================== U_SHAPE_LETTERS_SHAPE_TASHKEEL_ISOLATED ==================== */
errorCode=U_ZERO_ERROR;
length=u_shapeArabic(source, LENGTHOF(source),
dest, LENGTHOF(dest),
U_SHAPE_LETTERS_EXCEPT_TASHKEEL|U_SHAPE_LENGTH_FIXED_SPACES_NEAR|
U_SHAPE_LETTERS_SHAPE_TASHKEEL_ISOLATED|U_SHAPE_LENGTH_FIXED_SPACES_NEAR|
U_SHAPE_TEXT_DIRECTION_VISUAL_LTR,
&errorCode);
@ -904,7 +887,7 @@ doLamAlefSpecialVLTRArabicShapingTest() {
length=u_shapeArabic(source, LENGTHOF(source),
dest, LENGTHOF(dest),
U_SHAPE_LETTERS_EXCEPT_TASHKEEL|U_SHAPE_LENGTH_FIXED_SPACES_AT_END|
U_SHAPE_LETTERS_SHAPE_TASHKEEL_ISOLATED|U_SHAPE_LENGTH_FIXED_SPACES_AT_END|
U_SHAPE_TEXT_DIRECTION_VISUAL_LTR,
&errorCode);
@ -916,7 +899,7 @@ doLamAlefSpecialVLTRArabicShapingTest() {
length=u_shapeArabic(source, LENGTHOF(source),
dest, LENGTHOF(dest),
U_SHAPE_LETTERS_EXCEPT_TASHKEEL|U_SHAPE_LENGTH_FIXED_SPACES_AT_BEGINNING|
U_SHAPE_LETTERS_SHAPE_TASHKEEL_ISOLATED|U_SHAPE_LENGTH_FIXED_SPACES_AT_BEGINNING|
U_SHAPE_TEXT_DIRECTION_VISUAL_LTR,
&errorCode);
@ -928,14 +911,13 @@ doLamAlefSpecialVLTRArabicShapingTest() {
length=u_shapeArabic(source, LENGTHOF(source),
dest, LENGTHOF(dest),
U_SHAPE_LETTERS_EXCEPT_TASHKEEL|U_SHAPE_LENGTH_GROW_SHRINK|
U_SHAPE_LETTERS_SHAPE_TASHKEEL_ISOLATED|U_SHAPE_LENGTH_GROW_SHRINK|
U_SHAPE_TEXT_DIRECTION_VISUAL_LTR,
&errorCode);
if(U_FAILURE(errorCode) || uprv_memcmp(dest, shape_excepttashkeel_grow_shrink, length*U_SIZEOF_UCHAR)!=0) {
log_err("failure in u_shapeArabic(LAMALEF shape_excepttashkeel_grow_shrink)\n");
}
#endif
}
static void
@ -977,26 +959,17 @@ doTashkeelSpecialVLTRArabicShapingTest() {
log_err("failure in u_shapeArabic(TASHKEEL shape_near)\n");
}
#if 0
/*
* ### TODO
*
* The option U_SHAPE_LETTERS_EXCEPT_TASHKEEL was removed from ushape.h for
* ICU 1.8 because its semantics were not clear.
* See ushape.c.
*/
errorCode=U_ZERO_ERROR;
length=u_shapeArabic(source, LENGTHOF(source),
dest, LENGTHOF(dest),
U_SHAPE_LETTERS_EXCEPT_TASHKEEL|U_SHAPE_LENGTH_FIXED_SPACES_NEAR|
U_SHAPE_LETTERS_SHAPE_TASHKEEL_ISOLATED|U_SHAPE_LENGTH_FIXED_SPACES_NEAR|
U_SHAPE_TEXT_DIRECTION_VISUAL_LTR,
&errorCode);
if(U_FAILURE(errorCode) || length!=LENGTHOF(shape_excepttashkeel_near) || uprv_memcmp(dest, shape_excepttashkeel_near, length*U_SIZEOF_UCHAR)!=0) {
log_err("failure in u_shapeArabic(TASHKEEL shape_excepttashkeel_near)\n");
}
#endif
}
static void