| File: | root/firefox-clang/intl/icu/source/common/ubidiwrt.cpp |
| Warning: | line 203, column 24 Dereference of null pointer |
Press '?' to see keyboard shortcuts
Keyboard shortcuts:
| 1 | // © 2016 and later: Unicode, Inc. and others. | |||
| 2 | // License & terms of use: http://www.unicode.org/copyright.html | |||
| 3 | /* | |||
| 4 | ****************************************************************************** | |||
| 5 | * | |||
| 6 | * Copyright (C) 2000-2015, International Business Machines | |||
| 7 | * Corporation and others. All Rights Reserved. | |||
| 8 | * | |||
| 9 | ****************************************************************************** | |||
| 10 | * file name: ubidiwrt.c | |||
| 11 | * encoding: UTF-8 | |||
| 12 | * tab size: 8 (not used) | |||
| 13 | * indentation:4 | |||
| 14 | * | |||
| 15 | * created on: 1999aug06 | |||
| 16 | * created by: Markus W. Scherer, updated by Matitiahu Allouche | |||
| 17 | * | |||
| 18 | * This file contains implementations for BiDi functions that use | |||
| 19 | * the core algorithm and core API to write reordered text. | |||
| 20 | */ | |||
| 21 | ||||
| 22 | #include "unicode/utypes.h" | |||
| 23 | #include "unicode/ustring.h" | |||
| 24 | #include "unicode/uchar.h" | |||
| 25 | #include "unicode/ubidi.h" | |||
| 26 | #include "unicode/utf16.h" | |||
| 27 | #include "cmemory.h" | |||
| 28 | #include "ustr_imp.h" | |||
| 29 | #include "ubidiimp.h" | |||
| 30 | ||||
| 31 | /* | |||
| 32 | * The function implementations in this file are designed | |||
| 33 | * for UTF-16 and UTF-32, not for UTF-8. | |||
| 34 | * | |||
| 35 | * Assumptions that are not true for UTF-8: | |||
| 36 | * - Any code point always needs the same number of code units | |||
| 37 | * ("minimum-length-problem" of UTF-8) | |||
| 38 | * - The BiDi control characters need only one code unit each | |||
| 39 | * | |||
| 40 | * Further assumptions for all UTFs: | |||
| 41 | * - u_charMirror(c) needs the same number of code units as c | |||
| 42 | */ | |||
| 43 | #if defined(UTF_SIZE) && UTF_SIZE==8 | |||
| 44 | # error reimplement ubidi_writeReorderedubidi_writeReordered_77() for UTF-8, see comment above | |||
| 45 | #endif | |||
| 46 | ||||
| 47 | #define IS_COMBINING(type)((1UL<<(type))&(1UL<<U_NON_SPACING_MARK|1UL<< U_COMBINING_SPACING_MARK|1UL<<U_ENCLOSING_MARK)) ((1UL<<(type))&(1UL<<U_NON_SPACING_MARK|1UL<<U_COMBINING_SPACING_MARK|1UL<<U_ENCLOSING_MARK)) | |||
| 48 | ||||
| 49 | /* | |||
| 50 | * When we have UBIDI_OUTPUT_REVERSE set on ubidi_writeReordered(), then we | |||
| 51 | * semantically write RTL runs in reverse and later reverse them again. | |||
| 52 | * Instead, we actually write them in forward order to begin with. | |||
| 53 | * However, if the RTL run was to be mirrored, we need to mirror here now | |||
| 54 | * since the implicit second reversal must not do it. | |||
| 55 | * It looks strange to do mirroring in LTR output, but it is only because | |||
| 56 | * we are writing RTL output in reverse. | |||
| 57 | */ | |||
| 58 | static int32_t | |||
| 59 | doWriteForward(const char16_t *src, int32_t srcLength, | |||
| 60 | char16_t *dest, int32_t destSize, | |||
| 61 | uint16_t options, | |||
| 62 | UErrorCode *pErrorCode) { | |||
| 63 | /* optimize for several combinations of options */ | |||
| 64 | switch(options&(UBIDI_REMOVE_BIDI_CONTROLS8|UBIDI_DO_MIRRORING2)) { | |||
| 65 | case 0: { | |||
| 66 | /* simply copy the LTR run to the destination */ | |||
| 67 | int32_t length=srcLength; | |||
| 68 | if(destSize<length) { | |||
| 69 | *pErrorCode=U_BUFFER_OVERFLOW_ERROR; | |||
| 70 | return srcLength; | |||
| 71 | } | |||
| 72 | do { | |||
| 73 | *dest++=*src++; | |||
| 74 | } while(--length>0); | |||
| 75 | return srcLength; | |||
| 76 | } | |||
| 77 | case UBIDI_DO_MIRRORING2: { | |||
| 78 | /* do mirroring */ | |||
| 79 | int32_t i=0, j=0; | |||
| 80 | UChar32 c; | |||
| 81 | ||||
| 82 | if(destSize<srcLength) { | |||
| 83 | *pErrorCode=U_BUFFER_OVERFLOW_ERROR; | |||
| 84 | return srcLength; | |||
| 85 | } | |||
| 86 | do { | |||
| 87 | U16_NEXT(src, i, srcLength, c)do { (c)=(src)[(i)++]; if((((c)&0xfffffc00)==0xd800)) { uint16_t __c2; if((i)!=(srcLength) && (((__c2=(src)[(i)])& 0xfffffc00)==0xdc00)) { ++(i); (c)=(((UChar32)((c))<<10UL )+(UChar32)(__c2)-((0xd800<<10UL)+0xdc00-0x10000)); } } } while (false); | |||
| 88 | c=u_charMirroru_charMirror_77(c); | |||
| 89 | U16_APPEND_UNSAFE(dest, j, c)do { if((uint32_t)(c)<=0xffff) { (dest)[(j)++]=(uint16_t)( c); } else { (dest)[(j)++]=(uint16_t)(((c)>>10)+0xd7c0) ; (dest)[(j)++]=(uint16_t)(((c)&0x3ff)|0xdc00); } } while (false); | |||
| 90 | } while(i<srcLength); | |||
| 91 | return srcLength; | |||
| 92 | } | |||
| 93 | case UBIDI_REMOVE_BIDI_CONTROLS8: { | |||
| 94 | /* copy the LTR run and remove any BiDi control characters */ | |||
| 95 | int32_t remaining=destSize; | |||
| 96 | char16_t c; | |||
| 97 | do { | |||
| 98 | c=*src++; | |||
| 99 | if(!IS_BIDI_CONTROL_CHAR(c)(((uint32_t)(c)&0xfffffffc)==ZWNJ_CHAR || (uint32_t)((c)- LRE_CHAR)<5 || (uint32_t)((c)-LRI_CHAR)<4)) { | |||
| 100 | if(--remaining<0) { | |||
| 101 | *pErrorCode=U_BUFFER_OVERFLOW_ERROR; | |||
| 102 | ||||
| 103 | /* preflight the length */ | |||
| 104 | while(--srcLength>0) { | |||
| 105 | c=*src++; | |||
| 106 | if(!IS_BIDI_CONTROL_CHAR(c)(((uint32_t)(c)&0xfffffffc)==ZWNJ_CHAR || (uint32_t)((c)- LRE_CHAR)<5 || (uint32_t)((c)-LRI_CHAR)<4)) { | |||
| 107 | --remaining; | |||
| 108 | } | |||
| 109 | } | |||
| 110 | return destSize-remaining; | |||
| 111 | } | |||
| 112 | *dest++=c; | |||
| 113 | } | |||
| 114 | } while(--srcLength>0); | |||
| 115 | return destSize-remaining; | |||
| 116 | } | |||
| 117 | default: { | |||
| 118 | /* remove BiDi control characters and do mirroring */ | |||
| 119 | int32_t remaining=destSize; | |||
| 120 | int32_t i, j=0; | |||
| 121 | UChar32 c; | |||
| 122 | do { | |||
| 123 | i=0; | |||
| 124 | U16_NEXT(src, i, srcLength, c)do { (c)=(src)[(i)++]; if((((c)&0xfffffc00)==0xd800)) { uint16_t __c2; if((i)!=(srcLength) && (((__c2=(src)[(i)])& 0xfffffc00)==0xdc00)) { ++(i); (c)=(((UChar32)((c))<<10UL )+(UChar32)(__c2)-((0xd800<<10UL)+0xdc00-0x10000)); } } } while (false); | |||
| 125 | src+=i; | |||
| 126 | srcLength-=i; | |||
| 127 | if(!IS_BIDI_CONTROL_CHAR(c)(((uint32_t)(c)&0xfffffffc)==ZWNJ_CHAR || (uint32_t)((c)- LRE_CHAR)<5 || (uint32_t)((c)-LRI_CHAR)<4)) { | |||
| 128 | remaining-=i; | |||
| 129 | if(remaining<0) { | |||
| 130 | *pErrorCode=U_BUFFER_OVERFLOW_ERROR; | |||
| 131 | ||||
| 132 | /* preflight the length */ | |||
| 133 | while(srcLength>0) { | |||
| 134 | c=*src++; | |||
| 135 | if(!IS_BIDI_CONTROL_CHAR(c)(((uint32_t)(c)&0xfffffffc)==ZWNJ_CHAR || (uint32_t)((c)- LRE_CHAR)<5 || (uint32_t)((c)-LRI_CHAR)<4)) { | |||
| 136 | --remaining; | |||
| 137 | } | |||
| 138 | --srcLength; | |||
| 139 | } | |||
| 140 | return destSize-remaining; | |||
| 141 | } | |||
| 142 | c=u_charMirroru_charMirror_77(c); | |||
| 143 | U16_APPEND_UNSAFE(dest, j, c)do { if((uint32_t)(c)<=0xffff) { (dest)[(j)++]=(uint16_t)( c); } else { (dest)[(j)++]=(uint16_t)(((c)>>10)+0xd7c0) ; (dest)[(j)++]=(uint16_t)(((c)&0x3ff)|0xdc00); } } while (false); | |||
| 144 | } | |||
| 145 | } while(srcLength>0); | |||
| 146 | return j; | |||
| 147 | } | |||
| 148 | } /* end of switch */ | |||
| 149 | } | |||
| 150 | ||||
| 151 | static int32_t | |||
| 152 | doWriteReverse(const char16_t *src, int32_t srcLength, | |||
| 153 | char16_t *dest, int32_t destSize, | |||
| 154 | uint16_t options, | |||
| 155 | UErrorCode *pErrorCode) { | |||
| 156 | /* | |||
| 157 | * RTL run - | |||
| 158 | * | |||
| 159 | * RTL runs need to be copied to the destination in reverse order | |||
| 160 | * of code points, not code units, to keep Unicode characters intact. | |||
| 161 | * | |||
| 162 | * The general strategy for this is to read the source text | |||
| 163 | * in backward order, collect all code units for a code point | |||
| 164 | * (and optionally following combining characters, see below), | |||
| 165 | * and copy all these code units in ascending order | |||
| 166 | * to the destination for this run. | |||
| 167 | * | |||
| 168 | * Several options request whether combining characters | |||
| 169 | * should be kept after their base characters, | |||
| 170 | * whether BiDi control characters should be removed, and | |||
| 171 | * whether characters should be replaced by their mirror-image | |||
| 172 | * equivalent Unicode characters. | |||
| 173 | */ | |||
| 174 | int32_t i, j; | |||
| 175 | UChar32 c; | |||
| 176 | ||||
| 177 | /* optimize for several combinations of options */ | |||
| 178 | switch(options&(UBIDI_REMOVE_BIDI_CONTROLS8|UBIDI_DO_MIRRORING2|UBIDI_KEEP_BASE_COMBINING1)) { | |||
| 179 | case 0: | |||
| 180 | /* | |||
| 181 | * With none of the "complicated" options set, the destination | |||
| 182 | * run will have the same length as the source run, | |||
| 183 | * and there is no mirroring and no keeping combining characters | |||
| 184 | * with their base characters. | |||
| 185 | */ | |||
| 186 | if(destSize<srcLength) { | |||
| 187 | *pErrorCode=U_BUFFER_OVERFLOW_ERROR; | |||
| 188 | return srcLength; | |||
| 189 | } | |||
| 190 | destSize=srcLength; | |||
| 191 | ||||
| 192 | /* preserve character integrity */ | |||
| 193 | do { | |||
| 194 | /* i is always after the last code unit known to need to be kept in this segment */ | |||
| 195 | i=srcLength; | |||
| 196 | ||||
| 197 | /* collect code units for one base character */ | |||
| 198 | U16_BACK_1(src, 0, srcLength)do { if(((((src)[--(srcLength)])&0xfffffc00)==0xdc00) && (srcLength)>(0) && ((((src)[(srcLength)-1])&0xfffffc00 )==0xd800)) { --(srcLength); } } while (false); | |||
| 199 | ||||
| 200 | /* copy this base character */ | |||
| 201 | j=srcLength; | |||
| 202 | do { | |||
| 203 | *dest++=src[j++]; | |||
| ||||
| 204 | } while(j<i); | |||
| 205 | } while(srcLength>0); | |||
| 206 | break; | |||
| 207 | case UBIDI_KEEP_BASE_COMBINING1: | |||
| 208 | /* | |||
| 209 | * Here, too, the destination | |||
| 210 | * run will have the same length as the source run, | |||
| 211 | * and there is no mirroring. | |||
| 212 | * We do need to keep combining characters with their base characters. | |||
| 213 | */ | |||
| 214 | if(destSize<srcLength) { | |||
| 215 | *pErrorCode=U_BUFFER_OVERFLOW_ERROR; | |||
| 216 | return srcLength; | |||
| 217 | } | |||
| 218 | destSize=srcLength; | |||
| 219 | ||||
| 220 | /* preserve character integrity */ | |||
| 221 | do { | |||
| 222 | /* i is always after the last code unit known to need to be kept in this segment */ | |||
| 223 | i=srcLength; | |||
| 224 | ||||
| 225 | /* collect code units and modifier letters for one base character */ | |||
| 226 | do { | |||
| 227 | U16_PREV(src, 0, srcLength, c)do { (c)=(src)[--(srcLength)]; if((((c)&0xfffffc00)==0xdc00 )) { uint16_t __c2; if((srcLength)>(0) && (((__c2= (src)[(srcLength)-1])&0xfffffc00)==0xd800)) { --(srcLength ); (c)=(((UChar32)(__c2)<<10UL)+(UChar32)((c))-((0xd800 <<10UL)+0xdc00-0x10000)); } } } while (false); | |||
| 228 | } while(srcLength>0 && IS_COMBINING(u_charType(c))((1UL<<(u_charType_77(c)))&(1UL<<U_NON_SPACING_MARK |1UL<<U_COMBINING_SPACING_MARK|1UL<<U_ENCLOSING_MARK ))); | |||
| 229 | ||||
| 230 | /* copy this "user character" */ | |||
| 231 | j=srcLength; | |||
| 232 | do { | |||
| 233 | *dest++=src[j++]; | |||
| 234 | } while(j<i); | |||
| 235 | } while(srcLength>0); | |||
| 236 | break; | |||
| 237 | default: | |||
| 238 | /* | |||
| 239 | * With several "complicated" options set, this is the most | |||
| 240 | * general and the slowest copying of an RTL run. | |||
| 241 | * We will do mirroring, remove BiDi controls, and | |||
| 242 | * keep combining characters with their base characters | |||
| 243 | * as requested. | |||
| 244 | */ | |||
| 245 | if(!(options&UBIDI_REMOVE_BIDI_CONTROLS8)) { | |||
| 246 | i=srcLength; | |||
| 247 | } else { | |||
| 248 | /* we need to find out the destination length of the run, | |||
| 249 | which will not include the BiDi control characters */ | |||
| 250 | int32_t length=srcLength; | |||
| 251 | char16_t ch; | |||
| 252 | ||||
| 253 | i=0; | |||
| 254 | do { | |||
| 255 | ch=*src++; | |||
| 256 | if(!IS_BIDI_CONTROL_CHAR(ch)(((uint32_t)(ch)&0xfffffffc)==ZWNJ_CHAR || (uint32_t)((ch )-LRE_CHAR)<5 || (uint32_t)((ch)-LRI_CHAR)<4)) { | |||
| 257 | ++i; | |||
| 258 | } | |||
| 259 | } while(--length>0); | |||
| 260 | src-=srcLength; | |||
| 261 | } | |||
| 262 | ||||
| 263 | if(destSize<i) { | |||
| 264 | *pErrorCode=U_BUFFER_OVERFLOW_ERROR; | |||
| 265 | return i; | |||
| 266 | } | |||
| 267 | destSize=i; | |||
| 268 | ||||
| 269 | /* preserve character integrity */ | |||
| 270 | do { | |||
| 271 | /* i is always after the last code unit known to need to be kept in this segment */ | |||
| 272 | i=srcLength; | |||
| 273 | ||||
| 274 | /* collect code units for one base character */ | |||
| 275 | U16_PREV(src, 0, srcLength, c)do { (c)=(src)[--(srcLength)]; if((((c)&0xfffffc00)==0xdc00 )) { uint16_t __c2; if((srcLength)>(0) && (((__c2= (src)[(srcLength)-1])&0xfffffc00)==0xd800)) { --(srcLength ); (c)=(((UChar32)(__c2)<<10UL)+(UChar32)((c))-((0xd800 <<10UL)+0xdc00-0x10000)); } } } while (false); | |||
| 276 | if(options&UBIDI_KEEP_BASE_COMBINING1) { | |||
| 277 | /* collect modifier letters for this base character */ | |||
| 278 | while(srcLength>0 && IS_COMBINING(u_charType(c))((1UL<<(u_charType_77(c)))&(1UL<<U_NON_SPACING_MARK |1UL<<U_COMBINING_SPACING_MARK|1UL<<U_ENCLOSING_MARK ))) { | |||
| 279 | U16_PREV(src, 0, srcLength, c)do { (c)=(src)[--(srcLength)]; if((((c)&0xfffffc00)==0xdc00 )) { uint16_t __c2; if((srcLength)>(0) && (((__c2= (src)[(srcLength)-1])&0xfffffc00)==0xd800)) { --(srcLength ); (c)=(((UChar32)(__c2)<<10UL)+(UChar32)((c))-((0xd800 <<10UL)+0xdc00-0x10000)); } } } while (false); | |||
| 280 | } | |||
| 281 | } | |||
| 282 | ||||
| 283 | if(options&UBIDI_REMOVE_BIDI_CONTROLS8 && IS_BIDI_CONTROL_CHAR(c)(((uint32_t)(c)&0xfffffffc)==ZWNJ_CHAR || (uint32_t)((c)- LRE_CHAR)<5 || (uint32_t)((c)-LRI_CHAR)<4)) { | |||
| 284 | /* do not copy this BiDi control character */ | |||
| 285 | continue; | |||
| 286 | } | |||
| 287 | ||||
| 288 | /* copy this "user character" */ | |||
| 289 | j=srcLength; | |||
| 290 | if(options&UBIDI_DO_MIRRORING2) { | |||
| 291 | /* mirror only the base character */ | |||
| 292 | int32_t k=0; | |||
| 293 | c=u_charMirroru_charMirror_77(c); | |||
| 294 | U16_APPEND_UNSAFE(dest, k, c)do { if((uint32_t)(c)<=0xffff) { (dest)[(k)++]=(uint16_t)( c); } else { (dest)[(k)++]=(uint16_t)(((c)>>10)+0xd7c0) ; (dest)[(k)++]=(uint16_t)(((c)&0x3ff)|0xdc00); } } while (false); | |||
| 295 | dest+=k; | |||
| 296 | j+=k; | |||
| 297 | } | |||
| 298 | while(j<i) { | |||
| 299 | *dest++=src[j++]; | |||
| 300 | } | |||
| 301 | } while(srcLength>0); | |||
| 302 | break; | |||
| 303 | } /* end of switch */ | |||
| 304 | ||||
| 305 | return destSize; | |||
| 306 | } | |||
| 307 | ||||
| 308 | U_CAPIextern "C" int32_t U_EXPORT2 | |||
| 309 | ubidi_writeReverseubidi_writeReverse_77(const char16_t *src, int32_t srcLength, | |||
| 310 | char16_t *dest, int32_t destSize, | |||
| 311 | uint16_t options, | |||
| 312 | UErrorCode *pErrorCode) { | |||
| 313 | int32_t destLength; | |||
| 314 | ||||
| 315 | if(pErrorCode==nullptr || U_FAILURE(*pErrorCode)) { | |||
| 316 | return 0; | |||
| 317 | } | |||
| 318 | ||||
| 319 | /* more error checking */ | |||
| 320 | if( src==nullptr || srcLength<-1 || | |||
| 321 | destSize<0 || (destSize>0 && dest==nullptr)) | |||
| 322 | { | |||
| 323 | *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; | |||
| 324 | return 0; | |||
| 325 | } | |||
| 326 | ||||
| 327 | /* do input and output overlap? */ | |||
| 328 | if( dest!=nullptr && | |||
| 329 | ((src>=dest && src<dest+destSize) || | |||
| 330 | (dest>=src && dest<src+srcLength))) | |||
| 331 | { | |||
| 332 | *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; | |||
| 333 | return 0; | |||
| 334 | } | |||
| 335 | ||||
| 336 | if(srcLength==-1) { | |||
| 337 | srcLength=u_strlenu_strlen_77(src); | |||
| 338 | } | |||
| 339 | if(srcLength>0) { | |||
| 340 | destLength=doWriteReverse(src, srcLength, dest, destSize, options, pErrorCode); | |||
| 341 | } else { | |||
| 342 | /* nothing to do */ | |||
| 343 | destLength=0; | |||
| 344 | } | |||
| 345 | ||||
| 346 | return u_terminateUCharsu_terminateUChars_77(dest, destSize, destLength, pErrorCode); | |||
| 347 | } | |||
| 348 | ||||
| 349 | // Ticket 20907 - The optimizer in MSVC/Visual Studio versions below 16.4 has trouble with this | |||
| 350 | // function on Windows ARM64. As a work-around, we disable optimizations for this function. | |||
| 351 | // This work-around could/should be removed once the following versions of Visual Studio are no | |||
| 352 | // longer supported: All versions of VS2017, and versions of VS2019 below 16.4. | |||
| 353 | #if (defined(_MSC_VER) && (defined(_M_ARM64)) && (_MSC_VER < 1924)) | |||
| 354 | #pragma optimize( "", off ) | |||
| 355 | #endif | |||
| 356 | U_CAPIextern "C" int32_t U_EXPORT2 | |||
| 357 | ubidi_writeReorderedubidi_writeReordered_77(UBiDi *pBiDi, | |||
| 358 | char16_t *dest, int32_t destSize, | |||
| 359 | uint16_t options, | |||
| 360 | UErrorCode *pErrorCode) { | |||
| 361 | const char16_t *text; | |||
| 362 | char16_t *saveDest; | |||
| 363 | int32_t length, destCapacity; | |||
| 364 | int32_t run, runCount, logicalStart, runLength; | |||
| 365 | ||||
| 366 | if(pErrorCode==nullptr || U_FAILURE(*pErrorCode)) { | |||
| ||||
| 367 | return 0; | |||
| 368 | } | |||
| 369 | ||||
| 370 | /* more error checking */ | |||
| 371 | if( pBiDi==nullptr || | |||
| 372 | (text=pBiDi->text)==nullptr || (length=pBiDi->length)<0 || | |||
| 373 | destSize<0 || (destSize>0 && dest==nullptr)) | |||
| 374 | { | |||
| 375 | *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; | |||
| 376 | return 0; | |||
| 377 | } | |||
| 378 | ||||
| 379 | /* do input and output overlap? */ | |||
| 380 | if( dest!=nullptr && | |||
| 381 | ((text>=dest && text<dest+destSize) || | |||
| 382 | (dest>=text && dest<text+pBiDi->originalLength))) | |||
| 383 | { | |||
| 384 | *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR; | |||
| 385 | return 0; | |||
| 386 | } | |||
| 387 | ||||
| 388 | if(length==0) { | |||
| 389 | /* nothing to do */ | |||
| 390 | return u_terminateUCharsu_terminateUChars_77(dest, destSize, 0, pErrorCode); | |||
| 391 | } | |||
| 392 | ||||
| 393 | runCount=ubidi_countRunsubidi_countRuns_77(pBiDi, pErrorCode); | |||
| 394 | if(U_FAILURE(*pErrorCode)) { | |||
| 395 | return 0; | |||
| 396 | } | |||
| 397 | ||||
| 398 | /* destSize shrinks, later destination length=destCapacity-destSize */ | |||
| 399 | saveDest=dest; | |||
| 400 | destCapacity=destSize; | |||
| 401 | ||||
| 402 | /* | |||
| 403 | * Option "insert marks" implies UBIDI_INSERT_LRM_FOR_NUMERIC if the | |||
| 404 | * reordering mode (checked below) is appropriate. | |||
| 405 | */ | |||
| 406 | if(pBiDi->reorderingOptions & UBIDI_OPTION_INSERT_MARKS) { | |||
| 407 | options|=UBIDI_INSERT_LRM_FOR_NUMERIC4; | |||
| 408 | options&=~UBIDI_REMOVE_BIDI_CONTROLS8; | |||
| 409 | } | |||
| 410 | /* | |||
| 411 | * Option "remove controls" implies UBIDI_REMOVE_BIDI_CONTROLS | |||
| 412 | * and cancels UBIDI_INSERT_LRM_FOR_NUMERIC. | |||
| 413 | */ | |||
| 414 | if(pBiDi->reorderingOptions & UBIDI_OPTION_REMOVE_CONTROLS) { | |||
| 415 | options|=UBIDI_REMOVE_BIDI_CONTROLS8; | |||
| 416 | options&=~UBIDI_INSERT_LRM_FOR_NUMERIC4; | |||
| 417 | } | |||
| 418 | /* | |||
| 419 | * If we do not perform the "inverse BiDi" algorithm, then we | |||
| 420 | * don't need to insert any LRMs, and don't need to test for it. | |||
| 421 | */ | |||
| 422 | if((pBiDi->reorderingMode != UBIDI_REORDER_INVERSE_NUMBERS_AS_L) && | |||
| 423 | (pBiDi->reorderingMode != UBIDI_REORDER_INVERSE_LIKE_DIRECT) && | |||
| 424 | (pBiDi->reorderingMode != UBIDI_REORDER_INVERSE_FOR_NUMBERS_SPECIAL) && | |||
| 425 | (pBiDi->reorderingMode != UBIDI_REORDER_RUNS_ONLY)) { | |||
| 426 | options&=~UBIDI_INSERT_LRM_FOR_NUMERIC4; | |||
| 427 | } | |||
| 428 | /* | |||
| 429 | * Iterate through all visual runs and copy the run text segments to | |||
| 430 | * the destination, according to the options. | |||
| 431 | * | |||
| 432 | * The tests for where to insert LRMs ignore the fact that there may be | |||
| 433 | * BN codes or non-BMP code points at the beginning and end of a run; | |||
| 434 | * they may insert LRMs unnecessarily but the tests are faster this way | |||
| 435 | * (this would have to be improved for UTF-8). | |||
| 436 | * | |||
| 437 | * Note that the only errors that are set by doWriteXY() are buffer overflow | |||
| 438 | * errors. Ignore them until the end, and continue for preflighting. | |||
| 439 | */ | |||
| 440 | if(!(options&UBIDI_OUTPUT_REVERSE16)) { | |||
| 441 | /* forward output */ | |||
| 442 | if(!(options&UBIDI_INSERT_LRM_FOR_NUMERIC4)) { | |||
| 443 | /* do not insert BiDi controls */ | |||
| 444 | for(run=0; run<runCount; ++run) { | |||
| 445 | if(UBIDI_LTR==ubidi_getVisualRunubidi_getVisualRun_77(pBiDi, run, &logicalStart, &runLength)) { | |||
| 446 | runLength=doWriteForward(text+logicalStart, runLength, | |||
| 447 | dest, destSize, | |||
| 448 | (uint16_t)(options&~UBIDI_DO_MIRRORING2), pErrorCode); | |||
| 449 | } else { | |||
| 450 | runLength=doWriteReverse(text+logicalStart, runLength, | |||
| 451 | dest, destSize, | |||
| 452 | options, pErrorCode); | |||
| 453 | } | |||
| 454 | if(dest!=nullptr) { | |||
| 455 | dest+=runLength; | |||
| 456 | } | |||
| 457 | destSize-=runLength; | |||
| 458 | } | |||
| 459 | } else { | |||
| 460 | /* insert BiDi controls for "inverse BiDi" */ | |||
| 461 | const DirProp *dirProps=pBiDi->dirProps; | |||
| 462 | const char16_t *src; | |||
| 463 | char16_t uc; | |||
| 464 | UBiDiDirection dir; | |||
| 465 | int32_t markFlag; | |||
| 466 | ||||
| 467 | for(run=0; run<runCount; ++run) { | |||
| 468 | dir=ubidi_getVisualRunubidi_getVisualRun_77(pBiDi, run, &logicalStart, &runLength); | |||
| 469 | src=text+logicalStart; | |||
| 470 | /* check if something relevant in insertPoints */ | |||
| 471 | markFlag=pBiDi->runs[run].insertRemove; | |||
| 472 | if(markFlag<0) { /* BiDi controls count */ | |||
| 473 | markFlag=0; | |||
| 474 | } | |||
| 475 | ||||
| 476 | if(UBIDI_LTR==dir) { | |||
| 477 | if((pBiDi->isInverse) && | |||
| 478 | (/*run>0 &&*/ dirProps[logicalStart]!=L)) { | |||
| 479 | markFlag |= LRM_BEFORE; | |||
| 480 | } | |||
| 481 | if (markFlag & LRM_BEFORE) { | |||
| 482 | uc=LRM_CHAR; | |||
| 483 | } | |||
| 484 | else if (markFlag & RLM_BEFORE) { | |||
| 485 | uc=RLM_CHAR; | |||
| 486 | } | |||
| 487 | else uc=0; | |||
| 488 | if(uc) { | |||
| 489 | if(destSize>0) { | |||
| 490 | *dest++=uc; | |||
| 491 | } | |||
| 492 | --destSize; | |||
| 493 | } | |||
| 494 | ||||
| 495 | runLength=doWriteForward(src, runLength, | |||
| 496 | dest, destSize, | |||
| 497 | (uint16_t)(options&~UBIDI_DO_MIRRORING2), pErrorCode); | |||
| 498 | if(dest!=nullptr) { | |||
| 499 | dest+=runLength; | |||
| 500 | } | |||
| 501 | destSize-=runLength; | |||
| 502 | ||||
| 503 | if((pBiDi->isInverse) && | |||
| 504 | (/*run<runCount-1 &&*/ | |||
| 505 | runLength > 0 && // doWriteForward may return 0 if src | |||
| 506 | // only include bidi control chars | |||
| 507 | dirProps[logicalStart+runLength-1]!=L)) { | |||
| 508 | markFlag |= LRM_AFTER; | |||
| 509 | } | |||
| 510 | if (markFlag & LRM_AFTER) { | |||
| 511 | uc=LRM_CHAR; | |||
| 512 | } | |||
| 513 | else if (markFlag & RLM_AFTER) { | |||
| 514 | uc=RLM_CHAR; | |||
| 515 | } | |||
| 516 | else uc=0; | |||
| 517 | if(uc) { | |||
| 518 | if(destSize>0) { | |||
| 519 | *dest++=uc; | |||
| 520 | } | |||
| 521 | --destSize; | |||
| 522 | } | |||
| 523 | } else { /* RTL run */ | |||
| 524 | if((pBiDi->isInverse) && | |||
| 525 | (/*run>0 &&*/ !(MASK_R_AL((1UL<<(R))|(1UL<<(AL)))&DIRPROP_FLAG(dirProps[logicalStart+runLength-1])(1UL<<(dirProps[logicalStart+runLength-1]))))) { | |||
| 526 | markFlag |= RLM_BEFORE; | |||
| 527 | } | |||
| 528 | if (markFlag & LRM_BEFORE) { | |||
| 529 | uc=LRM_CHAR; | |||
| 530 | } | |||
| 531 | else if (markFlag & RLM_BEFORE) { | |||
| 532 | uc=RLM_CHAR; | |||
| 533 | } | |||
| 534 | else uc=0; | |||
| 535 | if(uc) { | |||
| 536 | if(destSize>0) { | |||
| 537 | *dest++=uc; | |||
| 538 | } | |||
| 539 | --destSize; | |||
| 540 | } | |||
| 541 | ||||
| 542 | runLength=doWriteReverse(src, runLength, | |||
| 543 | dest, destSize, | |||
| 544 | options, pErrorCode); | |||
| 545 | if(dest!=nullptr) { | |||
| 546 | dest+=runLength; | |||
| 547 | } | |||
| 548 | destSize-=runLength; | |||
| 549 | ||||
| 550 | if((pBiDi->isInverse) && | |||
| 551 | (/*run<runCount-1 &&*/ !(MASK_R_AL((1UL<<(R))|(1UL<<(AL)))&DIRPROP_FLAG(dirProps[logicalStart])(1UL<<(dirProps[logicalStart]))))) { | |||
| 552 | markFlag |= RLM_AFTER; | |||
| 553 | } | |||
| 554 | if (markFlag & LRM_AFTER) { | |||
| 555 | uc=LRM_CHAR; | |||
| 556 | } | |||
| 557 | else if (markFlag & RLM_AFTER) { | |||
| 558 | uc=RLM_CHAR; | |||
| 559 | } | |||
| 560 | else uc=0; | |||
| 561 | if(uc) { | |||
| 562 | if(destSize>0) { | |||
| 563 | *dest++=uc; | |||
| 564 | } | |||
| 565 | --destSize; | |||
| 566 | } | |||
| 567 | } | |||
| 568 | } | |||
| 569 | } | |||
| 570 | } else { | |||
| 571 | /* reverse output */ | |||
| 572 | if(!(options&UBIDI_INSERT_LRM_FOR_NUMERIC4)) { | |||
| 573 | /* do not insert BiDi controls */ | |||
| 574 | for(run=runCount; --run>=0;) { | |||
| 575 | if(UBIDI_LTR==ubidi_getVisualRunubidi_getVisualRun_77(pBiDi, run, &logicalStart, &runLength)) { | |||
| 576 | runLength=doWriteReverse(text+logicalStart, runLength, | |||
| 577 | dest, destSize, | |||
| 578 | (uint16_t)(options&~UBIDI_DO_MIRRORING2), pErrorCode); | |||
| 579 | } else { | |||
| 580 | runLength=doWriteForward(text+logicalStart, runLength, | |||
| 581 | dest, destSize, | |||
| 582 | options, pErrorCode); | |||
| 583 | } | |||
| 584 | if(dest!=nullptr) { | |||
| 585 | dest+=runLength; | |||
| 586 | } | |||
| 587 | destSize-=runLength; | |||
| 588 | } | |||
| 589 | } else { | |||
| 590 | /* insert BiDi controls for "inverse BiDi" */ | |||
| 591 | const DirProp *dirProps=pBiDi->dirProps; | |||
| 592 | const char16_t *src; | |||
| 593 | UBiDiDirection dir; | |||
| 594 | ||||
| 595 | for(run=runCount; --run>=0;) { | |||
| 596 | /* reverse output */ | |||
| 597 | dir=ubidi_getVisualRunubidi_getVisualRun_77(pBiDi, run, &logicalStart, &runLength); | |||
| 598 | src=text+logicalStart; | |||
| 599 | ||||
| 600 | if(UBIDI_LTR==dir) { | |||
| 601 | if(/*run<runCount-1 &&*/ dirProps[logicalStart+runLength-1]!=L) { | |||
| 602 | if(destSize>0) { | |||
| 603 | *dest++=LRM_CHAR; | |||
| 604 | } | |||
| 605 | --destSize; | |||
| 606 | } | |||
| 607 | ||||
| 608 | runLength=doWriteReverse(src, runLength, | |||
| 609 | dest, destSize, | |||
| 610 | (uint16_t)(options&~UBIDI_DO_MIRRORING2), pErrorCode); | |||
| 611 | if(dest!=nullptr) { | |||
| 612 | dest+=runLength; | |||
| 613 | } | |||
| 614 | destSize-=runLength; | |||
| 615 | ||||
| 616 | if(/*run>0 &&*/ dirProps[logicalStart]!=L) { | |||
| 617 | if(destSize>0) { | |||
| 618 | *dest++=LRM_CHAR; | |||
| 619 | } | |||
| 620 | --destSize; | |||
| 621 | } | |||
| 622 | } else { | |||
| 623 | if(/*run<runCount-1 &&*/ !(MASK_R_AL((1UL<<(R))|(1UL<<(AL)))&DIRPROP_FLAG(dirProps[logicalStart])(1UL<<(dirProps[logicalStart])))) { | |||
| 624 | if(destSize>0) { | |||
| 625 | *dest++=RLM_CHAR; | |||
| 626 | } | |||
| 627 | --destSize; | |||
| 628 | } | |||
| 629 | ||||
| 630 | runLength=doWriteForward(src, runLength, | |||
| 631 | dest, destSize, | |||
| 632 | options, pErrorCode); | |||
| 633 | if(dest!=nullptr) { | |||
| 634 | dest+=runLength; | |||
| 635 | } | |||
| 636 | destSize-=runLength; | |||
| 637 | ||||
| 638 | if(/*run>0 &&*/ | |||
| 639 | runLength > 0 && // doWriteForward may return 0 if src | |||
| 640 | // only include bidi control chars | |||
| 641 | !(MASK_R_AL((1UL<<(R))|(1UL<<(AL)))&DIRPROP_FLAG(dirProps[logicalStart+runLength-1])(1UL<<(dirProps[logicalStart+runLength-1])))) { | |||
| 642 | if(destSize>0) { | |||
| 643 | *dest++=RLM_CHAR; | |||
| 644 | } | |||
| 645 | --destSize; | |||
| 646 | } | |||
| 647 | } | |||
| 648 | } | |||
| 649 | } | |||
| 650 | } | |||
| 651 | ||||
| 652 | return u_terminateUCharsu_terminateUChars_77(saveDest, destCapacity, destCapacity-destSize, pErrorCode); | |||
| 653 | } | |||
| 654 | #if (defined(_MSC_VER) && (defined(_M_ARM64)) && (_MSC_VER < 1924)) | |||
| 655 | #pragma optimize( "", on ) | |||
| 656 | #endif |