Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 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) 2009-2014, International Business Machines |
| 7 | * Corporation and others. All Rights Reserved. |
| 8 | * |
| 9 | ******************************************************************************* |
| 10 | * file name: bidiconf.cpp |
| 11 | * encoding: UTF-8 |
| 12 | * tab size: 8 (not used) |
| 13 | * indentation:4 |
| 14 | * |
| 15 | * created on: 2009oct16 |
| 16 | * created by: Markus W. Scherer |
| 17 | * |
| 18 | * BiDi conformance test, using the Unicode BidiTest.txt and BidiCharacterTest.txt files. |
| 19 | */ |
| 20 | |
| 21 | #include <stdio.h> |
| 22 | #include <stdlib.h> |
| 23 | #include <string.h> |
| 24 | #include "unicode/utypes.h" |
| 25 | #include "unicode/ubidi.h" |
| 26 | #include "unicode/errorcode.h" |
| 27 | #include "unicode/localpointer.h" |
| 28 | #include "unicode/putil.h" |
| 29 | #include "unicode/unistr.h" |
| 30 | #include "intltest.h" |
| 31 | #include "uparse.h" |
| 32 | |
| 33 | class BiDiConformanceTest : public IntlTest { |
| 34 | public: |
| 35 | BiDiConformanceTest() : |
| 36 | directionBits(0), lineNumber(0), levelsCount(0), orderingCount(0), |
| 37 | errorCount(0) {} |
| 38 | |
| 39 | void runIndexedTest(int32_t index, UBool exec, const char *&name, char *par=NULL) override; |
| 40 | |
| 41 | void TestBidiTest(); |
| 42 | void TestBidiCharacterTest(); |
| 43 | private: |
| 44 | UBool parseLevels(const char *&start); |
| 45 | UBool parseOrdering(const char *start); |
| 46 | UBool parseInputStringFromBiDiClasses(const char *&start); |
| 47 | |
| 48 | UBool checkLevels(const UBiDiLevel actualLevels[], int32_t actualCount); |
| 49 | UBool checkOrdering(UBiDi *ubidi); |
| 50 | |
| 51 | void printErrorLine(); |
| 52 | |
| 53 | char line[10000]; |
| 54 | UBiDiLevel levels[1000]; |
| 55 | uint32_t directionBits; |
| 56 | int32_t ordering[1000]; |
| 57 | int32_t lineNumber; |
| 58 | int32_t levelsCount; |
| 59 | int32_t orderingCount; |
| 60 | int32_t errorCount; |
| 61 | UnicodeString inputString; |
| 62 | const char *paraLevelName; |
| 63 | char levelNameString[12]; |
| 64 | }; |
| 65 | |
| 66 | extern IntlTest *createBiDiConformanceTest() { |
| 67 | return new BiDiConformanceTest(); |
| 68 | } |
| 69 | |
| 70 | void BiDiConformanceTest::runIndexedTest(int32_t index, UBool exec, const char *&name, char * /*par*/) { |
| 71 | if(exec) { |
| 72 | logln("TestSuite BiDiConformanceTest: "); |
| 73 | } |
| 74 | TESTCASE_AUTO_BEGIN; |
| 75 | TESTCASE_AUTO(TestBidiTest); |
| 76 | TESTCASE_AUTO(TestBidiCharacterTest); |
| 77 | TESTCASE_AUTO_END; |
| 78 | } |
| 79 | |
| 80 | U_DEFINE_LOCAL_OPEN_POINTER(LocalStdioFilePointer, FILE, fclose); |
| 81 | |
| 82 | UBool BiDiConformanceTest::parseLevels(const char *&start) { |
| 83 | directionBits=0; |
| 84 | levelsCount=0; |
| 85 | while(*start!=0 && *(start=u_skipWhitespace(start))!=0 && *start!=';') { |
| 86 | if(*start=='x') { |
| 87 | levels[levelsCount++]=UBIDI_DEFAULT_LTR; |
| 88 | ++start; |
| 89 | } else { |
| 90 | char *end; |
| 91 | uint32_t value=(uint32_t)strtoul(start, &end, 10); |
| 92 | if(end<=start || (!U_IS_INV_WHITESPACE(*end) && *end!=0 && *end!=';') |
| 93 | || value>(UBIDI_MAX_EXPLICIT_LEVEL+1)) { |
| 94 | errln("\nError on line %d: Levels parse error at %s", (int)lineNumber, start); |
| 95 | printErrorLine(); |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 96 | return false; |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 97 | } |
| 98 | levels[levelsCount++]=(UBiDiLevel)value; |
| 99 | directionBits|=(1<<(value&1)); |
| 100 | start=end; |
| 101 | } |
| 102 | } |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 103 | return true; |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 104 | } |
| 105 | |
| 106 | UBool BiDiConformanceTest::parseOrdering(const char *start) { |
| 107 | orderingCount=0; |
| 108 | while(*start!=0 && *(start=u_skipWhitespace(start))!=0 && *start!=';') { |
| 109 | char *end; |
| 110 | uint32_t value=(uint32_t)strtoul(start, &end, 10); |
| 111 | if(end<=start || (!U_IS_INV_WHITESPACE(*end) && *end!=0 && *end!=';') || value>=1000) { |
| 112 | errln("\nError on line %d: Reorder parse error at %s", (int)lineNumber, start); |
| 113 | printErrorLine(); |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 114 | return false; |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 115 | } |
| 116 | ordering[orderingCount++]=(int32_t)value; |
| 117 | start=end; |
| 118 | } |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 119 | return true; |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 120 | } |
| 121 | |
| 122 | static const UChar charFromBiDiClass[U_CHAR_DIRECTION_COUNT]={ |
| 123 | 0x6c, // 'l' for L |
| 124 | 0x52, // 'R' for R |
| 125 | 0x33, // '3' for EN |
| 126 | 0x2d, // '-' for ES |
| 127 | 0x25, // '%' for ET |
| 128 | 0x39, // '9' for AN |
| 129 | 0x2c, // ',' for CS |
| 130 | 0x2f, // '/' for B |
| 131 | 0x5f, // '_' for S |
| 132 | 0x20, // ' ' for WS |
| 133 | 0x3d, // '=' for ON |
| 134 | 0x65, // 'e' for LRE |
| 135 | 0x6f, // 'o' for LRO |
| 136 | 0x41, // 'A' for AL |
| 137 | 0x45, // 'E' for RLE |
| 138 | 0x4f, // 'O' for RLO |
| 139 | 0x2a, // '*' for PDF |
| 140 | 0x60, // '`' for NSM |
| 141 | 0x7c, // '|' for BN |
| 142 | // new in Unicode 6.3/ICU 52 |
| 143 | 0x53, // 'S' for FSI |
| 144 | 0x69, // 'i' for LRI |
| 145 | 0x49, // 'I' for RLI |
| 146 | 0x2e // '.' for PDI |
| 147 | }; |
| 148 | |
| 149 | U_CDECL_BEGIN |
| 150 | |
| 151 | static UCharDirection U_CALLCONV |
| 152 | biDiConfUBiDiClassCallback(const void * /*context*/, UChar32 c) { |
| 153 | for(int i=0; i<U_CHAR_DIRECTION_COUNT; ++i) { |
| 154 | if(c==charFromBiDiClass[i]) { |
| 155 | return (UCharDirection)i; |
| 156 | } |
| 157 | } |
| 158 | // Character not in our hardcoded table. |
| 159 | // Should not occur during testing. |
| 160 | return U_BIDI_CLASS_DEFAULT; |
| 161 | } |
| 162 | |
| 163 | U_CDECL_END |
| 164 | |
| 165 | static const int8_t biDiClassNameLengths[U_CHAR_DIRECTION_COUNT+1]={ |
| 166 | 1, 1, 2, 2, 2, 2, 2, 1, 1, 2, 2, 3, 3, 2, 3, 3, 3, 3, 2, 3, 3, 3, 3, 0 |
| 167 | }; |
| 168 | |
| 169 | UBool BiDiConformanceTest::parseInputStringFromBiDiClasses(const char *&start) { |
| 170 | inputString.remove(); |
| 171 | /* |
| 172 | * Lengthy but fast BiDi class parser. |
| 173 | * A simple parser could terminate or extract the name string and use |
| 174 | * int32_t biDiClassInt=u_getPropertyValueEnum(UCHAR_BIDI_CLASS, bidiClassString); |
| 175 | * but that makes this test take significantly more time. |
| 176 | */ |
| 177 | while(*start!=0 && *(start=u_skipWhitespace(start))!=0 && *start!=';') { |
| 178 | UCharDirection biDiClass=U_CHAR_DIRECTION_COUNT; |
| 179 | // Compare each character once until we have a match on |
| 180 | // a complete, short BiDi class name. |
| 181 | if(start[0]=='L') { |
| 182 | if(start[1]=='R') { |
| 183 | if(start[2]=='E') { |
| 184 | biDiClass=U_LEFT_TO_RIGHT_EMBEDDING; |
| 185 | } else if(start[2]=='I') { |
| 186 | biDiClass=U_LEFT_TO_RIGHT_ISOLATE; |
| 187 | } else if(start[2]=='O') { |
| 188 | biDiClass=U_LEFT_TO_RIGHT_OVERRIDE; |
| 189 | } |
| 190 | } else { |
| 191 | biDiClass=U_LEFT_TO_RIGHT; |
| 192 | } |
| 193 | } else if(start[0]=='R') { |
| 194 | if(start[1]=='L') { |
| 195 | if(start[2]=='E') { |
| 196 | biDiClass=U_RIGHT_TO_LEFT_EMBEDDING; |
| 197 | } else if(start[2]=='I') { |
| 198 | biDiClass=U_RIGHT_TO_LEFT_ISOLATE; |
| 199 | } else if(start[2]=='O') { |
| 200 | biDiClass=U_RIGHT_TO_LEFT_OVERRIDE; |
| 201 | } |
| 202 | } else { |
| 203 | biDiClass=U_RIGHT_TO_LEFT; |
| 204 | } |
| 205 | } else if(start[0]=='E') { |
| 206 | if(start[1]=='N') { |
| 207 | biDiClass=U_EUROPEAN_NUMBER; |
| 208 | } else if(start[1]=='S') { |
| 209 | biDiClass=U_EUROPEAN_NUMBER_SEPARATOR; |
| 210 | } else if(start[1]=='T') { |
| 211 | biDiClass=U_EUROPEAN_NUMBER_TERMINATOR; |
| 212 | } |
| 213 | } else if(start[0]=='A') { |
| 214 | if(start[1]=='L') { |
| 215 | biDiClass=U_RIGHT_TO_LEFT_ARABIC; |
| 216 | } else if(start[1]=='N') { |
| 217 | biDiClass=U_ARABIC_NUMBER; |
| 218 | } |
| 219 | } else if(start[0]=='C' && start[1]=='S') { |
| 220 | biDiClass=U_COMMON_NUMBER_SEPARATOR; |
| 221 | } else if(start[0]=='B') { |
| 222 | if(start[1]=='N') { |
| 223 | biDiClass=U_BOUNDARY_NEUTRAL; |
| 224 | } else { |
| 225 | biDiClass=U_BLOCK_SEPARATOR; |
| 226 | } |
| 227 | } else if(start[0]=='S') { |
| 228 | biDiClass=U_SEGMENT_SEPARATOR; |
| 229 | } else if(start[0]=='W' && start[1]=='S') { |
| 230 | biDiClass=U_WHITE_SPACE_NEUTRAL; |
| 231 | } else if(start[0]=='O' && start[1]=='N') { |
| 232 | biDiClass=U_OTHER_NEUTRAL; |
| 233 | } else if(start[0]=='P' && start[1]=='D') { |
| 234 | if(start[2]=='F') { |
| 235 | biDiClass=U_POP_DIRECTIONAL_FORMAT; |
| 236 | } else if(start[2]=='I') { |
| 237 | biDiClass=U_POP_DIRECTIONAL_ISOLATE; |
| 238 | } |
| 239 | } else if(start[0]=='N' && start[1]=='S' && start[2]=='M') { |
| 240 | biDiClass=U_DIR_NON_SPACING_MARK; |
| 241 | } else if(start[0]=='F' && start[1]=='S' && start[2]=='I') { |
| 242 | biDiClass=U_FIRST_STRONG_ISOLATE; |
| 243 | } |
| 244 | // Now we verify that the class name is terminated properly, |
| 245 | // and not just the start of a longer word. |
| 246 | int8_t biDiClassNameLength=biDiClassNameLengths[biDiClass]; |
| 247 | char c=start[biDiClassNameLength]; |
| 248 | if(biDiClass<U_CHAR_DIRECTION_COUNT && (U_IS_INV_WHITESPACE(c) || c==';' || c==0)) { |
| 249 | inputString.append(charFromBiDiClass[biDiClass]); |
| 250 | start+=biDiClassNameLength; |
| 251 | continue; |
| 252 | } |
| 253 | errln("\nError on line %d: BiDi class string not recognized at %s", (int)lineNumber, start); |
| 254 | printErrorLine(); |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 255 | return false; |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 256 | } |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 257 | return true; |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 258 | } |
| 259 | |
| 260 | void BiDiConformanceTest::TestBidiTest() { |
| 261 | IcuTestErrorCode errorCode(*this, "TestBidiTest"); |
| 262 | const char *sourceTestDataPath=getSourceTestData(errorCode); |
| 263 | if(errorCode.errIfFailureAndReset("unable to find the source/test/testdata " |
| 264 | "folder (getSourceTestData())")) { |
| 265 | return; |
| 266 | } |
| 267 | char bidiTestPath[400]; |
| 268 | strcpy(bidiTestPath, sourceTestDataPath); |
| 269 | strcat(bidiTestPath, "BidiTest.txt"); |
| 270 | LocalStdioFilePointer bidiTestFile(fopen(bidiTestPath, "r")); |
| 271 | if(bidiTestFile.isNull()) { |
| 272 | errln("unable to open %s", bidiTestPath); |
| 273 | return; |
| 274 | } |
| 275 | LocalUBiDiPointer ubidi(ubidi_open()); |
| 276 | ubidi_setClassCallback(ubidi.getAlias(), biDiConfUBiDiClassCallback, NULL, |
| 277 | NULL, NULL, errorCode); |
| 278 | if(errorCode.errIfFailureAndReset("ubidi_setClassCallback()")) { |
| 279 | return; |
| 280 | } |
| 281 | lineNumber=0; |
| 282 | levelsCount=0; |
| 283 | orderingCount=0; |
| 284 | errorCount=0; |
| 285 | // paraLevelName must be initialized in case the first non-comment line is in error |
| 286 | paraLevelName="N/A"; |
| 287 | while(errorCount<10 && fgets(line, (int)sizeof(line), bidiTestFile.getAlias())!=NULL) { |
| 288 | ++lineNumber; |
| 289 | // Remove trailing comments and whitespace. |
| 290 | char *commentStart=strchr(line, '#'); |
| 291 | if(commentStart!=NULL) { |
| 292 | *commentStart=0; |
| 293 | } |
| 294 | u_rtrim(line); |
| 295 | const char *start=u_skipWhitespace(line); |
| 296 | if(*start==0) { |
| 297 | continue; // Skip empty and comment-only lines. |
| 298 | } |
| 299 | if(*start=='@') { |
| 300 | ++start; |
| 301 | if(0==strncmp(start, "Levels:", 7)) { |
| 302 | start+=7; |
| 303 | if(!parseLevels(start)) { |
| 304 | return; |
| 305 | } |
| 306 | } else if(0==strncmp(start, "Reorder:", 8)) { |
| 307 | if(!parseOrdering(start+8)) { |
| 308 | return; |
| 309 | } |
| 310 | } |
| 311 | // Skip unknown @Xyz: ... |
| 312 | } else { |
| 313 | if(!parseInputStringFromBiDiClasses(start)) { |
| 314 | return; |
| 315 | } |
| 316 | start=u_skipWhitespace(start); |
| 317 | if(*start!=';') { |
| 318 | errln("missing ; separator on input line %s", line); |
| 319 | return; |
| 320 | } |
| 321 | start=u_skipWhitespace(start+1); |
| 322 | char *end; |
| 323 | uint32_t bitset=(uint32_t)strtoul(start, &end, 16); |
| 324 | if(end<=start || (!U_IS_INV_WHITESPACE(*end) && *end!=';' && *end!=0)) { |
| 325 | errln("input bitset parse error at %s", start); |
| 326 | return; |
| 327 | } |
| 328 | // Loop over the bitset. |
| 329 | static const UBiDiLevel paraLevels[]={ UBIDI_DEFAULT_LTR, 0, 1, UBIDI_DEFAULT_RTL }; |
| 330 | static const char *const paraLevelNames[]={ "auto/LTR", "LTR", "RTL", "auto/RTL" }; |
| 331 | for(int i=0; i<=3; ++i) { |
| 332 | if(bitset&(1<<i)) { |
| 333 | ubidi_setPara(ubidi.getAlias(), inputString.getBuffer(), inputString.length(), |
| 334 | paraLevels[i], NULL, errorCode); |
| 335 | const UBiDiLevel *actualLevels=ubidi_getLevels(ubidi.getAlias(), errorCode); |
| 336 | if(errorCode.errIfFailureAndReset("ubidi_setPara() or ubidi_getLevels()")) { |
| 337 | errln("Input line %d: %s", (int)lineNumber, line); |
| 338 | return; |
| 339 | } |
| 340 | paraLevelName=paraLevelNames[i]; |
| 341 | if(!checkLevels(actualLevels, ubidi_getProcessedLength(ubidi.getAlias()))) { |
| 342 | // continue outerLoop; does not exist in C++ |
| 343 | // so just break out of the inner loop. |
| 344 | break; |
| 345 | } |
| 346 | if(!checkOrdering(ubidi.getAlias())) { |
| 347 | // continue outerLoop; does not exist in C++ |
| 348 | // so just break out of the inner loop. |
| 349 | break; |
| 350 | } |
| 351 | } |
| 352 | } |
| 353 | } |
| 354 | } |
| 355 | } |
| 356 | |
| 357 | /* |
| 358 | ******************************************************************************* |
| 359 | * |
| 360 | * created on: 2013jul01 |
| 361 | * created by: Matitiahu Allouche |
| 362 | |
| 363 | This function performs a conformance test for implementations of the |
| 364 | Unicode Bidirectional Algorithm, specified in UAX #9: Unicode |
| 365 | Bidirectional Algorithm, at https://www.unicode.org/reports/tr9/ |
| 366 | |
| 367 | Each test case is represented in a single line which is read from a file |
| 368 | named BidiCharacter.txt. Empty, blank and comment lines may also appear |
| 369 | in this file. |
| 370 | |
| 371 | The format of the test data is specified below. Note that each test |
| 372 | case constitutes a single line of text; reordering is applied within a |
| 373 | single line and independently of a rendering engine, and rules L3 and L4 |
| 374 | are out of scope. |
| 375 | |
| 376 | The number sign '#' is the comment character: everything is ignored from |
| 377 | the occurrence of '#' until the end of the line, |
| 378 | Empty lines and lines containing only spaces and/or comments are ignored. |
| 379 | |
| 380 | Lines which represent test cases consist of 4 or 5 fields separated by a |
| 381 | semicolon. Each field consists of tokens separated by whitespace (space |
| 382 | or Tab). Whitespace before and after semicolons is optional. |
| 383 | |
| 384 | Field 0: A sequence of hexadecimal code point values separated by space |
| 385 | |
| 386 | Field 1: A value representing the paragraph direction, as follows: |
| 387 | - 0 represents left-to-right |
| 388 | - 1 represents right-to-left |
| 389 | - 2 represents auto-LTR according to rules P2 and P3 of the algorithm |
| 390 | - 3 represents auto-RTL according to rules P2 and P3 of the algorithm |
| 391 | - a negative number whose absolute value is taken as paragraph level; |
| 392 | this may be useful to test cases where the embedding level approaches |
| 393 | or exceeds the maximum embedding level. |
| 394 | |
| 395 | Field 2: The resolved paragraph embedding level. If the input (field 0) |
| 396 | includes more than one paragraph, this field represents the |
| 397 | resolved level of the first paragraph. |
| 398 | |
| 399 | Field 3: An ordered list of resulting levels for each token in field 0 |
| 400 | (each token represents one source character). |
| 401 | The UBA does not assign levels to certain characters (e.g. LRO); |
| 402 | characters removed in rule X9 are indicated with an 'x'. |
| 403 | |
| 404 | Field 4: An ordered list of indices showing the resulting visual ordering |
| 405 | from left to right; characters with a resolved level of 'x' are |
| 406 | skipped. The number are zero-based. Each index corresponds to |
| 407 | a character in the reordered (visual) string. It represents the |
| 408 | index of the source character in the input (field 0). |
| 409 | This field is optional. When it is absent, the visual ordering |
| 410 | is not verified. |
| 411 | |
| 412 | Examples: |
| 413 | |
| 414 | # This is a comment line. |
| 415 | L L ON R ; 0 ; 0 ; 0 0 0 1 ; 0 1 2 3 |
| 416 | L L ON R;0;0;0 0 0 1;0 1 2 3 |
| 417 | |
| 418 | # Note: in the next line, 'B' represents a block separator, not the letter 'B'. |
| 419 | LRE A B C PDF;2;0;x 2 0 0 x;1 2 3 |
| 420 | # Note: in the next line, 'b' represents the letter 'b', not a block separator. |
| 421 | a b c 05d0 05d1 x ; 0 ; 0 ; 0 0 0 1 1 0 ; 0 1 2 4 3 5 |
| 422 | |
| 423 | a R R x ; 1 ; 1 ; 2 1 1 2 |
| 424 | L L R R R B R R L L L B ON ON ; 3 ; 0 ; 0 0 1 1 1 0 1 1 2 2 2 1 1 1 |
| 425 | |
| 426 | * |
| 427 | ******************************************************************************* |
| 428 | */ |
| 429 | void BiDiConformanceTest::TestBidiCharacterTest() { |
| 430 | IcuTestErrorCode errorCode(*this, "TestBidiCharacterTest"); |
| 431 | const char *sourceTestDataPath=getSourceTestData(errorCode); |
| 432 | if(errorCode.errIfFailureAndReset("unable to find the source/test/testdata " |
| 433 | "folder (getSourceTestData())")) { |
| 434 | return; |
| 435 | } |
| 436 | char bidiTestPath[400]; |
| 437 | strcpy(bidiTestPath, sourceTestDataPath); |
| 438 | strcat(bidiTestPath, "BidiCharacterTest.txt"); |
| 439 | LocalStdioFilePointer bidiTestFile(fopen(bidiTestPath, "r")); |
| 440 | if(bidiTestFile.isNull()) { |
| 441 | errln("unable to open %s", bidiTestPath); |
| 442 | return; |
| 443 | } |
| 444 | LocalUBiDiPointer ubidi(ubidi_open()); |
| 445 | lineNumber=0; |
| 446 | levelsCount=0; |
| 447 | orderingCount=0; |
| 448 | errorCount=0; |
| 449 | while(errorCount<20 && fgets(line, (int)sizeof(line), bidiTestFile.getAlias())!=NULL) { |
| 450 | ++lineNumber; |
| 451 | paraLevelName="N/A"; |
| 452 | inputString="N/A"; |
| 453 | // Remove trailing comments and whitespace. |
| 454 | char *commentStart=strchr(line, '#'); |
| 455 | if(commentStart!=NULL) { |
| 456 | *commentStart=0; |
| 457 | } |
| 458 | u_rtrim(line); |
| 459 | const char *start=u_skipWhitespace(line); |
| 460 | if(*start==0) { |
| 461 | continue; // Skip empty and comment-only lines. |
| 462 | } |
| 463 | // Parse the code point string in field 0. |
| 464 | UChar *buffer=inputString.getBuffer(200); |
| 465 | int32_t length=u_parseString(start, buffer, inputString.getCapacity(), NULL, errorCode); |
| 466 | if(errorCode.errIfFailureAndReset("Invalid string in field 0")) { |
| 467 | errln("Input line %d: %s", (int)lineNumber, line); |
| 468 | inputString.remove(); |
| 469 | continue; |
| 470 | } |
| 471 | inputString.releaseBuffer(length); |
| 472 | start=strchr(start, ';'); |
| 473 | if(start==NULL) { |
| 474 | errorCount++; |
| 475 | errln("\nError on line %d: Missing ; separator on line: %s", (int)lineNumber, line); |
| 476 | continue; |
| 477 | } |
| 478 | start=u_skipWhitespace(start+1); |
| 479 | char *end; |
| 480 | int32_t paraDirection=(int32_t)strtol(start, &end, 10); |
| 481 | UBiDiLevel paraLevel=UBIDI_MAX_EXPLICIT_LEVEL+2; |
| 482 | if(paraDirection==0) { |
| 483 | paraLevel=0; |
| 484 | paraLevelName="LTR"; |
| 485 | } |
| 486 | else if(paraDirection==1) { |
| 487 | paraLevel=1; |
| 488 | paraLevelName="RTL"; |
| 489 | } |
| 490 | else if(paraDirection==2) { |
| 491 | paraLevel=UBIDI_DEFAULT_LTR; |
| 492 | paraLevelName="Auto/LTR"; |
| 493 | } |
| 494 | else if(paraDirection==3) { |
| 495 | paraLevel=UBIDI_DEFAULT_RTL; |
| 496 | paraLevelName="Auto/RTL"; |
| 497 | } |
| 498 | else if(paraDirection<0 && -paraDirection<=(UBIDI_MAX_EXPLICIT_LEVEL+1)) { |
| 499 | paraLevel=(UBiDiLevel)(-paraDirection); |
| 500 | sprintf(levelNameString, "%d", (int)paraLevel); |
| 501 | paraLevelName=levelNameString; |
| 502 | } |
| 503 | if(end<=start || (!U_IS_INV_WHITESPACE(*end) && *end!=';' && *end!=0) || |
| 504 | paraLevel==(UBIDI_MAX_EXPLICIT_LEVEL+2)) { |
| 505 | errln("\nError on line %d: Input paragraph direction incorrect at %s", (int)lineNumber, start); |
| 506 | printErrorLine(); |
| 507 | continue; |
| 508 | } |
| 509 | start=u_skipWhitespace(end); |
| 510 | if(*start!=';') { |
| 511 | errorCount++; |
| 512 | errln("\nError on line %d: Missing ; separator on line: %s", (int)lineNumber, line); |
| 513 | continue; |
| 514 | } |
| 515 | start++; |
| 516 | uint32_t resolvedParaLevel=(uint32_t)strtoul(start, &end, 10); |
| 517 | if(end<=start || (!U_IS_INV_WHITESPACE(*end) && *end!=';' && *end!=0) || |
| 518 | resolvedParaLevel>1) { |
| 519 | errln("\nError on line %d: Resolved paragraph level incorrect at %s", (int)lineNumber, start); |
| 520 | printErrorLine(); |
| 521 | continue; |
| 522 | } |
| 523 | start=u_skipWhitespace(end); |
| 524 | if(*start!=';') { |
| 525 | errorCount++; |
| 526 | errln("\nError on line %d: Missing ; separator on line: %s", (int)lineNumber, line); |
| 527 | return; |
| 528 | } |
| 529 | start++; |
| 530 | if(!parseLevels(start)) { |
| 531 | continue; |
| 532 | } |
| 533 | start=u_skipWhitespace(start); |
| 534 | if(*start==';') { |
| 535 | if(!parseOrdering(start+1)) { |
| 536 | continue; |
| 537 | } |
| 538 | } |
| 539 | else |
| 540 | orderingCount=-1; |
| 541 | |
| 542 | ubidi_setPara(ubidi.getAlias(), inputString.getBuffer(), inputString.length(), |
| 543 | paraLevel, NULL, errorCode); |
| 544 | const UBiDiLevel *actualLevels=ubidi_getLevels(ubidi.getAlias(), errorCode); |
| 545 | if(errorCode.errIfFailureAndReset("ubidi_setPara() or ubidi_getLevels()")) { |
| 546 | errln("Input line %d: %s", (int)lineNumber, line); |
| 547 | continue; |
| 548 | } |
| 549 | UBiDiLevel actualLevel; |
| 550 | if((actualLevel=ubidi_getParaLevel(ubidi.getAlias()))!=resolvedParaLevel) { |
| 551 | printErrorLine(); |
| 552 | errln("\nError on line %d: Wrong resolved paragraph level; expected %d actual %d", |
| 553 | (int)lineNumber, resolvedParaLevel, actualLevel); |
| 554 | continue; |
| 555 | } |
| 556 | if(!checkLevels(actualLevels, ubidi_getProcessedLength(ubidi.getAlias()))) { |
| 557 | continue; |
| 558 | } |
| 559 | if(orderingCount>=0 && !checkOrdering(ubidi.getAlias())) { |
| 560 | continue; |
| 561 | } |
| 562 | } |
| 563 | } |
| 564 | |
| 565 | static UChar printLevel(UBiDiLevel level) { |
| 566 | if(level<UBIDI_DEFAULT_LTR) { |
| 567 | return 0x30+level; |
| 568 | } else { |
| 569 | return 0x78; // 'x' |
| 570 | } |
| 571 | } |
| 572 | |
| 573 | static uint32_t getDirectionBits(const UBiDiLevel actualLevels[], int32_t actualCount) { |
| 574 | uint32_t actualDirectionBits=0; |
| 575 | for(int32_t i=0; i<actualCount; ++i) { |
| 576 | actualDirectionBits|=(1<<(actualLevels[i]&1)); |
| 577 | } |
| 578 | return actualDirectionBits; |
| 579 | } |
| 580 | |
| 581 | UBool BiDiConformanceTest::checkLevels(const UBiDiLevel actualLevels[], int32_t actualCount) { |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 582 | UBool isOk=true; |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 583 | if(levelsCount!=actualCount) { |
| 584 | errln("\nError on line %d: Wrong number of level values; expected %d actual %d", |
| 585 | (int)lineNumber, (int)levelsCount, (int)actualCount); |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 586 | isOk=false; |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 587 | } else { |
| 588 | for(int32_t i=0; i<actualCount; ++i) { |
| 589 | if(levels[i]!=actualLevels[i] && levels[i]<UBIDI_DEFAULT_LTR) { |
| 590 | if(directionBits!=3 && directionBits==getDirectionBits(actualLevels, actualCount)) { |
| 591 | // ICU used a shortcut: |
| 592 | // Since the text is unidirectional, it did not store the resolved |
| 593 | // levels but just returns all levels as the paragraph level 0 or 1. |
| 594 | // The reordering result is the same, so this is fine. |
| 595 | break; |
| 596 | } else { |
| 597 | errln("\nError on line %d: Wrong level value at index %d; expected %d actual %d", |
| 598 | (int)lineNumber, (int)i, levels[i], actualLevels[i]); |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 599 | isOk=false; |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 600 | break; |
| 601 | } |
| 602 | } |
| 603 | } |
| 604 | } |
| 605 | if(!isOk) { |
| 606 | printErrorLine(); |
| 607 | UnicodeString els("Expected levels: "); |
| 608 | int32_t i; |
| 609 | for(i=0; i<levelsCount; ++i) { |
| 610 | els.append((UChar)0x20).append(printLevel(levels[i])); |
| 611 | } |
| 612 | UnicodeString als("Actual levels: "); |
| 613 | for(i=0; i<actualCount; ++i) { |
| 614 | als.append((UChar)0x20).append(printLevel(actualLevels[i])); |
| 615 | } |
| 616 | errln(els); |
| 617 | errln(als); |
| 618 | } |
| 619 | return isOk; |
| 620 | } |
| 621 | |
| 622 | // Note: ubidi_setReorderingOptions(ubidi, UBIDI_OPTION_REMOVE_CONTROLS); |
| 623 | // does not work for custom BiDi class assignments |
| 624 | // and anyway also removes LRM/RLM/ZWJ/ZWNJ which is not desirable here. |
| 625 | // Therefore we just skip the indexes for BiDi controls while comparing |
| 626 | // with the expected ordering that has them omitted. |
| 627 | UBool BiDiConformanceTest::checkOrdering(UBiDi *ubidi) { |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 628 | UBool isOk=true; |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 629 | IcuTestErrorCode errorCode(*this, "checkOrdering()"); |
| 630 | int32_t resultLength=ubidi_getResultLength(ubidi); // visual length including BiDi controls |
| 631 | int32_t i, visualIndex; |
| 632 | // Note: It should be faster to call ubidi_countRuns()/ubidi_getVisualRun() |
| 633 | // and loop over each run's indexes, but that seems unnecessary for this test code. |
| 634 | for(i=visualIndex=0; i<resultLength; ++i) { |
| 635 | int32_t logicalIndex=ubidi_getLogicalIndex(ubidi, i, errorCode); |
| 636 | if(errorCode.errIfFailureAndReset("ubidi_getLogicalIndex()")) { |
| 637 | errln("Input line %d: %s", (int)lineNumber, line); |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 638 | return false; |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 639 | } |
| 640 | if(levels[logicalIndex]>=UBIDI_DEFAULT_LTR) { |
| 641 | continue; // BiDi control, omitted from expected ordering. |
| 642 | } |
| 643 | if(visualIndex<orderingCount && logicalIndex!=ordering[visualIndex]) { |
| 644 | errln("\nError on line %d: Wrong ordering value at visual index %d; expected %d actual %d", |
| 645 | (int)lineNumber, (int)visualIndex, ordering[visualIndex], logicalIndex); |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 646 | isOk=false; |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 647 | break; |
| 648 | } |
| 649 | ++visualIndex; |
| 650 | } |
| 651 | // visualIndex is now the visual length minus the BiDi controls, |
| 652 | // which should match the length of the BidiTest.txt ordering. |
| 653 | if(isOk && orderingCount!=visualIndex) { |
| 654 | errln("\nError on line %d: Wrong number of ordering values; expected %d actual %d", |
| 655 | (int)lineNumber, (int)orderingCount, (int)visualIndex); |
Frank Tang | 1f164ee | 2022-11-08 12:31:27 -0800 | [diff] [blame^] | 656 | isOk=false; |
Frank Tang | 3e05d9d | 2021-11-08 14:04:04 -0800 | [diff] [blame] | 657 | } |
| 658 | if(!isOk) { |
| 659 | printErrorLine(); |
| 660 | UnicodeString eord("Expected ordering: "); |
| 661 | for(i=0; i<orderingCount; ++i) { |
| 662 | eord.append((UChar)0x20).append((UChar)(0x30+ordering[i])); |
| 663 | } |
| 664 | UnicodeString aord("Actual ordering: "); |
| 665 | for(i=0; i<resultLength; ++i) { |
| 666 | int32_t logicalIndex=ubidi_getLogicalIndex(ubidi, i, errorCode); |
| 667 | if(levels[logicalIndex]<UBIDI_DEFAULT_LTR) { |
| 668 | aord.append((UChar)0x20).append((UChar)(0x30+logicalIndex)); |
| 669 | } |
| 670 | } |
| 671 | errln(eord); |
| 672 | errln(aord); |
| 673 | } |
| 674 | return isOk; |
| 675 | } |
| 676 | |
| 677 | void BiDiConformanceTest::printErrorLine() { |
| 678 | ++errorCount; |
| 679 | errln("Input line %5d: %s", (int)lineNumber, line); |
| 680 | errln(UnicodeString("Input string: ")+inputString); |
| 681 | errln("Para level: %s", paraLevelName); |
| 682 | } |