projects / apple / icu.git / blame
| Commit | Line | Data |
|---|---|---|
| b75a7d8f A |
1 | |
| 2 | // | |
| 3 | // file: regexcmp.cpp | |
| 4 | // | |
| 73c04bcf | 5 | // Copyright (C) 2002-2006 International Business Machines Corporation and others. |
| b75a7d8f A |
6 | // All Rights Reserved. |
| 7 | // | |
| 8 | // This file contains the ICU regular expression compiler, which is responsible | |
| 9 | // for processing a regular expression pattern into the compiled form that | |
| 10 | // is used by the match finding engine. | |
| 11 | // | |
| 12 | ||
| 13 | #include "unicode/utypes.h" | |
| 14 | ||
| 15 | #if !UCONFIG_NO_REGULAR_EXPRESSIONS | |
| 16 | ||
| 17 | #include "unicode/unistr.h" | |
| 18 | #include "unicode/uniset.h" | |
| 19 | #include "unicode/uchar.h" | |
| 20 | #include "unicode/uchriter.h" | |
| 21 | #include "unicode/parsepos.h" | |
| 22 | #include "unicode/parseerr.h" | |
| 23 | #include "unicode/regex.h" | |
| 374ca955 | 24 | #include "util.h" |
| b75a7d8f A |
25 | #include "cmemory.h" |
| 26 | #include "cstring.h" | |
| 27 | #include "uvectr32.h" | |
| 28 | #include "uassert.h" | |
| 29 | #include "ucln_in.h" | |
| 30 | #include "mutex.h" | |
| 31 | ||
| 32 | #include "regeximp.h" | |
| 33 | #include "regexcst.h" // Contains state table for the regex pattern parser. | |
| 34 | // generated by a Perl script. | |
| 35 | #include "regexcmp.h" | |
| 36 | #include "regexst.h" | |
| 37 | ||
| 38 | ||
| 39 | ||
| 40 | U_NAMESPACE_BEGIN | |
| 41 | ||
| 42 | ||
| 43 | ||
| 44 | ||
| 45 | ||
| 73c04bcf | 46 | //------------------------------------------------------------------------------ |
| b75a7d8f A |
47 | // |
| 48 | // Constructor. | |
| 49 | // | |
| 73c04bcf | 50 | //------------------------------------------------------------------------------ |
| b75a7d8f A |
51 | RegexCompile::RegexCompile(RegexPattern *rxp, UErrorCode &status) : fParenStack(status) |
| 52 | { | |
| 53 | fStatus = &status; | |
| 54 | ||
| 55 | fRXPat = rxp; | |
| 56 | fScanIndex = 0; | |
| 57 | fNextIndex = 0; | |
| 58 | fPeekChar = -1; | |
| 59 | fLineNum = 1; | |
| 60 | fCharNum = 0; | |
| 61 | fQuoteMode = FALSE; | |
| 62 | fInBackslashQuote = FALSE; | |
| 73c04bcf | 63 | fModeFlags = fRXPat->fFlags | 0x80000000; |
| b75a7d8f A |
64 | fEOLComments = TRUE; |
| 65 | ||
| 66 | fMatchOpenParen = -1; | |
| 67 | fMatchCloseParen = -1; | |
| 68 | fStringOpStart = -1; | |
| 69 | ||
| 70 | if (U_SUCCESS(status) && U_FAILURE(rxp->fDeferredStatus)) { | |
| 71 | status = rxp->fDeferredStatus; | |
| 72 | } | |
| 73 | } | |
| 74 | ||
| 75 | ||
| 76 | ||
| 73c04bcf | 77 | //------------------------------------------------------------------------------ |
| b75a7d8f A |
78 | // |
| 79 | // Destructor | |
| 80 | // | |
| 73c04bcf | 81 | //------------------------------------------------------------------------------ |
| b75a7d8f A |
82 | RegexCompile::~RegexCompile() { |
| 83 | } | |
| 84 | ||
| 73c04bcf | 85 | //------------------------------------------------------------------------------ |
| b75a7d8f A |
86 | // |
| 87 | // Compile regex pattern. The state machine for rexexp pattern parsing is here. | |
| 88 | // The state tables are hand-written in the file regexcst.txt, | |
| 89 | // and converted to the form used here by a perl | |
| 90 | // script regexcst.pl | |
| 91 | // | |
| 73c04bcf | 92 | //------------------------------------------------------------------------------ |
| b75a7d8f A |
93 | void RegexCompile::compile( |
| 94 | const UnicodeString &pat, // Source pat to be compiled. | |
| 95 | UParseError &pp, // Error position info | |
| 96 | UErrorCode &e) // Error Code | |
| 97 | { | |
| 98 | fStatus = &e; | |
| 99 | fParseErr = &pp; | |
| 100 | fStackPtr = 0; | |
| 101 | fStack[fStackPtr] = 0; | |
| 102 | ||
| 103 | if (U_FAILURE(*fStatus)) { | |
| 104 | return; | |
| 105 | } | |
| 106 | ||
| 107 | // There should be no pattern stuff in the RegexPattern object. They can not be reused. | |
| 108 | U_ASSERT(fRXPat->fPattern.length() == 0); | |
| 109 | ||
| 110 | // Prepare the RegexPattern object to receive the compiled pattern. | |
| 111 | // TODO: remove per-instance field, and just use globals directly. (But check perf) | |
| 112 | fRXPat->fPattern = pat; | |
| 113 | fRXPat->fStaticSets = RegexStaticSets::gStaticSets->fPropSets; | |
| 114 | fRXPat->fStaticSets8 = RegexStaticSets::gStaticSets->fPropSets8; | |
| 115 | ||
| 116 | ||
| 117 | // Initialize the pattern scanning state machine | |
| 118 | fPatternLength = pat.length(); | |
| 119 | uint16_t state = 1; | |
| 120 | const RegexTableEl *tableEl; | |
| 121 | nextChar(fC); // Fetch the first char from the pattern string. | |
| 122 | ||
| 123 | // | |
| 124 | // Main loop for the regex pattern parsing state machine. | |
| 125 | // Runs once per state transition. | |
| 126 | // Each time through optionally performs, depending on the state table, | |
| 127 | // - an advance to the the next pattern char | |
| 128 | // - an action to be performed. | |
| 129 | // - pushing or popping a state to/from the local state return stack. | |
| 130 | // file regexcst.txt is the source for the state table. The logic behind | |
| 131 | // recongizing the pattern syntax is there, not here. | |
| 132 | // | |
| 133 | for (;;) { | |
| 134 | // Bail out if anything has gone wrong. | |
| 135 | // Regex pattern parsing stops on the first error encountered. | |
| 136 | if (U_FAILURE(*fStatus)) { | |
| 137 | break; | |
| 138 | } | |
| 139 | ||
| 140 | U_ASSERT(state != 0); | |
| 141 | ||
| 142 | // Find the state table element that matches the input char from the pattern, or the | |
| 143 | // class of the input character. Start with the first table row for this | |
| 144 | // state, then linearly scan forward until we find a row that matches the | |
| 145 | // character. The last row for each state always matches all characters, so | |
| 146 | // the search will stop there, if not before. | |
| 147 | // | |
| 148 | tableEl = &gRuleParseStateTable[state]; | |
| 374ca955 A |
149 | REGEX_SCAN_DEBUG_PRINTF(("char, line, col = (\'%c\', %d, %d) state=%s ", |
| 150 | fC.fChar, fLineNum, fCharNum, RegexStateNames[state])); | |
| b75a7d8f A |
151 | |
| 152 | for (;;) { // loop through table rows belonging to this state, looking for one | |
| 153 | // that matches the current input char. | |
| 374ca955 | 154 | REGEX_SCAN_DEBUG_PRINTF((".")); |
| b75a7d8f A |
155 | if (tableEl->fCharClass < 127 && fC.fQuoted == FALSE && tableEl->fCharClass == fC.fChar) { |
| 156 | // Table row specified an individual character, not a set, and | |
| 157 | // the input character is not quoted, and | |
| 158 | // the input character matched it. | |
| 159 | break; | |
| 160 | } | |
| 161 | if (tableEl->fCharClass == 255) { | |
| 162 | // Table row specified default, match anything character class. | |
| 163 | break; | |
| 164 | } | |
| 165 | if (tableEl->fCharClass == 254 && fC.fQuoted) { | |
| 166 | // Table row specified "quoted" and the char was quoted. | |
| 167 | break; | |
| 168 | } | |
| 169 | if (tableEl->fCharClass == 253 && fC.fChar == (UChar32)-1) { | |
| 170 | // Table row specified eof and we hit eof on the input. | |
| 171 | break; | |
| 172 | } | |
| 173 | ||
| 174 | if (tableEl->fCharClass >= 128 && tableEl->fCharClass < 240 && // Table specs a char class && | |
| 175 | fC.fQuoted == FALSE && // char is not escaped && | |
| 176 | fC.fChar != (UChar32)-1) { // char is not EOF | |
| 177 | UnicodeSet *uniset = RegexStaticSets::gStaticSets->fRuleSets[tableEl->fCharClass-128]; | |
| 178 | if (uniset->contains(fC.fChar)) { | |
| 179 | // Table row specified a character class, or set of characters, | |
| 180 | // and the current char matches it. | |
| 181 | break; | |
| 182 | } | |
| 183 | } | |
| 184 | ||
| 185 | // No match on this row, advance to the next row for this state, | |
| 186 | tableEl++; | |
| 187 | } | |
| 374ca955 | 188 | REGEX_SCAN_DEBUG_PRINTF(("\n")); |
| b75a7d8f A |
189 | |
| 190 | // | |
| 191 | // We've found the row of the state table that matches the current input | |
| 192 | // character from the rules string. | |
| 193 | // Perform any action specified by this row in the state table. | |
| 194 | if (doParseActions((EParseAction)tableEl->fAction) == FALSE) { | |
| 195 | // Break out of the state machine loop if the | |
| 196 | // the action signalled some kind of error, or | |
| 197 | // the action was to exit, occurs on normal end-of-rules-input. | |
| 198 | break; | |
| 199 | } | |
| 200 | ||
| 201 | if (tableEl->fPushState != 0) { | |
| 202 | fStackPtr++; | |
| 203 | if (fStackPtr >= kStackSize) { | |
| 204 | error(U_REGEX_INTERNAL_ERROR); | |
| 374ca955 | 205 | REGEX_SCAN_DEBUG_PRINTF(("RegexCompile::parse() - state stack overflow.\n")); |
| b75a7d8f A |
206 | fStackPtr--; |
| 207 | } | |
| 208 | fStack[fStackPtr] = tableEl->fPushState; | |
| 209 | } | |
| 210 | ||
| 211 | // | |
| 212 | // NextChar. This is where characters are actually fetched from the pattern. | |
| 213 | // Happens under control of the 'n' tag in the state table. | |
| 214 | // | |
| 215 | if (tableEl->fNextChar) { | |
| 216 | nextChar(fC); | |
| 217 | } | |
| 218 | ||
| 219 | // Get the next state from the table entry, or from the | |
| 220 | // state stack if the next state was specified as "pop". | |
| 221 | if (tableEl->fNextState != 255) { | |
| 222 | state = tableEl->fNextState; | |
| 223 | } else { | |
| 224 | state = fStack[fStackPtr]; | |
| 225 | fStackPtr--; | |
| 226 | if (fStackPtr < 0) { | |
| 227 | // state stack underflow | |
| 228 | // This will occur if the user pattern has mis-matched parentheses, | |
| 229 | // with extra close parens. | |
| 230 | // | |
| 231 | fStackPtr++; | |
| 232 | error(U_REGEX_MISMATCHED_PAREN); | |
| 233 | } | |
| 234 | } | |
| 235 | ||
| 236 | } | |
| 237 | ||
| 238 | // | |
| 239 | // The pattern has now been read and processed, and the compiled code generated. | |
| 240 | // | |
| 241 | ||
| 242 | // Back-reference fixup | |
| 243 | // | |
| 244 | int32_t loc; | |
| 245 | for (loc=0; loc<fRXPat->fCompiledPat->size(); loc++) { | |
| 246 | int32_t op = fRXPat->fCompiledPat->elementAti(loc); | |
| 247 | int32_t opType = URX_TYPE(op); | |
| 248 | if (opType == URX_BACKREF || opType == URX_BACKREF_I) { | |
| 249 | int32_t where = URX_VAL(op); | |
| 250 | if (where > fRXPat->fGroupMap->size()) { | |
| 251 | error(U_REGEX_INVALID_BACK_REF); | |
| 252 | break; | |
| 253 | } | |
| 254 | where = fRXPat->fGroupMap->elementAti(where-1); | |
| 255 | op = URX_BUILD(opType, where); | |
| 256 | fRXPat->fCompiledPat->setElementAt(op, loc); | |
| 257 | } | |
| 258 | } | |
| 259 | ||
| 260 | ||
| 261 | // | |
| 262 | // Compute the number of digits requried for the largest capture group number. | |
| 263 | // | |
| 264 | fRXPat->fMaxCaptureDigits = 1; | |
| 265 | int32_t n = 10; | |
| 266 | for (;;) { | |
| 267 | if (n > fRXPat->fGroupMap->size()) { | |
| 268 | break; | |
| 269 | } | |
| 270 | fRXPat->fMaxCaptureDigits++; | |
| 271 | n *= 10; | |
| 272 | } | |
| 273 | ||
| 274 | // | |
| 275 | // The pattern's fFrameSize so far has accumulated the requirements for | |
| 276 | // storage for capture parentheses, counters, etc. that are encountered | |
| 277 | // in the pattern. Add space for the two variables that are always | |
| 278 | // present in the saved state: the input string position and the | |
| 279 | // position in the compiled pattern. | |
| 280 | // | |
| 281 | fRXPat->fFrameSize+=2; | |
| 282 | ||
| 283 | // | |
| 284 | // Get bounds for the minimum and maximum length of a string that this | |
| 285 | // pattern can match. Used to avoid looking for matches in strings that | |
| 286 | // are too short. | |
| 287 | // | |
| 288 | fRXPat->fMinMatchLen = minMatchLength(3, fRXPat->fCompiledPat->size()-1); | |
| 289 | ||
| 290 | // | |
| 291 | // Optimization passes | |
| 292 | // | |
| 293 | matchStartType(); | |
| 294 | OptDotStar(); | |
| 295 | stripNOPs(); | |
| 296 | ||
| 297 | // | |
| 298 | // Set up fast latin-1 range sets | |
| 299 | // | |
| 300 | int32_t numSets = fRXPat->fSets->size(); | |
| 301 | fRXPat->fSets8 = new Regex8BitSet[numSets]; | |
| 302 | int32_t i; | |
| 303 | for (i=0; i<numSets; i++) { | |
| 304 | UnicodeSet *s = (UnicodeSet *)fRXPat->fSets->elementAt(i); | |
| 305 | fRXPat->fSets8[i].init(s); | |
| 306 | } | |
| 307 | ||
| b75a7d8f A |
308 | } |
| 309 | ||
| 310 | ||
| 311 | ||
| 312 | ||
| 313 | ||
| 73c04bcf | 314 | //------------------------------------------------------------------------------ |
| b75a7d8f A |
315 | // |
| 316 | // doParseAction Do some action during regex pattern parsing. | |
| 317 | // Called by the parse state machine. | |
| 318 | // | |
| 319 | // Generation of the match engine PCode happens here, or | |
| 320 | // in functions called from the parse actions defined here. | |
| 321 | // | |
| 322 | // | |
| 73c04bcf | 323 | //------------------------------------------------------------------------------ |
| b75a7d8f A |
324 | UBool RegexCompile::doParseActions(EParseAction action) |
| 325 | { | |
| 326 | UBool returnVal = TRUE; | |
| 327 | ||
| 328 | switch ((Regex_PatternParseAction)action) { | |
| 329 | ||
| 330 | case doPatStart: | |
| 331 | // Start of pattern compiles to: | |
| 332 | //0 SAVE 2 Fall back to position of FAIL | |
| 333 | //1 jmp 3 | |
| 334 | //2 FAIL Stop if we ever reach here. | |
| 335 | //3 NOP Dummy, so start of pattern looks the same as | |
| 336 | // the start of an ( grouping. | |
| 337 | //4 NOP Resreved, will be replaced by a save if there are | |
| 338 | // OR | operators at the top level | |
| 339 | fRXPat->fCompiledPat->addElement(URX_BUILD(URX_STATE_SAVE, 2), *fStatus); | |
| 340 | fRXPat->fCompiledPat->addElement(URX_BUILD(URX_JMP, 3), *fStatus); | |
| 341 | fRXPat->fCompiledPat->addElement(URX_BUILD(URX_FAIL, 0), *fStatus); | |
| b75a7d8f | 342 | |
| 73c04bcf A |
343 | // Standard open nonCapture paren action emits the two NOPs and |
| 344 | // sets up the paren stack frame. | |
| 345 | doParseActions((EParseAction)doOpenNonCaptureParen); | |
| b75a7d8f A |
346 | break; |
| 347 | ||
| 348 | case doPatFinish: | |
| 349 | // We've scanned to the end of the pattern | |
| 350 | // The end of pattern compiles to: | |
| 351 | // URX_END | |
| 352 | // which will stop the runtime match engine. | |
| 353 | // Encountering end of pattern also behaves like a close paren, | |
| 354 | // and forces fixups of the State Save at the beginning of the compiled pattern | |
| 355 | // and of any OR operations at the top level. | |
| 356 | // | |
| 357 | handleCloseParen(); | |
| 358 | if (fParenStack.size() > 0) { | |
| 359 | // Missing close paren in pattern. | |
| 360 | error(U_REGEX_MISMATCHED_PAREN); | |
| 361 | } | |
| 362 | ||
| 363 | // add the END operation to the compiled pattern. | |
| 364 | fRXPat->fCompiledPat->addElement(URX_BUILD(URX_END, 0), *fStatus); | |
| 365 | ||
| 366 | // Terminate the pattern compilation state machine. | |
| 367 | returnVal = FALSE; | |
| 368 | break; | |
| 369 | ||
| 370 | ||
| 371 | ||
| 372 | case doOrOperator: | |
| 373 | // Scanning a '|', as in (A|B) | |
| 374 | { | |
| 375 | // Insert a SAVE operation at the start of the pattern section preceding | |
| 376 | // this OR at this level. This SAVE will branch the match forward | |
| 377 | // to the right hand side of the OR in the event that the left hand | |
| 378 | // side fails to match and backtracks. Locate the position for the | |
| 379 | // save from the location on the top of the parentheses stack. | |
| 380 | int32_t savePosition = fParenStack.popi(); | |
| 381 | int32_t op = fRXPat->fCompiledPat->elementAti(savePosition); | |
| 382 | U_ASSERT(URX_TYPE(op) == URX_NOP); // original contents of reserved location | |
| 383 | op = URX_BUILD(URX_STATE_SAVE, fRXPat->fCompiledPat->size()+1); | |
| 384 | fRXPat->fCompiledPat->setElementAt(op, savePosition); | |
| 385 | ||
| 386 | // Append an JMP operation into the compiled pattern. The operand for | |
| 387 | // the JMP will eventually be the location following the ')' for the | |
| 388 | // group. This will be patched in later, when the ')' is encountered. | |
| 389 | op = URX_BUILD(URX_JMP, 0); | |
| 390 | fRXPat->fCompiledPat->addElement(op, *fStatus); | |
| 391 | ||
| 392 | // Push the position of the newly added JMP op onto the parentheses stack. | |
| 393 | // This registers if for fixup when this block's close paren is encountered. | |
| 394 | fParenStack.push(fRXPat->fCompiledPat->size()-1, *fStatus); | |
| 395 | ||
| 396 | // Append a NOP to the compiled pattern. This is the slot reserved | |
| 397 | // for a SAVE in the event that there is yet another '|' following | |
| 398 | // this one. | |
| 399 | fRXPat->fCompiledPat->addElement(URX_BUILD(URX_NOP, 0), *fStatus); | |
| 400 | fParenStack.push(fRXPat->fCompiledPat->size()-1, *fStatus); | |
| 401 | } | |
| 402 | break; | |
| 403 | ||
| 404 | ||
| 405 | case doOpenCaptureParen: | |
| 406 | // Open Paren. | |
| 407 | // Compile to a | |
| 408 | // - NOP, which later may be replaced by a save-state if the | |
| 409 | // parenthesized group gets a * quantifier, followed by | |
| 410 | // - START_CAPTURE n where n is stack frame offset to the capture group variables. | |
| 411 | // - NOP, which may later be replaced by a save-state if there | |
| 412 | // is an '|' alternation within the parens. | |
| 413 | // | |
| 414 | // Each capture group gets three slots in the save stack frame: | |
| 415 | // 0: Capture Group start position (in input string being matched.) | |
| 416 | // 1: Capture Group end positino. | |
| 417 | // 2: Start of Match-in-progress. | |
| 418 | // The first two locations are for a completed capture group, and are | |
| 419 | // referred to by back references and the like. | |
| 420 | // The third location stores the capture start position when an START_CAPTURE is | |
| 421 | // encountered. This will be promoted to a completed capture when (and if) the corresponding | |
| 422 | // END_CAPure is encountered. | |
| 423 | { | |
| 424 | fRXPat->fCompiledPat->addElement(URX_BUILD(URX_NOP, 0), *fStatus); | |
| 425 | int32_t varsLoc = fRXPat->fFrameSize; // Reserve three slots in match stack frame. | |
| 426 | fRXPat->fFrameSize += 3; | |
| 427 | int32_t cop = URX_BUILD(URX_START_CAPTURE, varsLoc); | |
| 428 | fRXPat->fCompiledPat->addElement(cop, *fStatus); | |
| 429 | fRXPat->fCompiledPat->addElement(URX_BUILD(URX_NOP, 0), *fStatus); | |
| 430 | ||
| 431 | // On the Parentheses stack, start a new frame and add the postions | |
| 432 | // of the two NOPs. Depending on what follows in the pattern, the | |
| 433 | // NOPs may be changed to SAVE_STATE or JMP ops, with a target | |
| 434 | // address of the end of the parenthesized group. | |
| 435 | fParenStack.push(fModeFlags, *fStatus); // Match mode state | |
| 436 | fParenStack.push(capturing, *fStatus); // Frame type. | |
| 437 | fParenStack.push(fRXPat->fCompiledPat->size()-3, *fStatus); // The first NOP location | |
| 438 | fParenStack.push(fRXPat->fCompiledPat->size()-1, *fStatus); // The second NOP loc | |
| 439 | ||
| 440 | // Save the mapping from group number to stack frame variable position. | |
| 441 | fRXPat->fGroupMap->addElement(varsLoc, *fStatus); | |
| 442 | } | |
| 443 | break; | |
| 444 | ||
| 445 | case doOpenNonCaptureParen: | |
| 446 | // Open non-caputuring (grouping only) Paren. | |
| 447 | // Compile to a | |
| 448 | // - NOP, which later may be replaced by a save-state if the | |
| 449 | // parenthesized group gets a * quantifier, followed by | |
| 450 | // - NOP, which may later be replaced by a save-state if there | |
| 451 | // is an '|' alternation within the parens. | |
| 452 | { | |
| 453 | fRXPat->fCompiledPat->addElement(URX_BUILD(URX_NOP, 0), *fStatus); | |
| 454 | fRXPat->fCompiledPat->addElement(URX_BUILD(URX_NOP, 0), *fStatus); | |
| 455 | ||
| 456 | // On the Parentheses stack, start a new frame and add the postions | |
| 457 | // of the two NOPs. | |
| 458 | fParenStack.push(fModeFlags, *fStatus); // Match mode state | |
| 459 | fParenStack.push(plain, *fStatus); // Begin a new frame. | |
| 460 | fParenStack.push(fRXPat->fCompiledPat->size()-2, *fStatus); // The first NOP location | |
| 461 | fParenStack.push(fRXPat->fCompiledPat->size()-1, *fStatus); // The second NOP loc | |
| 462 | } | |
| 463 | break; | |
| 464 | ||
| 465 | ||
| 466 | case doOpenAtomicParen: | |
| 467 | // Open Atomic Paren. (?> | |
| 468 | // Compile to a | |
| 469 | // - NOP, which later may be replaced if the parenthesized group | |
| 470 | // has a quantifier, followed by | |
| 471 | // - STO_SP save state stack position, so it can be restored at the ")" | |
| 472 | // - NOP, which may later be replaced by a save-state if there | |
| 473 | // is an '|' alternation within the parens. | |
| 474 | { | |
| 475 | fRXPat->fCompiledPat->addElement(URX_BUILD(URX_NOP, 0), *fStatus); | |
| 476 | int32_t varLoc = fRXPat->fDataSize; // Reserve a data location for saving the | |
| 477 | fRXPat->fDataSize += 1; // state stack ptr. | |
| 478 | int32_t stoOp = URX_BUILD(URX_STO_SP, varLoc); | |
| 479 | fRXPat->fCompiledPat->addElement(stoOp, *fStatus); | |
| 480 | fRXPat->fCompiledPat->addElement(URX_BUILD(URX_NOP, 0), *fStatus); | |
| 481 | ||
| 482 | // On the Parentheses stack, start a new frame and add the postions | |
| 483 | // of the two NOPs. Depending on what follows in the pattern, the | |
| 484 | // NOPs may be changed to SAVE_STATE or JMP ops, with a target | |
| 485 | // address of the end of the parenthesized group. | |
| 486 | fParenStack.push(fModeFlags, *fStatus); // Match mode state | |
| 487 | fParenStack.push(atomic, *fStatus); // Frame type. | |
| 488 | fParenStack.push(fRXPat->fCompiledPat->size()-3, *fStatus); // The first NOP | |
| 489 | fParenStack.push(fRXPat->fCompiledPat->size()-1, *fStatus); // The second NOP | |
| 490 | } | |
| 491 | break; | |
| 492 | ||
| 493 | ||
| 494 | case doOpenLookAhead: | |
| 495 | // Positive Look-ahead (?= stuff ) | |
| 496 | // Compiles to | |
| 497 | // 1 START_LA dataLoc | |
| 498 | // 2. NOP reserved for use by quantifiers on the block. | |
| 499 | // Look-ahead can't have quantifiers, but paren stack | |
| 500 | // compile time conventions require the slot anyhow. | |
| 501 | // 3. NOP may be replaced if there is are '|' ops in the block. | |
| 502 | // 4. code for parenthesized stuff. | |
| 503 | // 5. ENDLA | |
| 504 | // | |
| 505 | // Two data slots are reserved, for saving the stack ptr and the input position. | |
| 506 | { | |
| 507 | int32_t dataLoc = fRXPat->fDataSize; | |
| 508 | fRXPat->fDataSize += 2; | |
| 509 | int32_t op = URX_BUILD(URX_LA_START, dataLoc); | |
| 510 | fRXPat->fCompiledPat->addElement(op, *fStatus); | |
| 511 | ||
| 512 | op = URX_BUILD(URX_NOP, 0); | |
| 513 | fRXPat->fCompiledPat->addElement(op, *fStatus); | |
| 514 | fRXPat->fCompiledPat->addElement(op, *fStatus); | |
| 515 | ||
| 516 | // On the Parentheses stack, start a new frame and add the postions | |
| 517 | // of the NOPs. | |
| 518 | fParenStack.push(fModeFlags, *fStatus); // Match mode state | |
| 519 | fParenStack.push(lookAhead, *fStatus); // Frame type. | |
| 520 | fParenStack.push(fRXPat->fCompiledPat->size()-2, *fStatus); // The first NOP location | |
| 521 | fParenStack.push(fRXPat->fCompiledPat->size()-1, *fStatus); // The second NOP location | |
| 522 | } | |
| 523 | break; | |
| 524 | ||
| 525 | case doOpenLookAheadNeg: | |
| 526 | // Negated Lookahead. (?! stuff ) | |
| 527 | // Compiles to | |
| 528 | // 1. START_LA dataloc | |
| 529 | // 2. SAVE_STATE 7 // Fail within look-ahead block restores to this state, | |
| 530 | // // which continues with the match. | |
| 531 | // 3. NOP // Std. Open Paren sequence, for possible '|' | |
| 532 | // 4. code for parenthesized stuff. | |
| 533 | // 5. END_LA // Cut back stack, remove saved state from step 2. | |
| 534 | // 6. FAIL // code in block succeeded, so neg. lookahead fails. | |
| 535 | // 7. ... | |
| 536 | { | |
| 537 | int32_t dataLoc = fRXPat->fDataSize; | |
| 538 | fRXPat->fDataSize += 2; | |
| 539 | int32_t op = URX_BUILD(URX_LA_START, dataLoc); | |
| 540 | fRXPat->fCompiledPat->addElement(op, *fStatus); | |
| 541 | ||
| 542 | op = URX_BUILD(URX_STATE_SAVE, 0); // dest address will be patched later. | |
| 543 | fRXPat->fCompiledPat->addElement(op, *fStatus); | |
| 544 | ||
| 545 | op = URX_BUILD(URX_NOP, 0); | |
| 546 | fRXPat->fCompiledPat->addElement(op, *fStatus); | |
| 547 | ||
| 548 | // On the Parentheses stack, start a new frame and add the postions | |
| 549 | // of the StateSave and NOP. | |
| 550 | fParenStack.push(fModeFlags, *fStatus); // Match mode state | |
| 551 | fParenStack.push( negLookAhead, *fStatus); // Frame type | |
| 552 | fParenStack.push(fRXPat->fCompiledPat->size()-2, *fStatus); // The STATE_SAVE location | |
| 553 | fParenStack.push(fRXPat->fCompiledPat->size()-1, *fStatus); // The second NOP location | |
| 554 | ||
| 555 | // Instructions #5 and #6 will be added when the ')' is encountered. | |
| 556 | } | |
| 557 | break; | |
| 558 | ||
| 559 | case doOpenLookBehind: | |
| 560 | { | |
| 561 | // Compile a (?<= look-behind open paren. | |
| 562 | // | |
| 563 | // Compiles to | |
| 564 | // 0 URX_LB_START dataLoc | |
| 565 | // 1 URX_LB_CONT dataLoc | |
| 566 | // 2 MinMatchLen | |
| 567 | // 3 MaxMatchLen | |
| 568 | // 4 URX_NOP Standard '(' boilerplate. | |
| 569 | // 5 URX_NOP Reserved slot for use with '|' ops within (block). | |
| 570 | // 6 <code for LookBehind expression> | |
| 571 | // 7 URX_LB_END dataLoc # Check match len, restore input len | |
| 572 | // 8 URX_LA_END dataLoc # Restore stack, input pos | |
| 573 | // | |
| 574 | // Allocate a block of matcher data, to contain (when running a match) | |
| 575 | // 0: Stack ptr on entry | |
| 576 | // 1: Input Index on entry | |
| 577 | // 2: Start index of match current match attempt. | |
| 578 | // 3: Original Input String len. | |
| 579 | ||
| 580 | // Allocate data space | |
| 581 | int32_t dataLoc = fRXPat->fDataSize; | |
| 582 | fRXPat->fDataSize += 4; | |
| 583 | ||
| 584 | // Emit URX_LB_START | |
| 585 | int32_t op = URX_BUILD(URX_LB_START, dataLoc); | |
| 586 | fRXPat->fCompiledPat->addElement(op, *fStatus); | |
| 587 | ||
| 588 | // Emit URX_LB_CONT | |
| 589 | op = URX_BUILD(URX_LB_CONT, dataLoc); | |
| 590 | fRXPat->fCompiledPat->addElement(op, *fStatus); | |
| 591 | fRXPat->fCompiledPat->addElement(0, *fStatus); // MinMatchLength. To be filled later. | |
| 592 | fRXPat->fCompiledPat->addElement(0, *fStatus); // MaxMatchLength. To be filled later. | |
| 593 | ||
| 594 | // Emit the NOP | |
| 595 | op = URX_BUILD(URX_NOP, 0); | |
| 596 | fRXPat->fCompiledPat->addElement(op, *fStatus); | |
| 597 | fRXPat->fCompiledPat->addElement(op, *fStatus); | |
| 598 | ||
| 599 | // On the Parentheses stack, start a new frame and add the postions | |
| 600 | // of the URX_LB_CONT and the NOP. | |
| 601 | fParenStack.push(fModeFlags, *fStatus); // Match mode state | |
| 602 | fParenStack.push(lookBehind, *fStatus); // Frame type | |
| 603 | fParenStack.push(fRXPat->fCompiledPat->size()-2, *fStatus); // The first NOP location | |
| 604 | fParenStack.push(fRXPat->fCompiledPat->size()-1, *fStatus); // The 2nd NOP location | |
| 605 | ||
| 606 | // The final two instructions will be added when the ')' is encountered. | |
| 607 | } | |
| 608 | ||
| 609 | break; | |
| 610 | ||
| 611 | case doOpenLookBehindNeg: | |
| 612 | { | |
| 613 | // Compile a (?<! negated look-behind open paren. | |
| 614 | // | |
| 615 | // Compiles to | |
| 616 | // 0 URX_LB_START dataLoc # Save entry stack, input len | |
| 617 | // 1 URX_LBN_CONT dataLoc # Iterate possible match positions | |
| 618 | // 2 MinMatchLen | |
| 619 | // 3 MaxMatchLen | |
| 620 | // 4 continueLoc (9) | |
| 621 | // 5 URX_NOP Standard '(' boilerplate. | |
| 622 | // 6 URX_NOP Reserved slot for use with '|' ops within (block). | |
| 623 | // 7 <code for LookBehind expression> | |
| 624 | // 8 URX_LBN_END dataLoc # Check match len, cause a FAIL | |
| 625 | // 9 ... | |
| 626 | // | |
| 627 | // Allocate a block of matcher data, to contain (when running a match) | |
| 628 | // 0: Stack ptr on entry | |
| 629 | // 1: Input Index on entry | |
| 630 | // 2: Start index of match current match attempt. | |
| 631 | // 3: Original Input String len. | |
| 632 | ||
| 633 | // Allocate data space | |
| 634 | int32_t dataLoc = fRXPat->fDataSize; | |
| 635 | fRXPat->fDataSize += 4; | |
| 636 | ||
| 637 | // Emit URX_LB_START | |
| 638 | int32_t op = URX_BUILD(URX_LB_START, dataLoc); | |
| 639 | fRXPat->fCompiledPat->addElement(op, *fStatus); | |
| 640 | ||
| 641 | // Emit URX_LBN_CONT | |
| 642 | op = URX_BUILD(URX_LBN_CONT, dataLoc); | |
| 643 | fRXPat->fCompiledPat->addElement(op, *fStatus); | |
| 644 | fRXPat->fCompiledPat->addElement(0, *fStatus); // MinMatchLength. To be filled later. | |
| 645 | fRXPat->fCompiledPat->addElement(0, *fStatus); // MaxMatchLength. To be filled later. | |
| 646 | fRXPat->fCompiledPat->addElement(0, *fStatus); // Continue Loc. To be filled later. | |
| 647 | ||
| 648 | // Emit the NOP | |
| 649 | op = URX_BUILD(URX_NOP, 0); | |
| 650 | fRXPat->fCompiledPat->addElement(op, *fStatus); | |
| 651 | fRXPat->fCompiledPat->addElement(op, *fStatus); | |
| 652 | ||
| 653 | // On the Parentheses stack, start a new frame and add the postions | |
| 654 | // of the URX_LB_CONT and the NOP. | |
| 655 | fParenStack.push(fModeFlags, *fStatus); // Match mode state | |
| 656 | fParenStack.push(lookBehindN, *fStatus); // Frame type | |
| 657 | fParenStack.push(fRXPat->fCompiledPat->size()-2, *fStatus); // The first NOP location | |
| 658 | fParenStack.push(fRXPat->fCompiledPat->size()-1, *fStatus); // The 2nd NOP location | |
| 659 | ||
| 660 | // The final two instructions will be added when the ')' is encountered. | |
| 661 | } | |
| 662 | break; | |
| 663 | ||
| 664 | case doConditionalExpr: | |
| 665 | // Conditionals such as (?(1)a:b) | |
| 666 | case doPerlInline: | |
| 667 | // Perl inline-condtionals. (?{perl code}a|b) We're not perl, no way to do them. | |
| 668 | error(U_REGEX_UNIMPLEMENTED); | |
| 669 | break; | |
| 670 | ||
| 671 | ||
| 672 | case doCloseParen: | |
| 673 | handleCloseParen(); | |
| 674 | if (fParenStack.size() <= 0) { | |
| 675 | // Extra close paren, or missing open paren. | |
| 676 | error(U_REGEX_MISMATCHED_PAREN); | |
| 677 | } | |
| 678 | break; | |
| 679 | ||
| 680 | case doNOP: | |
| 681 | break; | |
| 682 | ||
| 683 | ||
| 684 | case doBadOpenParenType: | |
| 685 | case doRuleError: | |
| 686 | error(U_REGEX_RULE_SYNTAX); | |
| 687 | break; | |
| 688 | ||
| 689 | ||
| 690 | case doMismatchedParenErr: | |
| 691 | error(U_REGEX_MISMATCHED_PAREN); | |
| 692 | break; | |
| 693 | ||
| 694 | case doPlus: | |
| 695 | // Normal '+' compiles to | |
| 696 | // 1. stuff to be repeated (already built) | |
| 697 | // 2. jmp-sav 1 | |
| 698 | // 3. ... | |
| 699 | // | |
| 700 | // Or, if the item to be repeated can match a zero length string, | |
| 701 | // 1. STO_INP_LOC data-loc | |
| 702 | // 2. body of stuff to be repeated | |
| 703 | // 3. JMP_SAV_X 2 | |
| 704 | // 4. ... | |
| 705 | ||
| 706 | // | |
| 707 | // Or, if the item to be repeated is simple | |
| 708 | // 1. Item to be repeated. | |
| 709 | // 2. LOOP_SR_I set number (assuming repeated item is a set ref) | |
| 710 | // 3. LOOP_C stack location | |
| 711 | { | |
| 712 | int32_t topLoc = blockTopLoc(FALSE); // location of item #1 | |
| 713 | int32_t frameLoc; | |
| 714 | ||
| 715 | // Check for simple constructs, which may get special optimized code. | |
| 716 | if (topLoc == fRXPat->fCompiledPat->size() - 1) { | |
| 717 | int32_t repeatedOp = fRXPat->fCompiledPat->elementAti(topLoc); | |
| 718 | ||
| 719 | if (URX_TYPE(repeatedOp) == URX_SETREF) { | |
| 720 | // Emit optimized code for [char set]+ | |
| 721 | int32_t loopOpI = URX_BUILD(URX_LOOP_SR_I, URX_VAL(repeatedOp)); | |
| 722 | fRXPat->fCompiledPat->addElement(loopOpI, *fStatus); | |
| 723 | frameLoc = fRXPat->fFrameSize; | |
| 724 | fRXPat->fFrameSize++; | |
| 725 | int32_t loopOpC = URX_BUILD(URX_LOOP_C, frameLoc); | |
| 726 | fRXPat->fCompiledPat->addElement(loopOpC, *fStatus); | |
| 727 | break; | |
| 728 | } | |
| 729 | ||
| 730 | if (URX_TYPE(repeatedOp) == URX_DOTANY || | |
| 731 | URX_TYPE(repeatedOp) == URX_DOTANY_ALL) { | |
| 732 | // Emit Optimized code for .+ operations. | |
| 733 | int32_t loopOpI = URX_BUILD(URX_LOOP_DOT_I, 0); | |
| 734 | if (URX_TYPE(repeatedOp) == URX_DOTANY_ALL) { | |
| 735 | // URX_LOOP_DOT_I operand is a flag indicating . matches any mode. | |
| 736 | loopOpI |= 1; | |
| 737 | } | |
| 738 | fRXPat->fCompiledPat->addElement(loopOpI, *fStatus); | |
| 739 | frameLoc = fRXPat->fFrameSize; | |
| 740 | fRXPat->fFrameSize++; | |
| 741 | int32_t loopOpC = URX_BUILD(URX_LOOP_C, frameLoc); | |
| 742 | fRXPat->fCompiledPat->addElement(loopOpC, *fStatus); | |
| 743 | break; | |
| 744 | } | |
| 745 | ||
| 746 | } | |
| 747 | ||
| 748 | // General case. | |
| 749 | ||
| 750 | // Check for minimum match length of zero, which requires | |
| 751 | // extra loop-breaking code. | |
| 752 | if (minMatchLength(topLoc, fRXPat->fCompiledPat->size()-1) == 0) { | |
| 753 | // Zero length match is possible. | |
| 754 | // Emit the code sequence that can handle it. | |
| 755 | insertOp(topLoc); | |
| 756 | frameLoc = fRXPat->fFrameSize; | |
| 757 | fRXPat->fFrameSize++; | |
| 758 | ||
| 759 | int32_t op = URX_BUILD(URX_STO_INP_LOC, frameLoc); | |
| 760 | fRXPat->fCompiledPat->setElementAt(op, topLoc); | |
| 761 | ||
| 762 | op = URX_BUILD(URX_JMP_SAV_X, topLoc+1); | |
| 763 | fRXPat->fCompiledPat->addElement(op, *fStatus); | |
| 764 | } else { | |
| 765 | // Simpler code when the repeated body must match something non-empty | |
| 766 | int32_t jmpOp = URX_BUILD(URX_JMP_SAV, topLoc); | |
| 767 | fRXPat->fCompiledPat->addElement(jmpOp, *fStatus); | |
| 768 | } | |
| 769 | } | |
| 770 | break; | |
| 771 | ||
| 772 | case doNGPlus: | |
| 773 | // Non-greedy '+?' compiles to | |
| 774 | // 1. stuff to be repeated (already built) | |
| 775 | // 2. state-save 1 | |
| 776 | // 3. ... | |
| 777 | { | |
| 778 | int32_t topLoc = blockTopLoc(FALSE); | |
| 779 | int32_t saveStateOp = URX_BUILD(URX_STATE_SAVE, topLoc); | |
| 780 | fRXPat->fCompiledPat->addElement(saveStateOp, *fStatus); | |
| 781 | } | |
| 782 | break; | |
| 783 | ||
| 784 | ||
| 785 | case doOpt: | |
| 786 | // Normal (greedy) ? quantifier. | |
| 787 | // Compiles to | |
| 788 | // 1. state save 3 | |
| 789 | // 2. body of optional block | |
| 790 | // 3. ... | |
| 791 | // Insert the state save into the compiled pattern, and we're done. | |
| 792 | { | |
| 793 | int32_t saveStateLoc = blockTopLoc(TRUE); | |
| 794 | int32_t saveStateOp = URX_BUILD(URX_STATE_SAVE, fRXPat->fCompiledPat->size()); | |
| 795 | fRXPat->fCompiledPat->setElementAt(saveStateOp, saveStateLoc); | |
| 796 | } | |
| 797 | break; | |
| 798 | ||
| 799 | case doNGOpt: | |
| 800 | // Non-greedy ?? quantifier | |
| 801 | // compiles to | |
| 802 | // 1. jmp 4 | |
| 803 | // 2. body of optional block | |
| 804 | // 3 jmp 5 | |
| 805 | // 4. state save 2 | |
| 806 | // 5 ... | |
| 807 | // This code is less than ideal, with two jmps instead of one, because we can only | |
| 808 | // insert one instruction at the top of the block being iterated. | |
| 809 | { | |
| 810 | int32_t jmp1_loc = blockTopLoc(TRUE); | |
| 811 | int32_t jmp2_loc = fRXPat->fCompiledPat->size(); | |
| 812 | ||
| 813 | int32_t jmp1_op = URX_BUILD(URX_JMP, jmp2_loc+1); | |
| 814 | fRXPat->fCompiledPat->setElementAt(jmp1_op, jmp1_loc); | |
| 815 | ||
| 816 | int32_t jmp2_op = URX_BUILD(URX_JMP, jmp2_loc+2); | |
| 817 | fRXPat->fCompiledPat->addElement(jmp2_op, *fStatus); | |
| 818 | ||
| 819 | int32_t save_op = URX_BUILD(URX_STATE_SAVE, jmp1_loc+1); | |
| 820 | fRXPat->fCompiledPat->addElement(save_op, *fStatus); | |
| 821 | } | |
| 822 | break; | |
| 823 | ||
| 824 | ||
| 825 | case doStar: | |
| 826 | // Normal (greedy) * quantifier. | |
| 827 | // Compiles to | |
| 828 | // 1. STATE_SAVE 4 | |
| 829 | // 2. body of stuff being iterated over | |
| 830 | // 3. JMP_SAV 2 | |
| 831 | // 4. ... | |
| 832 | // | |
| 833 | // Or, if the body is a simple [Set], | |
| 834 | // 1. LOOP_SR_I set number | |
| 835 | // 2. LOOP_C stack location | |
| 836 | // ... | |
| 837 | // | |
| 838 | // Or if this is a .* | |
| 839 | // 1. LOOP_DOT_I (. matches all mode flag) | |
| 840 | // 2. LOOP_C stack location | |
| 841 | // | |
| 842 | // Or, if the body can match a zero-length string, to inhibit infinite loops, | |
| 843 | // 1. STATE_SAVE 5 | |
| 844 | // 2. STO_INP_LOC data-loc | |
| 845 | // 3. body of stuff | |
| 846 | // 4. JMP_SAV_X 2 | |
| 847 | // 5. ... | |
| 848 | { | |
| 849 | // location of item #1, the STATE_SAVE | |
| 850 | int32_t topLoc = blockTopLoc(FALSE); | |
| 851 | int32_t dataLoc = -1; | |
| 852 | ||
| 853 | // Check for simple *, where the construct being repeated | |
| 854 | // compiled to single opcode, and might be optimizable. | |
| 855 | if (topLoc == fRXPat->fCompiledPat->size() - 1) { | |
| 856 | int32_t repeatedOp = fRXPat->fCompiledPat->elementAti(topLoc); | |
| 857 | ||
| 858 | if (URX_TYPE(repeatedOp) == URX_SETREF) { | |
| 859 | // Emit optimized code for a [char set]* | |
| 860 | int32_t loopOpI = URX_BUILD(URX_LOOP_SR_I, URX_VAL(repeatedOp)); | |
| 861 | fRXPat->fCompiledPat->setElementAt(loopOpI, topLoc); | |
| 862 | dataLoc = fRXPat->fFrameSize; | |
| 863 | fRXPat->fFrameSize++; | |
| 864 | int32_t loopOpC = URX_BUILD(URX_LOOP_C, dataLoc); | |
| 865 | fRXPat->fCompiledPat->addElement(loopOpC, *fStatus); | |
| 866 | break; | |
| 867 | } | |
| 868 | ||
| 869 | if (URX_TYPE(repeatedOp) == URX_DOTANY || | |
| 870 | URX_TYPE(repeatedOp) == URX_DOTANY_ALL) { | |
| 871 | // Emit Optimized code for .* operations. | |
| 872 | int32_t loopOpI = URX_BUILD(URX_LOOP_DOT_I, 0); | |
| 873 | if (URX_TYPE(repeatedOp) == URX_DOTANY_ALL) { | |
| 874 | // URX_LOOP_DOT_I operand is a flag indicating . matches any mode. | |
| 875 | loopOpI |= 1; | |
| 876 | } | |
| 877 | fRXPat->fCompiledPat->setElementAt(loopOpI, topLoc); | |
| 878 | dataLoc = fRXPat->fFrameSize; | |
| 879 | fRXPat->fFrameSize++; | |
| 880 | int32_t loopOpC = URX_BUILD(URX_LOOP_C, dataLoc); | |
| 881 | fRXPat->fCompiledPat->addElement(loopOpC, *fStatus); | |
| 882 | break; | |
| 883 | } | |
| 884 | } | |
| 885 | ||
| 886 | // Emit general case code for this * | |
| 887 | // The optimizations did not apply. | |
| 888 | ||
| 889 | int32_t saveStateLoc = blockTopLoc(TRUE); | |
| 890 | int32_t jmpOp = URX_BUILD(URX_JMP_SAV, saveStateLoc+1); | |
| 891 | ||
| 892 | // Check for minimum match length of zero, which requires | |
| 893 | // extra loop-breaking code. | |
| 894 | if (minMatchLength(saveStateLoc, fRXPat->fCompiledPat->size()-1) == 0) { | |
| 895 | insertOp(saveStateLoc); | |
| 896 | dataLoc = fRXPat->fFrameSize; | |
| 897 | fRXPat->fFrameSize++; | |
| 898 | ||
| 899 | int32_t op = URX_BUILD(URX_STO_INP_LOC, dataLoc); | |
| 900 | fRXPat->fCompiledPat->setElementAt(op, saveStateLoc+1); | |
| 901 | jmpOp = URX_BUILD(URX_JMP_SAV_X, saveStateLoc+2); | |
| 902 | } | |
| 903 | ||
| 904 | // Locate the position in the compiled pattern where the match will continue | |
| 905 | // after completing the *. (4 or 5 in the comment above) | |
| 906 | int32_t continueLoc = fRXPat->fCompiledPat->size()+1; | |
| 907 | ||
| 908 | // Put together the save state op store it into the compiled code. | |
| 909 | int32_t saveStateOp = URX_BUILD(URX_STATE_SAVE, continueLoc); | |
| 910 | fRXPat->fCompiledPat->setElementAt(saveStateOp, saveStateLoc); | |
| 911 | ||
| 912 | // Append the URX_JMP_SAV or URX_JMPX operation to the compiled pattern. | |
| 913 | fRXPat->fCompiledPat->addElement(jmpOp, *fStatus); | |
| 914 | } | |
| 915 | break; | |
| 916 | ||
| 917 | case doNGStar: | |
| 918 | // Non-greedy *? quantifier | |
| 919 | // compiles to | |
| 920 | // 1. JMP 3 | |
| 921 | // 2. body of stuff being iterated over | |
| 922 | // 3. STATE_SAVE 2 | |
| 923 | // 4 ... | |
| 924 | { | |
| 925 | int32_t jmpLoc = blockTopLoc(TRUE); // loc 1. | |
| 926 | int32_t saveLoc = fRXPat->fCompiledPat->size(); // loc 3. | |
| 927 | int32_t jmpOp = URX_BUILD(URX_JMP, saveLoc); | |
| 928 | int32_t stateSaveOp = URX_BUILD(URX_STATE_SAVE, jmpLoc+1); | |
| 929 | fRXPat->fCompiledPat->setElementAt(jmpOp, jmpLoc); | |
| 930 | fRXPat->fCompiledPat->addElement(stateSaveOp, *fStatus); | |
| 931 | } | |
| 932 | break; | |
| 933 | ||
| 934 | ||
| 935 | case doIntervalInit: | |
| 936 | // The '{' opening an interval quantifier was just scanned. | |
| 937 | // Init the counter varaiables that will accumulate the values as the digits | |
| 938 | // are scanned. | |
| 939 | fIntervalLow = 0; | |
| 940 | fIntervalUpper = -1; | |
| 941 | break; | |
| 942 | ||
| 943 | case doIntevalLowerDigit: | |
| 944 | // Scanned a digit from the lower value of an {lower,upper} interval | |
| 945 | { | |
| 946 | int32_t digitValue = u_charDigitValue(fC.fChar); | |
| 947 | U_ASSERT(digitValue >= 0); | |
| 948 | fIntervalLow = fIntervalLow*10 + digitValue; | |
| 949 | if (fIntervalLow < 0) { | |
| 950 | error(U_REGEX_NUMBER_TOO_BIG); | |
| 951 | } | |
| 952 | } | |
| 953 | break; | |
| 954 | ||
| 955 | case doIntervalUpperDigit: | |
| 956 | // Scanned a digit from the upper value of an {lower,upper} interval | |
| 957 | { | |
| 958 | if (fIntervalUpper < 0) { | |
| 959 | fIntervalUpper = 0; | |
| 960 | } | |
| 961 | int32_t digitValue = u_charDigitValue(fC.fChar); | |
| 962 | U_ASSERT(digitValue >= 0); | |
| 963 | fIntervalUpper = fIntervalUpper*10 + digitValue; | |
| 374ca955 | 964 | if (fIntervalUpper < 0) { |
| b75a7d8f A |
965 | error(U_REGEX_NUMBER_TOO_BIG); |
| 966 | } | |
| 967 | } | |
| 968 | break; | |
| 969 | ||
| 970 | case doIntervalSame: | |
| 971 | // Scanned a single value interval like {27}. Upper = Lower. | |
| 972 | fIntervalUpper = fIntervalLow; | |
| 973 | break; | |
| 974 | ||
| 975 | case doInterval: | |
| 976 | // Finished scanning a normal {lower,upper} interval. Generate the code for it. | |
| 977 | if (compileInlineInterval() == FALSE) { | |
| 978 | compileInterval(URX_CTR_INIT, URX_CTR_LOOP); | |
| 979 | } | |
| 980 | break; | |
| 981 | ||
| 982 | case doPossessiveInterval: | |
| 983 | // Finished scanning a Possessive {lower,upper}+ interval. Generate the code for it. | |
| 984 | { | |
| 985 | // Remember the loc for the top of the block being looped over. | |
| 986 | // (Can not reserve a slot in the compiled pattern at this time, becuase | |
| 987 | // compileInterval needs to reserve also, and blockTopLoc can only reserve | |
| 988 | // once per block.) | |
| 989 | int32_t topLoc = blockTopLoc(FALSE); | |
| 990 | ||
| 991 | // Produce normal looping code. | |
| 992 | compileInterval(URX_CTR_INIT, URX_CTR_LOOP); | |
| 993 | ||
| 994 | // Surround the just-emitted normal looping code with a STO_SP ... LD_SP | |
| 995 | // just as if the loop was inclosed in atomic parentheses. | |
| 996 | ||
| 997 | // First the STO_SP before the start of the loop | |
| 998 | insertOp(topLoc); | |
| 999 | int32_t varLoc = fRXPat->fDataSize; // Reserve a data location for saving the | |
| 1000 | fRXPat->fDataSize += 1; // state stack ptr. | |
| 1001 | int32_t op = URX_BUILD(URX_STO_SP, varLoc); | |
| 1002 | fRXPat->fCompiledPat->setElementAt(op, topLoc); | |
| 1003 | ||
| 1004 | int32_t loopOp = fRXPat->fCompiledPat->popi(); | |
| 1005 | U_ASSERT(URX_TYPE(loopOp) == URX_CTR_LOOP && URX_VAL(loopOp) == topLoc); | |
| 1006 | loopOp++; // point LoopOp after the just-inserted STO_SP | |
| 1007 | fRXPat->fCompiledPat->push(loopOp, *fStatus); | |
| 1008 | ||
| 1009 | // Then the LD_SP after the end of the loop | |
| 1010 | op = URX_BUILD(URX_LD_SP, varLoc); | |
| 1011 | fRXPat->fCompiledPat->addElement(op, *fStatus); | |
| 1012 | } | |
| 1013 | ||
| 1014 | break; | |
| 1015 | ||
| 1016 | case doNGInterval: | |
| 1017 | // Finished scanning a non-greedy {lower,upper}? interval. Generate the code for it. | |
| 1018 | compileInterval(URX_CTR_INIT_NG, URX_CTR_LOOP_NG); | |
| 1019 | break; | |
| 1020 | ||
| 1021 | case doIntervalError: | |
| 1022 | error(U_REGEX_BAD_INTERVAL); | |
| 1023 | break; | |
| 1024 | ||
| 1025 | case doLiteralChar: | |
| 1026 | // We've just scanned a "normal" character from the pattern, | |
| 1027 | literalChar(fC.fChar); | |
| 1028 | break; | |
| 1029 | ||
| 1030 | ||
| 1031 | ||
| 1032 | case doDotAny: | |
| 1033 | // scanned a ".", match any single character. | |
| 1034 | { | |
| 1035 | int32_t op; | |
| 1036 | if (fModeFlags & UREGEX_DOTALL) { | |
| 1037 | op = URX_BUILD(URX_DOTANY_ALL, 0); | |
| 1038 | } else { | |
| 1039 | op = URX_BUILD(URX_DOTANY, 0); | |
| 1040 | } | |
| 1041 | fRXPat->fCompiledPat->addElement(op, *fStatus); | |
| 1042 | } | |
| 1043 | break; | |
| 1044 | ||
| 1045 | case doCaret: | |
| 1046 | { | |
| 1047 | int32_t op = (fModeFlags & UREGEX_MULTILINE)? URX_CARET_M : URX_CARET; | |
| 1048 | fRXPat->fCompiledPat->addElement(URX_BUILD(op, 0), *fStatus); | |
| 1049 | } | |
| 1050 | break; | |
| 1051 | ||
| 1052 | ||
| 1053 | case doDollar: | |
| 1054 | { | |
| 1055 | int32_t op = (fModeFlags & UREGEX_MULTILINE)? URX_DOLLAR_M : URX_DOLLAR; | |
| 1056 | fRXPat->fCompiledPat->addElement(URX_BUILD(op, 0), *fStatus); | |
| 1057 | } | |
| 1058 | break; | |
| 1059 | ||
| 1060 | case doBackslashA: | |
| 1061 | fRXPat->fCompiledPat->addElement(URX_BUILD(URX_CARET, 0), *fStatus); | |
| 1062 | break; | |
| 1063 | ||
| 1064 | case doBackslashB: | |
| 374ca955 A |
1065 | { |
| 1066 | #if UCONFIG_NO_BREAK_ITERATION==1 | |
| 1067 | if (fModeFlags & UREGEX_UWORD) { | |
| 1068 | error(U_UNSUPPORTED_ERROR); | |
| 1069 | } | |
| 1070 | #endif | |
| 1071 | int32_t op = (fModeFlags & UREGEX_UWORD)? URX_BACKSLASH_BU : URX_BACKSLASH_B; | |
| 1072 | fRXPat->fCompiledPat->addElement(URX_BUILD(op, 1), *fStatus); | |
| 1073 | } | |
| b75a7d8f A |
1074 | break; |
| 1075 | ||
| 1076 | case doBackslashb: | |
| 374ca955 A |
1077 | { |
| 1078 | #if UCONFIG_NO_BREAK_ITERATION==1 | |
| 1079 | if (fModeFlags & UREGEX_UWORD) { | |
| 1080 | error(U_UNSUPPORTED_ERROR); | |
| 1081 | } | |
| 1082 | #endif | |
| 1083 | int32_t op = (fModeFlags & UREGEX_UWORD)? URX_BACKSLASH_BU : URX_BACKSLASH_B; | |
| 1084 | fRXPat->fCompiledPat->addElement(URX_BUILD(op, 0), *fStatus); | |
| 1085 | } | |
| b75a7d8f A |
1086 | break; |
| 1087 | ||
| 1088 | case doBackslashD: | |
| 1089 | fRXPat->fCompiledPat->addElement(URX_BUILD(URX_BACKSLASH_D, 1), *fStatus); | |
| 1090 | break; | |
| 1091 | ||
| 1092 | case doBackslashd: | |
| 1093 | fRXPat->fCompiledPat->addElement(URX_BUILD(URX_BACKSLASH_D, 0), *fStatus); | |
| 1094 | break; | |
| 1095 | ||
| 1096 | case doBackslashG: | |
| 1097 | fRXPat->fCompiledPat->addElement(URX_BUILD(URX_BACKSLASH_G, 0), *fStatus); | |
| 1098 | break; | |
| 1099 | ||
| 1100 | case doBackslashS: | |
| 1101 | fRXPat->fCompiledPat->addElement( | |
| 1102 | URX_BUILD(URX_STAT_SETREF_N, URX_ISSPACE_SET), *fStatus); | |
| 1103 | break; | |
| 1104 | ||
| 1105 | case doBackslashs: | |
| 1106 | fRXPat->fCompiledPat->addElement( | |
| 1107 | URX_BUILD(URX_STATIC_SETREF, URX_ISSPACE_SET), *fStatus); | |
| 1108 | break; | |
| 1109 | ||
| 1110 | case doBackslashW: | |
| 1111 | fRXPat->fCompiledPat->addElement( | |
| 1112 | URX_BUILD(URX_STAT_SETREF_N, URX_ISWORD_SET), *fStatus); | |
| 1113 | break; | |
| 1114 | ||
| 1115 | case doBackslashw: | |
| 1116 | fRXPat->fCompiledPat->addElement( | |
| 1117 | URX_BUILD(URX_STATIC_SETREF, URX_ISWORD_SET), *fStatus); | |
| 1118 | break; | |
| 1119 | ||
| 1120 | case doBackslashX: | |
| 1121 | fRXPat->fCompiledPat->addElement(URX_BUILD(URX_BACKSLASH_X, 0), *fStatus); | |
| 1122 | break; | |
| 1123 | ||
| 1124 | ||
| 1125 | case doBackslashZ: | |
| 1126 | fRXPat->fCompiledPat->addElement(URX_BUILD(URX_DOLLAR, 0), *fStatus); | |
| 1127 | break; | |
| 1128 | ||
| 1129 | case doBackslashz: | |
| 1130 | fRXPat->fCompiledPat->addElement(URX_BUILD(URX_BACKSLASH_Z, 0), *fStatus); | |
| 1131 | break; | |
| 1132 | ||
| 1133 | case doEscapeError: | |
| 1134 | error(U_REGEX_BAD_ESCAPE_SEQUENCE); | |
| 1135 | break; | |
| 1136 | ||
| 1137 | case doExit: | |
| 1138 | returnVal = FALSE; | |
| 1139 | break; | |
| 1140 | ||
| 1141 | case doProperty: | |
| 1142 | { | |
| 1143 | UnicodeSet *theSet = scanProp(); | |
| 1144 | compileSet(theSet); | |
| 1145 | } | |
| 1146 | break; | |
| 1147 | ||
| 1148 | ||
| 1149 | case doScanUnicodeSet: | |
| 1150 | { | |
| 1151 | UnicodeSet *theSet = scanSet(); | |
| 1152 | compileSet(theSet); | |
| 1153 | } | |
| 1154 | break; | |
| 1155 | ||
| 1156 | case doEnterQuoteMode: | |
| 1157 | // Just scanned a \Q. Put character scanner into quote mode. | |
| 1158 | fQuoteMode = TRUE; | |
| 1159 | break; | |
| 1160 | ||
| 1161 | case doBackRef: | |
| 1162 | // BackReference. Somewhat unusual in that the front-end can not completely parse | |
| 1163 | // the regular expression, because the number of digits to be consumed | |
| 1164 | // depends on the number of capture groups that have been defined. So | |
| 1165 | // we have to do it here instead. | |
| 1166 | { | |
| 1167 | int32_t numCaptureGroups = fRXPat->fGroupMap->size(); | |
| 1168 | int32_t groupNum = 0; | |
| 1169 | UChar32 c = fC.fChar; | |
| 1170 | ||
| 1171 | for (;;) { | |
| 1172 | // Loop once per digit, for max allowed number of digits in a back reference. | |
| 1173 | int32_t digit = u_charDigitValue(c); | |
| 1174 | groupNum = groupNum * 10 + digit; | |
| 1175 | if (groupNum >= numCaptureGroups) { | |
| 1176 | break; | |
| 1177 | } | |
| 1178 | c = peekCharLL(); | |
| 1179 | if (RegexStaticSets::gStaticSets->fRuleDigits->contains(c) == FALSE) { | |
| 1180 | break; | |
| 1181 | } | |
| 1182 | nextCharLL(); | |
| 1183 | } | |
| 1184 | ||
| 1185 | // Scan of the back reference in the source regexp is complete. Now generate | |
| 1186 | // the compiled code for it. | |
| 1187 | // Because capture groups can be forward-referenced by back-references, | |
| 1188 | // we fill the operand with the capture group number. At the end | |
| 374ca955 | 1189 | // of compilation, it will be changed to the variable's location. |
| b75a7d8f A |
1190 | U_ASSERT(groupNum > 0); |
| 1191 | int32_t op; | |
| 1192 | if (fModeFlags & UREGEX_CASE_INSENSITIVE) { | |
| 1193 | op = URX_BUILD(URX_BACKREF_I, groupNum); | |
| 1194 | } else { | |
| 1195 | op = URX_BUILD(URX_BACKREF, groupNum); | |
| 1196 | } | |
| 1197 | fRXPat->fCompiledPat->addElement(op, *fStatus); | |
| 1198 | } | |
| 1199 | break; | |
| 1200 | ||
| 1201 | ||
| b75a7d8f A |
1202 | case doPossessivePlus: |
| 1203 | // Possessive ++ quantifier. | |
| 1204 | // Compiles to | |
| 1205 | // 1. STO_SP | |
| 1206 | // 2. body of stuff being iterated over | |
| 1207 | // 3. STATE_SAVE 5 | |
| 1208 | // 4. JMP 2 | |
| 1209 | // 5. LD_SP | |
| 1210 | // 6. ... | |
| 1211 | // | |
| 1212 | // Note: TODO: This is pretty inefficient. A mass of saved state is built up | |
| 1213 | // then unconditionally discarded. Perhaps introduce a new opcode | |
| 1214 | // | |
| 1215 | { | |
| 1216 | // Emit the STO_SP | |
| 1217 | int32_t topLoc = blockTopLoc(TRUE); | |
| 1218 | int32_t stoLoc = fRXPat->fDataSize; | |
| 1219 | fRXPat->fDataSize++; // Reserve the data location for storing save stack ptr. | |
| 1220 | int32_t op = URX_BUILD(URX_STO_SP, stoLoc); | |
| 1221 | fRXPat->fCompiledPat->setElementAt(op, topLoc); | |
| 1222 | ||
| 1223 | // Emit the STATE_SAVE | |
| 1224 | op = URX_BUILD(URX_STATE_SAVE, fRXPat->fCompiledPat->size()+2); | |
| 1225 | fRXPat->fCompiledPat->addElement(op, *fStatus); | |
| 1226 | ||
| 1227 | // Emit the JMP | |
| 1228 | op = URX_BUILD(URX_JMP, topLoc+1); | |
| 1229 | fRXPat->fCompiledPat->addElement(op, *fStatus); | |
| 1230 | ||
| 1231 | // Emit the LD_SP | |
| 1232 | op = URX_BUILD(URX_LD_SP, stoLoc); | |
| 1233 | fRXPat->fCompiledPat->addElement(op, *fStatus); | |
| 1234 | } | |
| 1235 | break; | |
| 1236 | ||
| 1237 | case doPossessiveStar: | |
| 1238 | // Possessive *+ quantifier. | |
| 1239 | // Compiles to | |
| 1240 | // 1. STO_SP loc | |
| 1241 | // 2. STATE_SAVE 5 | |
| 1242 | // 3. body of stuff being iterated over | |
| 1243 | // 4. JMP 2 | |
| 1244 | // 5. LD_SP loc | |
| 1245 | // 6 ... | |
| 1246 | // TODO: do something to cut back the state stack each time through the loop. | |
| 1247 | { | |
| 1248 | // Reserve two slots at the top of the block. | |
| 1249 | int32_t topLoc = blockTopLoc(TRUE); | |
| 1250 | insertOp(topLoc); | |
| 1251 | ||
| 1252 | // emit STO_SP loc | |
| 1253 | int32_t stoLoc = fRXPat->fDataSize; | |
| 1254 | fRXPat->fDataSize++; // Reserve the data location for storing save stack ptr. | |
| 1255 | int32_t op = URX_BUILD(URX_STO_SP, stoLoc); | |
| 1256 | fRXPat->fCompiledPat->setElementAt(op, topLoc); | |
| 1257 | ||
| 1258 | // Emit the SAVE_STATE 5 | |
| 1259 | int32_t L7 = fRXPat->fCompiledPat->size()+1; | |
| 1260 | op = URX_BUILD(URX_STATE_SAVE, L7); | |
| 1261 | fRXPat->fCompiledPat->setElementAt(op, topLoc+1); | |
| 1262 | ||
| 1263 | // Append the JMP operation. | |
| 1264 | op = URX_BUILD(URX_JMP, topLoc+1); | |
| 1265 | fRXPat->fCompiledPat->addElement(op, *fStatus); | |
| 1266 | ||
| 1267 | // Emit the LD_SP loc | |
| 1268 | op = URX_BUILD(URX_LD_SP, stoLoc); | |
| 1269 | fRXPat->fCompiledPat->addElement(op, *fStatus); | |
| 1270 | } | |
| 1271 | break; | |
| 1272 | ||
| 1273 | case doPossessiveOpt: | |
| 1274 | // Possessive ?+ quantifier. | |
| 1275 | // Compiles to | |
| 1276 | // 1. STO_SP loc | |
| 1277 | // 2. SAVE_STATE 5 | |
| 1278 | // 3. body of optional block | |
| 1279 | // 4. LD_SP loc | |
| 1280 | // 5. ... | |
| 1281 | // | |
| 1282 | { | |
| 1283 | // Reserve two slots at the top of the block. | |
| 1284 | int32_t topLoc = blockTopLoc(TRUE); | |
| 1285 | insertOp(topLoc); | |
| 1286 | ||
| 1287 | // Emit the STO_SP | |
| 1288 | int32_t stoLoc = fRXPat->fDataSize; | |
| 1289 | fRXPat->fDataSize++; // Reserve the data location for storing save stack ptr. | |
| 1290 | int32_t op = URX_BUILD(URX_STO_SP, stoLoc); | |
| 1291 | fRXPat->fCompiledPat->setElementAt(op, topLoc); | |
| 1292 | ||
| 1293 | // Emit the SAVE_STATE | |
| 1294 | int32_t continueLoc = fRXPat->fCompiledPat->size()+1; | |
| 1295 | op = URX_BUILD(URX_STATE_SAVE, continueLoc); | |
| 1296 | fRXPat->fCompiledPat->setElementAt(op, topLoc+1); | |
| 1297 | ||
| 1298 | // Emit the LD_SP | |
| 1299 | op = URX_BUILD(URX_LD_SP, stoLoc); | |
| 1300 | fRXPat->fCompiledPat->addElement(op, *fStatus); | |
| 1301 | } | |
| 1302 | break; | |
| 1303 | ||
| 1304 | ||
| 1305 | case doBeginMatchMode: | |
| 1306 | fNewModeFlags = fModeFlags; | |
| 1307 | fSetModeFlag = TRUE; | |
| 1308 | break; | |
| 1309 | ||
| 1310 | case doMatchMode: // (?i) and similar | |
| 1311 | { | |
| 1312 | int32_t bit = 0; | |
| 1313 | switch (fC.fChar) { | |
| 1314 | case 0x69: /* 'i' */ bit = UREGEX_CASE_INSENSITIVE; break; | |
| 1315 | case 0x6d: /* 'm' */ bit = UREGEX_MULTILINE; break; | |
| 1316 | case 0x73: /* 's' */ bit = UREGEX_DOTALL; break; | |
| 374ca955 | 1317 | case 0x77: /* 'w' */ bit = UREGEX_UWORD; break; |
| b75a7d8f A |
1318 | case 0x78: /* 'x' */ bit = UREGEX_COMMENTS; break; |
| 1319 | case 0x2d: /* '-' */ fSetModeFlag = FALSE; break; | |
| 1320 | default: | |
| 1321 | U_ASSERT(FALSE); // Should never happen. Other chars are filtered out | |
| 1322 | // by the scanner. | |
| 1323 | } | |
| 1324 | if (fSetModeFlag) { | |
| 1325 | fNewModeFlags |= bit; | |
| 1326 | } else { | |
| 1327 | fNewModeFlags &= ~bit; | |
| 1328 | } | |
| 1329 | } | |
| 1330 | break; | |
| 1331 | ||
| 1332 | case doSetMatchMode: | |
| 1333 | // We've got a (?i) or similar. The match mode is being changed, but | |
| 1334 | // the change is not scoped to a parenthesized block. | |
| 73c04bcf | 1335 | U_ASSERT(fNewModeFlags < 0); |
| b75a7d8f A |
1336 | fModeFlags = fNewModeFlags; |
| 1337 | ||
| 1338 | // Prevent any string from spanning across the change of match mode. | |
| 1339 | // Otherwise the pattern "abc(?i)def" would make a single string of "abcdef" | |
| 1340 | fixLiterals(); | |
| 1341 | break; | |
| 1342 | ||
| 1343 | ||
| 1344 | case doMatchModeParen: | |
| 1345 | // We've got a (?i: or similar. Begin a parenthesized block, save old | |
| 1346 | // mode flags so they can be restored at the close of the block. | |
| 1347 | // | |
| 1348 | // Compile to a | |
| 1349 | // - NOP, which later may be replaced by a save-state if the | |
| 1350 | // parenthesized group gets a * quantifier, followed by | |
| 1351 | // - NOP, which may later be replaced by a save-state if there | |
| 1352 | // is an '|' alternation within the parens. | |
| 1353 | { | |
| 1354 | fRXPat->fCompiledPat->addElement(URX_BUILD(URX_NOP, 0), *fStatus); | |
| 1355 | fRXPat->fCompiledPat->addElement(URX_BUILD(URX_NOP, 0), *fStatus); | |
| 1356 | ||
| 1357 | // On the Parentheses stack, start a new frame and add the postions | |
| 1358 | // of the two NOPs (a normal non-capturing () frame, except for the | |
| 1359 | // saving of the orignal mode flags.) | |
| 1360 | fParenStack.push(fModeFlags, *fStatus); | |
| 1361 | fParenStack.push(flags, *fStatus); // Frame Marker | |
| 1362 | fParenStack.push(fRXPat->fCompiledPat->size()-2, *fStatus); // The first NOP | |
| 1363 | fParenStack.push(fRXPat->fCompiledPat->size()-1, *fStatus); // The second NOP | |
| 1364 | ||
| 1365 | // Set the current mode flags to the new values. | |
| 73c04bcf | 1366 | U_ASSERT(fNewModeFlags < 0); |
| b75a7d8f A |
1367 | fModeFlags = fNewModeFlags; |
| 1368 | } | |
| 1369 | break; | |
| 1370 | ||
| 374ca955 A |
1371 | case doBadModeFlag: |
| 1372 | error(U_REGEX_INVALID_FLAG); | |
| 1373 | break; | |
| 1374 | ||
| b75a7d8f A |
1375 | case doSuppressComments: |
| 1376 | // We have just scanned a '(?'. We now need to prevent the character scanner from | |
| 1377 | // treating a '#' as a to-the-end-of-line comment. | |
| 1378 | // (This Perl compatibility just gets uglier and uglier to do...) | |
| 1379 | fEOLComments = FALSE; | |
| 1380 | break; | |
| 1381 | ||
| 1382 | ||
| 1383 | ||
| 1384 | default: | |
| 1385 | U_ASSERT(FALSE); | |
| 1386 | error(U_REGEX_INTERNAL_ERROR); | |
| 1387 | break; | |
| 1388 | } | |
| 1389 | ||
| 1390 | if (U_FAILURE(*fStatus)) { | |
| 1391 | returnVal = FALSE; | |
| 1392 | } | |
| 1393 | ||
| 1394 | return returnVal; | |
| 73c04bcf | 1395 | } |
| b75a7d8f A |
1396 | |
| 1397 | ||
| 1398 | ||
| 1399 | //------------------------------------------------------------------------------ | |
| 1400 | // | |
| 1401 | // literalChar We've encountered a literal character from the pattern, | |
| 1402 | // or an escape sequence that reduces to a character. | |
| 1403 | // Add it to the string containing all literal chars/strings from | |
| 1404 | // the pattern. | |
| 1405 | // If we are in a pattern string already, add the new char to it. | |
| 1406 | // If we aren't in a pattern string, begin one now. | |
| 1407 | // | |
| 1408 | //------------------------------------------------------------------------------ | |
| 1409 | void RegexCompile::literalChar(UChar32 c) { | |
| 1410 | int32_t op; // An operation in the compiled pattern. | |
| 1411 | int32_t opType; | |
| 1412 | int32_t patternLoc; // A position in the compiled pattern. | |
| 1413 | int32_t stringLen; | |
| 1414 | ||
| 1415 | ||
| 1416 | // If the last thing compiled into the pattern was not a literal char, | |
| 1417 | // force this new literal char to begin a new string, and not append to the previous. | |
| 1418 | op = fRXPat->fCompiledPat->lastElementi(); | |
| 1419 | opType = URX_TYPE(op); | |
| 1420 | if (!(opType == URX_STRING_LEN || opType == URX_ONECHAR || opType == URX_ONECHAR_I)) { | |
| 1421 | fixLiterals(); | |
| 1422 | } | |
| 1423 | ||
| 1424 | if (fStringOpStart == -1) { | |
| 1425 | // First char of a string in the pattern. | |
| 1426 | // Emit a OneChar op into the compiled pattern. | |
| 1427 | emitONE_CHAR(c); | |
| 1428 | ||
| 1429 | // Also add it to the string pool, in case we get a second adjacent literal | |
| 1430 | // and want to change form ONE_CHAR to STRING | |
| 1431 | fStringOpStart = fRXPat->fLiteralText.length(); | |
| 1432 | fRXPat->fLiteralText.append(c); | |
| 1433 | return; | |
| 1434 | } | |
| 1435 | ||
| 1436 | // We are adding onto an existing string | |
| 1437 | fRXPat->fLiteralText.append(c); | |
| 1438 | ||
| 1439 | op = fRXPat->fCompiledPat->lastElementi(); | |
| 1440 | opType = URX_TYPE(op); | |
| 1441 | U_ASSERT(opType == URX_ONECHAR || opType == URX_ONECHAR_I || opType == URX_STRING_LEN); | |
| 1442 | ||
| 1443 | // If the most recently emitted op is a URX_ONECHAR, | |
| 1444 | if (opType == URX_ONECHAR || opType == URX_ONECHAR_I) { | |
| 1445 | if (U16_IS_TRAIL(c) && U16_IS_LEAD(URX_VAL(op))) { | |
| 1446 | // The most recently emitted op is a ONECHAR that was the first half | |
| 1447 | // of a surrogate pair. Update the ONECHAR's operand to be the | |
| 1448 | // supplementary code point resulting from both halves of the pair. | |
| 1449 | c = U16_GET_SUPPLEMENTARY(URX_VAL(op), c); | |
| 1450 | op = URX_BUILD(opType, c); | |
| 1451 | patternLoc = fRXPat->fCompiledPat->size() - 1; | |
| 1452 | fRXPat->fCompiledPat->setElementAt(op, patternLoc); | |
| 1453 | return; | |
| 1454 | } | |
| 1455 | ||
| 1456 | // The most recently emitted op is a ONECHAR. | |
| 1457 | // We've now received another adjacent char. Change the ONECHAR op | |
| 1458 | // to a string op. | |
| 1459 | if (fModeFlags & UREGEX_CASE_INSENSITIVE) { | |
| 1460 | op = URX_BUILD(URX_STRING_I, fStringOpStart); | |
| 1461 | } else { | |
| 1462 | op = URX_BUILD(URX_STRING, fStringOpStart); | |
| 1463 | } | |
| 1464 | patternLoc = fRXPat->fCompiledPat->size() - 1; | |
| 1465 | fRXPat->fCompiledPat->setElementAt(op, patternLoc); | |
| 1466 | op = URX_BUILD(URX_STRING_LEN, 0); | |
| 1467 | fRXPat->fCompiledPat->addElement(op, *fStatus); | |
| 1468 | } | |
| 1469 | ||
| 1470 | // The pattern contains a URX_SRING / URX_STRING_LEN. Update the | |
| 1471 | // string length to reflect the new char we just added to the string. | |
| 1472 | stringLen = fRXPat->fLiteralText.length() - fStringOpStart; | |
| 1473 | op = URX_BUILD(URX_STRING_LEN, stringLen); | |
| 1474 | patternLoc = fRXPat->fCompiledPat->size() - 1; | |
| 1475 | fRXPat->fCompiledPat->setElementAt(op, patternLoc); | |
| 1476 | } | |
| 1477 | ||
| 1478 | ||
| 1479 | ||
| 1480 | //------------------------------------------------------------------------------ | |
| 1481 | // | |
| 1482 | // emitONE_CHAR emit a ONE_CHAR op into the generated code. | |
| 1483 | // Choose cased or uncased version, depending on the | |
| 1484 | // match mode and whether the character itself is cased. | |
| 1485 | // | |
| 1486 | //------------------------------------------------------------------------------ | |
| 1487 | void RegexCompile::emitONE_CHAR(UChar32 c) { | |
| 1488 | int32_t op; | |
| 1489 | if ((fModeFlags & UREGEX_CASE_INSENSITIVE) && | |
| 1490 | u_hasBinaryProperty(c, UCHAR_CASE_SENSITIVE)) { | |
| 1491 | // We have a cased character, and are in case insensitive matching mode. | |
| 1492 | c = u_foldCase(c, U_FOLD_CASE_DEFAULT); | |
| 1493 | op = URX_BUILD(URX_ONECHAR_I, c); | |
| 1494 | } else { | |
| 1495 | // Uncased char, or case sensitive match mode. | |
| 1496 | // Either way, just generate a literal compare of the char. | |
| 1497 | op = URX_BUILD(URX_ONECHAR, c); | |
| 1498 | } | |
| 1499 | fRXPat->fCompiledPat->addElement(op, *fStatus); | |
| 1500 | } | |
| 1501 | ||
| 1502 | ||
| 1503 | //------------------------------------------------------------------------------ | |
| 1504 | // | |
| 1505 | // fixLiterals When compiling something that can follow a literal | |
| 1506 | // string in a pattern, we need to "fix" any preceding | |
| 1507 | // string, which will cause any subsequent literals to | |
| 1508 | // begin a new string, rather than appending to the | |
| 1509 | // old one. | |
| 1510 | // | |
| 1511 | // Optionally, split the last char of the string off into | |
| 1512 | // a single "ONE_CHAR" operation, so that quantifiers can | |
| 1513 | // apply to that char alone. Example: abc* | |
| 1514 | // The * must apply to the 'c' only. | |
| 1515 | // | |
| 1516 | //------------------------------------------------------------------------------ | |
| 1517 | void RegexCompile::fixLiterals(UBool split) { | |
| 1518 | int32_t stringStart = fStringOpStart; // start index of the current literal string | |
| 1519 | int32_t op; // An op from/for the compiled pattern. | |
| 1520 | int32_t opType; // An opcode type from the compiled pattern. | |
| 1521 | int32_t stringLastCharIdx; | |
| 1522 | UChar32 lastChar; | |
| 1523 | int32_t stringNextToLastCharIdx; | |
| 1524 | UChar32 nextToLastChar; | |
| 1525 | int32_t stringLen; | |
| 1526 | ||
| 1527 | fStringOpStart = -1; | |
| 1528 | if (!split) { | |
| 1529 | return; | |
| 1530 | } | |
| 1531 | ||
| 1532 | // Split: We need to ensure that the last item in the compiled pattern does | |
| 1533 | // not refer to a literal string of more than one char. If it does, | |
| 1534 | // separate the last char from the rest of the string. | |
| 1535 | ||
| 1536 | // If the last operation from the compiled pattern is not a string, | |
| 1537 | // nothing needs to be done | |
| 1538 | op = fRXPat->fCompiledPat->lastElementi(); | |
| 1539 | opType = URX_TYPE(op); | |
| 1540 | if (opType != URX_STRING_LEN) { | |
| 1541 | return; | |
| 1542 | } | |
| 1543 | stringLen = URX_VAL(op); | |
| 1544 | ||
| 1545 | // | |
| 1546 | // Find the position of the last code point in the string (might be a surrogate pair) | |
| 1547 | // | |
| 1548 | stringLastCharIdx = fRXPat->fLiteralText.length(); | |
| 1549 | stringLastCharIdx = fRXPat->fLiteralText.moveIndex32(stringLastCharIdx, -1); | |
| 1550 | lastChar = fRXPat->fLiteralText.char32At(stringLastCharIdx); | |
| 1551 | ||
| 1552 | // The string should always be at least two code points long, meaning that there | |
| 1553 | // should be something before the last char position that we just found. | |
| 1554 | U_ASSERT(stringLastCharIdx > stringStart); | |
| 1555 | stringNextToLastCharIdx = fRXPat->fLiteralText.moveIndex32(stringLastCharIdx, -1); | |
| 1556 | U_ASSERT(stringNextToLastCharIdx >= stringStart); | |
| 1557 | nextToLastChar = fRXPat->fLiteralText.char32At(stringNextToLastCharIdx); | |
| 1558 | ||
| 1559 | if (stringNextToLastCharIdx > stringStart) { | |
| 1560 | // The length of string remaining after removing one char is two or more. | |
| 1561 | // Leave the string in the compiled pattern, shorten it by one char, | |
| 1562 | // and append a URX_ONECHAR op for the last char. | |
| 1563 | stringLen -= (fRXPat->fLiteralText.length() - stringLastCharIdx); | |
| 1564 | op = URX_BUILD(URX_STRING_LEN, stringLen); | |
| 1565 | fRXPat->fCompiledPat->setElementAt(op, fRXPat->fCompiledPat->size() -1); | |
| 1566 | emitONE_CHAR(lastChar); | |
| 1567 | } else { | |
| 1568 | // The original string consisted of exactly two characters. Replace | |
| 1569 | // the existing compiled URX_STRING/URX_STRING_LEN ops with a pair | |
| 1570 | // of URX_ONECHARs. | |
| 1571 | fRXPat->fCompiledPat->setSize(fRXPat->fCompiledPat->size() -2); | |
| 1572 | emitONE_CHAR(nextToLastChar); | |
| 1573 | emitONE_CHAR(lastChar); | |
| 1574 | } | |
| 1575 | } | |
| 1576 | ||
| 1577 | ||
| 1578 | ||
| 1579 | ||
| 1580 | ||
| 1581 | ||
| 1582 | //------------------------------------------------------------------------------ | |
| 1583 | // | |
| 1584 | // insertOp() Insert a slot for a new opcode into the already | |
| 1585 | // compiled pattern code. | |
| 1586 | // | |
| 1587 | // Fill the slot with a NOP. Our caller will replace it | |
| 1588 | // with what they really wanted. | |
| 1589 | // | |
| 1590 | //------------------------------------------------------------------------------ | |
| 1591 | void RegexCompile::insertOp(int32_t where) { | |
| 1592 | UVector32 *code = fRXPat->fCompiledPat; | |
| 1593 | U_ASSERT(where>0 && where < code->size()); | |
| 1594 | ||
| 1595 | int32_t nop = URX_BUILD(URX_NOP, 0); | |
| 1596 | code->insertElementAt(nop, where, *fStatus); | |
| 1597 | ||
| 1598 | // Walk through the pattern, looking for any ops with targets that | |
| 1599 | // were moved down by the insert. Fix them. | |
| 1600 | int32_t loc; | |
| 1601 | for (loc=0; loc<code->size(); loc++) { | |
| 1602 | int32_t op = code->elementAti(loc); | |
| 1603 | int32_t opType = URX_TYPE(op); | |
| 1604 | int32_t opValue = URX_VAL(op); | |
| 1605 | if ((opType == URX_JMP || | |
| 1606 | opType == URX_JMPX || | |
| 1607 | opType == URX_STATE_SAVE || | |
| 1608 | opType == URX_CTR_LOOP || | |
| 1609 | opType == URX_CTR_LOOP_NG || | |
| 1610 | opType == URX_JMP_SAV || | |
| 1611 | opType == URX_RELOC_OPRND) && opValue > where) { | |
| 1612 | // Target location for this opcode is after the insertion point and | |
| 1613 | // needs to be incremented to adjust for the insertion. | |
| 1614 | opValue++; | |
| 1615 | op = URX_BUILD(opType, opValue); | |
| 1616 | code->setElementAt(op, loc); | |
| 1617 | } | |
| 1618 | } | |
| 1619 | ||
| 1620 | // Now fix up the parentheses stack. All positive values in it are locations in | |
| 1621 | // the compiled pattern. (Negative values are frame boundaries, and don't need fixing.) | |
| 1622 | for (loc=0; loc<fParenStack.size(); loc++) { | |
| 1623 | int32_t x = fParenStack.elementAti(loc); | |
| 73c04bcf | 1624 | U_ASSERT(x < code->size()); |
| b75a7d8f A |
1625 | if (x>where) { |
| 1626 | x++; | |
| 1627 | fParenStack.setElementAt(x, loc); | |
| 1628 | } | |
| 1629 | } | |
| 1630 | ||
| 1631 | if (fMatchCloseParen > where) { | |
| 1632 | fMatchCloseParen++; | |
| 1633 | } | |
| 1634 | if (fMatchOpenParen > where) { | |
| 1635 | fMatchOpenParen++; | |
| 1636 | } | |
| 1637 | } | |
| 1638 | ||
| 1639 | ||
| 1640 | ||
| 1641 | //------------------------------------------------------------------------------ | |
| 1642 | // | |
| 1643 | // blockTopLoc() Find or create a location in the compiled pattern | |
| 1644 | // at the start of the operation or block that has | |
| 1645 | // just been compiled. Needed when a quantifier (* or | |
| 1646 | // whatever) appears, and we need to add an operation | |
| 1647 | // at the start of the thing being quantified. | |
| 1648 | // | |
| 1649 | // (Parenthesized Blocks) have a slot with a NOP that | |
| 1650 | // is reserved for this purpose. .* or similar don't | |
| 1651 | // and a slot needs to be added. | |
| 1652 | // | |
| 1653 | // parameter reserveLoc : TRUE - ensure that there is space to add an opcode | |
| 1654 | // at the returned location. | |
| 1655 | // FALSE - just return the address, | |
| 1656 | // do not reserve a location there. | |
| 1657 | // | |
| 1658 | //------------------------------------------------------------------------------ | |
| 1659 | int32_t RegexCompile::blockTopLoc(UBool reserveLoc) { | |
| 1660 | int32_t theLoc; | |
| 1661 | if (fRXPat->fCompiledPat->size() == fMatchCloseParen) | |
| 1662 | { | |
| 1663 | // The item just processed is a parenthesized block. | |
| 1664 | theLoc = fMatchOpenParen; // A slot is already reserved for us. | |
| 1665 | U_ASSERT(theLoc > 0); | |
| 374ca955 | 1666 | U_ASSERT(URX_TYPE(((uint32_t)fRXPat->fCompiledPat->elementAti(theLoc))) == URX_NOP); |
| b75a7d8f A |
1667 | } |
| 1668 | else { | |
| 1669 | // Item just compiled is a single thing, a ".", or a single char, or a set reference. | |
| 1670 | // No slot for STATE_SAVE was pre-reserved in the compiled code. | |
| 1671 | // We need to make space now. | |
| 1672 | fixLiterals(TRUE); // If last item was a string, separate the last char. | |
| 1673 | theLoc = fRXPat->fCompiledPat->size()-1; | |
| 1674 | if (reserveLoc) { | |
| 374ca955 | 1675 | /*int32_t opAtTheLoc = fRXPat->fCompiledPat->elementAti(theLoc);*/ |
| b75a7d8f A |
1676 | int32_t nop = URX_BUILD(URX_NOP, 0); |
| 1677 | fRXPat->fCompiledPat->insertElementAt(nop, theLoc, *fStatus); | |
| 1678 | } | |
| 1679 | } | |
| 1680 | return theLoc; | |
| 1681 | } | |
| 1682 | ||
| 1683 | ||
| 1684 | ||
| 1685 | //------------------------------------------------------------------------------ | |
| 1686 | // | |
| 1687 | // handleCloseParen When compiling a close paren, we need to go back | |
| 1688 | // and fix up any JMP or SAVE operations within the | |
| 1689 | // parenthesized block that need to target the end | |
| 1690 | // of the block. The locations of these are kept on | |
| 1691 | // the paretheses stack. | |
| 1692 | // | |
| 1693 | // This function is called both when encountering a | |
| 1694 | // real ) and at the end of the pattern. | |
| 1695 | // | |
| 73c04bcf | 1696 | //------------------------------------------------------------------------------ |
| b75a7d8f A |
1697 | void RegexCompile::handleCloseParen() { |
| 1698 | int32_t patIdx; | |
| 1699 | int32_t patOp; | |
| 1700 | if (fParenStack.size() <= 0) { | |
| 1701 | error(U_REGEX_MISMATCHED_PAREN); | |
| 1702 | return; | |
| 1703 | } | |
| 1704 | ||
| 1705 | // Force any literal chars that may follow the close paren to start a new string, | |
| 1706 | // and not attach to any preceding it. | |
| 1707 | fixLiterals(FALSE); | |
| 1708 | ||
| 1709 | // Fixup any operations within the just-closed parenthesized group | |
| 1710 | // that need to reference the end of the (block). | |
| 1711 | // (The first one popped from the stack is an unused slot for | |
| 1712 | // alternation (OR) state save, but applying the fixup to it does no harm.) | |
| 1713 | for (;;) { | |
| 1714 | patIdx = fParenStack.popi(); | |
| 1715 | if (patIdx < 0) { | |
| 1716 | // value < 0 flags the start of the frame on the paren stack. | |
| 1717 | break; | |
| 1718 | } | |
| 1719 | U_ASSERT(patIdx>0 && patIdx <= fRXPat->fCompiledPat->size()); | |
| 1720 | patOp = fRXPat->fCompiledPat->elementAti(patIdx); | |
| 1721 | U_ASSERT(URX_VAL(patOp) == 0); // Branch target for JMP should not be set. | |
| 1722 | patOp |= fRXPat->fCompiledPat->size(); // Set it now. | |
| 1723 | fRXPat->fCompiledPat->setElementAt(patOp, patIdx); | |
| 1724 | fMatchOpenParen = patIdx; | |
| 1725 | } | |
| 1726 | ||
| 1727 | // At the close of any parenthesized block, restore the match mode flags to | |
| 1728 | // the value they had at the open paren. Saved value is | |
| 1729 | // at the top of the paren stack. | |
| 1730 | fModeFlags = fParenStack.popi(); | |
| 73c04bcf | 1731 | U_ASSERT(fModeFlags < 0); |
| b75a7d8f A |
1732 | |
| 1733 | // DO any additional fixups, depending on the specific kind of | |
| 1734 | // parentesized grouping this is | |
| 1735 | ||
| 1736 | switch (patIdx) { | |
| 1737 | case plain: | |
| 1738 | case flags: | |
| 1739 | // No additional fixups required. | |
| 1740 | // (Grouping-only parentheses) | |
| 1741 | break; | |
| 1742 | case capturing: | |
| 1743 | // Capturing Parentheses. | |
| 1744 | // Insert a End Capture op into the pattern. | |
| 1745 | // The frame offset of the variables for this cg is obtained from the | |
| 1746 | // start capture op and put it into the end-capture op. | |
| 1747 | { | |
| 1748 | int32_t captureOp = fRXPat->fCompiledPat->elementAti(fMatchOpenParen+1); | |
| 1749 | U_ASSERT(URX_TYPE(captureOp) == URX_START_CAPTURE); | |
| 1750 | ||
| 1751 | int32_t frameVarLocation = URX_VAL(captureOp); | |
| 1752 | int32_t endCaptureOp = URX_BUILD(URX_END_CAPTURE, frameVarLocation); | |
| 1753 | fRXPat->fCompiledPat->addElement(endCaptureOp, *fStatus); | |
| 1754 | } | |
| 1755 | break; | |
| 1756 | case atomic: | |
| 1757 | // Atomic Parenthesis. | |
| 1758 | // Insert a LD_SP operation to restore the state stack to the position | |
| 1759 | // it was when the atomic parens were entered. | |
| 1760 | { | |
| 1761 | int32_t stoOp = fRXPat->fCompiledPat->elementAti(fMatchOpenParen+1); | |
| 1762 | U_ASSERT(URX_TYPE(stoOp) == URX_STO_SP); | |
| 1763 | int32_t stoLoc = URX_VAL(stoOp); | |
| 1764 | int32_t ldOp = URX_BUILD(URX_LD_SP, stoLoc); | |
| 1765 | fRXPat->fCompiledPat->addElement(ldOp, *fStatus); | |
| 1766 | } | |
| 1767 | break; | |
| 1768 | ||
| 1769 | case lookAhead: | |
| 1770 | { | |
| 1771 | int32_t startOp = fRXPat->fCompiledPat->elementAti(fMatchOpenParen-1); | |
| 1772 | U_ASSERT(URX_TYPE(startOp) == URX_LA_START); | |
| 1773 | int32_t dataLoc = URX_VAL(startOp); | |
| 1774 | int32_t op = URX_BUILD(URX_LA_END, dataLoc); | |
| 1775 | fRXPat->fCompiledPat->addElement(op, *fStatus); | |
| 1776 | } | |
| 1777 | break; | |
| 1778 | ||
| 1779 | case negLookAhead: | |
| 1780 | { | |
| 1781 | // See comment at doOpenLookAheadNeg | |
| 1782 | int32_t startOp = fRXPat->fCompiledPat->elementAti(fMatchOpenParen-1); | |
| 1783 | U_ASSERT(URX_TYPE(startOp) == URX_LA_START); | |
| 1784 | int32_t dataLoc = URX_VAL(startOp); | |
| 1785 | int32_t op = URX_BUILD(URX_LA_END, dataLoc); | |
| 1786 | fRXPat->fCompiledPat->addElement(op, *fStatus); | |
| 1787 | op = URX_BUILD(URX_FAIL, 0); | |
| 1788 | fRXPat->fCompiledPat->addElement(op, *fStatus); | |
| 1789 | ||
| 1790 | // Patch the URX_SAVE near the top of the block. | |
| 1791 | int32_t saveOp = fRXPat->fCompiledPat->elementAti(fMatchOpenParen); | |
| 1792 | U_ASSERT(URX_TYPE(saveOp) == URX_STATE_SAVE); | |
| 1793 | int32_t dest = fRXPat->fCompiledPat->size(); | |
| 1794 | saveOp = URX_BUILD(URX_STATE_SAVE, dest); | |
| 1795 | fRXPat->fCompiledPat->setElementAt(saveOp, fMatchOpenParen); | |
| 1796 | } | |
| 1797 | break; | |
| 1798 | ||
| 1799 | case lookBehind: | |
| 1800 | { | |
| 1801 | // See comment at doOpenLookBehind. | |
| 1802 | ||
| 1803 | // Append the URX_LB_END and URX_LA_END to the compiled pattern. | |
| 1804 | int32_t startOp = fRXPat->fCompiledPat->elementAti(fMatchOpenParen-4); | |
| 1805 | U_ASSERT(URX_TYPE(startOp) == URX_LB_START); | |
| 1806 | int32_t dataLoc = URX_VAL(startOp); | |
| 1807 | int32_t op = URX_BUILD(URX_LB_END, dataLoc); | |
| 1808 | fRXPat->fCompiledPat->addElement(op, *fStatus); | |
| 1809 | op = URX_BUILD(URX_LA_END, dataLoc); | |
| 1810 | fRXPat->fCompiledPat->addElement(op, *fStatus); | |
| 1811 | ||
| 1812 | // Determine the min and max bounds for the length of the | |
| 1813 | // string that the pattern can match. | |
| 1814 | // An unbounded upper limit is an error. | |
| 1815 | int32_t patEnd = fRXPat->fCompiledPat->size() - 1; | |
| 1816 | int32_t minML = minMatchLength(fMatchOpenParen, patEnd); | |
| 1817 | int32_t maxML = maxMatchLength(fMatchOpenParen, patEnd); | |
| 1818 | if (maxML == INT32_MAX) { | |
| 1819 | error(U_REGEX_LOOK_BEHIND_LIMIT); | |
| 1820 | break; | |
| 1821 | } | |
| 1822 | U_ASSERT(minML <= maxML); | |
| 1823 | ||
| 1824 | // Insert the min and max match len bounds into the URX_LB_CONT op that | |
| 1825 | // appears at the top of the look-behind block, at location fMatchOpenParen+1 | |
| 1826 | fRXPat->fCompiledPat->setElementAt(minML, fMatchOpenParen-2); | |
| 1827 | fRXPat->fCompiledPat->setElementAt(maxML, fMatchOpenParen-1); | |
| 1828 | ||
| 1829 | } | |
| 1830 | break; | |
| 1831 | ||
| 1832 | ||
| 1833 | ||
| 1834 | case lookBehindN: | |
| 1835 | { | |
| 1836 | // See comment at doOpenLookBehindNeg. | |
| 1837 | ||
| 1838 | // Append the URX_LBN_END to the compiled pattern. | |
| 1839 | int32_t startOp = fRXPat->fCompiledPat->elementAti(fMatchOpenParen-5); | |
| 1840 | U_ASSERT(URX_TYPE(startOp) == URX_LB_START); | |
| 1841 | int32_t dataLoc = URX_VAL(startOp); | |
| 1842 | int32_t op = URX_BUILD(URX_LBN_END, dataLoc); | |
| 1843 | fRXPat->fCompiledPat->addElement(op, *fStatus); | |
| 1844 | ||
| 1845 | // Determine the min and max bounds for the length of the | |
| 1846 | // string that the pattern can match. | |
| 1847 | // An unbounded upper limit is an error. | |
| 1848 | int32_t patEnd = fRXPat->fCompiledPat->size() - 1; | |
| 1849 | int32_t minML = minMatchLength(fMatchOpenParen, patEnd); | |
| 1850 | int32_t maxML = maxMatchLength(fMatchOpenParen, patEnd); | |
| 1851 | if (maxML == INT32_MAX) { | |
| 1852 | error(U_REGEX_LOOK_BEHIND_LIMIT); | |
| 1853 | break; | |
| 1854 | } | |
| 1855 | U_ASSERT(minML <= maxML); | |
| 1856 | ||
| 1857 | // Insert the min and max match len bounds into the URX_LB_CONT op that | |
| 1858 | // appears at the top of the look-behind block, at location fMatchOpenParen+1 | |
| 1859 | fRXPat->fCompiledPat->setElementAt(minML, fMatchOpenParen-3); | |
| 1860 | fRXPat->fCompiledPat->setElementAt(maxML, fMatchOpenParen-2); | |
| 1861 | ||
| 1862 | // Insert the pattern location to continue at after a successful match | |
| 1863 | // as the last operand of the URX_LBN_CONT | |
| 1864 | op = URX_BUILD(URX_RELOC_OPRND, fRXPat->fCompiledPat->size()); | |
| 1865 | fRXPat->fCompiledPat->setElementAt(op, fMatchOpenParen-1); | |
| 1866 | } | |
| 1867 | break; | |
| 1868 | ||
| 1869 | ||
| 1870 | ||
| 1871 | default: | |
| 1872 | U_ASSERT(FALSE); | |
| 1873 | } | |
| 1874 | ||
| 1875 | // remember the next location in the compiled pattern. | |
| 1876 | // The compilation of Quantifiers will look at this to see whether its looping | |
| 1877 | // over a parenthesized block or a single item | |
| 1878 | fMatchCloseParen = fRXPat->fCompiledPat->size(); | |
| 1879 | } | |
| 1880 | ||
| 1881 | ||
| 1882 | ||
| 73c04bcf | 1883 | //------------------------------------------------------------------------------ |
| b75a7d8f A |
1884 | // |
| 1885 | // compileSet Compile the pattern operations for a reference to a | |
| 1886 | // UnicodeSet. | |
| 1887 | // | |
| 73c04bcf | 1888 | //------------------------------------------------------------------------------ |
| b75a7d8f A |
1889 | void RegexCompile::compileSet(UnicodeSet *theSet) |
| 1890 | { | |
| 1891 | if (theSet == NULL) { | |
| 1892 | return; | |
| 1893 | } | |
| 1894 | int32_t setSize = theSet->size(); | |
| 1895 | UChar32 firstSetChar = theSet->charAt(0); | |
| 1896 | if (firstSetChar == -1) { | |
| 1897 | // Sets that contain only strings, but no individual chars, | |
| 1898 | // will end up here. | |
| 1899 | error(U_REGEX_SET_CONTAINS_STRING); | |
| 1900 | setSize = 0; | |
| 1901 | } | |
| 1902 | ||
| 1903 | switch (setSize) { | |
| 1904 | case 0: | |
| 1905 | { | |
| 1906 | // Set of no elements. Always fails to match. | |
| 1907 | fRXPat->fCompiledPat->addElement(URX_BUILD(URX_BACKTRACK, 0), *fStatus); | |
| 1908 | delete theSet; | |
| 1909 | } | |
| 1910 | break; | |
| 1911 | ||
| 1912 | case 1: | |
| 1913 | { | |
| 1914 | // The set contains only a single code point. Put it into | |
| 1915 | // the compiled pattern as a single char operation rather | |
| 1916 | // than a set, and discard the set itself. | |
| 1917 | literalChar(firstSetChar); | |
| 1918 | delete theSet; | |
| 1919 | } | |
| 1920 | break; | |
| 1921 | ||
| 1922 | default: | |
| 1923 | { | |
| 1924 | // The set contains two or more chars. (the normal case) | |
| 1925 | // Put it into the compiled pattern as a set. | |
| 1926 | int32_t setNumber = fRXPat->fSets->size(); | |
| 1927 | fRXPat->fSets->addElement(theSet, *fStatus); | |
| 1928 | int32_t setOp = URX_BUILD(URX_SETREF, setNumber); | |
| 1929 | fRXPat->fCompiledPat->addElement(setOp, *fStatus); | |
| 1930 | } | |
| 1931 | } | |
| 1932 | } | |
| 1933 | ||
| 1934 | ||
| 73c04bcf | 1935 | //------------------------------------------------------------------------------ |
| b75a7d8f A |
1936 | // |
| 1937 | // compileInterval Generate the code for a {min, max} style interval quantifier. | |
| 1938 | // Except for the specific opcodes used, the code is the same | |
| 1939 | // for all three types (greedy, non-greedy, possessive) of | |
| 1940 | // intervals. The opcodes are supplied as parameters. | |
| 1941 | // | |
| 1942 | // The code for interval loops has this form: | |
| 1943 | // 0 CTR_INIT counter loc (in stack frame) | |
| 1944 | // 1 5 patt address of CTR_LOOP at bottom of block | |
| 1945 | // 2 min count | |
| 1946 | // 3 max count (-1 for unbounded) | |
| 1947 | // 4 ... block to be iterated over | |
| 1948 | // 5 CTR_LOOP | |
| 1949 | // | |
| 1950 | // In | |
| 73c04bcf | 1951 | //------------------------------------------------------------------------------ |
| b75a7d8f A |
1952 | void RegexCompile::compileInterval(int32_t InitOp, int32_t LoopOp) |
| 1953 | { | |
| 1954 | // The CTR_INIT op at the top of the block with the {n,m} quantifier takes | |
| 1955 | // four slots in the compiled code. Reserve them. | |
| 1956 | int32_t topOfBlock = blockTopLoc(TRUE); | |
| 1957 | insertOp(topOfBlock); | |
| 1958 | insertOp(topOfBlock); | |
| 1959 | insertOp(topOfBlock); | |
| 1960 | ||
| 1961 | // The operands for the CTR_INIT opcode include the index in the matcher data | |
| 1962 | // of the counter. Allocate it now. | |
| 1963 | int32_t counterLoc = fRXPat->fFrameSize; | |
| 1964 | fRXPat->fFrameSize++; | |
| 1965 | ||
| 1966 | int32_t op = URX_BUILD(InitOp, counterLoc); | |
| 1967 | fRXPat->fCompiledPat->setElementAt(op, topOfBlock); | |
| 1968 | ||
| 1969 | // The second operand of CTR_INIT is the location following the end of the loop. | |
| 1970 | // Must put in as a URX_RELOC_OPRND so that the value will be adjusted if the | |
| 1971 | // compilation of something later on causes the code to grow and the target | |
| 1972 | // position to move. | |
| 1973 | int32_t loopEnd = fRXPat->fCompiledPat->size(); | |
| 1974 | op = URX_BUILD(URX_RELOC_OPRND, loopEnd); | |
| 1975 | fRXPat->fCompiledPat->setElementAt(op, topOfBlock+1); | |
| 1976 | ||
| 1977 | // Followed by the min and max counts. | |
| 1978 | fRXPat->fCompiledPat->setElementAt(fIntervalLow, topOfBlock+2); | |
| 1979 | fRXPat->fCompiledPat->setElementAt(fIntervalUpper, topOfBlock+3); | |
| 1980 | ||
| 1981 | // Apend the CTR_LOOP op. The operand is the location of the CTR_INIT op. | |
| 1982 | // Goes at end of the block being looped over, so just append to the code so far. | |
| 1983 | op = URX_BUILD(LoopOp, topOfBlock); | |
| 1984 | fRXPat->fCompiledPat->addElement(op, *fStatus); | |
| 1985 | ||
| 374ca955 A |
1986 | if ((fIntervalLow & 0xff000000) != 0 || |
| 1987 | fIntervalUpper > 0 && (fIntervalUpper & 0xff000000) != 0) { | |
| 1988 | error(U_REGEX_NUMBER_TOO_BIG); | |
| 1989 | } | |
| 1990 | ||
| b75a7d8f A |
1991 | if (fIntervalLow > fIntervalUpper && fIntervalUpper != -1) { |
| 1992 | error(U_REGEX_MAX_LT_MIN); | |
| 1993 | } | |
| b75a7d8f A |
1994 | } |
| 1995 | ||
| 1996 | ||
| 1997 | ||
| 1998 | UBool RegexCompile::compileInlineInterval() { | |
| 1999 | if (fIntervalUpper > 10 || fIntervalUpper < fIntervalLow) { | |
| 2000 | // Too big to inline. Fail, which will cause looping code to be generated. | |
| 2001 | // (Upper < Lower picks up unbounded upper and errors, both.) | |
| 2002 | return FALSE; | |
| 2003 | } | |
| 2004 | ||
| 2005 | int32_t topOfBlock = blockTopLoc(FALSE); | |
| 2006 | if (fIntervalUpper == 0) { | |
| 2007 | // Pathological case. Attempt no matches, as if the block doesn't exist. | |
| 2008 | fRXPat->fCompiledPat->setSize(topOfBlock); | |
| 2009 | return TRUE; | |
| 2010 | } | |
| 2011 | ||
| 2012 | if (topOfBlock != fRXPat->fCompiledPat->size()-1 && fIntervalUpper != 1) { | |
| 2013 | // The thing being repeated is not a single op, but some | |
| 2014 | // more complex block. Do it as a loop, not inlines. | |
| 2015 | // Note that things "repeated" a max of once are handled as inline, because | |
| 2016 | // the one copy of the code already generated is just fine. | |
| 2017 | return FALSE; | |
| 2018 | } | |
| 2019 | ||
| 2020 | // Pick up the opcode that is to be repeated | |
| 2021 | // | |
| 2022 | int32_t op = fRXPat->fCompiledPat->elementAti(topOfBlock); | |
| 2023 | ||
| 2024 | // Compute the pattern location where the inline sequence | |
| 2025 | // will end, and set up the state save op that will be needed. | |
| 2026 | // | |
| 2027 | int32_t endOfSequenceLoc = fRXPat->fCompiledPat->size()-1 | |
| 2028 | + fIntervalUpper + (fIntervalUpper-fIntervalLow); | |
| 2029 | int32_t saveOp = URX_BUILD(URX_STATE_SAVE, endOfSequenceLoc); | |
| 2030 | if (fIntervalLow == 0) { | |
| 2031 | insertOp(topOfBlock); | |
| 2032 | fRXPat->fCompiledPat->setElementAt(saveOp, topOfBlock); | |
| 2033 | } | |
| 2034 | ||
| 2035 | ||
| 2036 | ||
| 2037 | // Loop, emitting the op for the thing being repeated each time. | |
| 2038 | // Loop starts at 1 because one instance of the op already exists in the pattern, | |
| 2039 | // it was put there when it was originally encountered. | |
| 2040 | int32_t i; | |
| 2041 | for (i=1; i<fIntervalUpper; i++ ) { | |
| 2042 | if (i == fIntervalLow) { | |
| 2043 | fRXPat->fCompiledPat->addElement(saveOp, *fStatus); | |
| 2044 | } | |
| 2045 | if (i > fIntervalLow) { | |
| 2046 | fRXPat->fCompiledPat->addElement(saveOp, *fStatus); | |
| 2047 | } | |
| 2048 | fRXPat->fCompiledPat->addElement(op, *fStatus); | |
| 2049 | } | |
| 2050 | return TRUE; | |
| 2051 | } | |
| 2052 | ||
| 2053 | ||
| 2054 | ||
| 73c04bcf | 2055 | //------------------------------------------------------------------------------ |
| b75a7d8f A |
2056 | // |
| 2057 | // matchStartType Determine how a match can start. | |
| 2058 | // Used to optimize find() operations. | |
| 2059 | // | |
| 2060 | // Operation is very similar to minMatchLength(). Walk the compiled | |
| 2061 | // pattern, keeping an on-going minimum-match-length. For any | |
| 2062 | // op where the min match coming in is zero, add that ops possible | |
| 2063 | // starting matches to the possible starts for the overall pattern. | |
| 2064 | // | |
| 73c04bcf | 2065 | //------------------------------------------------------------------------------ |
| b75a7d8f A |
2066 | void RegexCompile::matchStartType() { |
| 2067 | if (U_FAILURE(*fStatus)) { | |
| 2068 | return; | |
| 2069 | } | |
| 2070 | ||
| 2071 | ||
| 2072 | int32_t loc; // Location in the pattern of the current op being processed. | |
| 2073 | int32_t op; // The op being processed | |
| 2074 | int32_t opType; // The opcode type of the op | |
| 2075 | int32_t currentLen = 0; // Minimum length of a match to this point (loc) in the pattern | |
| 2076 | int32_t numInitialStrings = 0; // Number of strings encountered that could match at start. | |
| 2077 | ||
| 2078 | UBool atStart = TRUE; // True if no part of the pattern yet encountered | |
| 2079 | // could have advanced the position in a match. | |
| 2080 | // (Maximum match length so far == 0) | |
| 2081 | ||
| 2082 | // forwardedLength is a vector holding minimum-match-length values that | |
| 2083 | // are propagated forward in the pattern by JMP or STATE_SAVE operations. | |
| 2084 | // It must be one longer than the pattern being checked because some ops | |
| 2085 | // will jmp to a end-of-block+1 location from within a block, and we must | |
| 2086 | // count those when checking the block. | |
| 2087 | int32_t end = fRXPat->fCompiledPat->size(); | |
| 2088 | UVector32 forwardedLength(end+1, *fStatus); | |
| 2089 | forwardedLength.setSize(end+1); | |
| 2090 | for (loc=3; loc<end; loc++) { | |
| 2091 | forwardedLength.setElementAt(INT32_MAX, loc); | |
| 2092 | } | |
| 2093 | ||
| 2094 | for (loc = 3; loc<end; loc++) { | |
| 2095 | op = fRXPat->fCompiledPat->elementAti(loc); | |
| 2096 | opType = URX_TYPE(op); | |
| 2097 | ||
| 2098 | // The loop is advancing linearly through the pattern. | |
| 2099 | // If the op we are now at was the destination of a branch in the pattern, | |
| 2100 | // and that path has a shorter minimum length than the current accumulated value, | |
| 2101 | // replace the current accumulated value. | |
| 2102 | U_ASSERT(currentLen>=0 && currentLen < INT32_MAX); | |
| 2103 | if (forwardedLength.elementAti(loc) < currentLen) { | |
| 2104 | currentLen = forwardedLength.elementAti(loc); | |
| 2105 | U_ASSERT(currentLen>=0 && currentLen < INT32_MAX); | |
| 2106 | } | |
| 2107 | ||
| 2108 | switch (opType) { | |
| 2109 | // Ops that don't change the total length matched | |
| 2110 | case URX_RESERVED_OP: | |
| 2111 | case URX_END: | |
| 2112 | case URX_STRING_LEN: | |
| 2113 | case URX_NOP: | |
| 2114 | case URX_START_CAPTURE: | |
| 2115 | case URX_END_CAPTURE: | |
| 2116 | case URX_BACKSLASH_B: | |
| 374ca955 | 2117 | case URX_BACKSLASH_BU: |
| b75a7d8f A |
2118 | case URX_BACKSLASH_G: |
| 2119 | case URX_BACKSLASH_Z: | |
| 2120 | case URX_DOLLAR: | |
| 2121 | case URX_RELOC_OPRND: | |
| 2122 | case URX_STO_INP_LOC: | |
| 2123 | case URX_DOLLAR_M: | |
| 2124 | case URX_BACKTRACK: | |
| 2125 | case URX_BACKREF: // BackRef. Must assume that it might be a zero length match | |
| 2126 | case URX_BACKREF_I: | |
| 2127 | ||
| 2128 | case URX_STO_SP: // Setup for atomic or possessive blocks. Doesn't change what can match. | |
| 2129 | case URX_LD_SP: | |
| 2130 | break; | |
| 2131 | ||
| 2132 | case URX_CARET: | |
| 2133 | if (atStart) { | |
| 2134 | fRXPat->fStartType = START_START; | |
| 2135 | } | |
| 2136 | break; | |
| 2137 | ||
| 2138 | case URX_CARET_M: | |
| 2139 | if (atStart) { | |
| 2140 | fRXPat->fStartType = START_LINE; | |
| 2141 | } | |
| 2142 | break; | |
| 2143 | ||
| 2144 | case URX_ONECHAR: | |
| 2145 | if (currentLen == 0) { | |
| 2146 | // This character could appear at the start of a match. | |
| 2147 | // Add it to the set of possible starting characters. | |
| 2148 | fRXPat->fInitialChars->add(URX_VAL(op)); | |
| 2149 | numInitialStrings += 2; | |
| 2150 | } | |
| 2151 | currentLen++; | |
| 2152 | atStart = FALSE; | |
| 2153 | break; | |
| 2154 | ||
| 2155 | ||
| 2156 | case URX_SETREF: | |
| 2157 | if (currentLen == 0) { | |
| 2158 | int32_t sn = URX_VAL(op); | |
| 2159 | U_ASSERT(sn > 0 && sn < fRXPat->fSets->size()); | |
| 2160 | const UnicodeSet *s = (UnicodeSet *)fRXPat->fSets->elementAt(sn); | |
| 2161 | fRXPat->fInitialChars->addAll(*s); | |
| 2162 | numInitialStrings += 2; | |
| 2163 | } | |
| 2164 | currentLen++; | |
| 2165 | atStart = FALSE; | |
| 2166 | break; | |
| 2167 | ||
| 2168 | case URX_LOOP_SR_I: | |
| 2169 | // [Set]*, like a SETREF, above, in what it can match, | |
| 2170 | // but may not match at all, so currentLen is not incremented. | |
| 2171 | if (currentLen == 0) { | |
| 2172 | int32_t sn = URX_VAL(op); | |
| 2173 | U_ASSERT(sn > 0 && sn < fRXPat->fSets->size()); | |
| 2174 | const UnicodeSet *s = (UnicodeSet *)fRXPat->fSets->elementAt(sn); | |
| 2175 | fRXPat->fInitialChars->addAll(*s); | |
| 2176 | numInitialStrings += 2; | |
| 2177 | } | |
| 2178 | atStart = FALSE; | |
| 2179 | break; | |
| 2180 | ||
| 2181 | case URX_LOOP_DOT_I: | |
| 2182 | if (currentLen == 0) { | |
| 2183 | // .* at the start of a pattern. | |
| 2184 | // Any character can begin the match. | |
| 2185 | fRXPat->fInitialChars->clear(); | |
| 2186 | fRXPat->fInitialChars->complement(); | |
| 2187 | numInitialStrings += 2; | |
| 2188 | } | |
| 2189 | atStart = FALSE; | |
| 2190 | break; | |
| 2191 | ||
| 2192 | ||
| 2193 | case URX_STATIC_SETREF: | |
| 2194 | if (currentLen == 0) { | |
| 2195 | int32_t sn = URX_VAL(op); | |
| 2196 | U_ASSERT(sn>0 && sn<URX_LAST_SET); | |
| 2197 | const UnicodeSet *s = fRXPat->fStaticSets[sn]; | |
| 2198 | fRXPat->fInitialChars->addAll(*s); | |
| 2199 | numInitialStrings += 2; | |
| 2200 | } | |
| 2201 | currentLen++; | |
| 2202 | atStart = FALSE; | |
| 2203 | break; | |
| 2204 | ||
| 2205 | ||
| 2206 | ||
| 2207 | case URX_STAT_SETREF_N: | |
| 2208 | if (currentLen == 0) { | |
| 2209 | int32_t sn = URX_VAL(op); | |
| 2210 | const UnicodeSet *s = fRXPat->fStaticSets[sn]; | |
| 2211 | UnicodeSet sc(*s); | |
| 2212 | sc.complement(); | |
| 2213 | fRXPat->fInitialChars->addAll(sc); | |
| 2214 | numInitialStrings += 2; | |
| 2215 | } | |
| 2216 | currentLen++; | |
| 2217 | atStart = FALSE; | |
| 2218 | break; | |
| 2219 | ||
| 2220 | ||
| 2221 | ||
| 2222 | case URX_BACKSLASH_D: | |
| 2223 | // Digit Char | |
| 2224 | if (currentLen == 0) { | |
| 2225 | UnicodeSet s; | |
| 2226 | s.applyIntPropertyValue(UCHAR_GENERAL_CATEGORY_MASK, U_GC_ND_MASK, *fStatus); | |
| 2227 | if (URX_VAL(op) != 0) { | |
| 2228 | s.complement(); | |
| 2229 | } | |
| 2230 | fRXPat->fInitialChars->addAll(s); | |
| 2231 | numInitialStrings += 2; | |
| 2232 | } | |
| 2233 | currentLen++; | |
| 2234 | atStart = FALSE; | |
| 2235 | break; | |
| 2236 | ||
| 2237 | ||
| 2238 | case URX_ONECHAR_I: | |
| 2239 | // Case Insensitive Single Character. | |
| 2240 | if (currentLen == 0) { | |
| 2241 | UChar32 c = URX_VAL(op); | |
| 2242 | if (u_hasBinaryProperty(c, UCHAR_CASE_SENSITIVE)) { | |
| 2243 | // character may have distinct cased forms. Add all of them | |
| 2244 | // to the set of possible starting match chars. | |
| 2245 | UnicodeSet s(c, c); | |
| 73c04bcf | 2246 | s.closeOver(USET_CASE_INSENSITIVE); |
| b75a7d8f A |
2247 | fRXPat->fInitialChars->addAll(s); |
| 2248 | } else { | |
| 2249 | // Char has no case variants. Just add it as-is to the | |
| 2250 | // set of possible starting chars. | |
| 2251 | fRXPat->fInitialChars->add(c); | |
| 2252 | } | |
| 2253 | numInitialStrings += 2; | |
| 2254 | } | |
| 2255 | currentLen++; | |
| 2256 | atStart = FALSE; | |
| 2257 | break; | |
| 2258 | ||
| 2259 | ||
| 2260 | case URX_BACKSLASH_X: // Grahpeme Cluster. Minimum is 1, max unbounded. | |
| 2261 | case URX_DOTANY_ALL: // . matches one or two. | |
| 2262 | case URX_DOTANY: | |
| 2263 | case URX_DOTANY_ALL_PL: | |
| 2264 | case URX_DOTANY_PL: | |
| 2265 | if (currentLen == 0) { | |
| 2266 | // These constructs are all bad news when they appear at the start | |
| 2267 | // of a match. Any character can begin the match. | |
| 2268 | fRXPat->fInitialChars->clear(); | |
| 2269 | fRXPat->fInitialChars->complement(); | |
| 2270 | numInitialStrings += 2; | |
| 2271 | } | |
| 2272 | currentLen++; | |
| 2273 | atStart = FALSE; | |
| 2274 | break; | |
| 2275 | ||
| 2276 | ||
| 2277 | case URX_JMPX: | |
| 2278 | loc++; // Except for extra operand on URX_JMPX, same as URX_JMP. | |
| 2279 | case URX_JMP: | |
| 2280 | { | |
| 2281 | int32_t jmpDest = URX_VAL(op); | |
| 2282 | if (jmpDest < loc) { | |
| 2283 | // Loop of some kind. Can safely ignore, the worst that will happen | |
| 2284 | // is that we understate the true minimum length | |
| 2285 | currentLen = forwardedLength.elementAti(loc+1); | |
| 2286 | ||
| 2287 | } else { | |
| 2288 | // Forward jump. Propagate the current min length to the target loc of the jump. | |
| 2289 | U_ASSERT(jmpDest <= end+1); | |
| 2290 | if (forwardedLength.elementAti(jmpDest) > currentLen) { | |
| 2291 | forwardedLength.setElementAt(currentLen, jmpDest); | |
| 2292 | } | |
| 2293 | } | |
| 2294 | } | |
| 2295 | atStart = FALSE; | |
| 2296 | break; | |
| 2297 | ||
| 2298 | case URX_JMP_SAV: | |
| 2299 | case URX_JMP_SAV_X: | |
| 2300 | // Combo of state save to the next loc, + jmp backwards. | |
| 2301 | // Net effect on min. length computation is nothing. | |
| 2302 | atStart = FALSE; | |
| 2303 | break; | |
| 2304 | ||
| 2305 | case URX_FAIL: | |
| 2306 | // Fails are kind of like a branch, except that the min length was | |
| 2307 | // propagated already, by the state save. | |
| 2308 | currentLen = forwardedLength.elementAti(loc+1); | |
| 2309 | atStart = FALSE; | |
| 2310 | break; | |
| 2311 | ||
| 2312 | ||
| 2313 | case URX_STATE_SAVE: | |
| 2314 | { | |
| 2315 | // State Save, for forward jumps, propagate the current minimum. | |
| 2316 | // of the state save. | |
| 2317 | int32_t jmpDest = URX_VAL(op); | |
| 2318 | if (jmpDest > loc) { | |
| 2319 | if (currentLen < forwardedLength.elementAti(jmpDest)) { | |
| 2320 | forwardedLength.setElementAt(currentLen, jmpDest); | |
| 2321 | } | |
| 2322 | } | |
| 2323 | } | |
| 2324 | atStart = FALSE; | |
| 2325 | break; | |
| 2326 | ||
| 2327 | ||
| 2328 | ||
| 2329 | ||
| 2330 | case URX_STRING: | |
| 2331 | { | |
| 2332 | loc++; | |
| 2333 | int32_t stringLenOp = fRXPat->fCompiledPat->elementAti(loc); | |
| 2334 | int32_t stringLen = URX_VAL(stringLenOp); | |
| 2335 | U_ASSERT(URX_TYPE(stringLenOp) == URX_STRING_LEN); | |
| 2336 | U_ASSERT(stringLenOp >= 2); | |
| 2337 | if (currentLen == 0) { | |
| 2338 | // Add the starting character of this string to the set of possible starting | |
| 2339 | // characters for this pattern. | |
| 2340 | int32_t stringStartIdx = URX_VAL(op); | |
| 2341 | UChar32 c = fRXPat->fLiteralText.char32At(stringStartIdx); | |
| 2342 | fRXPat->fInitialChars->add(c); | |
| 2343 | ||
| 2344 | // Remember this string. After the entire pattern has been checked, | |
| 2345 | // if nothing else is identified that can start a match, we'll use it. | |
| 2346 | numInitialStrings++; | |
| 2347 | fRXPat->fInitialStringIdx = stringStartIdx; | |
| 2348 | fRXPat->fInitialStringLen = stringLen; | |
| 2349 | } | |
| 2350 | ||
| 2351 | currentLen += stringLen; | |
| 2352 | atStart = FALSE; | |
| 2353 | } | |
| 2354 | break; | |
| 2355 | ||
| 2356 | case URX_STRING_I: | |
| 2357 | { | |
| 2358 | // Case-insensitive string. Unlike exact-match strings, we won't | |
| 2359 | // attempt a string search for possible match positions. But we | |
| 2360 | // do update the set of possible starting characters. | |
| 2361 | loc++; | |
| 2362 | int32_t stringLenOp = fRXPat->fCompiledPat->elementAti(loc); | |
| 2363 | int32_t stringLen = URX_VAL(stringLenOp); | |
| 2364 | U_ASSERT(URX_TYPE(stringLenOp) == URX_STRING_LEN); | |
| 2365 | U_ASSERT(stringLenOp >= 2); | |
| 2366 | if (currentLen == 0) { | |
| 2367 | // Add the starting character of this string to the set of possible starting | |
| 2368 | // characters for this pattern. | |
| 2369 | int32_t stringStartIdx = URX_VAL(op); | |
| 2370 | UChar32 c = fRXPat->fLiteralText.char32At(stringStartIdx); | |
| 2371 | UnicodeSet s(c, c); | |
| 73c04bcf | 2372 | s.closeOver(USET_CASE_INSENSITIVE); |
| b75a7d8f A |
2373 | fRXPat->fInitialChars->addAll(s); |
| 2374 | numInitialStrings += 2; // Matching on an initial string not possible. | |
| 2375 | } | |
| 2376 | currentLen += stringLen; | |
| 2377 | atStart = FALSE; | |
| 2378 | } | |
| 2379 | break; | |
| 2380 | ||
| 2381 | case URX_CTR_INIT: | |
| 2382 | case URX_CTR_INIT_NG: | |
| 2383 | { | |
| 2384 | // Loop Init Ops. These don't change the min length, but they are 4 word ops | |
| 2385 | // so location must be updated accordingly. | |
| 2386 | // Loop Init Ops. | |
| 2387 | // If the min loop count == 0 | |
| 2388 | // move loc forwards to the end of the loop, skipping over the body. | |
| 2389 | // If the min count is > 0, | |
| 2390 | // continue normal processing of the body of the loop. | |
| 2391 | int32_t loopEndLoc = fRXPat->fCompiledPat->elementAti(loc+1); | |
| 2392 | loopEndLoc = URX_VAL(loopEndLoc); | |
| 2393 | int32_t minLoopCount = fRXPat->fCompiledPat->elementAti(loc+2); | |
| 2394 | if (minLoopCount == 0) { | |
| 374ca955 A |
2395 | // Min Loop Count of 0, treat like a forward branch and |
| 2396 | // move the current minimum length up to the target | |
| 2397 | // (end of loop) location. | |
| 2398 | U_ASSERT(loopEndLoc <= end+1); | |
| 2399 | if (forwardedLength.elementAti(loopEndLoc) > currentLen) { | |
| 2400 | forwardedLength.setElementAt(currentLen, loopEndLoc); | |
| 2401 | } | |
| 2402 | } | |
| 2403 | loc+=3; // Skips over operands of CTR_INIT | |
| b75a7d8f A |
2404 | } |
| 2405 | atStart = FALSE; | |
| 2406 | break; | |
| 2407 | ||
| 2408 | ||
| 2409 | case URX_CTR_LOOP: | |
| 2410 | case URX_CTR_LOOP_NG: | |
| 2411 | // Loop ops. | |
| 2412 | // The jump is conditional, backwards only. | |
| 2413 | atStart = FALSE; | |
| 2414 | break; | |
| 2415 | ||
| 2416 | case URX_LOOP_C: | |
| 2417 | // More loop ops. These state-save to themselves. | |
| 2418 | // don't change the minimum match | |
| 2419 | atStart = FALSE; | |
| 2420 | break; | |
| 2421 | ||
| 2422 | ||
| 2423 | case URX_LA_START: | |
| 2424 | case URX_LB_START: | |
| 2425 | { | |
| 2426 | // Look-around. Scan forward until the matching look-ahead end, | |
| 2427 | // without processing the look-around block. This is overly pessimistic. | |
| 2428 | int32_t depth = 0; | |
| 2429 | for (;;) { | |
| 2430 | loc++; | |
| 2431 | op = fRXPat->fCompiledPat->elementAti(loc); | |
| 2432 | if (URX_TYPE(op) == URX_LA_START || URX_TYPE(op) == URX_LB_START) { | |
| 2433 | depth++; | |
| 2434 | } | |
| 2435 | if (URX_TYPE(op) == URX_LA_END || URX_TYPE(op)==URX_LBN_END) { | |
| 2436 | if (depth == 0) { | |
| 2437 | break; | |
| 2438 | } | |
| 2439 | depth--; | |
| 2440 | } | |
| 2441 | if (URX_TYPE(op) == URX_STATE_SAVE) { | |
| 2442 | // Need this because neg lookahead blocks will FAIL to outside | |
| 2443 | // of the block. | |
| 2444 | int32_t jmpDest = URX_VAL(op); | |
| 2445 | if (jmpDest > loc) { | |
| 2446 | if (currentLen < forwardedLength.elementAti(jmpDest)) { | |
| 2447 | forwardedLength.setElementAt(currentLen, jmpDest); | |
| 2448 | } | |
| 2449 | } | |
| 2450 | } | |
| 2451 | U_ASSERT(loc <= end); | |
| 2452 | } | |
| 2453 | } | |
| 2454 | break; | |
| 2455 | ||
| 2456 | case URX_LA_END: | |
| 2457 | case URX_LB_CONT: | |
| 2458 | case URX_LB_END: | |
| 2459 | case URX_LBN_CONT: | |
| 2460 | case URX_LBN_END: | |
| 2461 | U_ASSERT(FALSE); // Shouldn't get here. These ops should be | |
| 2462 | // consumed by the scan in URX_LA_START and LB_START | |
| 2463 | ||
| 2464 | break; | |
| 2465 | ||
| 2466 | default: | |
| 2467 | U_ASSERT(FALSE); | |
| 2468 | } | |
| 2469 | ||
| 2470 | } | |
| 2471 | ||
| 2472 | ||
| 2473 | // We have finished walking through the ops. Check whether some forward jump | |
| 2474 | // propagated a shorter length to location end+1. | |
| 2475 | if (forwardedLength.elementAti(end+1) < currentLen) { | |
| 2476 | currentLen = forwardedLength.elementAti(end+1); | |
| 2477 | } | |
| 2478 | ||
| 2479 | ||
| 2480 | fRXPat->fInitialChars8->init(fRXPat->fInitialChars); | |
| 2481 | ||
| 2482 | ||
| 2483 | // Sort out what we should check for when looking for candidate match start positions. | |
| 2484 | // In order of preference, | |
| 2485 | // 1. Start of input text buffer. | |
| 2486 | // 2. A literal string. | |
| 2487 | // 3. Start of line in multi-line mode. | |
| 2488 | // 4. A single literal character. | |
| 2489 | // 5. A character from a set of characters. | |
| 2490 | // | |
| 2491 | if (fRXPat->fStartType == START_START) { | |
| 2492 | // Match only at the start of an input text string. | |
| 2493 | // start type is already set. We're done. | |
| 2494 | } else if (numInitialStrings == 1 && fRXPat->fMinMatchLen > 0) { | |
| 2495 | // Match beginning only with a literal string. | |
| 2496 | UChar32 c = fRXPat->fLiteralText.char32At(fRXPat->fInitialStringIdx); | |
| 2497 | U_ASSERT(fRXPat->fInitialChars->contains(c)); | |
| 2498 | fRXPat->fStartType = START_STRING; | |
| 2499 | fRXPat->fInitialChar = c; | |
| 2500 | } else if (fRXPat->fStartType == START_LINE) { | |
| 374ca955 | 2501 | // Match at start of line in Multi-Line mode. |
| b75a7d8f A |
2502 | // Nothing to do here; everything is already set. |
| 2503 | } else if (fRXPat->fMinMatchLen == 0) { | |
| 2504 | // Zero length match possible. We could start anywhere. | |
| 2505 | fRXPat->fStartType = START_NO_INFO; | |
| 2506 | } else if (fRXPat->fInitialChars->size() == 1) { | |
| 2507 | // All matches begin with the same char. | |
| 2508 | fRXPat->fStartType = START_CHAR; | |
| 2509 | fRXPat->fInitialChar = fRXPat->fInitialChars->charAt(0); | |
| 2510 | U_ASSERT(fRXPat->fInitialChar != (UChar32)-1); | |
| 2511 | } else if (fRXPat->fInitialChars->contains((UChar32)0, (UChar32)0x10ffff) == FALSE && | |
| 2512 | fRXPat->fMinMatchLen > 0) { | |
| 2513 | // Matches start with a set of character smaller than the set of all chars. | |
| 2514 | fRXPat->fStartType = START_SET; | |
| 2515 | } else { | |
| 2516 | // Matches can start with anything | |
| 2517 | fRXPat->fStartType = START_NO_INFO; | |
| 2518 | } | |
| 2519 | ||
| 2520 | return; | |
| 2521 | } | |
| 2522 | ||
| 2523 | ||
| 2524 | ||
| 73c04bcf | 2525 | //------------------------------------------------------------------------------ |
| b75a7d8f A |
2526 | // |
| 2527 | // minMatchLength Calculate the length of the shortest string that could | |
| 2528 | // match the specified pattern. | |
| 2529 | // Length is in 16 bit code units, not code points. | |
| 2530 | // | |
| 2531 | // The calculated length may not be exact. The returned | |
| 2532 | // value may be shorter than the actual minimum; it must | |
| 2533 | // never be longer. | |
| 2534 | // | |
| 2535 | // start and end are the range of p-code operations to be | |
| 2536 | // examined. The endpoints are included in the range. | |
| 2537 | // | |
| 73c04bcf | 2538 | //------------------------------------------------------------------------------ |
| b75a7d8f A |
2539 | int32_t RegexCompile::minMatchLength(int32_t start, int32_t end) { |
| 2540 | if (U_FAILURE(*fStatus)) { | |
| 2541 | return 0; | |
| 2542 | } | |
| 2543 | ||
| 2544 | U_ASSERT(start <= end); | |
| 2545 | U_ASSERT(end < fRXPat->fCompiledPat->size()); | |
| 2546 | ||
| 2547 | ||
| 2548 | int32_t loc; | |
| 2549 | int32_t op; | |
| 2550 | int32_t opType; | |
| 2551 | int32_t currentLen = 0; | |
| 2552 | ||
| 2553 | ||
| 2554 | // forwardedLength is a vector holding minimum-match-length values that | |
| 2555 | // are propagated forward in the pattern by JMP or STATE_SAVE operations. | |
| 2556 | // It must be one longer than the pattern being checked because some ops | |
| 2557 | // will jmp to a end-of-block+1 location from within a block, and we must | |
| 2558 | // count those when checking the block. | |
| 2559 | UVector32 forwardedLength(end+2, *fStatus); | |
| 2560 | forwardedLength.setSize(end+2); | |
| 2561 | for (loc=start; loc<=end+1; loc++) { | |
| 2562 | forwardedLength.setElementAt(INT32_MAX, loc); | |
| 2563 | } | |
| 2564 | ||
| 2565 | for (loc = start; loc<=end; loc++) { | |
| 2566 | op = fRXPat->fCompiledPat->elementAti(loc); | |
| 2567 | opType = URX_TYPE(op); | |
| 2568 | ||
| 2569 | // The loop is advancing linearly through the pattern. | |
| 2570 | // If the op we are now at was the destination of a branch in the pattern, | |
| 2571 | // and that path has a shorter minimum length than the current accumulated value, | |
| 2572 | // replace the current accumulated value. | |
| 2573 | U_ASSERT(currentLen>=0 && currentLen < INT32_MAX); | |
| 2574 | if (forwardedLength.elementAti(loc) < currentLen) { | |
| 2575 | currentLen = forwardedLength.elementAti(loc); | |
| 2576 | U_ASSERT(currentLen>=0 && currentLen < INT32_MAX); | |
| 2577 | } | |
| 2578 | ||
| 2579 | switch (opType) { | |
| 2580 | // Ops that don't change the total length matched | |
| 2581 | case URX_RESERVED_OP: | |
| 2582 | case URX_END: | |
| 2583 | case URX_STRING_LEN: | |
| 2584 | case URX_NOP: | |
| 2585 | case URX_START_CAPTURE: | |
| 2586 | case URX_END_CAPTURE: | |
| 2587 | case URX_BACKSLASH_B: | |
| 374ca955 | 2588 | case URX_BACKSLASH_BU: |
| b75a7d8f A |
2589 | case URX_BACKSLASH_G: |
| 2590 | case URX_BACKSLASH_Z: | |
| 2591 | case URX_CARET: | |
| 2592 | case URX_DOLLAR: | |
| 2593 | case URX_RELOC_OPRND: | |
| 2594 | case URX_STO_INP_LOC: | |
| 2595 | case URX_DOLLAR_M: | |
| 2596 | case URX_CARET_M: | |
| 2597 | case URX_BACKTRACK: | |
| 2598 | case URX_BACKREF: // BackRef. Must assume that it might be a zero length match | |
| 2599 | case URX_BACKREF_I: | |
| 2600 | ||
| 2601 | case URX_STO_SP: // Setup for atomic or possessive blocks. Doesn't change what can match. | |
| 2602 | case URX_LD_SP: | |
| 2603 | ||
| 2604 | case URX_JMP_SAV: | |
| 2605 | case URX_JMP_SAV_X: | |
| 2606 | break; | |
| 2607 | ||
| 2608 | ||
| 2609 | // Ops that match a minimum of one character (one or two 16 bit code units.) | |
| 2610 | // | |
| 2611 | case URX_ONECHAR: | |
| 2612 | case URX_STATIC_SETREF: | |
| 2613 | case URX_STAT_SETREF_N: | |
| 2614 | case URX_SETREF: | |
| 2615 | case URX_BACKSLASH_D: | |
| 2616 | case URX_ONECHAR_I: | |
| 2617 | case URX_BACKSLASH_X: // Grahpeme Cluster. Minimum is 1, max unbounded. | |
| 2618 | case URX_DOTANY_ALL: // . matches one or two. | |
| 2619 | case URX_DOTANY: | |
| 2620 | case URX_DOTANY_PL: | |
| 2621 | case URX_DOTANY_ALL_PL: | |
| 2622 | currentLen++; | |
| 2623 | break; | |
| 2624 | ||
| 2625 | ||
| 2626 | case URX_JMPX: | |
| 2627 | loc++; // URX_JMPX has an extra operand, ignored here, | |
| 2628 | // otherwise processed identically to URX_JMP. | |
| 2629 | case URX_JMP: | |
| 2630 | { | |
| 2631 | int32_t jmpDest = URX_VAL(op); | |
| 2632 | if (jmpDest < loc) { | |
| 2633 | // Loop of some kind. Can safely ignore, the worst that will happen | |
| 2634 | // is that we understate the true minimum length | |
| 2635 | currentLen = forwardedLength.elementAti(loc+1); | |
| 2636 | } else { | |
| 2637 | // Forward jump. Propagate the current min length to the target loc of the jump. | |
| 2638 | U_ASSERT(jmpDest <= end+1); | |
| 2639 | if (forwardedLength.elementAti(jmpDest) > currentLen) { | |
| 2640 | forwardedLength.setElementAt(currentLen, jmpDest); | |
| 2641 | } | |
| 2642 | } | |
| 2643 | } | |
| 2644 | break; | |
| 2645 | ||
| 2646 | case URX_FAIL: | |
| 2647 | { | |
| 2648 | // Fails are kind of like a branch, except that the min length was | |
| 2649 | // propagated already, by the state save. | |
| 2650 | currentLen = forwardedLength.elementAti(loc+1); | |
| 2651 | U_ASSERT(currentLen>=0 && currentLen < INT32_MAX); | |
| 2652 | } | |
| 2653 | break; | |
| 2654 | ||
| 2655 | ||
| 2656 | case URX_STATE_SAVE: | |
| 2657 | { | |
| 2658 | // State Save, for forward jumps, propagate the current minimum. | |
| 2659 | // of the state save. | |
| 2660 | int32_t jmpDest = URX_VAL(op); | |
| 2661 | if (jmpDest > loc) { | |
| 2662 | if (currentLen < forwardedLength.elementAti(jmpDest)) { | |
| 2663 | forwardedLength.setElementAt(currentLen, jmpDest); | |
| 2664 | } | |
| 2665 | } | |
| 2666 | } | |
| 2667 | break; | |
| 2668 | ||
| 2669 | ||
| 2670 | case URX_STRING: | |
| 2671 | case URX_STRING_I: | |
| 2672 | { | |
| 2673 | loc++; | |
| 2674 | int32_t stringLenOp = fRXPat->fCompiledPat->elementAti(loc); | |
| 2675 | currentLen += URX_VAL(stringLenOp); | |
| 2676 | } | |
| 2677 | break; | |
| 2678 | ||
| 2679 | ||
| 2680 | case URX_CTR_INIT: | |
| 2681 | case URX_CTR_INIT_NG: | |
| 2682 | { | |
| 2683 | // Loop Init Ops. | |
| 2684 | // If the min loop count == 0 | |
| 2685 | // move loc forwards to the end of the loop, skipping over the body. | |
| 2686 | // If the min count is > 0, | |
| 2687 | // continue normal processing of the body of the loop. | |
| 2688 | int32_t loopEndLoc = fRXPat->fCompiledPat->elementAti(loc+1); | |
| 2689 | loopEndLoc = URX_VAL(loopEndLoc); | |
| 2690 | int32_t minLoopCount = fRXPat->fCompiledPat->elementAti(loc+2); | |
| 2691 | if (minLoopCount == 0) { | |
| 2692 | loc = loopEndLoc; | |
| 2693 | } else { | |
| 2694 | loc+=3; // Skips over operands of CTR_INIT | |
| 2695 | } | |
| 2696 | } | |
| 2697 | break; | |
| 2698 | ||
| 2699 | ||
| 2700 | case URX_CTR_LOOP: | |
| 2701 | case URX_CTR_LOOP_NG: | |
| 2702 | // Loop ops. | |
| 2703 | // The jump is conditional, backwards only. | |
| 2704 | break; | |
| 2705 | ||
| 2706 | case URX_LOOP_SR_I: | |
| 2707 | case URX_LOOP_DOT_I: | |
| 2708 | case URX_LOOP_C: | |
| 2709 | // More loop ops. These state-save to themselves. | |
| 2710 | // don't change the minimum match - could match nothing at all. | |
| 2711 | break; | |
| 2712 | ||
| 2713 | ||
| 2714 | case URX_LA_START: | |
| 2715 | case URX_LB_START: | |
| 2716 | { | |
| 2717 | // Look-around. Scan forward until the matching look-ahead end, | |
| 2718 | // without processing the look-around block. This is overly pessimistic. | |
| 2719 | // TODO: Positive lookahead could recursively do the block, then continue | |
| 2720 | // with the longer of the block or the value coming in. | |
| 2721 | int32_t depth = 0; | |
| 2722 | for (;;) { | |
| 2723 | loc++; | |
| 2724 | op = fRXPat->fCompiledPat->elementAti(loc); | |
| 2725 | if (URX_TYPE(op) == URX_LA_START || URX_TYPE(op) == URX_LB_START) { | |
| 2726 | depth++; | |
| 2727 | } | |
| 2728 | if (URX_TYPE(op) == URX_LA_END || URX_TYPE(op)==URX_LBN_END) { | |
| 2729 | if (depth == 0) { | |
| 2730 | break; | |
| 2731 | } | |
| 2732 | depth--; | |
| 2733 | } | |
| 2734 | if (URX_TYPE(op) == URX_STATE_SAVE) { | |
| 2735 | // Need this because neg lookahead blocks will FAIL to outside | |
| 2736 | // of the block. | |
| 2737 | int32_t jmpDest = URX_VAL(op); | |
| 2738 | if (jmpDest > loc) { | |
| 2739 | if (currentLen < forwardedLength.elementAti(jmpDest)) { | |
| 2740 | forwardedLength.setElementAt(currentLen, jmpDest); | |
| 2741 | } | |
| 2742 | } | |
| 2743 | } | |
| 2744 | ||
| 2745 | U_ASSERT(loc <= end); | |
| 2746 | } | |
| 2747 | } | |
| 2748 | break; | |
| 2749 | ||
| 2750 | case URX_LA_END: | |
| 2751 | case URX_LB_CONT: | |
| 2752 | case URX_LB_END: | |
| 2753 | case URX_LBN_CONT: | |
| 2754 | case URX_LBN_END: | |
| 2755 | // Only come here if the matching URX_LA_START or URX_LB_START was not in the | |
| 2756 | // range being sized, which happens when measuring size of look-behind blocks. | |
| 2757 | break; | |
| 2758 | ||
| 2759 | default: | |
| 2760 | U_ASSERT(FALSE); | |
| 2761 | } | |
| 2762 | ||
| 2763 | } | |
| 2764 | ||
| 2765 | // We have finished walking through the ops. Check whether some forward jump | |
| 2766 | // propagated a shorter length to location end+1. | |
| 2767 | if (forwardedLength.elementAti(end+1) < currentLen) { | |
| 2768 | currentLen = forwardedLength.elementAti(end+1); | |
| 2769 | U_ASSERT(currentLen>=0 && currentLen < INT32_MAX); | |
| 2770 | } | |
| 2771 | ||
| 2772 | return currentLen; | |
| 2773 | } | |
| 2774 | ||
| 2775 | ||
| 2776 | ||
| 73c04bcf | 2777 | //------------------------------------------------------------------------------ |
| b75a7d8f A |
2778 | // |
| 2779 | // maxMatchLength Calculate the length of the longest string that could | |
| 2780 | // match the specified pattern. | |
| 2781 | // Length is in 16 bit code units, not code points. | |
| 2782 | // | |
| 2783 | // The calculated length may not be exact. The returned | |
| 2784 | // value may be longer than the actual maximum; it must | |
| 2785 | // never be shorter. | |
| 2786 | // | |
| 73c04bcf | 2787 | //------------------------------------------------------------------------------ |
| b75a7d8f A |
2788 | int32_t RegexCompile::maxMatchLength(int32_t start, int32_t end) { |
| 2789 | if (U_FAILURE(*fStatus)) { | |
| 2790 | return 0; | |
| 2791 | } | |
| 2792 | U_ASSERT(start <= end); | |
| 2793 | U_ASSERT(end < fRXPat->fCompiledPat->size()); | |
| 2794 | ||
| 2795 | ||
| 2796 | int32_t loc; | |
| 2797 | int32_t op; | |
| 2798 | int32_t opType; | |
| 2799 | int32_t currentLen = 0; | |
| 2800 | UVector32 forwardedLength(end+1, *fStatus); | |
| 2801 | forwardedLength.setSize(end+1); | |
| 2802 | ||
| 2803 | for (loc=start; loc<=end; loc++) { | |
| 2804 | forwardedLength.setElementAt(0, loc); | |
| 2805 | } | |
| 2806 | ||
| 2807 | for (loc = start; loc<=end; loc++) { | |
| 2808 | op = fRXPat->fCompiledPat->elementAti(loc); | |
| 2809 | opType = URX_TYPE(op); | |
| 2810 | ||
| 2811 | // The loop is advancing linearly through the pattern. | |
| 2812 | // If the op we are now at was the destination of a branch in the pattern, | |
| 2813 | // and that path has a longer maximum length than the current accumulated value, | |
| 2814 | // replace the current accumulated value. | |
| 2815 | if (forwardedLength.elementAti(loc) > currentLen) { | |
| 2816 | currentLen = forwardedLength.elementAti(loc); | |
| 2817 | } | |
| 2818 | ||
| 2819 | switch (opType) { | |
| 2820 | // Ops that don't change the total length matched | |
| 2821 | case URX_RESERVED_OP: | |
| 2822 | case URX_END: | |
| 2823 | case URX_STRING_LEN: | |
| 2824 | case URX_NOP: | |
| 2825 | case URX_START_CAPTURE: | |
| 2826 | case URX_END_CAPTURE: | |
| 2827 | case URX_BACKSLASH_B: | |
| 374ca955 | 2828 | case URX_BACKSLASH_BU: |
| b75a7d8f A |
2829 | case URX_BACKSLASH_G: |
| 2830 | case URX_BACKSLASH_Z: | |
| 2831 | case URX_CARET: | |
| 2832 | case URX_DOLLAR: | |
| 2833 | case URX_RELOC_OPRND: | |
| 2834 | case URX_STO_INP_LOC: | |
| 2835 | case URX_DOLLAR_M: | |
| 2836 | case URX_CARET_M: | |
| 2837 | case URX_BACKTRACK: | |
| 2838 | ||
| 2839 | case URX_STO_SP: // Setup for atomic or possessive blocks. Doesn't change what can match. | |
| 2840 | case URX_LD_SP: | |
| 2841 | ||
| 2842 | case URX_LB_END: | |
| 2843 | case URX_LB_CONT: | |
| 2844 | case URX_LBN_CONT: | |
| 2845 | case URX_LBN_END: | |
| 2846 | break; | |
| 2847 | ||
| 2848 | ||
| 2849 | // Ops that increase that cause an unbounded increase in the length | |
| 2850 | // of a matched string, or that increase it a hard to characterize way. | |
| 2851 | // Call the max length unbounded, and stop further checking. | |
| 2852 | case URX_BACKREF: // BackRef. Must assume that it might be a zero length match | |
| 2853 | case URX_BACKREF_I: | |
| 2854 | case URX_BACKSLASH_X: // Grahpeme Cluster. Minimum is 1, max unbounded. | |
| 2855 | case URX_DOTANY_PL: | |
| 2856 | case URX_DOTANY_ALL_PL: | |
| 2857 | currentLen = INT32_MAX; | |
| 2858 | break; | |
| 2859 | ||
| 2860 | ||
| 2861 | // Ops that match a max of one character (possibly two 16 bit code units.) | |
| 2862 | // | |
| 2863 | case URX_STATIC_SETREF: | |
| 2864 | case URX_STAT_SETREF_N: | |
| 2865 | case URX_SETREF: | |
| 2866 | case URX_BACKSLASH_D: | |
| 2867 | case URX_ONECHAR_I: | |
| 2868 | case URX_DOTANY_ALL: | |
| 2869 | case URX_DOTANY: | |
| 2870 | currentLen+=2; | |
| 2871 | break; | |
| 2872 | ||
| 2873 | // Single literal character. Increase current max length by one or two, | |
| 2874 | // depending on whether the char is in the supplementary range. | |
| 2875 | case URX_ONECHAR: | |
| 2876 | currentLen++; | |
| 2877 | if (URX_VAL(op) > 0x10000) { | |
| 2878 | currentLen++; | |
| 2879 | } | |
| 2880 | break; | |
| 2881 | ||
| 2882 | // Jumps. | |
| 2883 | // | |
| 2884 | case URX_JMP: | |
| 2885 | case URX_JMPX: | |
| 2886 | case URX_JMP_SAV: | |
| 2887 | case URX_JMP_SAV_X: | |
| 2888 | { | |
| 2889 | int32_t jmpDest = URX_VAL(op); | |
| 2890 | if (jmpDest < loc) { | |
| 2891 | // Loop of some kind. Max match length is unbounded. | |
| 2892 | currentLen = INT32_MAX; | |
| 2893 | } else { | |
| 2894 | // Forward jump. Propagate the current min length to the target loc of the jump. | |
| 2895 | if (forwardedLength.elementAti(jmpDest) < currentLen) { | |
| 2896 | forwardedLength.setElementAt(currentLen, jmpDest); | |
| 2897 | } | |
| 2898 | currentLen = 0; | |
| 2899 | } | |
| 2900 | } | |
| 2901 | break; | |
| 2902 | ||
| 2903 | case URX_FAIL: | |
| 2904 | // Fails are kind of like a branch, except that the max length was | |
| 2905 | // propagated already, by the state save. | |
| 2906 | currentLen = forwardedLength.elementAti(loc+1); | |
| 2907 | break; | |
| 2908 | ||
| 2909 | ||
| 2910 | case URX_STATE_SAVE: | |
| 2911 | { | |
| 2912 | // State Save, for forward jumps, propagate the current minimum. | |
| 2913 | // of the state save. | |
| 2914 | // For backwards jumps, they create a loop, maximum | |
| 2915 | // match length is unbounded. | |
| 2916 | int32_t jmpDest = URX_VAL(op); | |
| 2917 | if (jmpDest > loc) { | |
| 2918 | if (currentLen > forwardedLength.elementAti(jmpDest)) { | |
| 2919 | forwardedLength.setElementAt(currentLen, jmpDest); | |
| 2920 | } | |
| 2921 | } else { | |
| 2922 | currentLen = INT32_MAX; | |
| 2923 | } | |
| 2924 | } | |
| 2925 | break; | |
| 2926 | ||
| 2927 | ||
| 2928 | ||
| 2929 | ||
| 2930 | case URX_STRING: | |
| 2931 | case URX_STRING_I: | |
| 2932 | { | |
| 2933 | loc++; | |
| 2934 | int32_t stringLenOp = fRXPat->fCompiledPat->elementAti(loc); | |
| 2935 | currentLen += URX_VAL(stringLenOp); | |
| 2936 | } | |
| 2937 | break; | |
| 2938 | ||
| 2939 | ||
| 2940 | case URX_CTR_INIT: | |
| 2941 | case URX_CTR_INIT_NG: | |
| 2942 | case URX_CTR_LOOP: | |
| 2943 | case URX_CTR_LOOP_NG: | |
| 2944 | case URX_LOOP_SR_I: | |
| 2945 | case URX_LOOP_DOT_I: | |
| 2946 | case URX_LOOP_C: | |
| 2947 | // For anything to do with loops, make the match length unbounded. | |
| 2948 | // Note: INIT instructions are multi-word. Can ignore because | |
| 2949 | // INT32_MAX length will stop the per-instruction loop. | |
| 2950 | currentLen = INT32_MAX; | |
| 2951 | break; | |
| 2952 | ||
| 2953 | ||
| 2954 | ||
| 2955 | case URX_LA_START: | |
| 2956 | case URX_LA_END: | |
| 2957 | // Look-ahead. Just ignore, treat the look-ahead block as if | |
| 2958 | // it were normal pattern. Gives a too-long match length, | |
| 2959 | // but good enough for now. | |
| 2960 | break; | |
| 2961 | ||
| 2962 | // End of look-ahead ops should always be consumed by the processing at | |
| 2963 | // the URX_LA_START op. | |
| 73c04bcf A |
2964 | // U_ASSERT(FALSE); |
| 2965 | // break; | |
| b75a7d8f A |
2966 | |
| 2967 | case URX_LB_START: | |
| 2968 | { | |
| 2969 | // Look-behind. Scan forward until the matching look-around end, | |
| 2970 | // without processing the look-behind block. | |
| 2971 | int32_t depth = 0; | |
| 2972 | for (;;) { | |
| 2973 | loc++; | |
| 2974 | op = fRXPat->fCompiledPat->elementAti(loc); | |
| 2975 | if (URX_TYPE(op) == URX_LA_START || URX_TYPE(op) == URX_LB_START) { | |
| 2976 | depth++; | |
| 2977 | } | |
| 2978 | if (URX_TYPE(op) == URX_LA_END || URX_TYPE(op)==URX_LBN_END) { | |
| 2979 | if (depth == 0) { | |
| 2980 | break; | |
| 2981 | } | |
| 2982 | depth--; | |
| 2983 | } | |
| 2984 | U_ASSERT(loc < end); | |
| 2985 | } | |
| 2986 | } | |
| 2987 | break; | |
| 2988 | ||
| 2989 | default: | |
| 2990 | U_ASSERT(FALSE); | |
| 2991 | } | |
| 2992 | ||
| 2993 | ||
| 2994 | if (currentLen == INT32_MAX) { | |
| 2995 | // The maximum length is unbounded. | |
| 2996 | // Stop further processing of the pattern. | |
| 2997 | break; | |
| 2998 | } | |
| 2999 | ||
| 3000 | } | |
| 3001 | return currentLen; | |
| 3002 | ||
| 3003 | } | |
| 3004 | ||
| 3005 | ||
| 73c04bcf | 3006 | //------------------------------------------------------------------------------ |
| b75a7d8f A |
3007 | // |
| 3008 | // stripNOPs Remove any NOP operations from the compiled pattern code. | |
| 3009 | // Extra NOPs are inserted for some constructs during the initial | |
| 3010 | // code generation to provide locations that may be patched later. | |
| 3011 | // Many end up unneeded, and are removed by this function. | |
| 3012 | // | |
| 73c04bcf | 3013 | //------------------------------------------------------------------------------ |
| b75a7d8f A |
3014 | void RegexCompile::stripNOPs() { |
| 3015 | ||
| 3016 | if (U_FAILURE(*fStatus)) { | |
| 3017 | return; | |
| 3018 | } | |
| 3019 | ||
| 3020 | int32_t end = fRXPat->fCompiledPat->size(); | |
| 3021 | UVector32 deltas(end, *fStatus); | |
| 3022 | ||
| 3023 | // Make a first pass over the code, computing the amount that things | |
| 3024 | // will be offset at each location in the original code. | |
| 3025 | int32_t loc; | |
| 3026 | int32_t d = 0; | |
| 3027 | for (loc=0; loc<end; loc++) { | |
| 3028 | deltas.addElement(d, *fStatus); | |
| 3029 | int32_t op = fRXPat->fCompiledPat->elementAti(loc); | |
| 3030 | if (URX_TYPE(op) == URX_NOP) { | |
| 3031 | d++; | |
| 3032 | } | |
| 3033 | } | |
| 3034 | ||
| 3035 | // Make a second pass over the code, removing the NOPs by moving following | |
| 3036 | // code up, and patching operands that refer to code locations that | |
| 3037 | // are being moved. The array of offsets from the first step is used | |
| 3038 | // to compute the new operand values. | |
| 3039 | int32_t src; | |
| 3040 | int32_t dst = 0; | |
| 3041 | for (src=0; src<end; src++) { | |
| 3042 | int32_t op = fRXPat->fCompiledPat->elementAti(src); | |
| 3043 | int32_t opType = URX_TYPE(op); | |
| 3044 | switch (opType) { | |
| 3045 | case URX_NOP: | |
| 3046 | break; | |
| 3047 | ||
| 3048 | case URX_STATE_SAVE: | |
| 3049 | case URX_JMP: | |
| 3050 | case URX_CTR_LOOP: | |
| 3051 | case URX_CTR_LOOP_NG: | |
| 3052 | case URX_RELOC_OPRND: | |
| 3053 | case URX_JMPX: | |
| 3054 | case URX_JMP_SAV: | |
| 3055 | case URX_JMP_SAV_X: | |
| 3056 | // These are instructions with operands that refer to code locations. | |
| 3057 | { | |
| 3058 | int32_t operandAddress = URX_VAL(op); | |
| 3059 | U_ASSERT(operandAddress>=0 && operandAddress<deltas.size()); | |
| 3060 | int32_t fixedOperandAddress = operandAddress - deltas.elementAti(operandAddress); | |
| 3061 | op = URX_BUILD(opType, fixedOperandAddress); | |
| 3062 | fRXPat->fCompiledPat->setElementAt(op, dst); | |
| 3063 | dst++; | |
| 3064 | break; | |
| 3065 | } | |
| 3066 | ||
| 3067 | case URX_RESERVED_OP: | |
| 3068 | case URX_RESERVED_OP_N: | |
| 3069 | case URX_BACKTRACK: | |
| 3070 | case URX_END: | |
| 3071 | case URX_ONECHAR: | |
| 3072 | case URX_STRING: | |
| 3073 | case URX_STRING_LEN: | |
| 3074 | case URX_START_CAPTURE: | |
| 3075 | case URX_END_CAPTURE: | |
| 3076 | case URX_STATIC_SETREF: | |
| 3077 | case URX_STAT_SETREF_N: | |
| 3078 | case URX_SETREF: | |
| 3079 | case URX_DOTANY: | |
| 3080 | case URX_FAIL: | |
| 3081 | case URX_BACKSLASH_B: | |
| 374ca955 | 3082 | case URX_BACKSLASH_BU: |
| b75a7d8f A |
3083 | case URX_BACKSLASH_G: |
| 3084 | case URX_BACKSLASH_X: | |
| 3085 | case URX_BACKSLASH_Z: | |
| 3086 | case URX_DOTANY_ALL: | |
| 3087 | case URX_DOTANY_ALL_PL: | |
| 3088 | case URX_DOTANY_PL: | |
| 3089 | case URX_BACKSLASH_D: | |
| 3090 | case URX_CARET: | |
| 3091 | case URX_DOLLAR: | |
| 3092 | case URX_CTR_INIT: | |
| 3093 | case URX_CTR_INIT_NG: | |
| 3094 | case URX_STO_SP: | |
| 3095 | case URX_LD_SP: | |
| 3096 | case URX_BACKREF: | |
| 3097 | case URX_STO_INP_LOC: | |
| 3098 | case URX_LA_START: | |
| 3099 | case URX_LA_END: | |
| 3100 | case URX_ONECHAR_I: | |
| 3101 | case URX_STRING_I: | |
| 3102 | case URX_BACKREF_I: | |
| 3103 | case URX_DOLLAR_M: | |
| 3104 | case URX_CARET_M: | |
| 3105 | case URX_LB_START: | |
| 3106 | case URX_LB_CONT: | |
| 3107 | case URX_LB_END: | |
| 3108 | case URX_LBN_CONT: | |
| 3109 | case URX_LBN_END: | |
| 3110 | case URX_LOOP_SR_I: | |
| 3111 | case URX_LOOP_DOT_I: | |
| 3112 | case URX_LOOP_C: | |
| 3113 | // These instructions are unaltered by the relocation. | |
| 3114 | fRXPat->fCompiledPat->setElementAt(op, dst); | |
| 3115 | dst++; | |
| 3116 | break; | |
| 3117 | ||
| 3118 | default: | |
| 3119 | // Some op is unaccounted for. | |
| 3120 | U_ASSERT(FALSE); | |
| 3121 | error(U_REGEX_INTERNAL_ERROR); | |
| 3122 | } | |
| 3123 | } | |
| 3124 | ||
| 3125 | fRXPat->fCompiledPat->setSize(dst); | |
| 3126 | } | |
| 3127 | ||
| 3128 | ||
| 3129 | ||
| 3130 | ||
| 73c04bcf | 3131 | //------------------------------------------------------------------------------ |
| b75a7d8f A |
3132 | // |
| 3133 | // OptDotStar Optimize patterns that end with a '.*' or '.+' to | |
| 3134 | // just advance the input to the end. | |
| 3135 | // | |
| 3136 | // Transform this compiled sequence | |
| 3137 | // [DOT_ANY | DOT_ANY_ALL] | |
| 3138 | // JMP_SAV to previous instruction | |
| 3139 | // [NOP | END_CAPTURE | DOLLAR | BACKSLASH_Z]* | |
| 3140 | // END | |
| 3141 | // | |
| 3142 | // To | |
| 3143 | // NOP | |
| 3144 | // [DOT_ANY_PL | DOT_ANY_ALL_PL] | |
| 3145 | // [NOP | END_CAPTURE | DOLLAR | BACKSLASH_Z]* | |
| 3146 | // END | |
| 3147 | // | |
| 73c04bcf | 3148 | //------------------------------------------------------------------------------ |
| b75a7d8f A |
3149 | void RegexCompile::OptDotStar() { |
| 3150 | // Scan backwards in the pattern, looking for a JMP_SAV near the end. | |
| 3151 | int32_t jmpLoc; | |
| 3152 | int32_t op = 0; | |
| 3153 | int32_t opType; | |
| 3154 | for (jmpLoc=fRXPat->fCompiledPat->size(); jmpLoc--;) { | |
| 3155 | U_ASSERT(jmpLoc>0); | |
| 3156 | op = fRXPat->fCompiledPat->elementAti(jmpLoc); | |
| 3157 | opType = URX_TYPE(op); | |
| 3158 | switch(opType) { | |
| 3159 | ||
| 3160 | ||
| 3161 | case URX_END: | |
| 3162 | case URX_NOP: | |
| 3163 | case URX_END_CAPTURE: | |
| 3164 | case URX_DOLLAR_M: | |
| 3165 | case URX_DOLLAR: | |
| 3166 | case URX_BACKSLASH_Z: | |
| 3167 | // These ops may follow the JMP_SAV without preventing us from | |
| 3168 | // doing this optimization. | |
| 3169 | continue; | |
| 3170 | ||
| 3171 | case URX_JMP_SAV: | |
| 3172 | // Got a trailing JMP_SAV that's a candidate for optimization. | |
| 3173 | break; | |
| 3174 | ||
| 3175 | default: | |
| 3176 | // This optimization not possible. | |
| 3177 | return; | |
| 3178 | } | |
| 3179 | break; // from the for loop. | |
| 3180 | } | |
| 3181 | ||
| 3182 | // We found in URX_JMP_SAV near the end that is a candidate for optimizing. | |
| 3183 | // Is the target address the previous instruction? | |
| 3184 | // Is the previous instruction a flavor of URX_DOTANY | |
| 3185 | int32_t loopTopLoc = URX_VAL(op); | |
| 3186 | if (loopTopLoc != jmpLoc-1) { | |
| 3187 | return; | |
| 3188 | } | |
| 3189 | int32_t newOp; | |
| 3190 | int32_t oldOp = fRXPat->fCompiledPat->elementAti(loopTopLoc); | |
| 3191 | int32_t oldOpType = opType = URX_TYPE(oldOp); | |
| 3192 | if (oldOpType == URX_DOTANY) { | |
| 3193 | newOp = URX_BUILD(URX_DOTANY_PL, 0); | |
| 3194 | } | |
| 3195 | else if (oldOpType == URX_DOTANY_ALL) { | |
| 3196 | newOp = URX_BUILD(URX_DOTANY_ALL_PL, 0); | |
| 3197 | } else { | |
| 3198 | return; // Sequence we were looking for isn't there. | |
| 3199 | } | |
| 3200 | ||
| 3201 | // Substitute the new instructions into the pattern. | |
| 3202 | // The NOP will be removed in a later optimization step. | |
| 3203 | fRXPat->fCompiledPat->setElementAt(URX_BUILD(URX_NOP, 0), loopTopLoc); | |
| 3204 | fRXPat->fCompiledPat->setElementAt(newOp, jmpLoc); | |
| 3205 | } | |
| 3206 | ||
| 3207 | ||
| 73c04bcf | 3208 | //------------------------------------------------------------------------------ |
| b75a7d8f A |
3209 | // |
| 3210 | // Error Report a rule parse error. | |
| 3211 | // Only report it if no previous error has been recorded. | |
| 3212 | // | |
| 73c04bcf | 3213 | //------------------------------------------------------------------------------ |
| b75a7d8f A |
3214 | void RegexCompile::error(UErrorCode e) { |
| 3215 | if (U_SUCCESS(*fStatus)) { | |
| 3216 | *fStatus = e; | |
| 3217 | fParseErr->line = fLineNum; | |
| 3218 | fParseErr->offset = fCharNum; | |
| 3219 | ||
| 3220 | // Fill in the context. | |
| 3221 | // Note: extractBetween() pins supplied indicies to the string bounds. | |
| 3222 | uprv_memset(fParseErr->preContext, 0, sizeof(fParseErr->preContext)); | |
| 3223 | uprv_memset(fParseErr->postContext, 0, sizeof(fParseErr->postContext)); | |
| 3224 | fRXPat->fPattern.extractBetween(fScanIndex-U_PARSE_CONTEXT_LEN+1, fScanIndex, | |
| 3225 | fParseErr->preContext, 0); | |
| 3226 | fRXPat->fPattern.extractBetween(fScanIndex, fScanIndex+U_PARSE_CONTEXT_LEN-1, | |
| 3227 | fParseErr->postContext, 0); | |
| 3228 | } | |
| 3229 | } | |
| 3230 | ||
| 3231 | ||
| 3232 | // | |
| 3233 | // Assorted Unicode character constants. | |
| 3234 | // Numeric because there is no portable way to enter them as literals. | |
| 3235 | // (Think EBCDIC). | |
| 3236 | // | |
| 3237 | static const UChar chCR = 0x0d; // New lines, for terminating comments. | |
| 3238 | static const UChar chLF = 0x0a; | |
| 3239 | static const UChar chNEL = 0x85; // NEL newline variant | |
| 3240 | static const UChar chLS = 0x2028; // Unicode Line Separator | |
| b75a7d8f A |
3241 | static const UChar chPound = 0x23; // '#', introduces a comment. |
| 3242 | static const UChar chE = 0x45; // 'E' | |
| 73c04bcf A |
3243 | static const UChar chUpperN = 0x4E; |
| 3244 | static const UChar chLowerP = 0x70; | |
| 3245 | static const UChar chUpperP = 0x50; | |
| b75a7d8f | 3246 | static const UChar chBackSlash = 0x5c; // '\' introduces a char escape |
| b75a7d8f A |
3247 | static const UChar chLBracket = 0x5b; |
| 3248 | static const UChar chRBracket = 0x5d; | |
| 3249 | static const UChar chRBrace = 0x7d; | |
| b75a7d8f A |
3250 | |
| 3251 | ||
| 73c04bcf | 3252 | //------------------------------------------------------------------------------ |
| b75a7d8f A |
3253 | // |
| 3254 | // nextCharLL Low Level Next Char from the regex pattern. | |
| 3255 | // Get a char from the string, keep track of input position | |
| 3256 | // for error reporting. | |
| 3257 | // | |
| 73c04bcf | 3258 | //------------------------------------------------------------------------------ |
| b75a7d8f A |
3259 | UChar32 RegexCompile::nextCharLL() { |
| 3260 | UChar32 ch; | |
| 3261 | UnicodeString &pattern = fRXPat->fPattern; | |
| 3262 | ||
| 3263 | if (fPeekChar != -1) { | |
| 3264 | ch = fPeekChar; | |
| 3265 | fPeekChar = -1; | |
| 3266 | return ch; | |
| 3267 | } | |
| 3268 | if (fPatternLength==0 || fNextIndex >= fPatternLength) { | |
| 3269 | return (UChar32)-1; | |
| 3270 | } | |
| 3271 | ch = pattern.char32At(fNextIndex); | |
| 3272 | fNextIndex = pattern.moveIndex32(fNextIndex, 1); | |
| 3273 | ||
| 3274 | if (ch == chCR || | |
| 3275 | ch == chNEL || | |
| 3276 | ch == chLS || | |
| 3277 | ch == chLF && fLastChar != chCR) { | |
| 3278 | // Character is starting a new line. Bump up the line number, and | |
| 3279 | // reset the column to 0. | |
| 3280 | fLineNum++; | |
| 3281 | fCharNum=0; | |
| 3282 | if (fQuoteMode) { | |
| 3283 | error(U_REGEX_RULE_SYNTAX); | |
| 3284 | fQuoteMode = FALSE; | |
| 3285 | } | |
| 3286 | } | |
| 3287 | else { | |
| 3288 | // Character is not starting a new line. Except in the case of a | |
| 3289 | // LF following a CR, increment the column position. | |
| 3290 | if (ch != chLF) { | |
| 3291 | fCharNum++; | |
| 3292 | } | |
| 3293 | } | |
| 3294 | fLastChar = ch; | |
| 3295 | return ch; | |
| 3296 | } | |
| 3297 | ||
| 73c04bcf | 3298 | //------------------------------------------------------------------------------ |
| b75a7d8f A |
3299 | // |
| 3300 | // peekCharLL Low Level Character Scanning, sneak a peek at the next | |
| 3301 | // character without actually getting it. | |
| 3302 | // | |
| 73c04bcf | 3303 | //------------------------------------------------------------------------------ |
| b75a7d8f A |
3304 | UChar32 RegexCompile::peekCharLL() { |
| 3305 | if (fPeekChar == -1) { | |
| 3306 | fPeekChar = nextCharLL(); | |
| 3307 | } | |
| 3308 | return fPeekChar; | |
| 3309 | } | |
| 3310 | ||
| 3311 | ||
| 73c04bcf | 3312 | //------------------------------------------------------------------------------ |
| b75a7d8f A |
3313 | // |
| 3314 | // nextChar for pattern scanning. At this level, we handle stripping | |
| 3315 | // out comments and processing some backslash character escapes. | |
| 3316 | // The rest of the pattern grammar is handled at the next level up. | |
| 3317 | // | |
| 73c04bcf | 3318 | //------------------------------------------------------------------------------ |
| b75a7d8f A |
3319 | void RegexCompile::nextChar(RegexPatternChar &c) { |
| 3320 | ||
| 3321 | fScanIndex = fNextIndex; | |
| 3322 | c.fChar = nextCharLL(); | |
| 3323 | c.fQuoted = FALSE; | |
| 3324 | ||
| 3325 | if (fQuoteMode) { | |
| 3326 | c.fQuoted = TRUE; | |
| 3327 | if ((c.fChar==chBackSlash && peekCharLL()==chE) || c.fChar == (UChar32)-1) { | |
| 3328 | fQuoteMode = FALSE; // Exit quote mode, | |
| 3329 | nextCharLL(); // discard the E | |
| 3330 | nextChar(c); // recurse to get the real next char | |
| 3331 | } | |
| 3332 | } | |
| 3333 | else if (fInBackslashQuote) { | |
| 3334 | // The current character immediately follows a '\' | |
| 3335 | // Don't check for any further escapes, just return it as-is. | |
| 3336 | // Don't set c.fQuoted, because that would prevent the state machine from | |
| 3337 | // dispatching on the character. | |
| 3338 | fInBackslashQuote = FALSE; | |
| 3339 | } | |
| 3340 | else | |
| 3341 | { | |
| 3342 | // We are not in a \Q quoted region \E of the source. | |
| 3343 | // | |
| 3344 | if (fModeFlags & UREGEX_COMMENTS) { | |
| 3345 | // | |
| 3346 | // We are in free-spacing and comments mode. | |
| 3347 | // Scan through any white space and comments, until we | |
| 3348 | // reach a significant character or the end of inut. | |
| 3349 | for (;;) { | |
| 3350 | if (c.fChar == (UChar32)-1) { | |
| 3351 | break; // End of Input | |
| 3352 | } | |
| 3353 | if (c.fChar == chPound && fEOLComments == TRUE) { | |
| 3354 | // Start of a comment. Consume the rest of it, until EOF or a new line | |
| 3355 | for (;;) { | |
| 3356 | c.fChar = nextCharLL(); | |
| 3357 | if (c.fChar == (UChar32)-1 || // EOF | |
| 3358 | c.fChar == chCR || | |
| 3359 | c.fChar == chLF || | |
| 3360 | c.fChar == chNEL || | |
| 3361 | c.fChar == chLS) { | |
| 3362 | break; | |
| 3363 | } | |
| 3364 | } | |
| 3365 | } | |
| 3366 | if (uprv_isRuleWhiteSpace(c.fChar) == FALSE) { | |
| 3367 | break; | |
| 3368 | } | |
| 3369 | c.fChar = nextCharLL(); | |
| 3370 | } | |
| 3371 | } | |
| 3372 | ||
| 3373 | // | |
| 3374 | // check for backslash escaped characters. | |
| 3375 | // | |
| 3376 | int32_t startX = fNextIndex; // start and end positions of the | |
| 3377 | int32_t endX = fNextIndex; // sequence following the '\' | |
| 3378 | if (c.fChar == chBackSlash) { | |
| 3379 | if (RegexStaticSets::gStaticSets->fUnescapeCharSet->contains(peekCharLL())) { | |
| 3380 | // | |
| 3381 | // A '\' sequence that is handled by ICU's standard unescapeAt function. | |
| 3382 | // Includes \uxxxx, \n, \r, many others. | |
| 3383 | // Return the single equivalent character. | |
| 3384 | // | |
| 3385 | nextCharLL(); // get & discard the peeked char. | |
| 3386 | c.fQuoted = TRUE; | |
| 3387 | c.fChar = fRXPat->fPattern.unescapeAt(endX); | |
| 3388 | if (startX == endX) { | |
| 3389 | error(U_REGEX_BAD_ESCAPE_SEQUENCE); | |
| 3390 | } | |
| 3391 | fCharNum += endX - startX; | |
| 3392 | fNextIndex = endX; | |
| 3393 | } | |
| 3394 | else | |
| 3395 | { | |
| 3396 | // We are in a '\' escape that will be handled by the state table scanner. | |
| 3397 | // Just return the backslash, but remember that the following char is to | |
| 3398 | // be taken literally. TODO: this is awkward, think about alternatives. | |
| 3399 | fInBackslashQuote = TRUE; | |
| 3400 | } | |
| 3401 | } | |
| 3402 | } | |
| 3403 | ||
| 3404 | // re-enable # to end-of-line comments, in case they were disabled. | |
| 3405 | // They are disabled by the parser upon seeing '(?', but this lasts for | |
| 3406 | // the fetching of the next character only. | |
| 3407 | fEOLComments = TRUE; | |
| 3408 | ||
| 3409 | // putc(c.fChar, stdout); | |
| 3410 | } | |
| 3411 | ||
| 3412 | ||
| 3413 | ||
| 73c04bcf | 3414 | //------------------------------------------------------------------------------ |
| b75a7d8f A |
3415 | // |
| 3416 | // scanSet Construct a UnicodeSet from the text at the current scan | |
| 3417 | // position. Advance the scan position to the first character | |
| 3418 | // after the set. | |
| 3419 | // | |
| 3420 | // The scan position is normally under the control of the state machine | |
| 3421 | // that controls pattern parsing. UnicodeSets, however, are parsed by | |
| 3422 | // the UnicodeSet constructor, not by the Regex pattern parser. | |
| 3423 | // | |
| 73c04bcf | 3424 | //------------------------------------------------------------------------------ |
| b75a7d8f A |
3425 | UnicodeSet *RegexCompile::scanSet() { |
| 3426 | UnicodeSet *uset = NULL; | |
| 3427 | ParsePosition pos; | |
| b75a7d8f A |
3428 | int i; |
| 3429 | ||
| 3430 | if (U_FAILURE(*fStatus)) { | |
| 3431 | return NULL; | |
| 3432 | } | |
| 3433 | ||
| 3434 | pos.setIndex(fScanIndex); | |
| b75a7d8f A |
3435 | UErrorCode localStatus = U_ZERO_ERROR; |
| 3436 | uint32_t usetFlags = 0; | |
| 3437 | if (fModeFlags & UREGEX_CASE_INSENSITIVE) { | |
| 3438 | usetFlags |= USET_CASE_INSENSITIVE; | |
| 3439 | } | |
| 3440 | if (fModeFlags & UREGEX_COMMENTS) { | |
| 3441 | usetFlags |= USET_IGNORE_SPACE; | |
| 3442 | } | |
| 3443 | ||
| 3444 | uset = new UnicodeSet(fRXPat->fPattern, pos, | |
| 374ca955 | 3445 | usetFlags, NULL, localStatus); |
| b75a7d8f A |
3446 | if (U_FAILURE(localStatus)) { |
| 3447 | // TODO: Get more accurate position of the error from UnicodeSet's return info. | |
| 3448 | // UnicodeSet appears to not be reporting correctly at this time. | |
| 374ca955 | 3449 | REGEX_SCAN_DEBUG_PRINTF(("UnicodeSet parse postion.ErrorIndex = %d\n", pos.getIndex())); |
| b75a7d8f A |
3450 | error(localStatus); |
| 3451 | delete uset; | |
| 3452 | return NULL; | |
| 3453 | } | |
| 3454 | ||
| 3455 | // Advance the current scan postion over the UnicodeSet. | |
| 3456 | // Don't just set fScanIndex because the line/char positions maintained | |
| 3457 | // for error reporting would be thrown off. | |
| 3458 | i = pos.getIndex(); | |
| 3459 | for (;;) { | |
| 3460 | if (fNextIndex >= i) { | |
| 3461 | break; | |
| 3462 | } | |
| 3463 | nextCharLL(); | |
| 3464 | } | |
| 3465 | ||
| 3466 | return uset; | |
| 73c04bcf | 3467 | } |
| b75a7d8f A |
3468 | |
| 3469 | ||
| 73c04bcf | 3470 | //------------------------------------------------------------------------------ |
| b75a7d8f A |
3471 | // |
| 3472 | // scanProp Construct a UnicodeSet from the text at the current scan | |
| 3473 | // position, which will be of the form \p{whaterver} | |
| 3474 | // | |
| 3475 | // The scan position will be at the 'p' or 'P'. On return | |
| 3476 | // the scan position should be just after the '}' | |
| 3477 | // | |
| 3478 | // Return a UnicodeSet, constructed from the \P pattern, | |
| 3479 | // or NULL if the pattern is invalid. | |
| 3480 | // | |
| 73c04bcf | 3481 | //------------------------------------------------------------------------------ |
| b75a7d8f A |
3482 | UnicodeSet *RegexCompile::scanProp() { |
| 3483 | UnicodeSet *uset = NULL; | |
| 3484 | ||
| 3485 | if (U_FAILURE(*fStatus)) { | |
| 3486 | return NULL; | |
| 3487 | } | |
| 3488 | ||
| 3489 | U_ASSERT(fC.fChar == chLowerP || fC.fChar == chUpperP || fC.fChar == chUpperN); | |
| 3490 | ||
| 3491 | // enclose the \p{property} from the regex pattern source in [brackets] | |
| 3492 | UnicodeString setPattern; | |
| 3493 | setPattern.append(chLBracket); | |
| 3494 | setPattern.append(chBackSlash); | |
| 3495 | for (;;) { | |
| 3496 | setPattern.append(fC.fChar); | |
| 3497 | if (fC.fChar == chRBrace) { | |
| 3498 | break; | |
| 3499 | } | |
| 3500 | nextChar(fC); | |
| 3501 | if (fC.fChar == -1) { | |
| 3502 | // Hit the end of the input string without finding the closing '}' | |
| 3503 | error(U_REGEX_PROPERTY_SYNTAX); | |
| 3504 | return NULL; | |
| 3505 | } | |
| 3506 | } | |
| 3507 | setPattern.append(chRBracket); | |
| 3508 | ||
| 3509 | uint32_t usetFlags = 0; | |
| 3510 | if (fModeFlags & UREGEX_CASE_INSENSITIVE) { | |
| 3511 | usetFlags |= USET_CASE_INSENSITIVE; | |
| 3512 | } | |
| 3513 | if (fModeFlags & UREGEX_COMMENTS) { | |
| 3514 | usetFlags |= USET_IGNORE_SPACE; | |
| 3515 | } | |
| 3516 | ||
| 3517 | // Build the UnicodeSet from the set pattern we just built up in a string. | |
| 374ca955 | 3518 | uset = new UnicodeSet(setPattern, usetFlags, NULL, *fStatus); |
| b75a7d8f A |
3519 | if (U_FAILURE(*fStatus)) { |
| 3520 | delete uset; | |
| 3521 | uset = NULL; | |
| 3522 | } | |
| 3523 | ||
| 3524 | nextChar(fC); // Continue overall regex pattern processing with char after the '}' | |
| 3525 | return uset; | |
| 73c04bcf | 3526 | } |
| b75a7d8f A |
3527 | |
| 3528 | U_NAMESPACE_END | |
| 3529 | #endif // !UCONFIG_NO_REGULAR_EXPRESSIONS | |
| 73c04bcf | 3530 |