1// © 2016 and later: Unicode, Inc. and others.
2// License & terms of use: http://www.unicode.org/copyright.html
3//
4// file: rbbiscan.cpp
5//
6// Copyright (C) 2002-2016, International Business Machines Corporation and others.
7// All Rights Reserved.
8//
9// This file contains the Rule Based Break Iterator Rule Builder functions for
10// scanning the rules and assembling a parse tree. This is the first phase
11// of compiling the rules.
12//
13// The overall of the rules is managed by class RBBIRuleBuilder, which will
14// create and use an instance of this class as part of the process.
15//
16
17#include "unicode/utypes.h"
18
19#if !UCONFIG_NO_BREAK_ITERATION
20
21#include "unicode/unistr.h"
22#include "unicode/uniset.h"
23#include "unicode/uchar.h"
24#include "unicode/uchriter.h"
25#include "unicode/parsepos.h"
26#include "unicode/parseerr.h"
27#include "cmemory.h"
28#include "cstring.h"
29
30#include "rbbirpt.h" // Contains state table for the rbbi rules parser.
31 // generated by a Perl script.
32#include "rbbirb.h"
33#include "rbbinode.h"
34#include "rbbiscan.h"
35#include "rbbitblb.h"
36
37#include "uassert.h"
38
39//------------------------------------------------------------------------------
40//
41// Unicode Set init strings for each of the character classes needed for parsing a rule file.
42// (Initialized with hex values for portability to EBCDIC based machines.
43// Really ugly, but there's no good way to avoid it.)
44//
45// The sets are referred to by name in the rbbirpt.txt, which is the
46// source form of the state transition table for the RBBI rule parser.
47//
48//------------------------------------------------------------------------------
49static const UChar gRuleSet_rule_char_pattern[] = {
50 // Characters that may appear as literals in patterns without escaping or quoting.
51 // [ ^ [ \ p { Z } \ u 0 0 2 0
52 0x5b, 0x5e, 0x5b, 0x5c, 0x70, 0x7b, 0x5a, 0x7d, 0x5c, 0x75, 0x30, 0x30, 0x32, 0x30,
53 // - \ u 0 0 7 f ] - [ \ p
54 0x2d, 0x5c, 0x75, 0x30, 0x30, 0x37, 0x66, 0x5d, 0x2d, 0x5b, 0x5c, 0x70,
55 // { L } ] - [ \ p { N } ] ]
56 0x7b, 0x4c, 0x7d, 0x5d, 0x2d, 0x5b, 0x5c, 0x70, 0x7b, 0x4e, 0x7d, 0x5d, 0x5d, 0};
57
58static const UChar gRuleSet_name_char_pattern[] = {
59// [ _ \ p { L } \ p { N } ]
60 0x5b, 0x5f, 0x5c, 0x70, 0x7b, 0x4c, 0x7d, 0x5c, 0x70, 0x7b, 0x4e, 0x7d, 0x5d, 0};
61
62static const UChar gRuleSet_digit_char_pattern[] = {
63// [ 0 - 9 ]
64 0x5b, 0x30, 0x2d, 0x39, 0x5d, 0};
65
66static const UChar gRuleSet_name_start_char_pattern[] = {
67// [ _ \ p { L } ]
68 0x5b, 0x5f, 0x5c, 0x70, 0x7b, 0x4c, 0x7d, 0x5d, 0 };
69
70static const UChar kAny[] = {0x61, 0x6e, 0x79, 0x00}; // "any"
71
72
73U_CDECL_BEGIN
74static void U_CALLCONV RBBISetTable_deleter(void *p) {
75 icu::RBBISetTableEl *px = (icu::RBBISetTableEl *)p;
76 delete px->key;
77 // Note: px->val is owned by the linked list "fSetsListHead" in scanner.
78 // Don't delete the value nodes here.
79 uprv_free(px);
80}
81U_CDECL_END
82
83U_NAMESPACE_BEGIN
84
85//------------------------------------------------------------------------------
86//
87// Constructor.
88//
89//------------------------------------------------------------------------------
90RBBIRuleScanner::RBBIRuleScanner(RBBIRuleBuilder *rb)
91{
92 fRB = rb;
93 fScanIndex = 0;
94 fNextIndex = 0;
95 fQuoteMode = FALSE;
96 fLineNum = 1;
97 fCharNum = 0;
98 fLastChar = 0;
99
100 fStateTable = NULL;
101 fStack[0] = 0;
102 fStackPtr = 0;
103 fNodeStack[0] = NULL;
104 fNodeStackPtr = 0;
105
106 fReverseRule = FALSE;
107 fLookAheadRule = FALSE;
108 fNoChainInRule = FALSE;
109
110 fSymbolTable = NULL;
111 fSetTable = NULL;
112 fRuleNum = 0;
113 fOptionStart = 0;
114
115 // Do not check status until after all critical fields are sufficiently initialized
116 // that the destructor can run cleanly.
117 if (U_FAILURE(*rb->fStatus)) {
118 return;
119 }
120
121 //
122 // Set up the constant Unicode Sets.
123 // Note: These could be made static, lazily initialized, and shared among
124 // all instances of RBBIRuleScanners. BUT this is quite a bit simpler,
125 // and the time to build these few sets should be small compared to a
126 // full break iterator build.
127 fRuleSets[kRuleSet_rule_char-128]
128 = UnicodeSet(UnicodeString(gRuleSet_rule_char_pattern), *rb->fStatus);
129 // fRuleSets[kRuleSet_white_space-128] = [:Pattern_White_Space:]
130 fRuleSets[kRuleSet_white_space-128].
131 add(9, 0xd).add(0x20).add(0x85).add(0x200e, 0x200f).add(0x2028, 0x2029);
132 fRuleSets[kRuleSet_name_char-128]
133 = UnicodeSet(UnicodeString(gRuleSet_name_char_pattern), *rb->fStatus);
134 fRuleSets[kRuleSet_name_start_char-128]
135 = UnicodeSet(UnicodeString(gRuleSet_name_start_char_pattern), *rb->fStatus);
136 fRuleSets[kRuleSet_digit_char-128]
137 = UnicodeSet(UnicodeString(gRuleSet_digit_char_pattern), *rb->fStatus);
138 if (*rb->fStatus == U_ILLEGAL_ARGUMENT_ERROR) {
139 // This case happens if ICU's data is missing. UnicodeSet tries to look up property
140 // names from the init string, can't find them, and claims an illegal argument.
141 // Change the error so that the actual problem will be clearer to users.
142 *rb->fStatus = U_BRK_INIT_ERROR;
143 }
144 if (U_FAILURE(*rb->fStatus)) {
145 return;
146 }
147
148 fSymbolTable = new RBBISymbolTable(this, rb->fRules, *rb->fStatus);
149 if (fSymbolTable == NULL) {
150 *rb->fStatus = U_MEMORY_ALLOCATION_ERROR;
151 return;
152 }
153 fSetTable = uhash_open(uhash_hashUnicodeString, uhash_compareUnicodeString, NULL, rb->fStatus);
154 if (U_FAILURE(*rb->fStatus)) {
155 return;
156 }
157 uhash_setValueDeleter(fSetTable, RBBISetTable_deleter);
158}
159
160
161
162//------------------------------------------------------------------------------
163//
164// Destructor
165//
166//------------------------------------------------------------------------------
167RBBIRuleScanner::~RBBIRuleScanner() {
168 delete fSymbolTable;
169 if (fSetTable != NULL) {
170 uhash_close(fSetTable);
171 fSetTable = NULL;
172
173 }
174
175
176 // Node Stack.
177 // Normally has one entry, which is the entire parse tree for the rules.
178 // If errors occured, there may be additional subtrees left on the stack.
179 while (fNodeStackPtr > 0) {
180 delete fNodeStack[fNodeStackPtr];
181 fNodeStackPtr--;
182 }
183
184}
185
186//------------------------------------------------------------------------------
187//
188// doParseAction Do some action during rule parsing.
189// Called by the parse state machine.
190// Actions build the parse tree and Unicode Sets,
191// and maintain the parse stack for nested expressions.
192//
193// TODO: unify EParseAction and RBBI_RuleParseAction enum types.
194// They represent exactly the same thing. They're separate
195// only to work around enum forward declaration restrictions
196// in some compilers, while at the same time avoiding multiple
197// definitions problems. I'm sure that there's a better way.
198//
199//------------------------------------------------------------------------------
200UBool RBBIRuleScanner::doParseActions(int32_t action)
201{
202 RBBINode *n = NULL;
203
204 UBool returnVal = TRUE;
205
206 switch (action) {
207
208 case doExprStart:
209 pushNewNode(RBBINode::opStart);
210 fRuleNum++;
211 break;
212
213
214 case doNoChain:
215 // Scanned a '^' while on the rule start state.
216 fNoChainInRule = TRUE;
217 break;
218
219
220 case doExprOrOperator:
221 {
222 fixOpStack(RBBINode::precOpCat);
223 RBBINode *operandNode = fNodeStack[fNodeStackPtr--];
224 RBBINode *orNode = pushNewNode(RBBINode::opOr);
225 if (U_FAILURE(*fRB->fStatus)) {
226 break;
227 }
228 orNode->fLeftChild = operandNode;
229 operandNode->fParent = orNode;
230 }
231 break;
232
233 case doExprCatOperator:
234 // concatenation operator.
235 // For the implicit concatenation of adjacent terms in an expression that are
236 // not separated by any other operator. Action is invoked between the
237 // actions for the two terms.
238 {
239 fixOpStack(RBBINode::precOpCat);
240 RBBINode *operandNode = fNodeStack[fNodeStackPtr--];
241 RBBINode *catNode = pushNewNode(RBBINode::opCat);
242 if (U_FAILURE(*fRB->fStatus)) {
243 break;
244 }
245 catNode->fLeftChild = operandNode;
246 operandNode->fParent = catNode;
247 }
248 break;
249
250 case doLParen:
251 // Open Paren.
252 // The openParen node is a dummy operation type with a low precedence,
253 // which has the affect of ensuring that any real binary op that
254 // follows within the parens binds more tightly to the operands than
255 // stuff outside of the parens.
256 pushNewNode(RBBINode::opLParen);
257 break;
258
259 case doExprRParen:
260 fixOpStack(RBBINode::precLParen);
261 break;
262
263 case doNOP:
264 break;
265
266 case doStartAssign:
267 // We've just scanned "$variable = "
268 // The top of the node stack has the $variable ref node.
269
270 // Save the start position of the RHS text in the StartExpression node
271 // that precedes the $variableReference node on the stack.
272 // This will eventually be used when saving the full $variable replacement
273 // text as a string.
274 n = fNodeStack[fNodeStackPtr-1];
275 n->fFirstPos = fNextIndex; // move past the '='
276
277 // Push a new start-of-expression node; needed to keep parse of the
278 // RHS expression happy.
279 pushNewNode(RBBINode::opStart);
280 break;
281
282
283
284
285 case doEndAssign:
286 {
287 // We have reached the end of an assignement statement.
288 // Current scan char is the ';' that terminates the assignment.
289
290 // Terminate expression, leaves expression parse tree rooted in TOS node.
291 fixOpStack(RBBINode::precStart);
292
293 RBBINode *startExprNode = fNodeStack[fNodeStackPtr-2];
294 RBBINode *varRefNode = fNodeStack[fNodeStackPtr-1];
295 RBBINode *RHSExprNode = fNodeStack[fNodeStackPtr];
296
297 // Save original text of right side of assignment, excluding the terminating ';'
298 // in the root of the node for the right-hand-side expression.
299 RHSExprNode->fFirstPos = startExprNode->fFirstPos;
300 RHSExprNode->fLastPos = fScanIndex;
301 fRB->fRules.extractBetween(RHSExprNode->fFirstPos, RHSExprNode->fLastPos, RHSExprNode->fText);
302
303 // Expression parse tree becomes l. child of the $variable reference node.
304 varRefNode->fLeftChild = RHSExprNode;
305 RHSExprNode->fParent = varRefNode;
306
307 // Make a symbol table entry for the $variableRef node.
308 fSymbolTable->addEntry(varRefNode->fText, varRefNode, *fRB->fStatus);
309 if (U_FAILURE(*fRB->fStatus)) {
310 // This is a round-about way to get the parse position set
311 // so that duplicate symbols error messages include a line number.
312 UErrorCode t = *fRB->fStatus;
313 *fRB->fStatus = U_ZERO_ERROR;
314 error(t);
315 }
316
317 // Clean up the stack.
318 delete startExprNode;
319 fNodeStackPtr-=3;
320 break;
321 }
322
323 case doEndOfRule:
324 {
325 fixOpStack(RBBINode::precStart); // Terminate expression, leaves expression
326 if (U_FAILURE(*fRB->fStatus)) { // parse tree rooted in TOS node.
327 break;
328 }
329#ifdef RBBI_DEBUG
330 if (fRB->fDebugEnv && uprv_strstr(fRB->fDebugEnv, "rtree")) {printNodeStack("end of rule");}
331#endif
332 U_ASSERT(fNodeStackPtr == 1);
333 RBBINode *thisRule = fNodeStack[fNodeStackPtr];
334
335 // If this rule includes a look-ahead '/', add a endMark node to the
336 // expression tree.
337 if (fLookAheadRule) {
338 RBBINode *endNode = pushNewNode(RBBINode::endMark);
339 RBBINode *catNode = pushNewNode(RBBINode::opCat);
340 if (U_FAILURE(*fRB->fStatus)) {
341 break;
342 }
343 fNodeStackPtr -= 2;
344 catNode->fLeftChild = thisRule;
345 catNode->fRightChild = endNode;
346 fNodeStack[fNodeStackPtr] = catNode;
347 endNode->fVal = fRuleNum;
348 endNode->fLookAheadEnd = TRUE;
349 thisRule = catNode;
350
351 // TODO: Disable chaining out of look-ahead (hard break) rules.
352 // The break on rule match is forced, so there is no point in building up
353 // the state table to chain into another rule for a longer match.
354 }
355
356 // Mark this node as being the root of a rule.
357 thisRule->fRuleRoot = TRUE;
358
359 // Flag if chaining into this rule is wanted.
360 //
361 if (fRB->fChainRules && // If rule chaining is enabled globally via !!chain
362 !fNoChainInRule) { // and no '^' chain-in inhibit was on this rule
363 thisRule->fChainIn = TRUE;
364 }
365
366
367 // All rule expressions are ORed together.
368 // The ';' that terminates an expression really just functions as a '|' with
369 // a low operator prededence.
370 //
371 // Each of the four sets of rules are collected separately.
372 // (forward, reverse, safe_forward, safe_reverse)
373 // OR this rule into the appropriate group of them.
374 //
375 RBBINode **destRules = (fReverseRule? &fRB->fSafeRevTree : fRB->fDefaultTree);
376
377 if (*destRules != NULL) {
378 // This is not the first rule encounted.
379 // OR previous stuff (from *destRules)
380 // with the current rule expression (on the Node Stack)
381 // with the resulting OR expression going to *destRules
382 //
383 thisRule = fNodeStack[fNodeStackPtr];
384 RBBINode *prevRules = *destRules;
385 RBBINode *orNode = pushNewNode(RBBINode::opOr);
386 if (U_FAILURE(*fRB->fStatus)) {
387 break;
388 }
389 orNode->fLeftChild = prevRules;
390 prevRules->fParent = orNode;
391 orNode->fRightChild = thisRule;
392 thisRule->fParent = orNode;
393 *destRules = orNode;
394 }
395 else
396 {
397 // This is the first rule encountered (for this direction).
398 // Just move its parse tree from the stack to *destRules.
399 *destRules = fNodeStack[fNodeStackPtr];
400 }
401 fReverseRule = FALSE; // in preparation for the next rule.
402 fLookAheadRule = FALSE;
403 fNoChainInRule = FALSE;
404 fNodeStackPtr = 0;
405 }
406 break;
407
408
409 case doRuleError:
410 error(U_BRK_RULE_SYNTAX);
411 returnVal = FALSE;
412 break;
413
414
415 case doVariableNameExpectedErr:
416 error(U_BRK_RULE_SYNTAX);
417 break;
418
419
420 //
421 // Unary operands + ? *
422 // These all appear after the operand to which they apply.
423 // When we hit one, the operand (may be a whole sub expression)
424 // will be on the top of the stack.
425 // Unary Operator becomes TOS, with the old TOS as its one child.
426 case doUnaryOpPlus:
427 {
428 RBBINode *operandNode = fNodeStack[fNodeStackPtr--];
429 RBBINode *plusNode = pushNewNode(RBBINode::opPlus);
430 if (U_FAILURE(*fRB->fStatus)) {
431 break;
432 }
433 plusNode->fLeftChild = operandNode;
434 operandNode->fParent = plusNode;
435 }
436 break;
437
438 case doUnaryOpQuestion:
439 {
440 RBBINode *operandNode = fNodeStack[fNodeStackPtr--];
441 RBBINode *qNode = pushNewNode(RBBINode::opQuestion);
442 if (U_FAILURE(*fRB->fStatus)) {
443 break;
444 }
445 qNode->fLeftChild = operandNode;
446 operandNode->fParent = qNode;
447 }
448 break;
449
450 case doUnaryOpStar:
451 {
452 RBBINode *operandNode = fNodeStack[fNodeStackPtr--];
453 RBBINode *starNode = pushNewNode(RBBINode::opStar);
454 if (U_FAILURE(*fRB->fStatus)) {
455 break;
456 }
457 starNode->fLeftChild = operandNode;
458 operandNode->fParent = starNode;
459 }
460 break;
461
462 case doRuleChar:
463 // A "Rule Character" is any single character that is a literal part
464 // of the regular expression. Like a, b and c in the expression "(abc*) | [:L:]"
465 // These are pretty uncommon in break rules; the terms are more commonly
466 // sets. To keep things uniform, treat these characters like as
467 // sets that just happen to contain only one character.
468 {
469 n = pushNewNode(RBBINode::setRef);
470 if (U_FAILURE(*fRB->fStatus)) {
471 break;
472 }
473 findSetFor(UnicodeString(fC.fChar), n);
474 n->fFirstPos = fScanIndex;
475 n->fLastPos = fNextIndex;
476 fRB->fRules.extractBetween(n->fFirstPos, n->fLastPos, n->fText);
477 break;
478 }
479
480 case doDotAny:
481 // scanned a ".", meaning match any single character.
482 {
483 n = pushNewNode(RBBINode::setRef);
484 if (U_FAILURE(*fRB->fStatus)) {
485 break;
486 }
487 findSetFor(UnicodeString(TRUE, kAny, 3), n);
488 n->fFirstPos = fScanIndex;
489 n->fLastPos = fNextIndex;
490 fRB->fRules.extractBetween(n->fFirstPos, n->fLastPos, n->fText);
491 break;
492 }
493
494 case doSlash:
495 // Scanned a '/', which identifies a look-ahead break position in a rule.
496 n = pushNewNode(RBBINode::lookAhead);
497 if (U_FAILURE(*fRB->fStatus)) {
498 break;
499 }
500 n->fVal = fRuleNum;
501 n->fFirstPos = fScanIndex;
502 n->fLastPos = fNextIndex;
503 fRB->fRules.extractBetween(n->fFirstPos, n->fLastPos, n->fText);
504 fLookAheadRule = TRUE;
505 break;
506
507
508 case doStartTagValue:
509 // Scanned a '{', the opening delimiter for a tag value within a rule.
510 n = pushNewNode(RBBINode::tag);
511 if (U_FAILURE(*fRB->fStatus)) {
512 break;
513 }
514 n->fVal = 0;
515 n->fFirstPos = fScanIndex;
516 n->fLastPos = fNextIndex;
517 break;
518
519 case doTagDigit:
520 // Just scanned a decimal digit that's part of a tag value
521 {
522 n = fNodeStack[fNodeStackPtr];
523 uint32_t v = u_charDigitValue(fC.fChar);
524 U_ASSERT(v < 10);
525 n->fVal = n->fVal*10 + v;
526 break;
527 }
528
529 case doTagValue:
530 n = fNodeStack[fNodeStackPtr];
531 n->fLastPos = fNextIndex;
532 fRB->fRules.extractBetween(n->fFirstPos, n->fLastPos, n->fText);
533 break;
534
535 case doTagExpectedError:
536 error(U_BRK_MALFORMED_RULE_TAG);
537 returnVal = FALSE;
538 break;
539
540 case doOptionStart:
541 // Scanning a !!option. At the start of string.
542 fOptionStart = fScanIndex;
543 break;
544
545 case doOptionEnd:
546 {
547 UnicodeString opt(fRB->fRules, fOptionStart, fScanIndex-fOptionStart);
548 if (opt == UNICODE_STRING("chain", 5)) {
549 fRB->fChainRules = TRUE;
550 } else if (opt == UNICODE_STRING("LBCMNoChain", 11)) {
551 fRB->fLBCMNoChain = TRUE;
552 } else if (opt == UNICODE_STRING("forward", 7)) {
553 fRB->fDefaultTree = &fRB->fForwardTree;
554 } else if (opt == UNICODE_STRING("reverse", 7)) {
555 fRB->fDefaultTree = &fRB->fReverseTree;
556 } else if (opt == UNICODE_STRING("safe_forward", 12)) {
557 fRB->fDefaultTree = &fRB->fSafeFwdTree;
558 } else if (opt == UNICODE_STRING("safe_reverse", 12)) {
559 fRB->fDefaultTree = &fRB->fSafeRevTree;
560 } else if (opt == UNICODE_STRING("lookAheadHardBreak", 18)) {
561 fRB->fLookAheadHardBreak = TRUE;
562 } else if (opt == UNICODE_STRING("quoted_literals_only", 20)) {
563 fRuleSets[kRuleSet_rule_char-128].clear();
564 } else if (opt == UNICODE_STRING("unquoted_literals", 17)) {
565 fRuleSets[kRuleSet_rule_char-128].applyPattern(UnicodeString(gRuleSet_rule_char_pattern), *fRB->fStatus);
566 } else {
567 error(U_BRK_UNRECOGNIZED_OPTION);
568 }
569 }
570 break;
571
572 case doReverseDir:
573 fReverseRule = TRUE;
574 break;
575
576 case doStartVariableName:
577 n = pushNewNode(RBBINode::varRef);
578 if (U_FAILURE(*fRB->fStatus)) {
579 break;
580 }
581 n->fFirstPos = fScanIndex;
582 break;
583
584 case doEndVariableName:
585 n = fNodeStack[fNodeStackPtr];
586 if (n==NULL || n->fType != RBBINode::varRef) {
587 error(U_BRK_INTERNAL_ERROR);
588 break;
589 }
590 n->fLastPos = fScanIndex;
591 fRB->fRules.extractBetween(n->fFirstPos+1, n->fLastPos, n->fText);
592 // Look the newly scanned name up in the symbol table
593 // If there's an entry, set the l. child of the var ref to the replacement expression.
594 // (We also pass through here when scanning assignments, but no harm is done, other
595 // than a slight wasted effort that seems hard to avoid. Lookup will be null)
596 n->fLeftChild = fSymbolTable->lookupNode(n->fText);
597 break;
598
599 case doCheckVarDef:
600 n = fNodeStack[fNodeStackPtr];
601 if (n->fLeftChild == NULL) {
602 error(U_BRK_UNDEFINED_VARIABLE);
603 returnVal = FALSE;
604 }
605 break;
606
607 case doExprFinished:
608 break;
609
610 case doRuleErrorAssignExpr:
611 error(U_BRK_ASSIGN_ERROR);
612 returnVal = FALSE;
613 break;
614
615 case doExit:
616 returnVal = FALSE;
617 break;
618
619 case doScanUnicodeSet:
620 scanSet();
621 break;
622
623 default:
624 error(U_BRK_INTERNAL_ERROR);
625 returnVal = FALSE;
626 break;
627 }
628 return returnVal && U_SUCCESS(*fRB->fStatus);
629}
630
631
632
633
634//------------------------------------------------------------------------------
635//
636// Error Report a rule parse error.
637// Only report it if no previous error has been recorded.
638//
639//------------------------------------------------------------------------------
640void RBBIRuleScanner::error(UErrorCode e) {
641 if (U_SUCCESS(*fRB->fStatus)) {
642 *fRB->fStatus = e;
643 if (fRB->fParseError) {
644 fRB->fParseError->line = fLineNum;
645 fRB->fParseError->offset = fCharNum;
646 fRB->fParseError->preContext[0] = 0;
647 fRB->fParseError->postContext[0] = 0;
648 }
649 }
650}
651
652
653
654
655//------------------------------------------------------------------------------
656//
657// fixOpStack The parse stack holds partially assembled chunks of the parse tree.
658// An entry on the stack may be as small as a single setRef node,
659// or as large as the parse tree
660// for an entire expression (this will be the one item left on the stack
661// when the parsing of an RBBI rule completes.
662//
663// This function is called when a binary operator is encountered.
664// It looks back up the stack for operators that are not yet associated
665// with a right operand, and if the precedence of the stacked operator >=
666// the precedence of the current operator, binds the operand left,
667// to the previously encountered operator.
668//
669//------------------------------------------------------------------------------
670void RBBIRuleScanner::fixOpStack(RBBINode::OpPrecedence p) {
671 RBBINode *n;
672 // printNodeStack("entering fixOpStack()");
673 for (;;) {
674 n = fNodeStack[fNodeStackPtr-1]; // an operator node
675 if (n->fPrecedence == 0) {
676 RBBIDebugPuts("RBBIRuleScanner::fixOpStack, bad operator node");
677 error(U_BRK_INTERNAL_ERROR);
678 return;
679 }
680
681 if (n->fPrecedence < p || n->fPrecedence <= RBBINode::precLParen) {
682 // The most recent operand goes with the current operator,
683 // not with the previously stacked one.
684 break;
685 }
686 // Stack operator is a binary op ( '|' or concatenation)
687 // TOS operand becomes right child of this operator.
688 // Resulting subexpression becomes the TOS operand.
689 n->fRightChild = fNodeStack[fNodeStackPtr];
690 fNodeStack[fNodeStackPtr]->fParent = n;
691 fNodeStackPtr--;
692 // printNodeStack("looping in fixOpStack() ");
693 }
694
695 if (p <= RBBINode::precLParen) {
696 // Scan is at a right paren or end of expression.
697 // The scanned item must match the stack, or else there was an error.
698 // Discard the left paren (or start expr) node from the stack,
699 // leaving the completed (sub)expression as TOS.
700 if (n->fPrecedence != p) {
701 // Right paren encountered matched start of expression node, or
702 // end of expression matched with a left paren node.
703 error(U_BRK_MISMATCHED_PAREN);
704 }
705 fNodeStack[fNodeStackPtr-1] = fNodeStack[fNodeStackPtr];
706 fNodeStackPtr--;
707 // Delete the now-discarded LParen or Start node.
708 delete n;
709 }
710 // printNodeStack("leaving fixOpStack()");
711}
712
713
714
715
716//------------------------------------------------------------------------------
717//
718// findSetFor given a UnicodeString,
719// - find the corresponding Unicode Set (uset node)
720// (create one if necessary)
721// - Set fLeftChild of the caller's node (should be a setRef node)
722// to the uset node
723// Maintain a hash table of uset nodes, so the same one is always used
724// for the same string.
725// If a "to adopt" set is provided and we haven't seen this key before,
726// add the provided set to the hash table.
727// If the string is one (32 bit) char in length, the set contains
728// just one element which is the char in question.
729// If the string is "any", return a set containing all chars.
730//
731//------------------------------------------------------------------------------
732void RBBIRuleScanner::findSetFor(const UnicodeString &s, RBBINode *node, UnicodeSet *setToAdopt) {
733
734 RBBISetTableEl *el;
735
736 // First check whether we've already cached a set for this string.
737 // If so, just use the cached set in the new node.
738 // delete any set provided by the caller, since we own it.
739 el = (RBBISetTableEl *)uhash_get(fSetTable, &s);
740 if (el != NULL) {
741 delete setToAdopt;
742 node->fLeftChild = el->val;
743 U_ASSERT(node->fLeftChild->fType == RBBINode::uset);
744 return;
745 }
746
747 // Haven't seen this set before.
748 // If the caller didn't provide us with a prebuilt set,
749 // create a new UnicodeSet now.
750 if (setToAdopt == NULL) {
751 if (s.compare(kAny, -1) == 0) {
752 setToAdopt = new UnicodeSet(0x000000, 0x10ffff);
753 } else {
754 UChar32 c;
755 c = s.char32At(0);
756 setToAdopt = new UnicodeSet(c, c);
757 }
758 }
759
760 //
761 // Make a new uset node to refer to this UnicodeSet
762 // This new uset node becomes the child of the caller's setReference node.
763 //
764 RBBINode *usetNode = new RBBINode(RBBINode::uset);
765 if (usetNode == NULL) {
766 error(U_MEMORY_ALLOCATION_ERROR);
767 return;
768 }
769 usetNode->fInputSet = setToAdopt;
770 usetNode->fParent = node;
771 node->fLeftChild = usetNode;
772 usetNode->fText = s;
773
774
775 //
776 // Add the new uset node to the list of all uset nodes.
777 //
778 fRB->fUSetNodes->addElement(usetNode, *fRB->fStatus);
779
780
781 //
782 // Add the new set to the set hash table.
783 //
784 el = (RBBISetTableEl *)uprv_malloc(sizeof(RBBISetTableEl));
785 UnicodeString *tkey = new UnicodeString(s);
786 if (tkey == NULL || el == NULL || setToAdopt == NULL) {
787 // Delete to avoid memory leak
788 delete tkey;
789 tkey = NULL;
790 uprv_free(el);
791 el = NULL;
792 delete setToAdopt;
793 setToAdopt = NULL;
794
795 error(U_MEMORY_ALLOCATION_ERROR);
796 return;
797 }
798 el->key = tkey;
799 el->val = usetNode;
800 uhash_put(fSetTable, el->key, el, fRB->fStatus);
801
802 return;
803}
804
805
806
807//
808// Assorted Unicode character constants.
809// Numeric because there is no portable way to enter them as literals.
810// (Think EBCDIC).
811//
812static const UChar chCR = 0x0d; // New lines, for terminating comments.
813static const UChar chLF = 0x0a;
814static const UChar chNEL = 0x85; // NEL newline variant
815static const UChar chLS = 0x2028; // Unicode Line Separator
816static const UChar chApos = 0x27; // single quote, for quoted chars.
817static const UChar chPound = 0x23; // '#', introduces a comment.
818static const UChar chBackSlash = 0x5c; // '\' introduces a char escape
819static const UChar chLParen = 0x28;
820static const UChar chRParen = 0x29;
821
822
823//------------------------------------------------------------------------------
824//
825// stripRules Return a rules string without extra spaces.
826// (Comments are removed separately, during rule parsing.)
827//
828//------------------------------------------------------------------------------
829UnicodeString RBBIRuleScanner::stripRules(const UnicodeString &rules) {
830 UnicodeString strippedRules;
831 int32_t rulesLength = rules.length();
832 bool skippingSpaces = false;
833
834 for (int32_t idx=0; idx<rulesLength; idx = rules.moveIndex32(idx, 1)) {
835 UChar32 cp = rules.char32At(idx);
836 bool whiteSpace = u_hasBinaryProperty(cp, UCHAR_PATTERN_WHITE_SPACE);
837 if (skippingSpaces && whiteSpace) {
838 continue;
839 }
840 strippedRules.append(cp);
841 skippingSpaces = whiteSpace;
842 }
843 return strippedRules;
844}
845
846
847//------------------------------------------------------------------------------
848//
849// nextCharLL Low Level Next Char from rule input source.
850// Get a char from the input character iterator,
851// keep track of input position for error reporting.
852//
853//------------------------------------------------------------------------------
854UChar32 RBBIRuleScanner::nextCharLL() {
855 UChar32 ch;
856
857 if (fNextIndex >= fRB->fRules.length()) {
858 return (UChar32)-1;
859 }
860 ch = fRB->fRules.char32At(fNextIndex);
861 fNextIndex = fRB->fRules.moveIndex32(fNextIndex, 1);
862
863 if (ch == chCR ||
864 ch == chNEL ||
865 ch == chLS ||
866 (ch == chLF && fLastChar != chCR)) {
867 // Character is starting a new line. Bump up the line number, and
868 // reset the column to 0.
869 fLineNum++;
870 fCharNum=0;
871 if (fQuoteMode) {
872 error(U_BRK_NEW_LINE_IN_QUOTED_STRING);
873 fQuoteMode = FALSE;
874 }
875 }
876 else {
877 // Character is not starting a new line. Except in the case of a
878 // LF following a CR, increment the column position.
879 if (ch != chLF) {
880 fCharNum++;
881 }
882 }
883 fLastChar = ch;
884 return ch;
885}
886
887
888//------------------------------------------------------------------------------
889//
890// nextChar for rules scanning. At this level, we handle stripping
891// out comments and processing backslash character escapes.
892// The rest of the rules grammar is handled at the next level up.
893//
894//------------------------------------------------------------------------------
895void RBBIRuleScanner::nextChar(RBBIRuleChar &c) {
896
897 // Unicode Character constants needed for the processing done by nextChar(),
898 // in hex because literals wont work on EBCDIC machines.
899
900 fScanIndex = fNextIndex;
901 c.fChar = nextCharLL();
902 c.fEscaped = FALSE;
903
904 //
905 // check for '' sequence.
906 // These are recognized in all contexts, whether in quoted text or not.
907 //
908 if (c.fChar == chApos) {
909 if (fRB->fRules.char32At(fNextIndex) == chApos) {
910 c.fChar = nextCharLL(); // get nextChar officially so character counts
911 c.fEscaped = TRUE; // stay correct.
912 }
913 else
914 {
915 // Single quote, by itself.
916 // Toggle quoting mode.
917 // Return either '(' or ')', because quotes cause a grouping of the quoted text.
918 fQuoteMode = !fQuoteMode;
919 if (fQuoteMode == TRUE) {
920 c.fChar = chLParen;
921 } else {
922 c.fChar = chRParen;
923 }
924 c.fEscaped = FALSE; // The paren that we return is not escaped.
925 return;
926 }
927 }
928
929 if (fQuoteMode) {
930 c.fEscaped = TRUE;
931 }
932 else
933 {
934 // We are not in a 'quoted region' of the source.
935 //
936 if (c.fChar == chPound) {
937 // Start of a comment. Consume the rest of it.
938 // The new-line char that terminates the comment is always returned.
939 // It will be treated as white-space, and serves to break up anything
940 // that might otherwise incorrectly clump together with a comment in
941 // the middle (a variable name, for example.)
942 int32_t commentStart = fScanIndex;
943 for (;;) {
944 c.fChar = nextCharLL();
945 if (c.fChar == (UChar32)-1 || // EOF
946 c.fChar == chCR ||
947 c.fChar == chLF ||
948 c.fChar == chNEL ||
949 c.fChar == chLS) {break;}
950 }
951 for (int32_t i=commentStart; i<fNextIndex-1; ++i) {
952 fRB->fStrippedRules.setCharAt(i, u' ');
953 }
954 }
955 if (c.fChar == (UChar32)-1) {
956 return;
957 }
958
959 //
960 // check for backslash escaped characters.
961 // Use UnicodeString::unescapeAt() to handle them.
962 //
963 if (c.fChar == chBackSlash) {
964 c.fEscaped = TRUE;
965 int32_t startX = fNextIndex;
966 c.fChar = fRB->fRules.unescapeAt(fNextIndex);
967 if (fNextIndex == startX) {
968 error(U_BRK_HEX_DIGITS_EXPECTED);
969 }
970 fCharNum += fNextIndex-startX;
971 }
972 }
973 // putc(c.fChar, stdout);
974}
975
976//------------------------------------------------------------------------------
977//
978// Parse RBBI rules. The state machine for rules parsing is here.
979// The state tables are hand-written in the file rbbirpt.txt,
980// and converted to the form used here by a perl
981// script rbbicst.pl
982//
983//------------------------------------------------------------------------------
984void RBBIRuleScanner::parse() {
985 uint16_t state;
986 const RBBIRuleTableEl *tableEl;
987
988 if (U_FAILURE(*fRB->fStatus)) {
989 return;
990 }
991
992 state = 1;
993 nextChar(fC);
994 //
995 // Main loop for the rule parsing state machine.
996 // Runs once per state transition.
997 // Each time through optionally performs, depending on the state table,
998 // - an advance to the the next input char
999 // - an action to be performed.
1000 // - pushing or popping a state to/from the local state return stack.
1001 //
1002 for (;;) {
1003 // Bail out if anything has gone wrong.
1004 // RBBI rule file parsing stops on the first error encountered.
1005 if (U_FAILURE(*fRB->fStatus)) {
1006 break;
1007 }
1008
1009 // Quit if state == 0. This is the normal way to exit the state machine.
1010 //
1011 if (state == 0) {
1012 break;
1013 }
1014
1015 // Find the state table element that matches the input char from the rule, or the
1016 // class of the input character. Start with the first table row for this
1017 // state, then linearly scan forward until we find a row that matches the
1018 // character. The last row for each state always matches all characters, so
1019 // the search will stop there, if not before.
1020 //
1021 tableEl = &gRuleParseStateTable[state];
1022 #ifdef RBBI_DEBUG
1023 if (fRB->fDebugEnv && uprv_strstr(fRB->fDebugEnv, "scan")) {
1024 RBBIDebugPrintf("char, line, col = (\'%c\', %d, %d) state=%s ",
1025 fC.fChar, fLineNum, fCharNum, RBBIRuleStateNames[state]);
1026 }
1027 #endif
1028
1029 for (;;) {
1030 #ifdef RBBI_DEBUG
1031 if (fRB->fDebugEnv && uprv_strstr(fRB->fDebugEnv, "scan")) { RBBIDebugPrintf("."); fflush(stdout);}
1032 #endif
1033 if (tableEl->fCharClass < 127 && fC.fEscaped == FALSE && tableEl->fCharClass == fC.fChar) {
1034 // Table row specified an individual character, not a set, and
1035 // the input character is not escaped, and
1036 // the input character matched it.
1037 break;
1038 }
1039 if (tableEl->fCharClass == 255) {
1040 // Table row specified default, match anything character class.
1041 break;
1042 }
1043 if (tableEl->fCharClass == 254 && fC.fEscaped) {
1044 // Table row specified "escaped" and the char was escaped.
1045 break;
1046 }
1047 if (tableEl->fCharClass == 253 && fC.fEscaped &&
1048 (fC.fChar == 0x50 || fC.fChar == 0x70 )) {
1049 // Table row specified "escaped P" and the char is either 'p' or 'P'.
1050 break;
1051 }
1052 if (tableEl->fCharClass == 252 && fC.fChar == (UChar32)-1) {
1053 // Table row specified eof and we hit eof on the input.
1054 break;
1055 }
1056
1057 if (tableEl->fCharClass >= 128 && tableEl->fCharClass < 240 && // Table specs a char class &&
1058 fC.fEscaped == FALSE && // char is not escaped &&
1059 fC.fChar != (UChar32)-1) { // char is not EOF
1060 U_ASSERT((tableEl->fCharClass-128) < UPRV_LENGTHOF(fRuleSets));
1061 if (fRuleSets[tableEl->fCharClass-128].contains(fC.fChar)) {
1062 // Table row specified a character class, or set of characters,
1063 // and the current char matches it.
1064 break;
1065 }
1066 }
1067
1068 // No match on this row, advance to the next row for this state,
1069 tableEl++;
1070 }
1071 if (fRB->fDebugEnv && uprv_strstr(fRB->fDebugEnv, "scan")) { RBBIDebugPuts("");}
1072
1073 //
1074 // We've found the row of the state table that matches the current input
1075 // character from the rules string.
1076 // Perform any action specified by this row in the state table.
1077 if (doParseActions((int32_t)tableEl->fAction) == FALSE) {
1078 // Break out of the state machine loop if the
1079 // the action signalled some kind of error, or
1080 // the action was to exit, occurs on normal end-of-rules-input.
1081 break;
1082 }
1083
1084 if (tableEl->fPushState != 0) {
1085 fStackPtr++;
1086 if (fStackPtr >= kStackSize) {
1087 error(U_BRK_INTERNAL_ERROR);
1088 RBBIDebugPuts("RBBIRuleScanner::parse() - state stack overflow.");
1089 fStackPtr--;
1090 }
1091 fStack[fStackPtr] = tableEl->fPushState;
1092 }
1093
1094 if (tableEl->fNextChar) {
1095 nextChar(fC);
1096 }
1097
1098 // Get the next state from the table entry, or from the
1099 // state stack if the next state was specified as "pop".
1100 if (tableEl->fNextState != 255) {
1101 state = tableEl->fNextState;
1102 } else {
1103 state = fStack[fStackPtr];
1104 fStackPtr--;
1105 if (fStackPtr < 0) {
1106 error(U_BRK_INTERNAL_ERROR);
1107 RBBIDebugPuts("RBBIRuleScanner::parse() - state stack underflow.");
1108 fStackPtr++;
1109 }
1110 }
1111
1112 }
1113
1114 if (U_FAILURE(*fRB->fStatus)) {
1115 return;
1116 }
1117
1118 // If there are no forward rules set an error.
1119 //
1120 if (fRB->fForwardTree == NULL) {
1121 error(U_BRK_RULE_SYNTAX);
1122 return;
1123 }
1124
1125 //
1126 // Parsing of the input RBBI rules is complete.
1127 // We now have a parse tree for the rule expressions
1128 // and a list of all UnicodeSets that are referenced.
1129 //
1130#ifdef RBBI_DEBUG
1131 if (fRB->fDebugEnv && uprv_strstr(fRB->fDebugEnv, "symbols")) {fSymbolTable->rbbiSymtablePrint();}
1132 if (fRB->fDebugEnv && uprv_strstr(fRB->fDebugEnv, "ptree")) {
1133 RBBIDebugPrintf("Completed Forward Rules Parse Tree...\n");
1134 RBBINode::printTree(fRB->fForwardTree, TRUE);
1135 RBBIDebugPrintf("\nCompleted Reverse Rules Parse Tree...\n");
1136 RBBINode::printTree(fRB->fReverseTree, TRUE);
1137 RBBIDebugPrintf("\nCompleted Safe Point Forward Rules Parse Tree...\n");
1138 RBBINode::printTree(fRB->fSafeFwdTree, TRUE);
1139 RBBIDebugPrintf("\nCompleted Safe Point Reverse Rules Parse Tree...\n");
1140 RBBINode::printTree(fRB->fSafeRevTree, TRUE);
1141 }
1142#endif
1143}
1144
1145
1146//------------------------------------------------------------------------------
1147//
1148// printNodeStack for debugging...
1149//
1150//------------------------------------------------------------------------------
1151#ifdef RBBI_DEBUG
1152void RBBIRuleScanner::printNodeStack(const char *title) {
1153 int i;
1154 RBBIDebugPrintf("%s. Dumping node stack...\n", title);
1155 for (i=fNodeStackPtr; i>0; i--) {RBBINode::printTree(fNodeStack[i], TRUE);}
1156}
1157#endif
1158
1159
1160
1161
1162//------------------------------------------------------------------------------
1163//
1164// pushNewNode create a new RBBINode of the specified type and push it
1165// onto the stack of nodes.
1166//
1167//------------------------------------------------------------------------------
1168RBBINode *RBBIRuleScanner::pushNewNode(RBBINode::NodeType t) {
1169 if (U_FAILURE(*fRB->fStatus)) {
1170 return NULL;
1171 }
1172 if (fNodeStackPtr >= kStackSize - 1) {
1173 error(U_BRK_RULE_SYNTAX);
1174 RBBIDebugPuts("RBBIRuleScanner::pushNewNode - stack overflow.");
1175 return NULL;
1176 }
1177 fNodeStackPtr++;
1178 fNodeStack[fNodeStackPtr] = new RBBINode(t);
1179 if (fNodeStack[fNodeStackPtr] == NULL) {
1180 *fRB->fStatus = U_MEMORY_ALLOCATION_ERROR;
1181 }
1182 return fNodeStack[fNodeStackPtr];
1183}
1184
1185
1186
1187//------------------------------------------------------------------------------
1188//
1189// scanSet Construct a UnicodeSet from the text at the current scan
1190// position. Advance the scan position to the first character
1191// after the set.
1192//
1193// A new RBBI setref node referring to the set is pushed onto the node
1194// stack.
1195//
1196// The scan position is normally under the control of the state machine
1197// that controls rule parsing. UnicodeSets, however, are parsed by
1198// the UnicodeSet constructor, not by the RBBI rule parser.
1199//
1200//------------------------------------------------------------------------------
1201void RBBIRuleScanner::scanSet() {
1202 UnicodeSet *uset;
1203 ParsePosition pos;
1204 int startPos;
1205 int i;
1206
1207 if (U_FAILURE(*fRB->fStatus)) {
1208 return;
1209 }
1210
1211 pos.setIndex(fScanIndex);
1212 startPos = fScanIndex;
1213 UErrorCode localStatus = U_ZERO_ERROR;
1214 uset = new UnicodeSet();
1215 if (uset == NULL) {
1216 localStatus = U_MEMORY_ALLOCATION_ERROR;
1217 } else {
1218 uset->applyPatternIgnoreSpace(fRB->fRules, pos, fSymbolTable, localStatus);
1219 }
1220 if (U_FAILURE(localStatus)) {
1221 // TODO: Get more accurate position of the error from UnicodeSet's return info.
1222 // UnicodeSet appears to not be reporting correctly at this time.
1223 #ifdef RBBI_DEBUG
1224 RBBIDebugPrintf("UnicodeSet parse postion.ErrorIndex = %d\n", pos.getIndex());
1225 #endif
1226 error(localStatus);
1227 delete uset;
1228 return;
1229 }
1230
1231 // Verify that the set contains at least one code point.
1232 //
1233 U_ASSERT(uset!=NULL);
1234 if (uset->isEmpty()) {
1235 // This set is empty.
1236 // Make it an error, because it almost certainly is not what the user wanted.
1237 // Also, avoids having to think about corner cases in the tree manipulation code
1238 // that occurs later on.
1239 error(U_BRK_RULE_EMPTY_SET);
1240 delete uset;
1241 return;
1242 }
1243
1244
1245 // Advance the RBBI parse postion over the UnicodeSet pattern.
1246 // Don't just set fScanIndex because the line/char positions maintained
1247 // for error reporting would be thrown off.
1248 i = pos.getIndex();
1249 for (;;) {
1250 if (fNextIndex >= i) {
1251 break;
1252 }
1253 nextCharLL();
1254 }
1255
1256 if (U_SUCCESS(*fRB->fStatus)) {
1257 RBBINode *n;
1258
1259 n = pushNewNode(RBBINode::setRef);
1260 if (U_FAILURE(*fRB->fStatus)) {
1261 return;
1262 }
1263 n->fFirstPos = startPos;
1264 n->fLastPos = fNextIndex;
1265 fRB->fRules.extractBetween(n->fFirstPos, n->fLastPos, n->fText);
1266 // findSetFor() serves several purposes here:
1267 // - Adopts storage for the UnicodeSet, will be responsible for deleting.
1268 // - Mantains collection of all sets in use, needed later for establishing
1269 // character categories for run time engine.
1270 // - Eliminates mulitiple instances of the same set.
1271 // - Creates a new uset node if necessary (if this isn't a duplicate.)
1272 findSetFor(n->fText, n, uset);
1273 }
1274
1275}
1276
1277U_NAMESPACE_END
1278
1279#endif /* #if !UCONFIG_NO_BREAK_ITERATION */
1280