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 //------------------------------------------------------------------------------
49 static 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 
58 static 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 
62 static const UChar gRuleSet_digit_char_pattern[] = {
63 //    [    0      -    9     ]
64     0x5b, 0x30, 0x2d, 0x39, 0x5d, 0};
65 
66 static const UChar gRuleSet_name_start_char_pattern[] = {
67 //    [    _      \    p     {     L      }     ]
68     0x5b, 0x5f, 0x5c, 0x70, 0x7b, 0x4c, 0x7d, 0x5d, 0 };
69 
70 static const UChar kAny[] = {0x61, 0x6e, 0x79, 0x00};  // "any"
71 
72 
73 U_CDECL_BEGIN
RBBISetTable_deleter(void * p)74 static 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 }
81 U_CDECL_END
82 
83 U_NAMESPACE_BEGIN
84 
85 //------------------------------------------------------------------------------
86 //
87 //  Constructor.
88 //
89 //------------------------------------------------------------------------------
RBBIRuleScanner(RBBIRuleBuilder * rb)90 RBBIRuleScanner::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 //------------------------------------------------------------------------------
~RBBIRuleScanner()167 RBBIRuleScanner::~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 //------------------------------------------------------------------------------
doParseActions(int32_t action)200 UBool 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 //------------------------------------------------------------------------------
error(UErrorCode e)640 void 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 //------------------------------------------------------------------------------
fixOpStack(RBBINode::OpPrecedence p)670 void 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 //------------------------------------------------------------------------------
findSetFor(const UnicodeString & s,RBBINode * node,UnicodeSet * setToAdopt)732 void 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 //
812 static const UChar      chCR        = 0x0d;      // New lines, for terminating comments.
813 static const UChar      chLF        = 0x0a;
814 static const UChar      chNEL       = 0x85;      //    NEL newline variant
815 static const UChar      chLS        = 0x2028;    //    Unicode Line Separator
816 static const UChar      chApos      = 0x27;      //  single quote, for quoted chars.
817 static const UChar      chPound     = 0x23;      // '#', introduces a comment.
818 static const UChar      chBackSlash = 0x5c;      // '\'  introduces a char escape
819 static const UChar      chLParen    = 0x28;
820 static 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 //------------------------------------------------------------------------------
stripRules(const UnicodeString & rules)829 UnicodeString 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 //------------------------------------------------------------------------------
nextCharLL()854 UChar32  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 //------------------------------------------------------------------------------
nextChar(RBBIRuleChar & c)895 void 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 //------------------------------------------------------------------------------
parse()984 void 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
printNodeStack(const char * title)1152 void 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 //------------------------------------------------------------------------------
pushNewNode(RBBINode::NodeType t)1168 RBBINode  *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 //------------------------------------------------------------------------------
scanSet()1201 void 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 
1277 U_NAMESPACE_END
1278 
1279 #endif /* #if !UCONFIG_NO_BREAK_ITERATION */
1280