• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1 
2 //
3 //  file:  rbbiscan.cpp
4 //
5 //  Copyright (C) 2002-2010, International Business Machines Corporation and others.
6 //  All Rights Reserved.
7 //
8 //  This file contains the Rule Based Break Iterator Rule Builder functions for
9 //   scanning the rules and assembling a parse tree.  This is the first phase
10 //   of compiling the rules.
11 //
12 //  The overall of the rules is managed by class RBBIRuleBuilder, which will
13 //  create and use an instance of this class as part of the process.
14 //
15 
16 #include "unicode/utypes.h"
17 
18 #if !UCONFIG_NO_BREAK_ITERATION
19 
20 #include "unicode/unistr.h"
21 #include "unicode/uniset.h"
22 #include "unicode/uchar.h"
23 #include "unicode/uchriter.h"
24 #include "unicode/parsepos.h"
25 #include "unicode/parseerr.h"
26 #include "util.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 //
42 // Unicode Set init strings for each of the character classes needed for parsing a rule file.
43 //               (Initialized with hex values for portability to EBCDIC based machines.
44 //                Really ugly, but there's no good way to avoid it.)
45 //
46 //              The sets are referred to by name in the rbbirpt.txt, which is the
47 //              source form of the state transition table for the RBBI rule parser.
48 //
49 //------------------------------------------------------------------------------
50 static const UChar gRuleSet_rule_char_pattern[]       = {
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     U_NAMESPACE_QUALIFIER RBBISetTableEl *px = (U_NAMESPACE_QUALIFIER 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     fStackPtr           = 0;
94     fStack[fStackPtr]   = 0;
95     fNodeStackPtr       = 0;
96     fRuleNum            = 0;
97     fNodeStack[0]       = NULL;
98 
99     fSymbolTable                            = NULL;
100     fSetTable                               = NULL;
101 
102     fScanIndex = 0;
103     fNextIndex = 0;
104 
105     fReverseRule        = FALSE;
106     fLookAheadRule      = FALSE;
107 
108     fLineNum    = 1;
109     fCharNum    = 0;
110     fQuoteMode  = FALSE;
111 
112     // Do not check status until after all critical fields are sufficiently initialized
113     //   that the destructor can run cleanly.
114     if (U_FAILURE(*rb->fStatus)) {
115         return;
116     }
117 
118     //
119     //  Set up the constant Unicode Sets.
120     //     Note:  These could be made static, lazily initialized, and shared among
121     //            all instances of RBBIRuleScanners.  BUT this is quite a bit simpler,
122     //            and the time to build these few sets should be small compared to a
123     //            full break iterator build.
124     fRuleSets[kRuleSet_rule_char-128]       = UnicodeSet(gRuleSet_rule_char_pattern,       *rb->fStatus);
125     UnicodeSet *whitespaceSet = uprv_openRuleWhiteSpaceSet(rb->fStatus);
126     if (U_FAILURE(*rb->fStatus)) {
127         return;
128     }
129     fRuleSets[kRuleSet_white_space-128]     = *whitespaceSet;
130     delete whitespaceSet;
131     fRuleSets[kRuleSet_name_char-128]       = UnicodeSet(gRuleSet_name_char_pattern,       *rb->fStatus);
132     fRuleSets[kRuleSet_name_start_char-128] = UnicodeSet(gRuleSet_name_start_char_pattern, *rb->fStatus);
133     fRuleSets[kRuleSet_digit_char-128]      = UnicodeSet(gRuleSet_digit_char_pattern,      *rb->fStatus);
134     if (*rb->fStatus == U_ILLEGAL_ARGUMENT_ERROR) {
135         // This case happens if ICU's data is missing.  UnicodeSet tries to look up property
136         //   names from the init string, can't find them, and claims an illegal arguement.
137         //   Change the error so that the actual problem will be clearer to users.
138         *rb->fStatus = U_BRK_INIT_ERROR;
139     }
140     if (U_FAILURE(*rb->fStatus)) {
141         return;
142     }
143 
144     fSymbolTable = new RBBISymbolTable(this, rb->fRules, *rb->fStatus);
145     if (fSymbolTable == NULL) {
146         *rb->fStatus = U_MEMORY_ALLOCATION_ERROR;
147         return;
148     }
149     fSetTable    = uhash_open(uhash_hashUnicodeString, uhash_compareUnicodeString, NULL, rb->fStatus);
150     if (U_FAILURE(*rb->fStatus)) {
151         return;
152     }
153     uhash_setValueDeleter(fSetTable, RBBISetTable_deleter);
154 }
155 
156 
157 
158 //------------------------------------------------------------------------------
159 //
160 //  Destructor
161 //
162 //------------------------------------------------------------------------------
~RBBIRuleScanner()163 RBBIRuleScanner::~RBBIRuleScanner() {
164     delete fSymbolTable;
165     if (fSetTable != NULL) {
166          uhash_close(fSetTable);
167          fSetTable = NULL;
168 
169     }
170 
171 
172     // Node Stack.
173     //   Normally has one entry, which is the entire parse tree for the rules.
174     //   If errors occured, there may be additional subtrees left on the stack.
175     while (fNodeStackPtr > 0) {
176         delete fNodeStack[fNodeStackPtr];
177         fNodeStackPtr--;
178     }
179 
180 }
181 
182 //------------------------------------------------------------------------------
183 //
184 //  doParseAction        Do some action during rule parsing.
185 //                       Called by the parse state machine.
186 //                       Actions build the parse tree and Unicode Sets,
187 //                       and maintain the parse stack for nested expressions.
188 //
189 //                       TODO:  unify EParseAction and RBBI_RuleParseAction enum types.
190 //                              They represent exactly the same thing.  They're separate
191 //                              only to work around enum forward declaration restrictions
192 //                              in some compilers, while at the same time avoiding multiple
193 //                              definitions problems.  I'm sure that there's a better way.
194 //
195 //------------------------------------------------------------------------------
doParseActions(int32_t action)196 UBool RBBIRuleScanner::doParseActions(int32_t action)
197 {
198     RBBINode *n       = NULL;
199 
200     UBool   returnVal = TRUE;
201 
202     switch (action) {
203 
204     case doExprStart:
205         pushNewNode(RBBINode::opStart);
206         fRuleNum++;
207         break;
208 
209 
210     case doExprOrOperator:
211         {
212             fixOpStack(RBBINode::precOpCat);
213             RBBINode  *operandNode = fNodeStack[fNodeStackPtr--];
214             RBBINode  *orNode      = pushNewNode(RBBINode::opOr);
215             orNode->fLeftChild     = operandNode;
216             operandNode->fParent   = orNode;
217         }
218         break;
219 
220     case doExprCatOperator:
221         // concatenation operator.
222         // For the implicit concatenation of adjacent terms in an expression that are
223         //   not separated by any other operator.  Action is invoked between the
224         //   actions for the two terms.
225         {
226             fixOpStack(RBBINode::precOpCat);
227             RBBINode  *operandNode = fNodeStack[fNodeStackPtr--];
228             RBBINode  *catNode     = pushNewNode(RBBINode::opCat);
229             catNode->fLeftChild    = operandNode;
230             operandNode->fParent   = catNode;
231         }
232         break;
233 
234     case doLParen:
235         // Open Paren.
236         //   The openParen node is a dummy operation type with a low precedence,
237         //     which has the affect of ensuring that any real binary op that
238         //     follows within the parens binds more tightly to the operands than
239         //     stuff outside of the parens.
240         pushNewNode(RBBINode::opLParen);
241         break;
242 
243     case doExprRParen:
244         fixOpStack(RBBINode::precLParen);
245         break;
246 
247     case doNOP:
248         break;
249 
250     case doStartAssign:
251         // We've just scanned "$variable = "
252         // The top of the node stack has the $variable ref node.
253 
254         // Save the start position of the RHS text in the StartExpression node
255         //   that precedes the $variableReference node on the stack.
256         //   This will eventually be used when saving the full $variable replacement
257         //   text as a string.
258         n = fNodeStack[fNodeStackPtr-1];
259         n->fFirstPos = fNextIndex;              // move past the '='
260 
261         // Push a new start-of-expression node; needed to keep parse of the
262         //   RHS expression happy.
263         pushNewNode(RBBINode::opStart);
264         break;
265 
266 
267 
268 
269     case doEndAssign:
270         {
271             // We have reached the end of an assignement statement.
272             //   Current scan char is the ';' that terminates the assignment.
273 
274             // Terminate expression, leaves expression parse tree rooted in TOS node.
275             fixOpStack(RBBINode::precStart);
276 
277             RBBINode *startExprNode  = fNodeStack[fNodeStackPtr-2];
278             RBBINode *varRefNode     = fNodeStack[fNodeStackPtr-1];
279             RBBINode *RHSExprNode    = fNodeStack[fNodeStackPtr];
280 
281             // Save original text of right side of assignment, excluding the terminating ';'
282             //  in the root of the node for the right-hand-side expression.
283             RHSExprNode->fFirstPos = startExprNode->fFirstPos;
284             RHSExprNode->fLastPos  = fScanIndex;
285             fRB->fRules.extractBetween(RHSExprNode->fFirstPos, RHSExprNode->fLastPos, RHSExprNode->fText);
286 
287             // Expression parse tree becomes l. child of the $variable reference node.
288             varRefNode->fLeftChild = RHSExprNode;
289             RHSExprNode->fParent   = varRefNode;
290 
291             // Make a symbol table entry for the $variableRef node.
292             fSymbolTable->addEntry(varRefNode->fText, varRefNode, *fRB->fStatus);
293             if (U_FAILURE(*fRB->fStatus)) {
294                 // This is a round-about way to get the parse position set
295                 //  so that duplicate symbols error messages include a line number.
296                 UErrorCode t = *fRB->fStatus;
297                 *fRB->fStatus = U_ZERO_ERROR;
298                 error(t);
299             }
300 
301             // Clean up the stack.
302             delete startExprNode;
303             fNodeStackPtr-=3;
304             break;
305         }
306 
307     case doEndOfRule:
308         {
309         fixOpStack(RBBINode::precStart);      // Terminate expression, leaves expression
310         if (U_FAILURE(*fRB->fStatus)) {       //   parse tree rooted in TOS node.
311             break;
312         }
313 #ifdef RBBI_DEBUG
314         if (fRB->fDebugEnv && uprv_strstr(fRB->fDebugEnv, "rtree")) {printNodeStack("end of rule");}
315 #endif
316         U_ASSERT(fNodeStackPtr == 1);
317 
318         // If this rule includes a look-ahead '/', add a endMark node to the
319         //   expression tree.
320         if (fLookAheadRule) {
321             RBBINode  *thisRule       = fNodeStack[fNodeStackPtr];
322             RBBINode  *endNode        = pushNewNode(RBBINode::endMark);
323             RBBINode  *catNode        = pushNewNode(RBBINode::opCat);
324             fNodeStackPtr -= 2;
325             catNode->fLeftChild       = thisRule;
326             catNode->fRightChild      = endNode;
327             fNodeStack[fNodeStackPtr] = catNode;
328             endNode->fVal             = fRuleNum;
329             endNode->fLookAheadEnd    = TRUE;
330         }
331 
332         // All rule expressions are ORed together.
333         // The ';' that terminates an expression really just functions as a '|' with
334         //   a low operator prededence.
335         //
336         // Each of the four sets of rules are collected separately.
337         //  (forward, reverse, safe_forward, safe_reverse)
338         //  OR this rule into the appropriate group of them.
339         //
340         RBBINode **destRules = (fReverseRule? &fRB->fReverseTree : fRB->fDefaultTree);
341 
342         if (*destRules != NULL) {
343             // This is not the first rule encounted.
344             // OR previous stuff  (from *destRules)
345             // with the current rule expression (on the Node Stack)
346             //  with the resulting OR expression going to *destRules
347             //
348             RBBINode  *thisRule    = fNodeStack[fNodeStackPtr];
349             RBBINode  *prevRules   = *destRules;
350             RBBINode  *orNode      = pushNewNode(RBBINode::opOr);
351             orNode->fLeftChild     = prevRules;
352             prevRules->fParent     = orNode;
353             orNode->fRightChild    = thisRule;
354             thisRule->fParent      = orNode;
355             *destRules             = orNode;
356         }
357         else
358         {
359             // This is the first rule encountered (for this direction).
360             // Just move its parse tree from the stack to *destRules.
361             *destRules = fNodeStack[fNodeStackPtr];
362         }
363         fReverseRule   = FALSE;   // in preparation for the next rule.
364         fLookAheadRule = FALSE;
365         fNodeStackPtr  = 0;
366         }
367         break;
368 
369 
370     case doRuleError:
371         error(U_BRK_RULE_SYNTAX);
372         returnVal = FALSE;
373         break;
374 
375 
376     case doVariableNameExpectedErr:
377         error(U_BRK_RULE_SYNTAX);
378         break;
379 
380 
381     //
382     //  Unary operands  + ? *
383     //    These all appear after the operand to which they apply.
384     //    When we hit one, the operand (may be a whole sub expression)
385     //    will be on the top of the stack.
386     //    Unary Operator becomes TOS, with the old TOS as its one child.
387     case doUnaryOpPlus:
388         {
389             RBBINode  *operandNode = fNodeStack[fNodeStackPtr--];
390             RBBINode  *plusNode    = pushNewNode(RBBINode::opPlus);
391             plusNode->fLeftChild   = operandNode;
392             operandNode->fParent   = plusNode;
393         }
394         break;
395 
396     case doUnaryOpQuestion:
397         {
398             RBBINode  *operandNode = fNodeStack[fNodeStackPtr--];
399             RBBINode  *qNode       = pushNewNode(RBBINode::opQuestion);
400             qNode->fLeftChild      = operandNode;
401             operandNode->fParent   = qNode;
402         }
403         break;
404 
405     case doUnaryOpStar:
406         {
407             RBBINode  *operandNode = fNodeStack[fNodeStackPtr--];
408             RBBINode  *starNode    = pushNewNode(RBBINode::opStar);
409             starNode->fLeftChild   = operandNode;
410             operandNode->fParent   = starNode;
411         }
412         break;
413 
414     case doRuleChar:
415         // A "Rule Character" is any single character that is a literal part
416         // of the regular expression.  Like a, b and c in the expression "(abc*) | [:L:]"
417         // These are pretty uncommon in break rules; the terms are more commonly
418         //  sets.  To keep things uniform, treat these characters like as
419         // sets that just happen to contain only one character.
420         {
421             n = pushNewNode(RBBINode::setRef);
422             findSetFor(fC.fChar, n);
423             n->fFirstPos = fScanIndex;
424             n->fLastPos  = fNextIndex;
425             fRB->fRules.extractBetween(n->fFirstPos, n->fLastPos, n->fText);
426             break;
427         }
428 
429     case doDotAny:
430         // scanned a ".", meaning match any single character.
431         {
432             n = pushNewNode(RBBINode::setRef);
433             findSetFor(kAny, n);
434             n->fFirstPos = fScanIndex;
435             n->fLastPos  = fNextIndex;
436             fRB->fRules.extractBetween(n->fFirstPos, n->fLastPos, n->fText);
437             break;
438         }
439 
440     case doSlash:
441         // Scanned a '/', which identifies a look-ahead break position in a rule.
442         n = pushNewNode(RBBINode::lookAhead);
443         n->fVal      = fRuleNum;
444         n->fFirstPos = fScanIndex;
445         n->fLastPos  = fNextIndex;
446         fRB->fRules.extractBetween(n->fFirstPos, n->fLastPos, n->fText);
447         fLookAheadRule = TRUE;
448         break;
449 
450 
451     case doStartTagValue:
452         // Scanned a '{', the opening delimiter for a tag value within a rule.
453         n = pushNewNode(RBBINode::tag);
454         n->fVal      = 0;
455         n->fFirstPos = fScanIndex;
456         n->fLastPos  = fNextIndex;
457         break;
458 
459     case doTagDigit:
460         // Just scanned a decimal digit that's part of a tag value
461         {
462             n = fNodeStack[fNodeStackPtr];
463             uint32_t v = u_charDigitValue(fC.fChar);
464             U_ASSERT(v < 10);
465             n->fVal = n->fVal*10 + v;
466             break;
467         }
468 
469     case doTagValue:
470         n = fNodeStack[fNodeStackPtr];
471         n->fLastPos = fNextIndex;
472         fRB->fRules.extractBetween(n->fFirstPos, n->fLastPos, n->fText);
473         break;
474 
475     case doTagExpectedError:
476         error(U_BRK_MALFORMED_RULE_TAG);
477         returnVal = FALSE;
478         break;
479 
480     case doOptionStart:
481         // Scanning a !!option.   At the start of string.
482         fOptionStart = fScanIndex;
483         break;
484 
485     case doOptionEnd:
486         {
487             UnicodeString opt(fRB->fRules, fOptionStart, fScanIndex-fOptionStart);
488             if (opt == UNICODE_STRING("chain", 5)) {
489                 fRB->fChainRules = TRUE;
490             } else if (opt == UNICODE_STRING("LBCMNoChain", 11)) {
491                 fRB->fLBCMNoChain = TRUE;
492             } else if (opt == UNICODE_STRING("forward", 7)) {
493                 fRB->fDefaultTree   = &fRB->fForwardTree;
494             } else if (opt == UNICODE_STRING("reverse", 7)) {
495                 fRB->fDefaultTree   = &fRB->fReverseTree;
496             } else if (opt == UNICODE_STRING("safe_forward", 12)) {
497                 fRB->fDefaultTree   = &fRB->fSafeFwdTree;
498             } else if (opt == UNICODE_STRING("safe_reverse", 12)) {
499                 fRB->fDefaultTree   = &fRB->fSafeRevTree;
500             } else if (opt == UNICODE_STRING("lookAheadHardBreak", 18)) {
501                 fRB->fLookAheadHardBreak = TRUE;
502             } else {
503                 error(U_BRK_UNRECOGNIZED_OPTION);
504             }
505         }
506         break;
507 
508     case doReverseDir:
509         fReverseRule = TRUE;
510         break;
511 
512     case doStartVariableName:
513         n = pushNewNode(RBBINode::varRef);
514         if (U_FAILURE(*fRB->fStatus)) {
515             break;
516         }
517         n->fFirstPos = fScanIndex;
518         break;
519 
520     case doEndVariableName:
521         n = fNodeStack[fNodeStackPtr];
522         if (n==NULL || n->fType != RBBINode::varRef) {
523             error(U_BRK_INTERNAL_ERROR);
524             break;
525         }
526         n->fLastPos = fScanIndex;
527         fRB->fRules.extractBetween(n->fFirstPos+1, n->fLastPos, n->fText);
528         // Look the newly scanned name up in the symbol table
529         //   If there's an entry, set the l. child of the var ref to the replacement expression.
530         //   (We also pass through here when scanning assignments, but no harm is done, other
531         //    than a slight wasted effort that seems hard to avoid.  Lookup will be null)
532         n->fLeftChild = fSymbolTable->lookupNode(n->fText);
533         break;
534 
535     case doCheckVarDef:
536         n = fNodeStack[fNodeStackPtr];
537         if (n->fLeftChild == NULL) {
538             error(U_BRK_UNDEFINED_VARIABLE);
539             returnVal = FALSE;
540         }
541         break;
542 
543     case doExprFinished:
544         break;
545 
546     case doRuleErrorAssignExpr:
547         error(U_BRK_ASSIGN_ERROR);
548         returnVal = FALSE;
549         break;
550 
551     case doExit:
552         returnVal = FALSE;
553         break;
554 
555     case doScanUnicodeSet:
556         scanSet();
557         break;
558 
559     default:
560         error(U_BRK_INTERNAL_ERROR);
561         returnVal = FALSE;
562         break;
563     }
564     return returnVal;
565 }
566 
567 
568 
569 
570 //------------------------------------------------------------------------------
571 //
572 //  Error         Report a rule parse error.
573 //                Only report it if no previous error has been recorded.
574 //
575 //------------------------------------------------------------------------------
error(UErrorCode e)576 void RBBIRuleScanner::error(UErrorCode e) {
577     if (U_SUCCESS(*fRB->fStatus)) {
578         *fRB->fStatus = e;
579         if (fRB->fParseError) {
580             fRB->fParseError->line  = fLineNum;
581             fRB->fParseError->offset = fCharNum;
582             fRB->fParseError->preContext[0] = 0;
583             fRB->fParseError->preContext[0] = 0;
584         }
585     }
586 }
587 
588 
589 
590 
591 //------------------------------------------------------------------------------
592 //
593 //  fixOpStack   The parse stack holds partially assembled chunks of the parse tree.
594 //               An entry on the stack may be as small as a single setRef node,
595 //               or as large as the parse tree
596 //               for an entire expression (this will be the one item left on the stack
597 //               when the parsing of an RBBI rule completes.
598 //
599 //               This function is called when a binary operator is encountered.
600 //               It looks back up the stack for operators that are not yet associated
601 //               with a right operand, and if the precedence of the stacked operator >=
602 //               the precedence of the current operator, binds the operand left,
603 //               to the previously encountered operator.
604 //
605 //------------------------------------------------------------------------------
fixOpStack(RBBINode::OpPrecedence p)606 void RBBIRuleScanner::fixOpStack(RBBINode::OpPrecedence p) {
607     RBBINode *n;
608     // printNodeStack("entering fixOpStack()");
609     for (;;) {
610         n = fNodeStack[fNodeStackPtr-1];   // an operator node
611         if (n->fPrecedence == 0) {
612             RBBIDebugPuts("RBBIRuleScanner::fixOpStack, bad operator node");
613             error(U_BRK_INTERNAL_ERROR);
614             return;
615         }
616 
617         if (n->fPrecedence < p || n->fPrecedence <= RBBINode::precLParen) {
618             // The most recent operand goes with the current operator,
619             //   not with the previously stacked one.
620             break;
621         }
622             // Stack operator is a binary op  ( '|' or concatenation)
623             //   TOS operand becomes right child of this operator.
624             //   Resulting subexpression becomes the TOS operand.
625             n->fRightChild = fNodeStack[fNodeStackPtr];
626             fNodeStack[fNodeStackPtr]->fParent = n;
627             fNodeStackPtr--;
628         // printNodeStack("looping in fixOpStack()   ");
629     }
630 
631     if (p <= RBBINode::precLParen) {
632         // Scan is at a right paren or end of expression.
633         //  The scanned item must match the stack, or else there was an error.
634         //  Discard the left paren (or start expr) node from the stack,
635             //  leaving the completed (sub)expression as TOS.
636             if (n->fPrecedence != p) {
637                 // Right paren encountered matched start of expression node, or
638                 // end of expression matched with a left paren node.
639                 error(U_BRK_MISMATCHED_PAREN);
640             }
641             fNodeStack[fNodeStackPtr-1] = fNodeStack[fNodeStackPtr];
642             fNodeStackPtr--;
643             // Delete the now-discarded LParen or Start node.
644             delete n;
645     }
646     // printNodeStack("leaving fixOpStack()");
647 }
648 
649 
650 
651 
652 //------------------------------------------------------------------------------
653 //
654 //   findSetFor    given a UnicodeString,
655 //                  - find the corresponding Unicode Set  (uset node)
656 //                         (create one if necessary)
657 //                  - Set fLeftChild of the caller's node (should be a setRef node)
658 //                         to the uset node
659 //                 Maintain a hash table of uset nodes, so the same one is always used
660 //                    for the same string.
661 //                 If a "to adopt" set is provided and we haven't seen this key before,
662 //                    add the provided set to the hash table.
663 //                 If the string is one (32 bit) char in length, the set contains
664 //                    just one element which is the char in question.
665 //                 If the string is "any", return a set containing all chars.
666 //
667 //------------------------------------------------------------------------------
findSetFor(const UnicodeString & s,RBBINode * node,UnicodeSet * setToAdopt)668 void RBBIRuleScanner::findSetFor(const UnicodeString &s, RBBINode *node, UnicodeSet *setToAdopt) {
669 
670     RBBISetTableEl   *el;
671 
672     // First check whether we've already cached a set for this string.
673     // If so, just use the cached set in the new node.
674     //   delete any set provided by the caller, since we own it.
675     el = (RBBISetTableEl *)uhash_get(fSetTable, &s);
676     if (el != NULL) {
677         delete setToAdopt;
678         node->fLeftChild = el->val;
679         U_ASSERT(node->fLeftChild->fType == RBBINode::uset);
680         return;
681     }
682 
683     // Haven't seen this set before.
684     // If the caller didn't provide us with a prebuilt set,
685     //   create a new UnicodeSet now.
686     if (setToAdopt == NULL) {
687         if (s.compare(kAny, -1) == 0) {
688             setToAdopt = new UnicodeSet(0x000000, 0x10ffff);
689         } else {
690             UChar32 c;
691             c = s.char32At(0);
692             setToAdopt = new UnicodeSet(c, c);
693         }
694     }
695 
696     //
697     // Make a new uset node to refer to this UnicodeSet
698     // This new uset node becomes the child of the caller's setReference node.
699     //
700     RBBINode *usetNode    = new RBBINode(RBBINode::uset);
701     if (usetNode == NULL) {
702         error(U_MEMORY_ALLOCATION_ERROR);
703         return;
704     }
705     usetNode->fInputSet   = setToAdopt;
706     usetNode->fParent     = node;
707     node->fLeftChild      = usetNode;
708     usetNode->fText = s;
709 
710 
711     //
712     // Add the new uset node to the list of all uset nodes.
713     //
714     fRB->fUSetNodes->addElement(usetNode, *fRB->fStatus);
715 
716 
717     //
718     // Add the new set to the set hash table.
719     //
720     el      = (RBBISetTableEl *)uprv_malloc(sizeof(RBBISetTableEl));
721     UnicodeString *tkey = new UnicodeString(s);
722     if (tkey == NULL || el == NULL || setToAdopt == NULL) {
723         // Delete to avoid memory leak
724         delete tkey;
725         tkey = NULL;
726         uprv_free(el);
727         el = NULL;
728         delete setToAdopt;
729         setToAdopt = NULL;
730 
731         error(U_MEMORY_ALLOCATION_ERROR);
732         return;
733     }
734     el->key = tkey;
735     el->val = usetNode;
736     uhash_put(fSetTable, el->key, el, fRB->fStatus);
737 
738     return;
739 }
740 
741 
742 
743 //
744 //  Assorted Unicode character constants.
745 //     Numeric because there is no portable way to enter them as literals.
746 //     (Think EBCDIC).
747 //
748 static const UChar      chCR        = 0x0d;      // New lines, for terminating comments.
749 static const UChar      chLF        = 0x0a;
750 static const UChar      chNEL       = 0x85;      //    NEL newline variant
751 static const UChar      chLS        = 0x2028;    //    Unicode Line Separator
752 static const UChar      chApos      = 0x27;      //  single quote, for quoted chars.
753 static const UChar      chPound     = 0x23;      // '#', introduces a comment.
754 static const UChar      chBackSlash = 0x5c;      // '\'  introduces a char escape
755 static const UChar      chLParen    = 0x28;
756 static const UChar      chRParen    = 0x29;
757 
758 
759 //------------------------------------------------------------------------------
760 //
761 //  stripRules    Return a rules string without unnecessary
762 //                characters.
763 //
764 //------------------------------------------------------------------------------
stripRules(const UnicodeString & rules)765 UnicodeString RBBIRuleScanner::stripRules(const UnicodeString &rules) {
766     UnicodeString strippedRules;
767     int rulesLength = rules.length();
768     for (int idx = 0; idx < rulesLength; ) {
769         UChar ch = rules[idx++];
770         if (ch == chPound) {
771             while (idx < rulesLength
772                 && ch != chCR && ch != chLF && ch != chNEL)
773             {
774                 ch = rules[idx++];
775             }
776         }
777         if (!u_isISOControl(ch)) {
778             strippedRules.append(ch);
779         }
780     }
781     // strippedRules = strippedRules.unescape();
782     return strippedRules;
783 }
784 
785 
786 //------------------------------------------------------------------------------
787 //
788 //  nextCharLL    Low Level Next Char from rule input source.
789 //                Get a char from the input character iterator,
790 //                keep track of input position for error reporting.
791 //
792 //------------------------------------------------------------------------------
nextCharLL()793 UChar32  RBBIRuleScanner::nextCharLL() {
794     UChar32  ch;
795 
796     if (fNextIndex >= fRB->fRules.length()) {
797         return (UChar32)-1;
798     }
799     ch         = fRB->fRules.char32At(fNextIndex);
800     fNextIndex = fRB->fRules.moveIndex32(fNextIndex, 1);
801 
802     if (ch == chCR ||
803         ch == chNEL ||
804         ch == chLS   ||
805         (ch == chLF && fLastChar != chCR)) {
806         // Character is starting a new line.  Bump up the line number, and
807         //  reset the column to 0.
808         fLineNum++;
809         fCharNum=0;
810         if (fQuoteMode) {
811             error(U_BRK_NEW_LINE_IN_QUOTED_STRING);
812             fQuoteMode = FALSE;
813         }
814     }
815     else {
816         // Character is not starting a new line.  Except in the case of a
817         //   LF following a CR, increment the column position.
818         if (ch != chLF) {
819             fCharNum++;
820         }
821     }
822     fLastChar = ch;
823     return ch;
824 }
825 
826 
827 //------------------------------------------------------------------------------
828 //
829 //   nextChar     for rules scanning.  At this level, we handle stripping
830 //                out comments and processing backslash character escapes.
831 //                The rest of the rules grammar is handled at the next level up.
832 //
833 //------------------------------------------------------------------------------
nextChar(RBBIRuleChar & c)834 void RBBIRuleScanner::nextChar(RBBIRuleChar &c) {
835 
836     // Unicode Character constants needed for the processing done by nextChar(),
837     //   in hex because literals wont work on EBCDIC machines.
838 
839     fScanIndex = fNextIndex;
840     c.fChar    = nextCharLL();
841     c.fEscaped = FALSE;
842 
843     //
844     //  check for '' sequence.
845     //  These are recognized in all contexts, whether in quoted text or not.
846     //
847     if (c.fChar == chApos) {
848         if (fRB->fRules.char32At(fNextIndex) == chApos) {
849             c.fChar    = nextCharLL();        // get nextChar officially so character counts
850             c.fEscaped = TRUE;                //   stay correct.
851         }
852         else
853         {
854             // Single quote, by itself.
855             //   Toggle quoting mode.
856             //   Return either '('  or ')', because quotes cause a grouping of the quoted text.
857             fQuoteMode = !fQuoteMode;
858             if (fQuoteMode == TRUE) {
859                 c.fChar = chLParen;
860             } else {
861                 c.fChar = chRParen;
862             }
863             c.fEscaped = FALSE;      // The paren that we return is not escaped.
864             return;
865         }
866     }
867 
868     if (fQuoteMode) {
869         c.fEscaped = TRUE;
870     }
871     else
872     {
873         // We are not in a 'quoted region' of the source.
874         //
875         if (c.fChar == chPound) {
876             // Start of a comment.  Consume the rest of it.
877             //  The new-line char that terminates the comment is always returned.
878             //  It will be treated as white-space, and serves to break up anything
879             //    that might otherwise incorrectly clump together with a comment in
880             //    the middle (a variable name, for example.)
881             for (;;) {
882                 c.fChar = nextCharLL();
883                 if (c.fChar == (UChar32)-1 ||  // EOF
884                     c.fChar == chCR     ||
885                     c.fChar == chLF     ||
886                     c.fChar == chNEL    ||
887                     c.fChar == chLS)       {break;}
888             }
889         }
890         if (c.fChar == (UChar32)-1) {
891             return;
892         }
893 
894         //
895         //  check for backslash escaped characters.
896         //  Use UnicodeString::unescapeAt() to handle them.
897         //
898         if (c.fChar == chBackSlash) {
899             c.fEscaped = TRUE;
900             int32_t startX = fNextIndex;
901             c.fChar = fRB->fRules.unescapeAt(fNextIndex);
902             if (fNextIndex == startX) {
903                 error(U_BRK_HEX_DIGITS_EXPECTED);
904             }
905             fCharNum += fNextIndex-startX;
906         }
907     }
908     // putc(c.fChar, stdout);
909 }
910 
911 //------------------------------------------------------------------------------
912 //
913 //  Parse RBBI rules.   The state machine for rules parsing is here.
914 //                      The state tables are hand-written in the file rbbirpt.txt,
915 //                      and converted to the form used here by a perl
916 //                      script rbbicst.pl
917 //
918 //------------------------------------------------------------------------------
parse()919 void RBBIRuleScanner::parse() {
920     uint16_t                state;
921     const RBBIRuleTableEl  *tableEl;
922 
923     if (U_FAILURE(*fRB->fStatus)) {
924         return;
925     }
926 
927     state = 1;
928     nextChar(fC);
929     //
930     // Main loop for the rule parsing state machine.
931     //   Runs once per state transition.
932     //   Each time through optionally performs, depending on the state table,
933     //      - an advance to the the next input char
934     //      - an action to be performed.
935     //      - pushing or popping a state to/from the local state return stack.
936     //
937     for (;;) {
938         //  Bail out if anything has gone wrong.
939         //  RBBI rule file parsing stops on the first error encountered.
940         if (U_FAILURE(*fRB->fStatus)) {
941             break;
942         }
943 
944         // Quit if state == 0.  This is the normal way to exit the state machine.
945         //
946         if (state == 0) {
947             break;
948         }
949 
950         // Find the state table element that matches the input char from the rule, or the
951         //    class of the input character.  Start with the first table row for this
952         //    state, then linearly scan forward until we find a row that matches the
953         //    character.  The last row for each state always matches all characters, so
954         //    the search will stop there, if not before.
955         //
956         tableEl = &gRuleParseStateTable[state];
957         #ifdef RBBI_DEBUG
958             if (fRB->fDebugEnv && uprv_strstr(fRB->fDebugEnv, "scan")) {
959                 RBBIDebugPrintf("char, line, col = (\'%c\', %d, %d)    state=%s ",
960                     fC.fChar, fLineNum, fCharNum, RBBIRuleStateNames[state]);
961             }
962         #endif
963 
964         for (;;) {
965             #ifdef RBBI_DEBUG
966                 if (fRB->fDebugEnv && uprv_strstr(fRB->fDebugEnv, "scan")) { RBBIDebugPrintf(".");}
967             #endif
968             if (tableEl->fCharClass < 127 && fC.fEscaped == FALSE &&   tableEl->fCharClass == fC.fChar) {
969                 // Table row specified an individual character, not a set, and
970                 //   the input character is not escaped, and
971                 //   the input character matched it.
972                 break;
973             }
974             if (tableEl->fCharClass == 255) {
975                 // Table row specified default, match anything character class.
976                 break;
977             }
978             if (tableEl->fCharClass == 254 && fC.fEscaped)  {
979                 // Table row specified "escaped" and the char was escaped.
980                 break;
981             }
982             if (tableEl->fCharClass == 253 && fC.fEscaped &&
983                 (fC.fChar == 0x50 || fC.fChar == 0x70 ))  {
984                 // Table row specified "escaped P" and the char is either 'p' or 'P'.
985                 break;
986             }
987             if (tableEl->fCharClass == 252 && fC.fChar == (UChar32)-1)  {
988                 // Table row specified eof and we hit eof on the input.
989                 break;
990             }
991 
992             if (tableEl->fCharClass >= 128 && tableEl->fCharClass < 240 &&   // Table specs a char class &&
993                 fC.fEscaped == FALSE &&                                      //   char is not escaped &&
994                 fC.fChar != (UChar32)-1) {                                   //   char is not EOF
995                 if (fRuleSets[tableEl->fCharClass-128].contains(fC.fChar)) {
996                     // Table row specified a character class, or set of characters,
997                     //   and the current char matches it.
998                     break;
999                 }
1000             }
1001 
1002             // No match on this row, advance to the next  row for this state,
1003             tableEl++;
1004         }
1005         if (fRB->fDebugEnv && uprv_strstr(fRB->fDebugEnv, "scan")) { RBBIDebugPuts("");}
1006 
1007         //
1008         // We've found the row of the state table that matches the current input
1009         //   character from the rules string.
1010         // Perform any action specified  by this row in the state table.
1011         if (doParseActions((int32_t)tableEl->fAction) == FALSE) {
1012             // Break out of the state machine loop if the
1013             //   the action signalled some kind of error, or
1014             //   the action was to exit, occurs on normal end-of-rules-input.
1015             break;
1016         }
1017 
1018         if (tableEl->fPushState != 0) {
1019             fStackPtr++;
1020             if (fStackPtr >= kStackSize) {
1021                 error(U_BRK_INTERNAL_ERROR);
1022                 RBBIDebugPuts("RBBIRuleScanner::parse() - state stack overflow.");
1023                 fStackPtr--;
1024             }
1025             fStack[fStackPtr] = tableEl->fPushState;
1026         }
1027 
1028         if (tableEl->fNextChar) {
1029             nextChar(fC);
1030         }
1031 
1032         // Get the next state from the table entry, or from the
1033         //   state stack if the next state was specified as "pop".
1034         if (tableEl->fNextState != 255) {
1035             state = tableEl->fNextState;
1036         } else {
1037             state = fStack[fStackPtr];
1038             fStackPtr--;
1039             if (fStackPtr < 0) {
1040                 error(U_BRK_INTERNAL_ERROR);
1041                 RBBIDebugPuts("RBBIRuleScanner::parse() - state stack underflow.");
1042                 fStackPtr++;
1043             }
1044         }
1045 
1046     }
1047 
1048     //
1049     // If there were NO user specified reverse rules, set up the equivalent of ".*;"
1050     //
1051     if (fRB->fReverseTree == NULL) {
1052         fRB->fReverseTree  = pushNewNode(RBBINode::opStar);
1053         RBBINode  *operand = pushNewNode(RBBINode::setRef);
1054         findSetFor(kAny, operand);
1055         fRB->fReverseTree->fLeftChild = operand;
1056         operand->fParent              = fRB->fReverseTree;
1057         fNodeStackPtr -= 2;
1058     }
1059 
1060 
1061     //
1062     // Parsing of the input RBBI rules is complete.
1063     // We now have a parse tree for the rule expressions
1064     // and a list of all UnicodeSets that are referenced.
1065     //
1066 #ifdef RBBI_DEBUG
1067     if (fRB->fDebugEnv && uprv_strstr(fRB->fDebugEnv, "symbols")) {fSymbolTable->rbbiSymtablePrint();}
1068     if (fRB->fDebugEnv && uprv_strstr(fRB->fDebugEnv, "ptree"))
1069     {
1070         RBBIDebugPrintf("Completed Forward Rules Parse Tree...\n");
1071         fRB->fForwardTree->printTree(TRUE);
1072         RBBIDebugPrintf("\nCompleted Reverse Rules Parse Tree...\n");
1073         fRB->fReverseTree->printTree(TRUE);
1074         RBBIDebugPrintf("\nCompleted Safe Point Forward Rules Parse Tree...\n");
1075         fRB->fSafeFwdTree->printTree(TRUE);
1076         RBBIDebugPrintf("\nCompleted Safe Point Reverse Rules Parse Tree...\n");
1077         fRB->fSafeRevTree->printTree(TRUE);
1078     }
1079 #endif
1080 }
1081 
1082 
1083 //------------------------------------------------------------------------------
1084 //
1085 //  printNodeStack     for debugging...
1086 //
1087 //------------------------------------------------------------------------------
1088 #ifdef RBBI_DEBUG
printNodeStack(const char * title)1089 void RBBIRuleScanner::printNodeStack(const char *title) {
1090     int i;
1091     RBBIDebugPrintf("%s.  Dumping node stack...\n", title);
1092     for (i=fNodeStackPtr; i>0; i--) {fNodeStack[i]->printTree(TRUE);}
1093 }
1094 #endif
1095 
1096 
1097 
1098 
1099 //------------------------------------------------------------------------------
1100 //
1101 //  pushNewNode   create a new RBBINode of the specified type and push it
1102 //                onto the stack of nodes.
1103 //
1104 //------------------------------------------------------------------------------
pushNewNode(RBBINode::NodeType t)1105 RBBINode  *RBBIRuleScanner::pushNewNode(RBBINode::NodeType  t) {
1106     fNodeStackPtr++;
1107     if (fNodeStackPtr >= kStackSize) {
1108         error(U_BRK_INTERNAL_ERROR);
1109         RBBIDebugPuts("RBBIRuleScanner::pushNewNode - stack overflow.");
1110         *fRB->fStatus = U_BRK_INTERNAL_ERROR;
1111         return NULL;
1112     }
1113     fNodeStack[fNodeStackPtr] = new RBBINode(t);
1114     if (fNodeStack[fNodeStackPtr] == NULL) {
1115         *fRB->fStatus = U_MEMORY_ALLOCATION_ERROR;
1116     }
1117     return fNodeStack[fNodeStackPtr];
1118 }
1119 
1120 
1121 
1122 //------------------------------------------------------------------------------
1123 //
1124 //  scanSet    Construct a UnicodeSet from the text at the current scan
1125 //             position.  Advance the scan position to the first character
1126 //             after the set.
1127 //
1128 //             A new RBBI setref node referring to the set is pushed onto the node
1129 //             stack.
1130 //
1131 //             The scan position is normally under the control of the state machine
1132 //             that controls rule parsing.  UnicodeSets, however, are parsed by
1133 //             the UnicodeSet constructor, not by the RBBI rule parser.
1134 //
1135 //------------------------------------------------------------------------------
scanSet()1136 void RBBIRuleScanner::scanSet() {
1137     UnicodeSet    *uset;
1138     ParsePosition  pos;
1139     int            startPos;
1140     int            i;
1141 
1142     if (U_FAILURE(*fRB->fStatus)) {
1143         return;
1144     }
1145 
1146     pos.setIndex(fScanIndex);
1147     startPos = fScanIndex;
1148     UErrorCode localStatus = U_ZERO_ERROR;
1149     uset = new UnicodeSet(fRB->fRules, pos, USET_IGNORE_SPACE,
1150                          fSymbolTable,
1151                          localStatus);
1152     if (uset == NULL) {
1153         localStatus = U_MEMORY_ALLOCATION_ERROR;
1154     }
1155     if (U_FAILURE(localStatus)) {
1156         //  TODO:  Get more accurate position of the error from UnicodeSet's return info.
1157         //         UnicodeSet appears to not be reporting correctly at this time.
1158         #ifdef RBBI_DEBUG
1159             RBBIDebugPrintf("UnicodeSet parse postion.ErrorIndex = %d\n", pos.getIndex());
1160         #endif
1161         error(localStatus);
1162         delete uset;
1163         return;
1164     }
1165 
1166     // Verify that the set contains at least one code point.
1167     //
1168     if (uset->isEmpty()) {
1169         // This set is empty.
1170         //  Make it an error, because it almost certainly is not what the user wanted.
1171         //  Also, avoids having to think about corner cases in the tree manipulation code
1172         //   that occurs later on.
1173         error(U_BRK_RULE_EMPTY_SET);
1174         delete uset;
1175         return;
1176     }
1177 
1178 
1179     // Advance the RBBI parse postion over the UnicodeSet pattern.
1180     //   Don't just set fScanIndex because the line/char positions maintained
1181     //   for error reporting would be thrown off.
1182     i = pos.getIndex();
1183     for (;;) {
1184         if (fNextIndex >= i) {
1185             break;
1186         }
1187         nextCharLL();
1188     }
1189 
1190     if (U_SUCCESS(*fRB->fStatus)) {
1191         RBBINode         *n;
1192 
1193         n = pushNewNode(RBBINode::setRef);
1194         n->fFirstPos = startPos;
1195         n->fLastPos  = fNextIndex;
1196         fRB->fRules.extractBetween(n->fFirstPos, n->fLastPos, n->fText);
1197         //  findSetFor() serves several purposes here:
1198         //     - Adopts storage for the UnicodeSet, will be responsible for deleting.
1199         //     - Mantains collection of all sets in use, needed later for establishing
1200         //          character categories for run time engine.
1201         //     - Eliminates mulitiple instances of the same set.
1202         //     - Creates a new uset node if necessary (if this isn't a duplicate.)
1203         findSetFor(n->fText, n, uset);
1204     }
1205 
1206 }
1207 
1208 U_NAMESPACE_END
1209 
1210 #endif /* #if !UCONFIG_NO_BREAK_ITERATION */
1211