src/theory/strings/theory_strings.h

   1 /*********************                                                        */
   2 /*! \file theory_strings.h
   3  ** \verbatim
   4  ** Original author: Tianyi Liang
   5  ** Major contributors: Andrew Reynolds
   6  ** Minor contributors (to current version): Martin Brain <>, Morgan Deters
   7  ** This file is part of the CVC4 project.
   8  ** Copyright (c) 2009-2014  New York University and The University of Iowa
   9  ** See the file COPYING in the top-level source directory for licensing
  10  ** information.\endverbatim
  11  **
  12  ** \brief Theory of strings
  13  **
  14  ** Theory of strings.
  15  **/
  16
  17 #include "cvc4_private.h"
  18
  19 #ifndef __CVC4__THEORY__STRINGS__THEORY_STRINGS_H
  20 #define __CVC4__THEORY__STRINGS__THEORY_STRINGS_H
  21
  22 #include "theory/theory.h"
  23 #include "theory/uf/equality_engine.h"
  24 #include "theory/strings/theory_strings_preprocess.h"
  25 #include "theory/strings/regexp_operation.h"
  26
  27 #include "context/cdchunk_list.h"
  28 #include "context/cdhashset.h"
  29
  30 #include <climits>
  31
  32 namespace CVC4 {
  33 namespace theory {
  34 namespace strings {
  35
  36 /**
  37  * Decision procedure for strings.
  38  *
  39  */
  40
  41 class TheoryStrings : public Theory {
  42   typedef context::CDChunkList<Node> NodeList;
  43   typedef context::CDHashMap<Node, NodeList*, NodeHashFunction> NodeListMap;
  44   typedef context::CDHashMap<Node, bool, NodeHashFunction> NodeBoolMap;
  45   typedef context::CDHashMap<Node, int, NodeHashFunction> NodeIntMap;
  46   typedef context::CDHashMap<Node, Node, NodeHashFunction> NodeNodeMap;
  47   typedef context::CDHashSet<Node, NodeHashFunction> NodeSet;
  48
  49 public:
  50   TheoryStrings(context::Context* c, context::UserContext* u, OutputChannel& out, Valuation valuation, const LogicInfo& logicInfo);
  51   ~TheoryStrings();
  52
  53   void setMasterEqualityEngine(eq::EqualityEngine* eq);
  54
  55   std::string identify() const { return std::string("TheoryStrings"); }
  56
  57 public:
  58   void propagate(Effort e);
  59   bool propagate(TNode literal);
  60   void explain( TNode literal, std::vector<TNode>& assumptions );
  61   Node explain( TNode literal );
  62
  63
  64   // NotifyClass for equality engine
  65   class NotifyClass : public eq::EqualityEngineNotify {
  66     TheoryStrings& d_str;
  67   public:
  68     NotifyClass(TheoryStrings& t_str): d_str(t_str) {}
  69     bool eqNotifyTriggerEquality(TNode equality, bool value) {
  70       Debug("strings") << "NotifyClass::eqNotifyTriggerEquality(" << equality << ", " << (value ? "true" : "false" )<< ")" << std::endl;
  71       if (value) {
  72         return d_str.propagate(equality);
  73       } else {
  74         // We use only literal triggers so taking not is safe
  75         return d_str.propagate(equality.notNode());
  76       }
  77     }
  78     bool eqNotifyTriggerPredicate(TNode predicate, bool value) {
  79       Debug("strings") << "NotifyClass::eqNotifyTriggerPredicate(" << predicate << ", " << (value ? "true" : "false") << ")" << std::endl;
  80       if (value) {
  81         return d_str.propagate(predicate);
  82       } else {
  83          return d_str.propagate(predicate.notNode());
  84       }
  85     }
  86     bool eqNotifyTriggerTermEquality(TheoryId tag, TNode t1, TNode t2, bool value) {
  87       Debug("strings") << "NotifyClass::eqNotifyTriggerTermMerge(" << tag << ", " << t1 << ", " << t2 << ")" << std::endl;
  88       if (value) {
  89       return d_str.propagate(t1.eqNode(t2));
  90       } else {
  91       return d_str.propagate(t1.eqNode(t2).notNode());
  92       }
  93     }
  94     void eqNotifyConstantTermMerge(TNode t1, TNode t2) {
  95       Debug("strings") << "NotifyClass::eqNotifyConstantTermMerge(" << t1 << ", " << t2 << ")" << std::endl;
  96       d_str.conflict(t1, t2);
  97     }
  98     void eqNotifyNewClass(TNode t) {
  99       Debug("strings") << "NotifyClass::eqNotifyNewClass(" << t << std::endl;
 100       d_str.eqNotifyNewClass(t);
 101     }
 102     void eqNotifyPreMerge(TNode t1, TNode t2) {
 103       Debug("strings") << "NotifyClass::eqNotifyPreMerge(" << t1 << ", " << t2 << std::endl;
 104       d_str.eqNotifyPreMerge(t1, t2);
 105     }
 106     void eqNotifyPostMerge(TNode t1, TNode t2) {
 107       Debug("strings") << "NotifyClass::eqNotifyPostMerge(" << t1 << ", " << t2 << std::endl;
 108       d_str.eqNotifyPostMerge(t1, t2);
 109     }
 110     void eqNotifyDisequal(TNode t1, TNode t2, TNode reason) {
 111       Debug("strings") << "NotifyClass::eqNotifyDisequal(" << t1 << ", " << t2 << ", " << reason << std::endl;
 112       d_str.eqNotifyDisequal(t1, t2, reason);
 113     }
 114   };/* class TheoryStrings::NotifyClass */
 115
 116 private:
 117   /**
 118    * Function symbol used to implement uninterpreted undefined string
 119    * semantics.  Needed to deal with partial charat/substr function.
 120    */
 121   Node d_ufSubstr;
 122
 123   // Constants
 124   Node d_emptyString;
 125   Node d_emptyRegexp;
 126   Node d_true;
 127   Node d_false;
 128   Node d_zero;
 129   Node d_one;
 130   CVC4::Rational RMAXINT;
 131   unsigned d_card_size;
 132   // Options
 133   bool d_opt_fmf;
 134   bool d_opt_regexp_gcd;
 135   // Helper functions
 136   Node getRepresentative( Node t );
 137   bool hasTerm( Node a );
 138   bool areEqual( Node a, Node b );
 139   bool areDisequal( Node a, Node b );
 140   Node getLengthTerm( Node t );
 141   Node getLength( Node t );
 142
 143 private:
 144   /** The notify class */
 145   NotifyClass d_notify;
 146   /** Equaltity engine */
 147   eq::EqualityEngine d_equalityEngine;
 148   /** Are we in conflict */
 149   context::CDO<bool> d_conflict;
 150   //list of pairs of nodes to merge
 151   std::map< Node, Node > d_pending_exp;
 152   std::vector< Node > d_pending;
 153   std::vector< Node > d_lemma_cache;
 154   std::map< Node, bool > d_pending_req_phase;
 155   /** inferences */
 156   NodeList d_infer;
 157   NodeList d_infer_exp;
 158   /** normal forms */
 159   std::map< Node, Node > d_normal_forms_base;
 160   std::map< Node, std::vector< Node > > d_normal_forms;
 161   std::map< Node, std::vector< Node > > d_normal_forms_exp;
 162   //map of pairs of terms that have the same normal form
 163   NodeListMap d_nf_pairs;
 164   void addNormalFormPair( Node n1, Node n2 );
 165   bool isNormalFormPair( Node n1, Node n2 );
 166   bool isNormalFormPair2( Node n1, Node n2 );
 167   // loop ant
 168   NodeSet d_loop_antec;
 169   NodeSet d_length_intro_vars;
 170   // preReg cache
 171   NodeSet d_registed_terms_cache;
 172   // term cache
 173   std::vector< Node > d_terms_cache;
 174   void collectTerm( Node n );
 175   void appendTermLemma();
 176
 177   /////////////////////////////////////////////////////////////////////////////
 178   // MODEL GENERATION
 179   /////////////////////////////////////////////////////////////////////////////
 180 public:
 181   void collectModelInfo(TheoryModel* m, bool fullModel);
 182
 183   /////////////////////////////////////////////////////////////////////////////
 184   // NOTIFICATIONS
 185   /////////////////////////////////////////////////////////////////////////////
 186 public:
 187   void presolve();
 188   void shutdown() { }
 189
 190   /////////////////////////////////////////////////////////////////////////////
 191   // MAIN SOLVER
 192   /////////////////////////////////////////////////////////////////////////////
 193 private:
 194   void addSharedTerm(TNode n);
 195   EqualityStatus getEqualityStatus(TNode a, TNode b);
 196
 197 private:
 198   class EqcInfo {
 199   public:
 200     EqcInfo( context::Context* c );
 201     ~EqcInfo(){}
 202     //constant in this eqc
 203     context::CDO< Node > d_const_term;
 204     context::CDO< Node > d_length_term;
 205     context::CDO< unsigned > d_cardinality_lem_k;
 206     // 1 = added length lemma
 207     context::CDO< Node > d_normalized_length;
 208   };
 209   /** map from representatives to information necessary for equivalence classes */
 210   std::map< Node, EqcInfo* > d_eqc_info;
 211   EqcInfo * getOrMakeEqcInfo( Node eqc, bool doMake = true );
 212   //maintain which concat terms have the length lemma instantiated
 213   NodeNodeMap d_length_inst;
 214 private:
 215   void mergeCstVec(std::vector< Node > &vec_strings);
 216   bool getNormalForms(Node &eqc, std::vector< Node > & visited, std::vector< Node > & nf,
 217         std::vector< std::vector< Node > > &normal_forms,
 218         std::vector< std::vector< Node > > &normal_forms_exp,
 219         std::vector< Node > &normal_form_src);
 220   bool detectLoop(std::vector< std::vector< Node > > &normal_forms,
 221         int i, int j, int index_i, int index_j,
 222         int &loop_in_i, int &loop_in_j);
 223   bool processLoop(std::vector< Node > &antec,
 224         std::vector< std::vector< Node > > &normal_forms,
 225         std::vector< Node > &normal_form_src,
 226         int i, int j, int loop_n_index, int other_n_index,
 227         int loop_index, int index, int other_index);
 228   bool processNEqc(std::vector< std::vector< Node > > &normal_forms,
 229         std::vector< std::vector< Node > > &normal_forms_exp,
 230         std::vector< Node > &normal_form_src);
 231   bool processReverseNEq(std::vector< std::vector< Node > > &normal_forms,
 232         std::vector< Node > &normal_form_src, std::vector< Node > &curr_exp, unsigned i, unsigned j );
 233   bool processSimpleNEq( std::vector< std::vector< Node > > &normal_forms,
 234         std::vector< Node > &normal_form_src, std::vector< Node > &curr_exp, unsigned i, unsigned j,
 235         unsigned& index_i, unsigned& index_j, bool isRev );
 236   bool normalizeEquivalenceClass( Node n, std::vector< Node > & visited, std::vector< Node > & nf, std::vector< Node > & nf_exp );
 237   bool processDeq( Node n1, Node n2 );
 238   int processReverseDeq( std::vector< Node >& nfi, std::vector< Node >& nfj, Node ni, Node nj );
 239   int processSimpleDeq( std::vector< Node >& nfi, std::vector< Node >& nfj, Node ni, Node nj, unsigned& index, bool isRev );
 240   //bool unrollStar( Node atom );
 241   Node mkRegExpAntec(Node atom, Node ant);
 242
 243   //bool checkSimple();
 244   bool checkNormalForms();
 245   void checkDeqNF();
 246   bool checkLengthsEqc();
 247   bool checkCardinality();
 248   bool checkInductiveEquations();
 249   //check membership constraints
 250   Node normalizeRegexp(Node r);
 251   bool normalizePosMemberships(std::map< Node, std::vector< Node > > &memb_with_exps);
 252   bool applyRConsume( CVC4::String &s, Node &r);
 253   Node applyRSplit(Node s1, Node s2, Node r);
 254   bool applyRLen(std::map< Node, std::vector< Node > > &XinR_with_exps);
 255   bool checkMembershipsWithoutLength(
 256     std::map< Node, std::vector< Node > > &memb_with_exps,
 257     std::map< Node, std::vector< Node > > &XinR_with_exps);
 258   bool checkMemberships();
 259   //temp
 260   bool checkMemberships2();
 261   bool checkPDerivative(Node x, Node r, Node atom, bool &addedLemma,
 262     std::vector< Node > &processed, std::vector< Node > &cprocessed,
 263     std::vector< Node > &nf_exp);
 264   bool checkContains();
 265   bool checkPosContains();
 266   bool checkNegContains();
 267
 268 public:
 269   void preRegisterTerm(TNode n);
 270   Node expandDefinition(LogicRequest &logicRequest, Node n);
 271   void check(Effort e);
 272
 273   /** Conflict when merging two constants */
 274   void conflict(TNode a, TNode b);
 275   /** called when a new equivalence class is created */
 276   void eqNotifyNewClass(TNode t);
 277   /** called when two equivalence classes will merge */
 278   void eqNotifyPreMerge(TNode t1, TNode t2);
 279   /** called when two equivalence classes have merged */
 280   void eqNotifyPostMerge(TNode t1, TNode t2);
 281   /** called when two equivalence classes are made disequal */
 282   void eqNotifyDisequal(TNode t1, TNode t2, TNode reason);
 283 protected:
 284   /** compute care graph */
 285   void computeCareGraph();
 286
 287   //do pending merges
 288   void assertPendingFact(Node fact, Node exp);
 289   void doPendingFacts();
 290   void doPendingLemmas();
 291
 292   //register term
 293   bool registerTerm( Node n );
 294   //send lemma
 295   void sendLemma( Node ant, Node conc, const char * c );
 296   void sendInfer( Node eq_exp, Node eq, const char * c );
 297   void sendSplit( Node a, Node b, const char * c, bool preq = true );
 298   void sendLengthLemma( Node n );
 299   /** mkConcat **/
 300   inline Node mkConcat( Node n1, Node n2 );
 301   inline Node mkConcat( Node n1, Node n2, Node n3 );
 302   inline Node mkConcat( const std::vector< Node >& c );
 303   //mkSkolem
 304   inline Node mkSkolemS(const char * c, int isLenSplit = 0);
 305   //inline Node mkSkolemI(const char * c);
 306   /** mkExplain **/
 307   Node mkExplain( std::vector< Node >& a );
 308   Node mkExplain( std::vector< Node >& a, std::vector< Node >& an );
 309   /** mkAnd **/
 310   Node mkAnd( std::vector< Node >& a );
 311   /** get concat vector */
 312   void getConcatVec( Node n, std::vector< Node >& c );
 313
 314   //get equivalence classes
 315   void getEquivalenceClasses( std::vector< Node >& eqcs );
 316   //get final normal form
 317   void getFinalNormalForm( Node n, std::vector< Node >& nf, std::vector< Node >& exp );
 318
 319   //separate into collections with equal length
 320   void separateByLength( std::vector< Node >& n, std::vector< std::vector< Node > >& col, std::vector< Node >& lts );
 321   void printConcat( std::vector< Node >& n, const char * c );
 322
 323 private:
 324   Node mkSplitEq( const char * c, const char * info, Node lhs, Node rhs, bool lgtZero );
 325
 326   // Special String Functions
 327   NodeList d_str_pos_ctn;
 328   NodeList d_str_neg_ctn;
 329   NodeSet d_neg_ctn_eqlen;
 330   NodeSet d_neg_ctn_ulen;
 331   NodeSet d_pos_ctn_cached;
 332   NodeSet d_neg_ctn_cached;
 333
 334   // Symbolic Regular Expression
 335 private:
 336   // regular expression memberships
 337   NodeList d_regexp_memberships;
 338   NodeSet d_regexp_ucached;
 339   NodeSet d_regexp_ccached;
 340   // stored assertions
 341   NodeListMap d_pos_memberships;
 342   NodeListMap d_neg_memberships;
 343   // semi normal forms for symbolic expression
 344   std::map< Node, Node > d_nf_regexps;
 345   std::map< Node, std::vector< Node > > d_nf_regexps_exp;
 346   // intersection
 347   NodeNodeMap d_inter_cache;
 348   NodeIntMap d_inter_index;
 349   // processed memberships
 350   NodeSet d_processed_memberships;
 351   // antecedant for why regexp membership must be true
 352   NodeNodeMap d_regexp_ant;
 353   // membership length
 354   //std::map< Node, bool > d_membership_length;
 355   // regular expression operations
 356   RegExpOpr d_regexp_opr;
 357
 358   CVC4::String getHeadConst( Node x );
 359   bool deriveRegExp( Node x, Node r, Node ant );
 360   bool addMembershipLength(Node atom);
 361   void addMembership(Node assertion);
 362   Node getNormalString(Node x, std::vector<Node> &nf_exp);
 363   Node getNormalSymRegExp(Node r, std::vector<Node> &nf_exp);
 364
 365
 366   // Finite Model Finding
 367 private:
 368   NodeSet d_input_vars;
 369   context::CDO< Node > d_input_var_lsum;
 370   context::CDHashMap< int, Node > d_cardinality_lits;
 371   context::CDO< int > d_curr_cardinality;
 372 public:
 373   //for finite model finding
 374   Node getNextDecisionRequest();
 375   void assertNode( Node lit );
 376
 377 public:
 378 /** statistics class */
 379   class Statistics {
 380   public:
 381     IntStat d_splits;
 382     IntStat d_eq_splits;
 383     IntStat d_deq_splits;
 384     IntStat d_loop_lemmas;
 385     IntStat d_new_skolems;
 386     Statistics();
 387     ~Statistics();
 388   };/* class TheoryStrings::Statistics */
 389   Statistics d_statistics;
 390 };/* class TheoryStrings */
 391
 392 }/* CVC4::theory::strings namespace */
 393 }/* CVC4::theory namespace */
 394 }/* CVC4 namespace */
 395
 396 #endif /* __CVC4__THEORY__STRINGS__THEORY_STRINGS_H */