00001
00002
00003
00004
00005 #ifndef IBIS_QUERY_H
00006 #define IBIS_QUERY_H
00007
00008
00009
00010 #include "part.h"
00011 #include "whereClause.h"
00012 #include "selectClause.h"
00013
00053 class FASTBIT_CXX_DLLSPEC ibis::query {
00054 public:
00055 enum QUERY_STATE {
00056 UNINITIALIZED,
00057 SET_COMPONENTS,
00058 SET_RIDS,
00059 SET_PREDICATE,
00060 SPECIFIED,
00061 QUICK_ESTIMATE,
00062 FULL_EVALUATE,
00063 BUNDLES_TRUNCATED,
00064 HITS_TRUNCATED
00065 };
00066
00067 virtual ~query();
00071 query(const char* dir, const ibis::partList& tl);
00073 query(const char* uid=0, const part* et=0, const char* pref=0);
00074
00076 const char* id() const {return myID;};
00077 const char* dir() const {return myDir;}
00078 const char* userName() const {return user;}
00079
00080 time_t timestamp() const {return dstime;}
00082 const part* partition() const {return mypart;}
00084 const selectClause& components() const {return comps;};
00085
00087 int setRIDs(const RIDSet& set);
00089 int setWhereClause(const char *str);
00091 int setWhereClause(const std::vector<const char*>& names,
00092 const std::vector<double>& lbounds,
00093 const std::vector<double>& rbounds);
00095 int setWhereClause(const ibis::qExpr* qexp);
00097 virtual int setSelectClause(const char *str);
00099 int setPartition(const ibis::part* tbl);
00101 int setTable(const ibis::part* tbl) {return setPartition(tbl);}
00103 virtual const char* getWhereClause() const {return conds.getString();}
00105 virtual const char* getSelectClause() const {return *comps;}
00106
00107 void expandQuery();
00108 void contractQuery();
00109 std::string removeComplexConditions();
00110
00112 const RIDSet* getUserRIDs() const {return rids_in;}
00113
00114
00115
00116 int estimate();
00117 long getMinNumHits() const;
00118 long getMaxNumHits() const;
00119
00120
00121
00122 int evaluate(const bool evalSelect=false);
00127 const ibis::bitvector* getHitVector() const {return hits;}
00128 long getNumHits() const;
00129 long getHitRows(std::vector<uint32_t> &rids) const;
00130 long countHits() const;
00131
00132 int orderby(const char *names) const;
00133 long limit(const char *names, uint32_t keep,
00134 bool updateHits = true);
00135
00146 array_t<signed char>* getQualifiedBytes(const char* column_name);
00148 array_t<unsigned char>* getQualifiedUBytes(const char* column_name);
00150 array_t<int16_t>* getQualifiedShorts(const char* column_name);
00152 array_t<uint16_t>* getQualifiedUShorts(const char* column_name);
00154 array_t<int32_t>* getQualifiedInts(const char* column_name);
00157 array_t<uint32_t>* getQualifiedUInts(const char* column_name);
00159 array_t<int64_t>* getQualifiedLongs(const char* column_name);
00161 array_t<uint64_t>* getQualifiedULongs(const char* column_name);
00164 array_t<float>* getQualifiedFloats(const char* column_name);
00167 array_t<double>* getQualifiedDoubles(const char* column_name);
00169 std::vector<std::string>* getQualifiedStrings(const char* column_name);
00171 RIDSet* getRIDs() const;
00173 RIDSet* getRIDs(const ibis::bitvector& mask) const;
00175 const RIDSet* getRIDsInBundle(const uint32_t bid) const;
00177
00184 void printSelected(std::ostream& out) const;
00188 void printSelectedWithRID(std::ostream& out) const;
00189
00195 long sequentialScan(ibis::bitvector& bv) const;
00196
00200 long getExpandedHits(ibis::bitvector&) const;
00201
00202
00203 RIDSet* readRIDs() const;
00204 void writeRIDs(const RIDSet* rids) const;
00205
00208 void logMessage(const char* event, const char* fmt, ...) const;
00209
00210
00211
00212
00214 void clear();
00216 QUERY_STATE getState() const;
00218 const char* getLastError() const {return lastError;}
00220 void clearErrorMessage() const {*lastError=0;}
00221
00224 static bool isValidToken(const char* tok);
00226
00227
00228 static unsigned tokenLength() {return 16;}
00229
00231 static void removeQueryRecords()
00232 {ibis::gParameters().add("query.purgeTempFiles", "true");}
00234 static void keepQueryRecords()
00235 {ibis::gParameters().add("query.purgeTempFiles", "false");}
00236
00237 class result;
00238 class weight;
00239 class readLock;
00240 class writeLock;
00241 friend class readLock;
00242 friend class writeLock;
00243
00244 protected:
00245 char* user;
00246 whereClause conds;
00247 selectClause comps;
00248 QUERY_STATE state;
00249 ibis::bitvector* hits;
00250 ibis::bitvector* sup;
00251 mutable ibis::part::readLock* dslock;
00252 mutable char lastError[MAX_LINE+PATH_MAX];
00253
00254 void logError(const char* event, const char* fmt, ...) const;
00255 void logWarning(const char* event, const char* fmt, ...) const;
00256
00257 void reorderExpr();
00258
00259 bool hasBundles() const;
00260 int computeHits();
00261 void getBounds();
00262
00263 void doEstimate(const qExpr* term, ibis::bitvector& low,
00264 ibis::bitvector& high) const;
00266 int doScan(const qExpr* term, const ibis::bitvector& mask,
00267 ibis::bitvector& hits) const;
00269 int doScan(const qExpr* term, ibis::bitvector& hits) const;
00271 int doEvaluate(const qExpr* term, ibis::bitvector& hits) const;
00273 int doEvaluate(const qExpr* term, const ibis::bitvector& mask,
00274 ibis::bitvector& hits) const;
00275
00277 int64_t processJoin();
00278
00280 virtual void writeQuery();
00282 void readQuery(const ibis::partList& tl);
00284 void removeFiles();
00285
00287 void readHits();
00289 void writeHits() const;
00291 void printRIDs(const RIDSet& ridset) const;
00294 uint32_t countPages(unsigned wordsize) const;
00295
00297 int doExpand(ibis::qExpr* exp0) const;
00299 int doContract(ibis::qExpr* exp0) const;
00300
00301
00302
00303 int64_t sortJoin(const std::vector<const ibis::deprecatedJoin*>& terms,
00304 const ibis::bitvector& mask) const;
00305 int64_t sortJoin(const ibis::deprecatedJoin& cmp,
00306 const ibis::bitvector& mask) const;
00307 int64_t sortEquiJoin(const ibis::deprecatedJoin& cmp,
00308 const ibis::bitvector& mask) const;
00309 int64_t sortRangeJoin(const ibis::deprecatedJoin& cmp,
00310 const ibis::bitvector& mask) const;
00311 int64_t sortEquiJoin(const ibis::deprecatedJoin& cmp,
00312 const ibis::bitvector& mask,
00313 const char* pairfile) const;
00314 int64_t sortRangeJoin(const ibis::deprecatedJoin& cmp,
00315 const ibis::bitvector& mask,
00316 const char* pairfile) const;
00317 void orderPairs(const char* pairfile) const;
00318 int64_t mergePairs(const char* pairfile) const;
00319
00320 template <typename T1, typename T2>
00321 int64_t countEqualPairs(const array_t<T1>& val1,
00322 const array_t<T2>& val2) const;
00323 template <typename T1, typename T2>
00324 int64_t countDeltaPairs(const array_t<T1>& val1,
00325 const array_t<T2>& val2, const T1& delta) const;
00326 template <typename T1, typename T2>
00327 int64_t recordEqualPairs(const array_t<T1>& val1,
00328 const array_t<T2>& val2,
00329 const array_t<uint32_t>& ind1,
00330 const array_t<uint32_t>& ind2,
00331 const char* pairfile) const;
00332 template <typename T1, typename T2>
00333 int64_t recordDeltaPairs(const array_t<T1>& val1,
00334 const array_t<T2>& val2,
00335 const array_t<uint32_t>& ind1,
00336 const array_t<uint32_t>& ind2,
00337 const T1& delta, const char* pairfile) const;
00338
00339
00340 void gainReadAccess(const char* mesg) const {
00341 if (ibis::gVerbose > 10)
00342 logMessage("gainReadAccess", "acquiring a read lock for %s",
00343 mesg);
00344 if (0 != pthread_rwlock_rdlock(&lock))
00345 logMessage("gainReadAccess",
00346 "unable to gain read access to rwlock for %s", mesg);
00347 }
00348 void gainWriteAccess(const char* mesg) const {
00349 if (ibis::gVerbose > 10)
00350 logMessage("gainWriteAccess", "acquiring a write lock for %s",
00351 mesg);
00352 if (0 != pthread_rwlock_wrlock(&lock))
00353 logMessage("gainWriteAccess",
00354 "unable to gain write access to rwlock for %s", mesg);
00355 }
00356 void releaseAccess(const char* mesg) const {
00357 if (ibis::gVerbose > 10)
00358 logMessage("releaseAccess", "releasing rwlock for %s", mesg);
00359 if (0 != pthread_rwlock_unlock(&lock))
00360 logMessage("releaseAccess", "unable to unlock the rwlock for %s",
00361 mesg);
00362 }
00363
00364 private:
00365 char* myID;
00366 char* myDir;
00367 RIDSet* rids_in;
00368 const part* mypart;
00369 time_t dstime;
00370 mutable pthread_rwlock_t lock;
00371
00372
00373 static char* newToken(const char*);
00374
00375 void setMyDir(const char *pref);
00376
00377 query(const query&);
00378 query& operator=(const query&);
00379 };
00380
00381 namespace ibis {
00389 template <>
00390 int64_t query::countEqualPairs(const array_t<int32_t>& val1,
00391 const array_t<uint32_t>& val2) const;
00392 template <>
00393 int64_t query::countEqualPairs(const array_t<uint32_t>& val1,
00394 const array_t<int32_t>& val2) const;
00395 template <>
00396 int64_t query::countDeltaPairs(const array_t<int32_t>& val1,
00397 const array_t<uint32_t>& val2,
00398 const int32_t& delta) const;
00399 template <>
00400 int64_t query::countDeltaPairs(const array_t<uint32_t>& val1,
00401 const array_t<int32_t>& val2,
00402 const uint32_t& delta) const;
00403 template <>
00404 int64_t query::recordEqualPairs(const array_t<int32_t>& val1,
00405 const array_t<uint32_t>& val2,
00406 const array_t<uint32_t>& ind1,
00407 const array_t<uint32_t>& ind2,
00408 const char *pairfile) const;
00409 template <>
00410 int64_t query::recordEqualPairs(const array_t<uint32_t>& val1,
00411 const array_t<int32_t>& val2,
00412 const array_t<uint32_t>& ind1,
00413 const array_t<uint32_t>& ind2,
00414 const char *pairfile) const;
00415 template <>
00416 int64_t query::recordDeltaPairs(const array_t<int32_t>& val1,
00417 const array_t<uint32_t>& val2,
00418 const array_t<uint32_t>& ind1,
00419 const array_t<uint32_t>& ind2,
00420 const int32_t& delta,
00421 const char *pairfile) const;
00422 template <>
00423 int64_t query::recordDeltaPairs(const array_t<uint32_t>& val1,
00424 const array_t<int32_t>& val2,
00425 const array_t<uint32_t>& ind1,
00426 const array_t<uint32_t>& ind2,
00427 const uint32_t& delta,
00428 const char *pairfile) const;
00430 }
00431
00433 class ibis::query::weight : public ibis::qExpr::weight {
00434 public:
00435 virtual double operator()(const ibis::qExpr* ex) const;
00436 weight(const ibis::part* ds) : dataset(ds) {}
00437
00438 private:
00439 const ibis::part* dataset;
00440 };
00441
00446 class ibis::query::readLock {
00447 public:
00448 readLock(const query* q, const char* m) : theQuery(q), mesg(m) {
00449 theQuery->gainReadAccess(m);
00450 };
00451 ~readLock() {theQuery->releaseAccess(mesg);}
00452 private:
00453 const query* theQuery;
00454 const char* mesg;
00455
00456 readLock() {};
00457 readLock(const readLock&) {};
00458 };
00459
00464 class ibis::query::writeLock {
00465 public:
00466 writeLock(const query* q, const char* m) : theQuery(q), mesg(m) {
00467 theQuery->gainWriteAccess(m);
00468 };
00469 ~writeLock() {theQuery->releaseAccess(mesg);}
00470 private:
00471 const query* theQuery;
00472 const char* mesg;
00473
00474 writeLock() {};
00475 writeLock(const writeLock&) {};
00476 };
00477 #endif // IBIS_QUERY_H