00001
00004
00005
00006
00007
00008
00009
00010
00011
00012
00013
00014
00015
00016
00017
00018
00019
00020
00021
00022
00023
00024 #ifndef XAPIAN_INCLUDED_ENQUIRE_H
00025 #define XAPIAN_INCLUDED_ENQUIRE_H
00026
00027 #include <string>
00028
00029 #include <xapian/base.h>
00030 #include <xapian/deprecated.h>
00031 #include <xapian/types.h>
00032 #include <xapian/termiterator.h>
00033 #include <xapian/visibility.h>
00034
00035 namespace Xapian {
00036
00037 class Database;
00038 class Document;
00039 class ErrorHandler;
00040 class ExpandDecider;
00041 class MSetIterator;
00042 class Query;
00043 class Weight;
00044
00048 class XAPIAN_VISIBILITY_DEFAULT MSet {
00049 public:
00050 class Internal;
00052 Xapian::Internal::RefCntPtr<Internal> internal;
00053
00055 explicit MSet(MSet::Internal * internal_);
00056
00058 MSet();
00059
00061 ~MSet();
00062
00064 MSet(const MSet & other);
00065
00067 void operator=(const MSet &other);
00068
00084 void fetch(const MSetIterator &begin, const MSetIterator &end) const;
00085
00088 void fetch(const MSetIterator &item) const;
00089
00092 void fetch() const;
00093
00098 Xapian::percent convert_to_percent(Xapian::weight wt) const;
00099
00101 Xapian::percent convert_to_percent(const MSetIterator &it) const;
00102
00110 Xapian::doccount get_termfreq(const std::string &tname) const;
00111
00119 Xapian::weight get_termweight(const std::string &tname) const;
00120
00128 Xapian::doccount get_firstitem() const;
00129
00139 Xapian::doccount get_matches_lower_bound() const;
00140
00153 Xapian::doccount get_matches_estimated() const;
00154
00164 Xapian::doccount get_matches_upper_bound() const;
00165
00171 Xapian::weight get_max_possible() const;
00172
00186 Xapian::weight get_max_attained() const;
00187
00189 Xapian::doccount size() const;
00190
00192 Xapian::doccount max_size() const { return size(); }
00193
00195 bool empty() const;
00196
00198 void swap(MSet & other);
00199
00201 MSetIterator begin() const;
00202
00204 MSetIterator end() const;
00205
00207 MSetIterator back() const;
00208
00218 MSetIterator operator[](Xapian::doccount i) const;
00219
00221
00222 typedef MSetIterator value_type;
00223 typedef MSetIterator iterator;
00224 typedef MSetIterator const_iterator;
00225 typedef MSetIterator & reference;
00226 typedef MSetIterator & const_reference;
00227 typedef MSetIterator * pointer;
00228 typedef Xapian::doccount_diff difference_type;
00229 typedef Xapian::doccount size_type;
00231
00235 std::string get_description() const;
00236 };
00237
00241 class XAPIAN_VISIBILITY_DEFAULT MSetIterator {
00242 private:
00243 friend class MSet;
00244 friend bool operator==(const MSetIterator &a, const MSetIterator &b);
00245 friend bool operator!=(const MSetIterator &a, const MSetIterator &b);
00246
00247 MSetIterator(Xapian::doccount index_, const MSet & mset_)
00248 : index(index_), mset(mset_) { }
00249
00250 Xapian::doccount index;
00251 MSet mset;
00252
00253 public:
00257 MSetIterator() : index(0), mset() { }
00258
00259 ~MSetIterator() { }
00260
00262 MSetIterator(const MSetIterator &other) {
00263 index = other.index;
00264 mset = other.mset;
00265 }
00266
00268 void operator=(const MSetIterator &other) {
00269 index = other.index;
00270 mset = other.mset;
00271 }
00272
00274 MSetIterator & operator++() {
00275 ++index;
00276 return *this;
00277 }
00278
00280 MSetIterator operator++(int) {
00281 MSetIterator tmp = *this;
00282 ++index;
00283 return tmp;
00284 }
00285
00287 MSetIterator & operator--() {
00288 --index;
00289 return *this;
00290 }
00291
00293 MSetIterator operator--(int) {
00294 MSetIterator tmp = *this;
00295 --index;
00296 return tmp;
00297 }
00298
00300 Xapian::docid operator*() const;
00301
00318 Xapian::Document get_document() const;
00319
00326 Xapian::doccount get_rank() const {
00327 return mset.get_firstitem() + index;
00328 }
00329
00331 Xapian::weight get_weight() const;
00332
00335 std::string get_collapse_key() const;
00336
00353 Xapian::doccount get_collapse_count() const;
00354
00360 Xapian::percent get_percent() const;
00361
00365 std::string get_description() const;
00366
00368
00369 typedef std::bidirectional_iterator_tag iterator_category;
00370 typedef Xapian::docid value_type;
00371 typedef Xapian::doccount_diff difference_type;
00372 typedef Xapian::docid * pointer;
00373 typedef Xapian::docid & reference;
00375 };
00376
00377 inline bool operator==(const MSetIterator &a, const MSetIterator &b)
00378 {
00379 return (a.index == b.index);
00380 }
00381
00382 inline bool operator!=(const MSetIterator &a, const MSetIterator &b)
00383 {
00384 return (a.index != b.index);
00385 }
00386
00387 class ESetIterator;
00388
00393 class XAPIAN_VISIBILITY_DEFAULT ESet {
00394 public:
00395 class Internal;
00397 Xapian::Internal::RefCntPtr<Internal> internal;
00398
00400 ESet();
00401
00403 ~ESet();
00404
00406 ESet(const ESet & other);
00407
00409 void operator=(const ESet &other);
00410
00415 Xapian::termcount get_ebound() const;
00416
00418 Xapian::termcount size() const;
00419
00421 Xapian::termcount max_size() const { return size(); }
00422
00424 bool empty() const;
00425
00427 void swap(ESet & other);
00428
00430 ESetIterator begin() const;
00431
00433 ESetIterator end() const;
00434
00436 ESetIterator back() const;
00437
00439 ESetIterator operator[](Xapian::termcount i) const;
00440
00445 std::string get_description() const;
00446 };
00447
00449 class XAPIAN_VISIBILITY_DEFAULT ESetIterator {
00450 private:
00451 friend class ESet;
00452 friend bool operator==(const ESetIterator &a, const ESetIterator &b);
00453 friend bool operator!=(const ESetIterator &a, const ESetIterator &b);
00454
00455 ESetIterator(Xapian::termcount index_, const ESet & eset_)
00456 : index(index_), eset(eset_) { }
00457
00458 Xapian::termcount index;
00459 ESet eset;
00460
00461 public:
00465 ESetIterator() : index(0), eset() { }
00466
00467 ~ESetIterator() { }
00468
00470 ESetIterator(const ESetIterator &other) {
00471 index = other.index;
00472 eset = other.eset;
00473 }
00474
00476 void operator=(const ESetIterator &other) {
00477 index = other.index;
00478 eset = other.eset;
00479 }
00480
00482 ESetIterator & operator++() {
00483 ++index;
00484 return *this;
00485 }
00486
00488 ESetIterator operator++(int) {
00489 ESetIterator tmp = *this;
00490 ++index;
00491 return tmp;
00492 }
00493
00495 ESetIterator & operator--() {
00496 --index;
00497 return *this;
00498 }
00499
00501 ESetIterator operator--(int) {
00502 ESetIterator tmp = *this;
00503 --index;
00504 return tmp;
00505 }
00506
00508 const std::string & operator *() const;
00509
00511 Xapian::weight get_weight() const;
00512
00516 std::string get_description() const;
00517
00519
00520 typedef std::bidirectional_iterator_tag iterator_category;
00521 typedef std::string value_type;
00522 typedef Xapian::termcount_diff difference_type;
00523 typedef std::string * pointer;
00524 typedef std::string & reference;
00526 };
00527
00528 inline bool operator==(const ESetIterator &a, const ESetIterator &b)
00529 {
00530 return (a.index == b.index);
00531 }
00532
00533 inline bool operator!=(const ESetIterator &a, const ESetIterator &b)
00534 {
00535 return (a.index != b.index);
00536 }
00537
00542 class XAPIAN_VISIBILITY_DEFAULT RSet {
00543 public:
00545 class Internal;
00546
00548 Xapian::Internal::RefCntPtr<Internal> internal;
00549
00551 RSet(const RSet &rset);
00552
00554 void operator=(const RSet &rset);
00555
00557 RSet();
00558
00560 ~RSet();
00561
00563 Xapian::doccount size() const;
00564
00566 bool empty() const;
00567
00569 void add_document(Xapian::docid did);
00570
00572 void add_document(const Xapian::MSetIterator & i) { add_document(*i); }
00573
00575 void remove_document(Xapian::docid did);
00576
00578 void remove_document(const Xapian::MSetIterator & i) { remove_document(*i); }
00579
00581 bool contains(Xapian::docid did) const;
00582
00584 bool contains(const Xapian::MSetIterator & i) const { return contains(*i); }
00585
00590 std::string get_description() const;
00591 };
00592
00595 class XAPIAN_VISIBILITY_DEFAULT MatchDecider {
00596 public:
00599 virtual bool operator()(const Xapian::Document &doc) const = 0;
00600
00602 virtual ~MatchDecider();
00603 };
00604
00615 class XAPIAN_VISIBILITY_DEFAULT Enquire {
00616 private:
00618 Enquire(const Enquire &);
00619
00621 void operator=(const Enquire &);
00622
00623 public:
00624 class Internal;
00626 Xapian::Internal::RefCntPtr<Internal> internal;
00627
00652 explicit Enquire(const Database &database, ErrorHandler * errorhandler_ = 0);
00653
00656 ~Enquire();
00657
00664 void set_query(const Xapian::Query & query, Xapian::termcount qlen = 0);
00665
00672 const Xapian::Query & get_query() const;
00673
00680 void set_weighting_scheme(const Weight &weight_);
00681
00708 void set_collapse_key(Xapian::valueno collapse_key);
00709
00710 typedef enum {
00711 ASCENDING = 1,
00712 DESCENDING = 0,
00713 DONT_CARE = 2
00714 } docid_order;
00715
00739 void set_docid_order(docid_order order);
00740
00759 void set_cutoff(Xapian::percent percent_cutoff, Xapian::weight weight_cutoff = 0);
00760
00765 void set_sort_by_relevance();
00766
00779 void set_sort_by_value(Xapian::valueno sort_key, bool ascending = true);
00780
00794 void set_sort_by_value_then_relevance(Xapian::valueno sort_key,
00795 bool ascending = true);
00796
00816 void set_sort_by_relevance_then_value(Xapian::valueno sort_key,
00817 bool ascending = true);
00818
00851 MSet get_mset(Xapian::doccount first, Xapian::doccount maxitems,
00852 Xapian::doccount checkatleast = 0,
00853 const RSet * omrset = 0,
00854 const MatchDecider * mdecider = 0) const;
00855 MSet get_mset(Xapian::doccount first, Xapian::doccount maxitems,
00856 Xapian::doccount checkatleast,
00857 const RSet * omrset,
00858 const MatchDecider * mdecider,
00859 const MatchDecider * matchspy) const;
00860 MSet get_mset(Xapian::doccount first, Xapian::doccount maxitems,
00861 const RSet * omrset,
00862 const MatchDecider * mdecider = 0) const {
00863 return get_mset(first, maxitems, 0, omrset, mdecider);
00864 }
00865
00866 static const int INCLUDE_QUERY_TERMS = 1;
00867 static const int USE_EXACT_TERMFREQ = 2;
00868 #ifndef _MSC_VER
00870 XAPIAN_DEPRECATED(static const int include_query_terms) = 1;
00872 XAPIAN_DEPRECATED(static const int use_exact_termfreq) = 2;
00873 #else
00874
00875
00876 static const int include_query_terms = 1;
00877 static const int use_exact_termfreq = 2;
00878 #pragma deprecated("Xapian::Enquire::include_query_terms", "Xapian::Enquire::use_exact_termfreq")
00879 #endif
00880
00903 ESet get_eset(Xapian::termcount maxitems,
00904 const RSet & omrset,
00905 int flags = 0,
00906 double k = 1.0,
00907 const Xapian::ExpandDecider * edecider = 0) const;
00908
00922 inline ESet get_eset(Xapian::termcount maxitems, const RSet & omrset,
00923 const Xapian::ExpandDecider * edecider) const {
00924 return get_eset(maxitems, omrset, 0, 1.0, edecider);
00925 }
00926
00955 TermIterator get_matching_terms_begin(Xapian::docid did) const;
00956
00958 TermIterator get_matching_terms_end(Xapian::docid ) const {
00959 return TermIterator(NULL);
00960 }
00961
00984 TermIterator get_matching_terms_begin(const MSetIterator &it) const;
00985
00987 TermIterator get_matching_terms_end(const MSetIterator &) const {
00988 return TermIterator(NULL);
00989 }
00990
00997 void register_match_decider(const std::string &name,
00998 const MatchDecider *mdecider = NULL);
00999
01003 std::string get_description() const;
01004 };
01005
01006 }
01007
01008 class RemoteServer;
01009
01010 namespace Xapian {
01011
01013 class XAPIAN_VISIBILITY_DEFAULT Weight {
01014 friend class Enquire;
01015 friend class ::RemoteServer;
01016 public:
01017 class Internal;
01018 protected:
01019 Weight(const Weight &);
01020 private:
01021 void operator=(Weight &);
01022
01032 virtual Weight * clone() const = 0;
01033
01034 protected:
01035 const Internal * internal;
01036 Xapian::doclength querysize;
01037 Xapian::termcount wqf;
01038 std::string tname;
01039
01040 public:
01041 Weight() { }
01042 virtual ~Weight();
01043
01056 Weight * create(const Internal * internal_, Xapian::doclength querysize_,
01057 Xapian::termcount wqf_, const std::string & tname_) const;
01058
01063 virtual std::string name() const = 0;
01064
01066 virtual std::string serialise() const = 0;
01067
01069 virtual Weight * unserialise(const std::string &s) const = 0;
01070
01078 virtual Xapian::weight get_sumpart(Xapian::termcount wdf,
01079 Xapian::doclength len) const = 0;
01080
01086 virtual Xapian::weight get_maxpart() const = 0;
01087
01096 virtual Xapian::weight get_sumextra(Xapian::doclength len) const = 0;
01097
01101 virtual Xapian::weight get_maxextra() const = 0;
01102
01104 virtual bool get_sumpart_needs_doclength() const;
01105 };
01106
01108 class XAPIAN_VISIBILITY_DEFAULT BoolWeight : public Weight {
01109 public:
01110 BoolWeight * clone() const;
01111 BoolWeight() { }
01112 ~BoolWeight();
01113 std::string name() const;
01114 std::string serialise() const;
01115 BoolWeight * unserialise(const std::string & s) const;
01116 Xapian::weight get_sumpart(Xapian::termcount wdf, Xapian::doclength len) const;
01117 Xapian::weight get_maxpart() const;
01118
01119 Xapian::weight get_sumextra(Xapian::doclength len) const;
01120 Xapian::weight get_maxextra() const;
01121
01122 bool get_sumpart_needs_doclength() const;
01123 };
01124
01137 class XAPIAN_VISIBILITY_DEFAULT BM25Weight : public Weight {
01138 private:
01139 mutable Xapian::weight termweight;
01140 mutable Xapian::doclength lenpart;
01141
01142 double k1, k2, k3, b;
01143 Xapian::doclength min_normlen;
01144
01145 mutable bool weight_calculated;
01146
01147 void calc_termweight() const;
01148
01149 public:
01168 BM25Weight(double k1_, double k2_, double k3_, double b_,
01169 double min_normlen_)
01170 : k1(k1_), k2(k2_), k3(k3_), b(b_), min_normlen(min_normlen_),
01171 weight_calculated(false)
01172 {
01173 if (k1 < 0) k1 = 0;
01174 if (k2 < 0) k2 = 0;
01175 if (k3 < 0) k3 = 0;
01176 if (b < 0) b = 0; else if (b > 1) b = 1;
01177 }
01178 BM25Weight() : k1(1), k2(0), k3(1), b(0.5), min_normlen(0.5),
01179 weight_calculated(false) { }
01180
01181 BM25Weight * clone() const;
01182 ~BM25Weight() { }
01183 std::string name() const;
01184 std::string serialise() const;
01185 BM25Weight * unserialise(const std::string & s) const;
01186 Xapian::weight get_sumpart(Xapian::termcount wdf, Xapian::doclength len) const;
01187 Xapian::weight get_maxpart() const;
01188
01189 Xapian::weight get_sumextra(Xapian::doclength len) const;
01190 Xapian::weight get_maxextra() const;
01191
01192 bool get_sumpart_needs_doclength() const;
01193 };
01194
01212 class XAPIAN_VISIBILITY_DEFAULT TradWeight : public Weight {
01213 private:
01214 mutable Xapian::weight termweight;
01215 mutable Xapian::doclength lenpart;
01216
01217 double param_k;
01218
01219 mutable bool weight_calculated;
01220
01221 void calc_termweight() const;
01222
01223 public:
01231 explicit TradWeight(double k) : param_k(k), weight_calculated(false) {
01232 if (param_k < 0) param_k = 0;
01233 }
01234
01235 TradWeight() : param_k(1.0), weight_calculated(false) { }
01236
01237 TradWeight * clone() const;
01238 ~TradWeight() { }
01239 std::string name() const;
01240 std::string serialise() const;
01241 TradWeight * unserialise(const std::string & s) const;
01242
01243 Xapian::weight get_sumpart(Xapian::termcount wdf, Xapian::doclength len) const;
01244 Xapian::weight get_maxpart() const;
01245
01246 Xapian::weight get_sumextra(Xapian::doclength len) const;
01247 Xapian::weight get_maxextra() const;
01248
01249 bool get_sumpart_needs_doclength() const;
01250 };
01251
01252 }
01253
01254 #endif