smlar.h

   1 #ifndef _SMLAR_H_
   2 #define _SMLAR_H_
   3
   4 #include "postgres.h"
   5 #include "utils/array.h"
   6 #include "access/tupdesc.h"
   7 #include "catalog/pg_collation.h"
   8
   9 #include <math.h>
  10
  11 typedef struct ProcTypeInfoData *ProcTypeInfo;
  12
  13 typedef struct ProcTypeInfoData {
  14         Oid                             typid;
  15         Oid                             hashFuncOid;
  16         Oid                             cmpFuncOid;
  17         int16                   typlen;
  18         bool                    typbyval;
  19         char                    typalign;
  20
  21         /* support of composite type */
  22         char                    typtype;
  23         TupleDesc               tupDesc;
  24
  25         /*
  26          * Following member can become invalid,
  27          * so fill it just before using
  28          */
  29         bool                    hashFuncInited;
  30         FmgrInfo                hashFunc;
  31         bool                    cmpFuncInited;
  32         FmgrInfo                cmpFunc;
  33 } ProcTypeInfoData;
  34
  35 ProcTypeInfo findProcs(Oid typid);
  36 void getFmgrInfoHash(ProcTypeInfo info);
  37 void getFmgrInfoCmp(ProcTypeInfo info);
  38
  39 #define NDIM 1
  40 /* reject arrays we can't handle; but allow a NULL or empty array */
  41 #define CHECKARRVALID(x) \
  42         do { \
  43                 if (x) { \
  44                         if (ARR_NDIM(x) != NDIM && ARR_NDIM(x) != 0) \
  45                                 ereport(ERROR, \
  46                                                 (errcode(ERRCODE_ARRAY_SUBSCRIPT_ERROR), \
  47                                                  errmsg("array must be one-dimensional"))); \
  48                         if (ARR_HASNULL(x)) \
  49                                 ereport(ERROR, \
  50                                                 (errcode(ERRCODE_NULL_VALUE_NOT_ALLOWED), \
  51                                                  errmsg("array must not contain nulls"))); \
  52                 } \
  53         } while(0)
  54
  55 #define ARRISVOID(x)  ((x) == NULL || ARRNELEMS(x) == 0)
  56 #define ARRNELEMS(x)  ArrayGetNItems(ARR_NDIM(x), ARR_DIMS(x))
  57
  58
  59 typedef struct SimpleArray {
  60         Datum              *elems;
  61         double             *df;  /* frequency in current doc */
  62         uint32             *hash;
  63         int                             nelems;
  64         ProcTypeInfo    info;
  65 } SimpleArray;
  66
  67 SimpleArray     * Array2SimpleArray(ProcTypeInfo info, ArrayType *a);
  68 SimpleArray     * Array2SimpleArrayS(ProcTypeInfo info, ArrayType *a);
  69 SimpleArray     * Array2SimpleArrayU(ProcTypeInfo info, ArrayType *a, void *cache);
  70 void allocateHash(void *cache, SimpleArray *a);
  71
  72 /*
  73  * GUC vars
  74  */
  75 double GetSmlarLimit(void);
  76 const char* GetSmlarTable(void);
  77 bool GetSmlarUsePersistent(void);
  78 double getOneAdd(void);
  79 int getTFMethod(void);
  80 int getSmlType(void);
  81 /*
  82  * GiST
  83  */
  84
  85 #define SmlarOverlapStrategy            1
  86 #define SmlarSimilarityStrategy         2
  87
  88 struct SmlSign;
  89 struct SmlSign* Array2HashedArray(ProcTypeInfo info, ArrayType *a);
  90 /*
  91  * Cache subsystem
  92  */
  93 void*   SearchArrayCache( void *cache, MemoryContext ctx, Datum a, ArrayType **da, SimpleArray **sa,  struct SmlSign  **ss );
  94
  95 typedef struct StatElem {
  96         Datum           datum;
  97         double          idf; /*  log(d/df) */
  98 } StatElem;
  99
 100 typedef struct HashedElem {
 101         uint32          hash;
 102         double          idfMin;
 103         double          idfMax;
 104 } HashedElem;
 105
 106 typedef struct SignedElem {
 107         double          idfMin;
 108         double          idfMax;
 109 } SignedElem;
 110
 111 typedef struct StatCache {
 112         StatElem                *elems;
 113         int                             nelems;
 114         int64_t                 ndoc;
 115         HashedElem              *helems;
 116         int                             nhelems;
 117         SignedElem              *selems;
 118         ProcTypeInfo    info;
 119 } StatCache;
 120
 121 StatCache *initStatCache(MemoryContext ctx);
 122 void getHashStatCache(StatCache *stat, MemoryContext ctx, size_t n);
 123
 124 void    resetStatCache(void);
 125 StatElem  *findStat(StatCache *stat, Datum query, StatElem *low);
 126 StatElem  *fingArrayStat(void *cache, Oid typoid, Datum query, StatElem *low);
 127 StatCache *getStat(void *cache, size_t n);
 128
 129 /*
 130  * Formula's type of similarity
 131  */
 132 #define ST_COSINE       1
 133 #define ST_TFIDF        2
 134 #define ST_OVERLAP      3
 135 /*
 136  * TF methods
 137  */
 138 #define TF_N            1
 139 #define TF_LOG          2
 140 #define TF_CONST        3
 141
 142 #define FCall2(f, x1, x2)   FunctionCall2Coll((f), C_COLLATION_OID, (x1), (x2))
 143
 144 #endif