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