granicus.if.org Git - postgresql/blob - contrib/tsearch2/dict_snowball.c

   1 /*
   2  * example of Snowball dictionary
   3  * http://snowball.tartarus.org/
   4  * Teodor Sigaev <teodor@sigaev.ru>
   5  */
   6 #include <stdlib.h>
   7 #include <string.h>
   8
   9 #include "postgres.h"
  10
  11 #include "dict.h"
  12 #include "common.h"
  13 #include "snowball/header.h"
  14 #include "snowball/english_stem.h"
  15 #include "snowball/russian_stem.h"
  16
  17 typedef struct
  18 {
  19         struct SN_env *z;
  20         StopList        stoplist;
  21         int                     (*stem) (struct SN_env * z);
  22 }       DictSnowball;
  23
  24
  25 PG_FUNCTION_INFO_V1(snb_en_init);
  26 Datum           snb_en_init(PG_FUNCTION_ARGS);
  27
  28 PG_FUNCTION_INFO_V1(snb_ru_init);
  29 Datum           snb_ru_init(PG_FUNCTION_ARGS);
  30
  31 PG_FUNCTION_INFO_V1(snb_lexize);
  32 Datum           snb_lexize(PG_FUNCTION_ARGS);
  33
  34 Datum
  35 snb_en_init(PG_FUNCTION_ARGS)
  36 {
  37         DictSnowball *d = (DictSnowball *) malloc(sizeof(DictSnowball));
  38
  39         if (!d)
  40                 ereport(ERROR,
  41                                 (errcode(ERRCODE_OUT_OF_MEMORY),
  42                                  errmsg("out of memory")));
  43         memset(d, 0, sizeof(DictSnowball));
  44         d->stoplist.wordop = lowerstr;
  45
  46         if (!PG_ARGISNULL(0) && PG_GETARG_POINTER(0) != NULL)
  47         {
  48                 text       *in = PG_GETARG_TEXT_P(0);
  49
  50                 readstoplist(in, &(d->stoplist));
  51                 sortstoplist(&(d->stoplist));
  52                 PG_FREE_IF_COPY(in, 0);
  53         }
  54
  55         d->z = english_create_env();
  56         if (!d->z)
  57         {
  58                 freestoplist(&(d->stoplist));
  59                 ereport(ERROR,
  60                                 (errcode(ERRCODE_OUT_OF_MEMORY),
  61                                  errmsg("out of memory")));
  62         }
  63         d->stem = english_stem;
  64
  65         PG_RETURN_POINTER(d);
  66 }
  67
  68 Datum
  69 snb_ru_init(PG_FUNCTION_ARGS)
  70 {
  71         DictSnowball *d = (DictSnowball *) malloc(sizeof(DictSnowball));
  72
  73         if (!d)
  74                 ereport(ERROR,
  75                                 (errcode(ERRCODE_OUT_OF_MEMORY),
  76                                  errmsg("out of memory")));
  77         memset(d, 0, sizeof(DictSnowball));
  78         d->stoplist.wordop = lowerstr;
  79
  80         if (!PG_ARGISNULL(0) && PG_GETARG_POINTER(0) != NULL)
  81         {
  82                 text       *in = PG_GETARG_TEXT_P(0);
  83
  84                 readstoplist(in, &(d->stoplist));
  85                 sortstoplist(&(d->stoplist));
  86                 PG_FREE_IF_COPY(in, 0);
  87         }
  88
  89         d->z = russian_create_env();
  90         if (!d->z)
  91         {
  92                 freestoplist(&(d->stoplist));
  93                 ereport(ERROR,
  94                                 (errcode(ERRCODE_OUT_OF_MEMORY),
  95                                  errmsg("out of memory")));
  96         }
  97         d->stem = russian_stem;
  98
  99         PG_RETURN_POINTER(d);
 100 }
 101
 102 Datum
 103 snb_lexize(PG_FUNCTION_ARGS)
 104 {
 105         DictSnowball *d = (DictSnowball *) PG_GETARG_POINTER(0);
 106         char       *in = (char *) PG_GETARG_POINTER(1);
 107         char       *txt = pnstrdup(in, PG_GETARG_INT32(2));
 108         TSLexeme          *res = palloc(sizeof(TSLexeme) * 2);
 109
 110         memset(res, 0, sizeof(TSLexeme) * 2);
 111         if (*txt == '\0' || searchstoplist(&(d->stoplist), txt))
 112         {
 113                 pfree(txt);
 114         }
 115         else
 116         {
 117                 SN_set_current(d->z, strlen(txt), txt);
 118                 (d->stem) (d->z);
 119                 if (d->z->p && d->z->l)
 120                 {
 121                         txt = repalloc(txt, d->z->l + 1);
 122                         memcpy(txt, d->z->p, d->z->l);
 123                         txt[d->z->l] = '\0';
 124                 }
 125                 res->lexeme = txt;
 126         }
 127
 128         PG_RETURN_POINTER(res);
 129 }