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

   1 /* $PostgreSQL: pgsql/contrib/tsearch2/dict_syn.c,v 1.11 2006/12/04 09:26:57 teodor Exp $ */
   2
   3 /*
   4  * ISpell interface
   5  * Teodor Sigaev <teodor@sigaev.ru>
   6  */
   7 #include "postgres.h"
   8
   9 #include <ctype.h>
  10
  11 #include "dict.h"
  12 #include "common.h"
  13 #include "ts_locale.h"
  14
  15 #define SYNBUFLEN       4096
  16 typedef struct
  17 {
  18         char       *in;
  19         char       *out;
  20 }       Syn;
  21
  22 typedef struct
  23 {
  24         int                     len;
  25         Syn                *syn;
  26 }       DictSyn;
  27
  28 PG_FUNCTION_INFO_V1(syn_init);
  29 Datum           syn_init(PG_FUNCTION_ARGS);
  30
  31 PG_FUNCTION_INFO_V1(syn_lexize);
  32 Datum           syn_lexize(PG_FUNCTION_ARGS);
  33
  34 static char *
  35 findwrd(char *in, char **end)
  36 {
  37         char       *start;
  38
  39         *end = NULL;
  40         while (*in && isspace((unsigned char) *in))
  41                 in++;
  42
  43         if (!in)
  44                 return NULL;
  45         start = in;
  46
  47         while (*in && !isspace((unsigned char) *in))
  48                 in++;
  49
  50         *end = in;
  51         return start;
  52 }
  53
  54 static int
  55 compareSyn(const void *a, const void *b)
  56 {
  57         return strcmp(((Syn *) a)->in, ((Syn *) b)->in);
  58 }
  59
  60
  61 Datum
  62 syn_init(PG_FUNCTION_ARGS)
  63 {
  64         text       *in;
  65         DictSyn    *d;
  66         int                     cur = 0;
  67         FILE       *fin;
  68         char       *filename;
  69         char            buf[SYNBUFLEN];
  70         char       *starti,
  71                            *starto,
  72                            *end = NULL;
  73         int                     slen;
  74
  75         if (PG_ARGISNULL(0) || PG_GETARG_POINTER(0) == NULL)
  76                 ereport(ERROR,
  77                                 (errcode(ERRCODE_INVALID_PARAMETER_VALUE),
  78                                  errmsg("NULL config")));
  79
  80         in = PG_GETARG_TEXT_P(0);
  81         if (VARSIZE(in) - VARHDRSZ == 0)
  82                 ereport(ERROR,
  83                                 (errcode(ERRCODE_INVALID_PARAMETER_VALUE),
  84                                  errmsg("VOID config")));
  85
  86         filename = text2char(in);
  87         PG_FREE_IF_COPY(in, 0);
  88         if ((fin = fopen(filename, "r")) == NULL)
  89                 ereport(ERROR,
  90                                 (errcode_for_file_access(),
  91                                  errmsg("could not open file \"%s\": %m",
  92                                                 filename)));
  93
  94         d = (DictSyn *) malloc(sizeof(DictSyn));
  95         if (!d)
  96         {
  97                 fclose(fin);
  98                 ereport(ERROR,
  99                                 (errcode(ERRCODE_OUT_OF_MEMORY),
 100                                  errmsg("out of memory")));
 101         }
 102         memset(d, 0, sizeof(DictSyn));
 103
 104         while (fgets(buf, SYNBUFLEN, fin))
 105         {
 106                 slen = strlen(buf) - 1;
 107                 buf[slen] = '\0';
 108                 if (*buf == '\0')
 109                         continue;
 110                 if (cur == d->len)
 111                 {
 112                         d->len = (d->len) ? 2 * d->len : 16;
 113                         d->syn = (Syn *) realloc(d->syn, sizeof(Syn) * d->len);
 114                         if (!d->syn)
 115                         {
 116                                 fclose(fin);
 117                                 ereport(ERROR,
 118                                                 (errcode(ERRCODE_OUT_OF_MEMORY),
 119                                                  errmsg("out of memory")));
 120                         }
 121                 }
 122
 123                 starti = findwrd(buf, &end);
 124                 if (!starti)
 125                         continue;
 126                 *end = '\0';
 127                 if (end >= buf + slen)
 128                         continue;
 129
 130                 starto = findwrd(end + 1, &end);
 131                 if (!starto)
 132                         continue;
 133                 *end = '\0';
 134
 135                 d->syn[cur].in = strdup(lowerstr(starti));
 136                 d->syn[cur].out = strdup(lowerstr(starto));
 137                 if (!(d->syn[cur].in && d->syn[cur].out))
 138                 {
 139                         fclose(fin);
 140                         ereport(ERROR,
 141                                         (errcode(ERRCODE_OUT_OF_MEMORY),
 142                                          errmsg("out of memory")));
 143                 }
 144
 145                 cur++;
 146         }
 147
 148         fclose(fin);
 149
 150         d->len = cur;
 151         if (cur > 1)
 152                 qsort(d->syn, d->len, sizeof(Syn), compareSyn);
 153
 154         pfree(filename);
 155         PG_RETURN_POINTER(d);
 156 }
 157
 158 Datum
 159 syn_lexize(PG_FUNCTION_ARGS)
 160 {
 161         DictSyn    *d = (DictSyn *) PG_GETARG_POINTER(0);
 162         char       *in = (char *) PG_GETARG_POINTER(1);
 163         Syn                     key,
 164                            *found;
 165         TSLexeme   *res = NULL;
 166         char       *wrd;
 167
 168         if (!PG_GETARG_INT32(2))
 169                 PG_RETURN_POINTER(NULL);
 170
 171         key.out = NULL;
 172         wrd = pnstrdup(in, PG_GETARG_INT32(2));
 173         key.in = lowerstr(wrd);
 174         pfree(wrd);
 175
 176         found = (Syn *) bsearch(&key, d->syn, d->len, sizeof(Syn), compareSyn);
 177         pfree(key.in);
 178
 179         if (!found)
 180                 PG_RETURN_POINTER(NULL);
 181
 182         res = palloc(sizeof(TSLexeme) * 2);
 183         memset(res, 0, sizeof(TSLexeme) * 2);
 184         res[0].lexeme = pstrdup(found->out);
 185
 186         PG_RETURN_POINTER(res);
 187 }