| 1 | /*------------------------------------------------------------------------- |
| 2 | * |
| 3 | * dict_snowball.c |
| 4 | * Snowball dictionary |
| 5 | * |
| 6 | * Portions Copyright (c) 1996-2019, PostgreSQL Global Development Group |
| 7 | * |
| 8 | * IDENTIFICATION |
| 9 | * src/backend/snowball/dict_snowball.c |
| 10 | * |
| 11 | *------------------------------------------------------------------------- |
| 12 | */ |
| 13 | #include "postgres.h" |
| 14 | |
| 15 | #include "commands/defrem.h" |
| 16 | #include "tsearch/ts_locale.h" |
| 17 | #include "tsearch/ts_utils.h" |
| 18 | |
| 19 | /* Some platforms define MAXINT and/or MININT, causing conflicts */ |
| 20 | #ifdef MAXINT |
| 21 | #undef MAXINT |
| 22 | #endif |
| 23 | #ifdef MININT |
| 24 | #undef MININT |
| 25 | #endif |
| 26 | |
| 27 | /* Now we can include the original Snowball header.h */ |
| 28 | #include "snowball/libstemmer/header.h" |
| 29 | #include "snowball/libstemmer/stem_ISO_8859_1_danish.h" |
| 30 | #include "snowball/libstemmer/stem_ISO_8859_1_dutch.h" |
| 31 | #include "snowball/libstemmer/stem_ISO_8859_1_english.h" |
| 32 | #include "snowball/libstemmer/stem_ISO_8859_1_finnish.h" |
| 33 | #include "snowball/libstemmer/stem_ISO_8859_1_french.h" |
| 34 | #include "snowball/libstemmer/stem_ISO_8859_1_german.h" |
| 35 | #include "snowball/libstemmer/stem_ISO_8859_1_indonesian.h" |
| 36 | #include "snowball/libstemmer/stem_ISO_8859_1_irish.h" |
| 37 | #include "snowball/libstemmer/stem_ISO_8859_1_italian.h" |
| 38 | #include "snowball/libstemmer/stem_ISO_8859_1_norwegian.h" |
| 39 | #include "snowball/libstemmer/stem_ISO_8859_1_porter.h" |
| 40 | #include "snowball/libstemmer/stem_ISO_8859_1_portuguese.h" |
| 41 | #include "snowball/libstemmer/stem_ISO_8859_1_spanish.h" |
| 42 | #include "snowball/libstemmer/stem_ISO_8859_1_swedish.h" |
| 43 | #include "snowball/libstemmer/stem_ISO_8859_2_hungarian.h" |
| 44 | #include "snowball/libstemmer/stem_ISO_8859_2_romanian.h" |
| 45 | #include "snowball/libstemmer/stem_KOI8_R_russian.h" |
| 46 | #include "snowball/libstemmer/stem_UTF_8_arabic.h" |
| 47 | #include "snowball/libstemmer/stem_UTF_8_danish.h" |
| 48 | #include "snowball/libstemmer/stem_UTF_8_dutch.h" |
| 49 | #include "snowball/libstemmer/stem_UTF_8_english.h" |
| 50 | #include "snowball/libstemmer/stem_UTF_8_finnish.h" |
| 51 | #include "snowball/libstemmer/stem_UTF_8_french.h" |
| 52 | #include "snowball/libstemmer/stem_UTF_8_german.h" |
| 53 | #include "snowball/libstemmer/stem_UTF_8_hungarian.h" |
| 54 | #include "snowball/libstemmer/stem_UTF_8_indonesian.h" |
| 55 | #include "snowball/libstemmer/stem_UTF_8_irish.h" |
| 56 | #include "snowball/libstemmer/stem_UTF_8_italian.h" |
| 57 | #include "snowball/libstemmer/stem_UTF_8_lithuanian.h" |
| 58 | #include "snowball/libstemmer/stem_UTF_8_nepali.h" |
| 59 | #include "snowball/libstemmer/stem_UTF_8_norwegian.h" |
| 60 | #include "snowball/libstemmer/stem_UTF_8_porter.h" |
| 61 | #include "snowball/libstemmer/stem_UTF_8_portuguese.h" |
| 62 | #include "snowball/libstemmer/stem_UTF_8_romanian.h" |
| 63 | #include "snowball/libstemmer/stem_UTF_8_russian.h" |
| 64 | #include "snowball/libstemmer/stem_UTF_8_spanish.h" |
| 65 | #include "snowball/libstemmer/stem_UTF_8_swedish.h" |
| 66 | #include "snowball/libstemmer/stem_UTF_8_tamil.h" |
| 67 | #include "snowball/libstemmer/stem_UTF_8_turkish.h" |
| 68 | |
| 69 | PG_MODULE_MAGIC; |
| 70 | |
| 71 | PG_FUNCTION_INFO_V1(dsnowball_init); |
| 72 | |
| 73 | PG_FUNCTION_INFO_V1(dsnowball_lexize); |
| 74 | |
| 75 | /* List of supported modules */ |
| 76 | typedef struct stemmer_module |
| 77 | { |
| 78 | const char *name; |
| 79 | pg_enc enc; |
| 80 | struct SN_env *(*create) (void); |
| 81 | void (*close) (struct SN_env *); |
| 82 | int (*stem) (struct SN_env *); |
| 83 | } stemmer_module; |
| 84 | |
| 85 | /* Args: stemmer name, PG code for encoding, Snowball's name for encoding */ |
| 86 | #define STEMMER_MODULE(name,enc,senc) \ |
| 87 | {#name, enc, name##_##senc##_create_env, name##_##senc##_close_env, name##_##senc##_stem} |
| 88 | |
| 89 | static const stemmer_module stemmer_modules[] = |
| 90 | { |
| 91 | /* |
| 92 | * Stemmers list from Snowball distribution |
| 93 | */ |
| 94 | STEMMER_MODULE(danish, PG_LATIN1, ISO_8859_1), |
| 95 | STEMMER_MODULE(dutch, PG_LATIN1, ISO_8859_1), |
| 96 | STEMMER_MODULE(english, PG_LATIN1, ISO_8859_1), |
| 97 | STEMMER_MODULE(finnish, PG_LATIN1, ISO_8859_1), |
| 98 | STEMMER_MODULE(french, PG_LATIN1, ISO_8859_1), |
| 99 | STEMMER_MODULE(german, PG_LATIN1, ISO_8859_1), |
| 100 | STEMMER_MODULE(indonesian, PG_LATIN1, ISO_8859_1), |
| 101 | STEMMER_MODULE(irish, PG_LATIN1, ISO_8859_1), |
| 102 | STEMMER_MODULE(italian, PG_LATIN1, ISO_8859_1), |
| 103 | STEMMER_MODULE(norwegian, PG_LATIN1, ISO_8859_1), |
| 104 | STEMMER_MODULE(porter, PG_LATIN1, ISO_8859_1), |
| 105 | STEMMER_MODULE(portuguese, PG_LATIN1, ISO_8859_1), |
| 106 | STEMMER_MODULE(spanish, PG_LATIN1, ISO_8859_1), |
| 107 | STEMMER_MODULE(swedish, PG_LATIN1, ISO_8859_1), |
| 108 | STEMMER_MODULE(hungarian, PG_LATIN2, ISO_8859_2), |
| 109 | STEMMER_MODULE(romanian, PG_LATIN2, ISO_8859_2), |
| 110 | STEMMER_MODULE(russian, PG_KOI8R, KOI8_R), |
| 111 | STEMMER_MODULE(arabic, PG_UTF8, UTF_8), |
| 112 | STEMMER_MODULE(danish, PG_UTF8, UTF_8), |
| 113 | STEMMER_MODULE(dutch, PG_UTF8, UTF_8), |
| 114 | STEMMER_MODULE(english, PG_UTF8, UTF_8), |
| 115 | STEMMER_MODULE(finnish, PG_UTF8, UTF_8), |
| 116 | STEMMER_MODULE(french, PG_UTF8, UTF_8), |
| 117 | STEMMER_MODULE(german, PG_UTF8, UTF_8), |
| 118 | STEMMER_MODULE(hungarian, PG_UTF8, UTF_8), |
| 119 | STEMMER_MODULE(indonesian, PG_UTF8, UTF_8), |
| 120 | STEMMER_MODULE(irish, PG_UTF8, UTF_8), |
| 121 | STEMMER_MODULE(italian, PG_UTF8, UTF_8), |
| 122 | STEMMER_MODULE(lithuanian, PG_UTF8, UTF_8), |
| 123 | STEMMER_MODULE(nepali, PG_UTF8, UTF_8), |
| 124 | STEMMER_MODULE(norwegian, PG_UTF8, UTF_8), |
| 125 | STEMMER_MODULE(porter, PG_UTF8, UTF_8), |
| 126 | STEMMER_MODULE(portuguese, PG_UTF8, UTF_8), |
| 127 | STEMMER_MODULE(romanian, PG_UTF8, UTF_8), |
| 128 | STEMMER_MODULE(russian, PG_UTF8, UTF_8), |
| 129 | STEMMER_MODULE(spanish, PG_UTF8, UTF_8), |
| 130 | STEMMER_MODULE(swedish, PG_UTF8, UTF_8), |
| 131 | STEMMER_MODULE(tamil, PG_UTF8, UTF_8), |
| 132 | STEMMER_MODULE(turkish, PG_UTF8, UTF_8), |
| 133 | |
| 134 | /* |
| 135 | * Stemmer with PG_SQL_ASCII encoding should be valid for any server |
| 136 | * encoding |
| 137 | */ |
| 138 | STEMMER_MODULE(english, PG_SQL_ASCII, ISO_8859_1), |
| 139 | |
| 140 | {NULL, 0, NULL, NULL, NULL} /* list end marker */ |
| 141 | }; |
| 142 | |
| 143 | |
| 144 | typedef struct DictSnowball |
| 145 | { |
| 146 | struct SN_env *z; |
| 147 | StopList stoplist; |
| 148 | bool needrecode; /* needs recoding before/after call stem */ |
| 149 | int (*stem) (struct SN_env *z); |
| 150 | |
| 151 | /* |
| 152 | * snowball saves alloced memory between calls, so we should run it in our |
| 153 | * private memory context. Note, init function is executed in long lived |
| 154 | * context, so we just remember CurrentMemoryContext |
| 155 | */ |
| 156 | MemoryContext dictCtx; |
| 157 | } DictSnowball; |
| 158 | |
| 159 | |
| 160 | static void |
| 161 | locate_stem_module(DictSnowball *d, const char *lang) |
| 162 | { |
| 163 | const stemmer_module *m; |
| 164 | |
| 165 | /* |
| 166 | * First, try to find exact match of stemmer module. Stemmer with |
| 167 | * PG_SQL_ASCII encoding is treated as working with any server encoding |
| 168 | */ |
| 169 | for (m = stemmer_modules; m->name; m++) |
| 170 | { |
| 171 | if ((m->enc == PG_SQL_ASCII || m->enc == GetDatabaseEncoding()) && |
| 172 | pg_strcasecmp(m->name, lang) == 0) |
| 173 | { |
| 174 | d->stem = m->stem; |
| 175 | d->z = m->create(); |
| 176 | d->needrecode = false; |
| 177 | return; |
| 178 | } |
| 179 | } |
| 180 | |
| 181 | /* |
| 182 | * Second, try to find stemmer for needed language for UTF8 encoding. |
| 183 | */ |
| 184 | for (m = stemmer_modules; m->name; m++) |
| 185 | { |
| 186 | if (m->enc == PG_UTF8 && pg_strcasecmp(m->name, lang) == 0) |
| 187 | { |
| 188 | d->stem = m->stem; |
| 189 | d->z = m->create(); |
| 190 | d->needrecode = true; |
| 191 | return; |
| 192 | } |
| 193 | } |
| 194 | |
| 195 | ereport(ERROR, |
| 196 | (errcode(ERRCODE_UNDEFINED_OBJECT), |
| 197 | errmsg("no Snowball stemmer available for language \"%s\" and encoding \"%s\"" , |
| 198 | lang, GetDatabaseEncodingName()))); |
| 199 | } |
| 200 | |
| 201 | Datum |
| 202 | dsnowball_init(PG_FUNCTION_ARGS) |
| 203 | { |
| 204 | List *dictoptions = (List *) PG_GETARG_POINTER(0); |
| 205 | DictSnowball *d; |
| 206 | bool stoploaded = false; |
| 207 | ListCell *l; |
| 208 | |
| 209 | d = (DictSnowball *) palloc0(sizeof(DictSnowball)); |
| 210 | |
| 211 | foreach(l, dictoptions) |
| 212 | { |
| 213 | DefElem *defel = (DefElem *) lfirst(l); |
| 214 | |
| 215 | if (strcmp(defel->defname, "stopwords" ) == 0) |
| 216 | { |
| 217 | if (stoploaded) |
| 218 | ereport(ERROR, |
| 219 | (errcode(ERRCODE_INVALID_PARAMETER_VALUE), |
| 220 | errmsg("multiple StopWords parameters" ))); |
| 221 | readstoplist(defGetString(defel), &d->stoplist, lowerstr); |
| 222 | stoploaded = true; |
| 223 | } |
| 224 | else if (strcmp(defel->defname, "language" ) == 0) |
| 225 | { |
| 226 | if (d->stem) |
| 227 | ereport(ERROR, |
| 228 | (errcode(ERRCODE_INVALID_PARAMETER_VALUE), |
| 229 | errmsg("multiple Language parameters" ))); |
| 230 | locate_stem_module(d, defGetString(defel)); |
| 231 | } |
| 232 | else |
| 233 | { |
| 234 | ereport(ERROR, |
| 235 | (errcode(ERRCODE_INVALID_PARAMETER_VALUE), |
| 236 | errmsg("unrecognized Snowball parameter: \"%s\"" , |
| 237 | defel->defname))); |
| 238 | } |
| 239 | } |
| 240 | |
| 241 | if (!d->stem) |
| 242 | ereport(ERROR, |
| 243 | (errcode(ERRCODE_INVALID_PARAMETER_VALUE), |
| 244 | errmsg("missing Language parameter" ))); |
| 245 | |
| 246 | d->dictCtx = CurrentMemoryContext; |
| 247 | |
| 248 | PG_RETURN_POINTER(d); |
| 249 | } |
| 250 | |
| 251 | Datum |
| 252 | dsnowball_lexize(PG_FUNCTION_ARGS) |
| 253 | { |
| 254 | DictSnowball *d = (DictSnowball *) PG_GETARG_POINTER(0); |
| 255 | char *in = (char *) PG_GETARG_POINTER(1); |
| 256 | int32 len = PG_GETARG_INT32(2); |
| 257 | char *txt = lowerstr_with_len(in, len); |
| 258 | TSLexeme *res = palloc0(sizeof(TSLexeme) * 2); |
| 259 | |
| 260 | if (*txt == '\0' || searchstoplist(&(d->stoplist), txt)) |
| 261 | { |
| 262 | pfree(txt); |
| 263 | } |
| 264 | else |
| 265 | { |
| 266 | MemoryContext saveCtx; |
| 267 | |
| 268 | /* |
| 269 | * recode to utf8 if stemmer is utf8 and doesn't match server encoding |
| 270 | */ |
| 271 | if (d->needrecode) |
| 272 | { |
| 273 | char *recoded; |
| 274 | |
| 275 | recoded = pg_server_to_any(txt, strlen(txt), PG_UTF8); |
| 276 | if (recoded != txt) |
| 277 | { |
| 278 | pfree(txt); |
| 279 | txt = recoded; |
| 280 | } |
| 281 | } |
| 282 | |
| 283 | /* see comment about d->dictCtx */ |
| 284 | saveCtx = MemoryContextSwitchTo(d->dictCtx); |
| 285 | SN_set_current(d->z, strlen(txt), (symbol *) txt); |
| 286 | d->stem(d->z); |
| 287 | MemoryContextSwitchTo(saveCtx); |
| 288 | |
| 289 | if (d->z->p && d->z->l) |
| 290 | { |
| 291 | txt = repalloc(txt, d->z->l + 1); |
| 292 | memcpy(txt, d->z->p, d->z->l); |
| 293 | txt[d->z->l] = '\0'; |
| 294 | } |
| 295 | |
| 296 | /* back recode if needed */ |
| 297 | if (d->needrecode) |
| 298 | { |
| 299 | char *recoded; |
| 300 | |
| 301 | recoded = pg_any_to_server(txt, strlen(txt), PG_UTF8); |
| 302 | if (recoded != txt) |
| 303 | { |
| 304 | pfree(txt); |
| 305 | txt = recoded; |
| 306 | } |
| 307 | } |
| 308 | |
| 309 | res->lexeme = txt; |
| 310 | } |
| 311 | |
| 312 | PG_RETURN_POINTER(res); |
| 313 | } |
| 314 | |