PostgreSQL Source Code git master
Data Structures | Macros | Typedefs | Functions | Variables
dict_snowball.c File Reference
#include "postgres.h"
#include "catalog/pg_collation_d.h"
#include "commands/defrem.h"
#include "mb/pg_wchar.h"
#include "tsearch/ts_public.h"
#include "utils/formatting.h"
#include "snowball/libstemmer/header.h"
#include "snowball/libstemmer/stem_ISO_8859_1_basque.h"
#include "snowball/libstemmer/stem_ISO_8859_1_catalan.h"
#include "snowball/libstemmer/stem_ISO_8859_1_danish.h"
#include "snowball/libstemmer/stem_ISO_8859_1_dutch.h"
#include "snowball/libstemmer/stem_ISO_8859_1_english.h"
#include "snowball/libstemmer/stem_ISO_8859_1_finnish.h"
#include "snowball/libstemmer/stem_ISO_8859_1_french.h"
#include "snowball/libstemmer/stem_ISO_8859_1_german.h"
#include "snowball/libstemmer/stem_ISO_8859_1_indonesian.h"
#include "snowball/libstemmer/stem_ISO_8859_1_irish.h"
#include "snowball/libstemmer/stem_ISO_8859_1_italian.h"
#include "snowball/libstemmer/stem_ISO_8859_1_norwegian.h"
#include "snowball/libstemmer/stem_ISO_8859_1_porter.h"
#include "snowball/libstemmer/stem_ISO_8859_1_portuguese.h"
#include "snowball/libstemmer/stem_ISO_8859_1_spanish.h"
#include "snowball/libstemmer/stem_ISO_8859_1_swedish.h"
#include "snowball/libstemmer/stem_ISO_8859_2_hungarian.h"
#include "snowball/libstemmer/stem_ISO_8859_2_romanian.h"
#include "snowball/libstemmer/stem_KOI8_R_russian.h"
#include "snowball/libstemmer/stem_UTF_8_arabic.h"
#include "snowball/libstemmer/stem_UTF_8_armenian.h"
#include "snowball/libstemmer/stem_UTF_8_basque.h"
#include "snowball/libstemmer/stem_UTF_8_catalan.h"
#include "snowball/libstemmer/stem_UTF_8_danish.h"
#include "snowball/libstemmer/stem_UTF_8_dutch.h"
#include "snowball/libstemmer/stem_UTF_8_english.h"
#include "snowball/libstemmer/stem_UTF_8_finnish.h"
#include "snowball/libstemmer/stem_UTF_8_french.h"
#include "snowball/libstemmer/stem_UTF_8_german.h"
#include "snowball/libstemmer/stem_UTF_8_greek.h"
#include "snowball/libstemmer/stem_UTF_8_hindi.h"
#include "snowball/libstemmer/stem_UTF_8_hungarian.h"
#include "snowball/libstemmer/stem_UTF_8_indonesian.h"
#include "snowball/libstemmer/stem_UTF_8_irish.h"
#include "snowball/libstemmer/stem_UTF_8_italian.h"
#include "snowball/libstemmer/stem_UTF_8_lithuanian.h"
#include "snowball/libstemmer/stem_UTF_8_nepali.h"
#include "snowball/libstemmer/stem_UTF_8_norwegian.h"
#include "snowball/libstemmer/stem_UTF_8_porter.h"
#include "snowball/libstemmer/stem_UTF_8_portuguese.h"
#include "snowball/libstemmer/stem_UTF_8_romanian.h"
#include "snowball/libstemmer/stem_UTF_8_russian.h"
#include "snowball/libstemmer/stem_UTF_8_serbian.h"
#include "snowball/libstemmer/stem_UTF_8_spanish.h"
#include "snowball/libstemmer/stem_UTF_8_swedish.h"
#include "snowball/libstemmer/stem_UTF_8_tamil.h"
#include "snowball/libstemmer/stem_UTF_8_turkish.h"
#include "snowball/libstemmer/stem_UTF_8_yiddish.h"
Include dependency graph for dict_snowball.c:

Go to the source code of this file.

Data Structures

struct  stemmer_module
 
struct  DictSnowball
 

Macros

#define STEMMER_MODULE(name, enc, senc)    {#name, enc, name##_##senc##_create_env, name##_##senc##_close_env, name##_##senc##_stem}
 

Typedefs

typedef struct stemmer_module stemmer_module
 
typedef struct DictSnowball DictSnowball
 

Functions

 PG_FUNCTION_INFO_V1 (dsnowball_init)
 
 PG_FUNCTION_INFO_V1 (dsnowball_lexize)
 
static void locate_stem_module (DictSnowball *d, const char *lang)
 
Datum dsnowball_init (PG_FUNCTION_ARGS)
 
Datum dsnowball_lexize (PG_FUNCTION_ARGS)
 

Variables

 PG_MODULE_MAGIC
 
static const stemmer_module stemmer_modules []
 

Macro Definition Documentation

◆ STEMMER_MODULE

#define STEMMER_MODULE (   name,
  enc,
  senc 
)     {#name, enc, name##_##senc##_create_env, name##_##senc##_close_env, name##_##senc##_stem}

Definition at line 97 of file dict_snowball.c.

Typedef Documentation

◆ DictSnowball

typedef struct DictSnowball DictSnowball

◆ stemmer_module

typedef struct stemmer_module stemmer_module

Function Documentation

◆ dsnowball_init()

Datum dsnowball_init ( PG_FUNCTION_ARGS  )

Definition at line 222 of file dict_snowball.c.

223{
224 List *dictoptions = (List *) PG_GETARG_POINTER(0);
225 DictSnowball *d;
226 bool stoploaded = false;
227 ListCell *l;
228
229 d = (DictSnowball *) palloc0(sizeof(DictSnowball));
230
231 foreach(l, dictoptions)
232 {
233 DefElem *defel = (DefElem *) lfirst(l);
234
235 if (strcmp(defel->defname, "stopwords") == 0)
236 {
237 if (stoploaded)
238 ereport(ERROR,
239 (errcode(ERRCODE_INVALID_PARAMETER_VALUE),
240 errmsg("multiple StopWords parameters")));
241 readstoplist(defGetString(defel), &d->stoplist, str_tolower);
242 stoploaded = true;
243 }
244 else if (strcmp(defel->defname, "language") == 0)
245 {
246 if (d->stem)
247 ereport(ERROR,
248 (errcode(ERRCODE_INVALID_PARAMETER_VALUE),
249 errmsg("multiple Language parameters")));
250 locate_stem_module(d, defGetString(defel));
251 }
252 else
253 {
254 ereport(ERROR,
255 (errcode(ERRCODE_INVALID_PARAMETER_VALUE),
256 errmsg("unrecognized Snowball parameter: \"%s\"",
257 defel->defname)));
258 }
259 }
260
261 if (!d->stem)
262 ereport(ERROR,
263 (errcode(ERRCODE_INVALID_PARAMETER_VALUE),
264 errmsg("missing Language parameter")));
265
266 d->dictCtx = CurrentMemoryContext;
267
268 PG_RETURN_POINTER(d);
269}
defGetString
char * defGetString(DefElem *def)
Definition: define.c:35
locate_stem_module
static void locate_stem_module(DictSnowball *d, const char *lang)
Definition: dict_snowball.c:181
errcode
int errcode(int sqlerrcode)
Definition: elog.c:853
errmsg
int errmsg(const char *fmt,...)
Definition: elog.c:1070
ERROR
#define ERROR
Definition: elog.h:39
ereport
#define ereport(elevel,...)
Definition: elog.h:149
PG_GETARG_POINTER
#define PG_GETARG_POINTER(n)
Definition: fmgr.h:276
PG_RETURN_POINTER
#define PG_RETURN_POINTER(x)
Definition: fmgr.h:361
str_tolower
char * str_tolower(const char *buff, size_t nbytes, Oid collid)
Definition: formatting.c:1637
palloc0
void * palloc0(Size size)
Definition: mcxt.c:1347
CurrentMemoryContext
MemoryContext CurrentMemoryContext
Definition: mcxt.c:143
lfirst
#define lfirst(lc)
Definition: pg_list.h:172
DefElem::defname
char * defname
Definition: parsenodes.h:826
DictSnowball::dictCtx
MemoryContext dictCtx
Definition: dict_snowball.c:176
DictSnowball::stoplist
StopList stoplist
Definition: dict_snowball.c:167
DictSnowball::stem
int(* stem)(struct SN_env *z)
Definition: dict_snowball.c:169
List
Definition: pg_list.h:54
readstoplist
void readstoplist(const char *fname, StopList *s, char *(*wordop)(const char *, size_t, Oid))
Definition: ts_utils.c:69

References CurrentMemoryContext, defGetString(), DefElem::defname, DictSnowball::dictCtx, ereport, errcode(), errmsg(), ERROR, SN_env::l, lfirst, locate_stem_module(), palloc0(), PG_GETARG_POINTER, PG_RETURN_POINTER, readstoplist(), DictSnowball::stem, DictSnowball::stoplist, and str_tolower().

◆ dsnowball_lexize()

Datum dsnowball_lexize ( PG_FUNCTION_ARGS  )

Definition at line 272 of file dict_snowball.c.

273{
274 DictSnowball *d = (DictSnowball *) PG_GETARG_POINTER(0);
275 char *in = (char *) PG_GETARG_POINTER(1);
276 int32 len = PG_GETARG_INT32(2);
277 char *txt = str_tolower(in, len, DEFAULT_COLLATION_OID);
278 TSLexeme *res = palloc0(sizeof(TSLexeme) * 2);
279
280 /*
281 * Do not pass strings exceeding 1000 bytes to the stemmer, as they're
282 * surely not words in any human language. This restriction avoids
283 * wasting cycles on stuff like base64-encoded data, and it protects us
284 * against possible inefficiency or misbehavior in the stemmer. (For
285 * example, the Turkish stemmer has an indefinite recursion, so it can
286 * crash on long-enough strings.) However, Snowball dictionaries are
287 * defined to recognize all strings, so we can't reject the string as an
288 * unknown word.
289 */
290 if (len > 1000)
291 {
292 /* return the lexeme lowercased, but otherwise unmodified */
293 res->lexeme = txt;
294 }
295 else if (*txt == '\0' || searchstoplist(&(d->stoplist), txt))
296 {
297 /* empty or stopword, so report as stopword */
298 pfree(txt);
299 }
300 else
301 {
302 MemoryContext saveCtx;
303
304 /*
305 * recode to utf8 if stemmer is utf8 and doesn't match server encoding
306 */
307 if (d->needrecode)
308 {
309 char *recoded;
310
311 recoded = pg_server_to_any(txt, strlen(txt), PG_UTF8);
312 if (recoded != txt)
313 {
314 pfree(txt);
315 txt = recoded;
316 }
317 }
318
319 /* see comment about d->dictCtx */
320 saveCtx = MemoryContextSwitchTo(d->dictCtx);
321 SN_set_current(d->z, strlen(txt), (symbol *) txt);
322 d->stem(d->z);
323 MemoryContextSwitchTo(saveCtx);
324
325 if (d->z->p && d->z->l)
326 {
327 txt = repalloc(txt, d->z->l + 1);
328 memcpy(txt, d->z->p, d->z->l);
329 txt[d->z->l] = '\0';
330 }
331
332 /* back recode if needed */
333 if (d->needrecode)
334 {
335 char *recoded;
336
337 recoded = pg_any_to_server(txt, strlen(txt), PG_UTF8);
338 if (recoded != txt)
339 {
340 pfree(txt);
341 txt = recoded;
342 }
343 }
344
345 res->lexeme = txt;
346 }
347
348 PG_RETURN_POINTER(res);
349}
SN_set_current
int SN_set_current(struct SN_env *z, int size, const symbol *s)
Definition: api.c:51
symbol
unsigned char symbol
Definition: api.h:2
int32
int32_t int32
Definition: c.h:484
PG_GETARG_INT32
#define PG_GETARG_INT32(n)
Definition: fmgr.h:269
pg_any_to_server
char * pg_any_to_server(const char *s, int len, int encoding)
Definition: mbutils.c:676
pg_server_to_any
char * pg_server_to_any(const char *s, int len, int encoding)
Definition: mbutils.c:749
repalloc
void * repalloc(void *pointer, Size size)
Definition: mcxt.c:1541
pfree
void pfree(void *pointer)
Definition: mcxt.c:1521
MemoryContextSwitchTo
static MemoryContext MemoryContextSwitchTo(MemoryContext context)
Definition: palloc.h:124
len
const void size_t len
Definition: pg_crc32c_sse42.c:25
PG_UTF8
@ PG_UTF8
Definition: pg_wchar.h:232
DictSnowball::z
struct SN_env * z
Definition: dict_snowball.c:166
SN_env::p
symbol * p
Definition: api.h:15
SN_env::l
int l
Definition: api.h:16
searchstoplist
bool searchstoplist(StopList *s, char *key)
Definition: ts_utils.c:141

References DictSnowball::dictCtx, SN_env::l, len, MemoryContextSwitchTo(), DictSnowball::needrecode, SN_env::p, palloc0(), pfree(), pg_any_to_server(), PG_GETARG_INT32, PG_GETARG_POINTER, PG_RETURN_POINTER, pg_server_to_any(), PG_UTF8, repalloc(), res, searchstoplist(), SN_set_current(), DictSnowball::stem, DictSnowball::stoplist, str_tolower(), and DictSnowball::z.

◆ locate_stem_module()

static void locate_stem_module ( DictSnowball *  d,
const char *  lang 
)
static

Definition at line 181 of file dict_snowball.c.

182{
183 const stemmer_module *m;
184
185 /*
186 * First, try to find exact match of stemmer module. Stemmer with
187 * PG_SQL_ASCII encoding is treated as working with any server encoding
188 */
189 for (m = stemmer_modules; m->name; m++)
190 {
191 if ((m->enc == PG_SQL_ASCII || m->enc == GetDatabaseEncoding()) &&
192 pg_strcasecmp(m->name, lang) == 0)
193 {
194 d->stem = m->stem;
195 d->z = m->create();
196 d->needrecode = false;
197 return;
198 }
199 }
200
201 /*
202 * Second, try to find stemmer for needed language for UTF8 encoding.
203 */
204 for (m = stemmer_modules; m->name; m++)
205 {
206 if (m->enc == PG_UTF8 && pg_strcasecmp(m->name, lang) == 0)
207 {
208 d->stem = m->stem;
209 d->z = m->create();
210 d->needrecode = true;
211 return;
212 }
213 }
214
215 ereport(ERROR,
216 (errcode(ERRCODE_UNDEFINED_OBJECT),
217 errmsg("no Snowball stemmer available for language \"%s\" and encoding \"%s\"",
218 lang, GetDatabaseEncodingName())));
219}
stemmer_modules
static const stemmer_module stemmer_modules[]
Definition: dict_snowball.c:100
GetDatabaseEncoding
int GetDatabaseEncoding(void)
Definition: mbutils.c:1261
GetDatabaseEncodingName
const char * GetDatabaseEncodingName(void)
Definition: mbutils.c:1267
PG_SQL_ASCII
@ PG_SQL_ASCII
Definition: pg_wchar.h:226
pg_strcasecmp
int pg_strcasecmp(const char *s1, const char *s2)
Definition: pgstrcasecmp.c:36
stemmer_module
Definition: dict_snowball.c:88
stemmer_module::create
struct SN_env *(* create)(void)
Definition: dict_snowball.c:91
stemmer_module::name
const char * name
Definition: dict_snowball.c:89
stemmer_module::enc
pg_enc enc
Definition: dict_snowball.c:90
stemmer_module::stem
int(* stem)(struct SN_env *)
Definition: dict_snowball.c:93

References stemmer_module::create, stemmer_module::enc, ereport, errcode(), errmsg(), ERROR, GetDatabaseEncoding(), GetDatabaseEncodingName(), stemmer_module::name, DictSnowball::needrecode, PG_SQL_ASCII, pg_strcasecmp(), PG_UTF8, stemmer_module::stem, DictSnowball::stem, stemmer_modules, and DictSnowball::z.

Referenced by dsnowball_init().

◆ PG_FUNCTION_INFO_V1() [1/2]

PG_FUNCTION_INFO_V1 ( dsnowball_init  )

◆ PG_FUNCTION_INFO_V1() [2/2]

PG_FUNCTION_INFO_V1 ( dsnowball_lexize  )

Variable Documentation

◆ PG_MODULE_MAGIC

PG_MODULE_MAGIC

Definition at line 80 of file dict_snowball.c.

◆ stemmer_modules

const stemmer_module stemmer_modules[]
static

Definition at line 100 of file dict_snowball.c.

Referenced by locate_stem_module().