PostgreSQL Source Code  git master
dict_snowball.c File Reference
#include "postgres.h"
#include "commands/defrem.h"
#include "tsearch/ts_locale.h"
#include "tsearch/ts_utils.h"
#include "snowball/libstemmer/header.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_hungarian.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_romanian.h"
#include "snowball/libstemmer/stem_KOI8_R_russian.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_hungarian.h"
#include "snowball/libstemmer/stem_UTF_8_italian.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_spanish.h"
#include "snowball/libstemmer/stem_UTF_8_swedish.h"
#include "snowball/libstemmer/stem_UTF_8_turkish.h"
Include dependency graph for dict_snowball.c:

Go to the source code of this file.

Data Structures

struct  stemmer_module
 
struct  DictSnowball
 

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 []
 

Typedef Documentation

◆ DictSnowball

◆ stemmer_module

Function Documentation

◆ dsnowball_init()

Datum dsnowball_init ( PG_FUNCTION_ARGS  )

Definition at line 182 of file dict_snowball.c.

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

183 {
184  List *dictoptions = (List *) PG_GETARG_POINTER(0);
185  DictSnowball *d;
186  bool stoploaded = false;
187  ListCell *l;
188 
189  d = (DictSnowball *) palloc0(sizeof(DictSnowball));
190 
191  foreach(l, dictoptions)
192  {
193  DefElem *defel = (DefElem *) lfirst(l);
194 
195  if (pg_strcasecmp("StopWords", defel->defname) == 0)
196  {
197  if (stoploaded)
198  ereport(ERROR,
199  (errcode(ERRCODE_INVALID_PARAMETER_VALUE),
200  errmsg("multiple StopWords parameters")));
202  stoploaded = true;
203  }
204  else if (pg_strcasecmp("Language", defel->defname) == 0)
205  {
206  if (d->stem)
207  ereport(ERROR,
208  (errcode(ERRCODE_INVALID_PARAMETER_VALUE),
209  errmsg("multiple Language parameters")));
210  locate_stem_module(d, defGetString(defel));
211  }
212  else
213  {
214  ereport(ERROR,
215  (errcode(ERRCODE_INVALID_PARAMETER_VALUE),
216  errmsg("unrecognized Snowball parameter: \"%s\"",
217  defel->defname)));
218  }
219  }
220 
221  if (!d->stem)
222  ereport(ERROR,
223  (errcode(ERRCODE_INVALID_PARAMETER_VALUE),
224  errmsg("missing Language parameter")));
225 
227 
229 }
#define PG_RETURN_POINTER(x)
Definition: fmgr.h:321
StopList stoplist
int errcode(int sqlerrcode)
Definition: elog.c:575
#define PG_GETARG_POINTER(n)
Definition: fmgr.h:241
char * lowerstr(const char *str)
Definition: ts_locale.c:228
int(* stem)(struct SN_env *z)
int pg_strcasecmp(const char *s1, const char *s2)
Definition: pgstrcasecmp.c:36
#define ERROR
Definition: elog.h:43
char * defGetString(DefElem *def)
Definition: define.c:49
MemoryContext CurrentMemoryContext
Definition: mcxt.c:37
#define ereport(elevel, rest)
Definition: elog.h:122
void readstoplist(const char *fname, StopList *s, char *(*wordop)(const char *))
Definition: ts_utils.c:68
void * palloc0(Size size)
Definition: mcxt.c:877
static void locate_stem_module(DictSnowball *d, const char *lang)
#define lfirst(lc)
Definition: pg_list.h:106
int errmsg(const char *fmt,...)
Definition: elog.c:797
MemoryContext dictCtx
char * defname
Definition: parsenodes.h:719
Definition: pg_list.h:45

◆ dsnowball_lexize()

Datum dsnowball_lexize ( PG_FUNCTION_ARGS  )

Definition at line 232 of file dict_snowball.c.

References DictSnowball::dictCtx, SN_env::l, TSLexeme::lexeme, lowerstr_with_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(), searchstoplist(), SN_set_current(), DictSnowball::stem, DictSnowball::stoplist, and DictSnowball::z.

233 {
235  char *in = (char *) PG_GETARG_POINTER(1);
236  int32 len = PG_GETARG_INT32(2);
237  char *txt = lowerstr_with_len(in, len);
238  TSLexeme *res = palloc0(sizeof(TSLexeme) * 2);
239 
240  if (*txt == '\0' || searchstoplist(&(d->stoplist), txt))
241  {
242  pfree(txt);
243  }
244  else
245  {
246  MemoryContext saveCtx;
247 
248  /*
249  * recode to utf8 if stemmer is utf8 and doesn't match server encoding
250  */
251  if (d->needrecode)
252  {
253  char *recoded;
254 
255  recoded = pg_server_to_any(txt, strlen(txt), PG_UTF8);
256  if (recoded != txt)
257  {
258  pfree(txt);
259  txt = recoded;
260  }
261  }
262 
263  /* see comment about d->dictCtx */
264  saveCtx = MemoryContextSwitchTo(d->dictCtx);
265  SN_set_current(d->z, strlen(txt), (symbol *) txt);
266  d->stem(d->z);
267  MemoryContextSwitchTo(saveCtx);
268 
269  if (d->z->p && d->z->l)
270  {
271  txt = repalloc(txt, d->z->l + 1);
272  memcpy(txt, d->z->p, d->z->l);
273  txt[d->z->l] = '\0';
274  }
275 
276  /* back recode if needed */
277  if (d->needrecode)
278  {
279  char *recoded;
280 
281  recoded = pg_any_to_server(txt, strlen(txt), PG_UTF8);
282  if (recoded != txt)
283  {
284  pfree(txt);
285  txt = recoded;
286  }
287  }
288 
289  res->lexeme = txt;
290  }
291 
292  PG_RETURN_POINTER(res);
293 }
#define PG_RETURN_POINTER(x)
Definition: fmgr.h:321
#define PG_GETARG_INT32(n)
Definition: fmgr.h:234
StopList stoplist
symbol * p
Definition: api.h:15
int SN_set_current(struct SN_env *z, int size, const symbol *s)
Definition: api.c:58
char * lowerstr_with_len(const char *str, int len)
Definition: ts_locale.c:241
static MemoryContext MemoryContextSwitchTo(MemoryContext context)
Definition: palloc.h:109
#define PG_GETARG_POINTER(n)
Definition: fmgr.h:241
int(* stem)(struct SN_env *z)
char * pg_server_to_any(const char *s, int len, int encoding)
Definition: mbutils.c:634
signed int int32
Definition: c.h:284
void pfree(void *pointer)
Definition: mcxt.c:949
char * lexeme
Definition: ts_public.h:111
void * palloc0(Size size)
Definition: mcxt.c:877
void * repalloc(void *pointer, Size size)
Definition: mcxt.c:962
struct SN_env * z
bool searchstoplist(StopList *s, char *key)
Definition: ts_utils.c:141
MemoryContext dictCtx
int l
Definition: api.h:16
char * pg_any_to_server(const char *s, int len, int encoding)
Definition: mbutils.c:561
unsigned char symbol
Definition: api.h:2

◆ locate_stem_module()

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

Definition at line 141 of file dict_snowball.c.

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, and DictSnowball::z.

Referenced by dsnowball_init().

142 {
143  const stemmer_module *m;
144 
145  /*
146  * First, try to find exact match of stemmer module. Stemmer with
147  * PG_SQL_ASCII encoding is treated as working with any server encoding
148  */
149  for (m = stemmer_modules; m->name; m++)
150  {
151  if ((m->enc == PG_SQL_ASCII || m->enc == GetDatabaseEncoding()) &&
152  pg_strcasecmp(m->name, lang) == 0)
153  {
154  d->stem = m->stem;
155  d->z = m->create();
156  d->needrecode = false;
157  return;
158  }
159  }
160 
161  /*
162  * Second, try to find stemmer for needed language for UTF8 encoding.
163  */
164  for (m = stemmer_modules; m->name; m++)
165  {
166  if (m->enc == PG_UTF8 && pg_strcasecmp(m->name, lang) == 0)
167  {
168  d->stem = m->stem;
169  d->z = m->create();
170  d->needrecode = true;
171  return;
172  }
173  }
174 
175  ereport(ERROR,
176  (errcode(ERRCODE_UNDEFINED_OBJECT),
177  errmsg("no Snowball stemmer available for language \"%s\" and encoding \"%s\"",
178  lang, GetDatabaseEncodingName())));
179 }
int errcode(int sqlerrcode)
Definition: elog.c:575
int(* stem)(struct SN_env *z)
int pg_strcasecmp(const char *s1, const char *s2)
Definition: pgstrcasecmp.c:36
#define ERROR
Definition: elog.h:43
struct SN_env *(* create)(void)
Definition: dict_snowball.c:72
#define ereport(elevel, rest)
Definition: elog.h:122
int GetDatabaseEncoding(void)
Definition: mbutils.c:1004
const char * GetDatabaseEncodingName(void)
Definition: mbutils.c:1010
static const stemmer_module stemmer_modules[]
Definition: dict_snowball.c:77
int(* stem)(struct SN_env *)
Definition: dict_snowball.c:74
struct SN_env * z
int errmsg(const char *fmt,...)
Definition: elog.c:797
const char * name
Definition: dict_snowball.c:70

◆ 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 61 of file dict_snowball.c.

◆ stemmer_modules

const stemmer_module stemmer_modules[]
static

Definition at line 77 of file dict_snowball.c.