aurelian-ruby-stemmer 0.5.4
Sign up to get free protection for your applications and to get access to all the features.
- data/MIT-LICENSE +21 -0
- data/Manifest +87 -0
- data/README.textile +66 -0
- data/Rakefile +39 -0
- data/extconf.rb +13 -0
- data/libstemmer_c/MANIFEST +72 -0
- data/libstemmer_c/Makefile +9 -0
- data/libstemmer_c/README +125 -0
- data/libstemmer_c/examples/stemwords.c +209 -0
- data/libstemmer_c/include/libstemmer.h +79 -0
- data/libstemmer_c/libstemmer/libstemmer.c +93 -0
- data/libstemmer_c/libstemmer/libstemmer_utf8.c +93 -0
- data/libstemmer_c/libstemmer/modules.h +190 -0
- data/libstemmer_c/libstemmer/modules.txt +50 -0
- data/libstemmer_c/libstemmer/modules_utf8.h +121 -0
- data/libstemmer_c/libstemmer/modules_utf8.txt +49 -0
- data/libstemmer_c/mkinc.mak +82 -0
- data/libstemmer_c/mkinc_utf8.mak +52 -0
- data/libstemmer_c/runtime/api.c +66 -0
- data/libstemmer_c/runtime/api.h +26 -0
- data/libstemmer_c/runtime/header.h +58 -0
- data/libstemmer_c/runtime/utilities.c +478 -0
- data/libstemmer_c/src_c/stem_ISO_8859_1_danish.c +337 -0
- data/libstemmer_c/src_c/stem_ISO_8859_1_danish.h +16 -0
- data/libstemmer_c/src_c/stem_ISO_8859_1_dutch.c +624 -0
- data/libstemmer_c/src_c/stem_ISO_8859_1_dutch.h +16 -0
- data/libstemmer_c/src_c/stem_ISO_8859_1_english.c +1117 -0
- data/libstemmer_c/src_c/stem_ISO_8859_1_english.h +16 -0
- data/libstemmer_c/src_c/stem_ISO_8859_1_finnish.c +762 -0
- data/libstemmer_c/src_c/stem_ISO_8859_1_finnish.h +16 -0
- data/libstemmer_c/src_c/stem_ISO_8859_1_french.c +1246 -0
- data/libstemmer_c/src_c/stem_ISO_8859_1_french.h +16 -0
- data/libstemmer_c/src_c/stem_ISO_8859_1_german.c +503 -0
- data/libstemmer_c/src_c/stem_ISO_8859_1_german.h +16 -0
- data/libstemmer_c/src_c/stem_ISO_8859_1_hungarian.c +1230 -0
- data/libstemmer_c/src_c/stem_ISO_8859_1_hungarian.h +16 -0
- data/libstemmer_c/src_c/stem_ISO_8859_1_italian.c +1065 -0
- data/libstemmer_c/src_c/stem_ISO_8859_1_italian.h +16 -0
- data/libstemmer_c/src_c/stem_ISO_8859_1_norwegian.c +297 -0
- data/libstemmer_c/src_c/stem_ISO_8859_1_norwegian.h +16 -0
- data/libstemmer_c/src_c/stem_ISO_8859_1_porter.c +749 -0
- data/libstemmer_c/src_c/stem_ISO_8859_1_porter.h +16 -0
- data/libstemmer_c/src_c/stem_ISO_8859_1_portuguese.c +1017 -0
- data/libstemmer_c/src_c/stem_ISO_8859_1_portuguese.h +16 -0
- data/libstemmer_c/src_c/stem_ISO_8859_1_spanish.c +1093 -0
- data/libstemmer_c/src_c/stem_ISO_8859_1_spanish.h +16 -0
- data/libstemmer_c/src_c/stem_ISO_8859_1_swedish.c +307 -0
- data/libstemmer_c/src_c/stem_ISO_8859_1_swedish.h +16 -0
- data/libstemmer_c/src_c/stem_ISO_8859_2_romanian.c +998 -0
- data/libstemmer_c/src_c/stem_ISO_8859_2_romanian.h +16 -0
- data/libstemmer_c/src_c/stem_KOI8_R_russian.c +700 -0
- data/libstemmer_c/src_c/stem_KOI8_R_russian.h +16 -0
- data/libstemmer_c/src_c/stem_UTF_8_danish.c +339 -0
- data/libstemmer_c/src_c/stem_UTF_8_danish.h +16 -0
- data/libstemmer_c/src_c/stem_UTF_8_dutch.c +634 -0
- data/libstemmer_c/src_c/stem_UTF_8_dutch.h +16 -0
- data/libstemmer_c/src_c/stem_UTF_8_english.c +1125 -0
- data/libstemmer_c/src_c/stem_UTF_8_english.h +16 -0
- data/libstemmer_c/src_c/stem_UTF_8_finnish.c +768 -0
- data/libstemmer_c/src_c/stem_UTF_8_finnish.h +16 -0
- data/libstemmer_c/src_c/stem_UTF_8_french.c +1256 -0
- data/libstemmer_c/src_c/stem_UTF_8_french.h +16 -0
- data/libstemmer_c/src_c/stem_UTF_8_german.c +509 -0
- data/libstemmer_c/src_c/stem_UTF_8_german.h +16 -0
- data/libstemmer_c/src_c/stem_UTF_8_hungarian.c +1234 -0
- data/libstemmer_c/src_c/stem_UTF_8_hungarian.h +16 -0
- data/libstemmer_c/src_c/stem_UTF_8_italian.c +1073 -0
- data/libstemmer_c/src_c/stem_UTF_8_italian.h +16 -0
- data/libstemmer_c/src_c/stem_UTF_8_norwegian.c +299 -0
- data/libstemmer_c/src_c/stem_UTF_8_norwegian.h +16 -0
- data/libstemmer_c/src_c/stem_UTF_8_porter.c +755 -0
- data/libstemmer_c/src_c/stem_UTF_8_porter.h +16 -0
- data/libstemmer_c/src_c/stem_UTF_8_portuguese.c +1023 -0
- data/libstemmer_c/src_c/stem_UTF_8_portuguese.h +16 -0
- data/libstemmer_c/src_c/stem_UTF_8_romanian.c +1004 -0
- data/libstemmer_c/src_c/stem_UTF_8_romanian.h +16 -0
- data/libstemmer_c/src_c/stem_UTF_8_russian.c +694 -0
- data/libstemmer_c/src_c/stem_UTF_8_russian.h +16 -0
- data/libstemmer_c/src_c/stem_UTF_8_spanish.c +1097 -0
- data/libstemmer_c/src_c/stem_UTF_8_spanish.h +16 -0
- data/libstemmer_c/src_c/stem_UTF_8_swedish.c +309 -0
- data/libstemmer_c/src_c/stem_UTF_8_swedish.h +16 -0
- data/libstemmer_c/src_c/stem_UTF_8_turkish.c +2205 -0
- data/libstemmer_c/src_c/stem_UTF_8_turkish.h +16 -0
- data/ruby-stemmer.c +142 -0
- data/ruby-stemmer.gemspec +31 -0
- data/test.rb +26 -0
- metadata +224 -0
@@ -0,0 +1,79 @@
|
|
1
|
+
|
2
|
+
/* Make header file work when included from C++ */
|
3
|
+
#ifdef __cplusplus
|
4
|
+
extern "C" {
|
5
|
+
#endif
|
6
|
+
|
7
|
+
struct sb_stemmer;
|
8
|
+
typedef unsigned char sb_symbol;
|
9
|
+
|
10
|
+
/* FIXME - should be able to get a version number for each stemming
|
11
|
+
* algorithm (which will be incremented each time the output changes). */
|
12
|
+
|
13
|
+
/** Returns an array of the names of the available stemming algorithms.
|
14
|
+
* Note that these are the canonical names - aliases (ie, other names for
|
15
|
+
* the same algorithm) will not be included in the list.
|
16
|
+
* The list is terminated with a null pointer.
|
17
|
+
*
|
18
|
+
* The list must not be modified in any way.
|
19
|
+
*/
|
20
|
+
const char ** sb_stemmer_list(void);
|
21
|
+
|
22
|
+
/** Create a new stemmer object, using the specified algorithm, for the
|
23
|
+
* specified character encoding.
|
24
|
+
*
|
25
|
+
* All algorithms will usually be available in UTF-8, but may also be
|
26
|
+
* available in other character encodings.
|
27
|
+
*
|
28
|
+
* @param algorithm The algorithm name. This is either the english
|
29
|
+
* name of the algorithm, or the 2 or 3 letter ISO 639 codes for the
|
30
|
+
* language. Note that case is significant in this parameter - the
|
31
|
+
* value should be supplied in lower case.
|
32
|
+
*
|
33
|
+
* @param charenc The character encoding. NULL may be passed as
|
34
|
+
* this value, in which case UTF-8 encoding will be assumed. Otherwise,
|
35
|
+
* the argument may be one of "UTF_8", "ISO_8859_1" (ie, Latin 1),
|
36
|
+
* "CP850" (ie, MS-DOS Latin 1) or "KOI8_R" (Russian). Note that
|
37
|
+
* case is significant in this parameter.
|
38
|
+
*
|
39
|
+
* @return NULL if the specified algorithm is not recognised, or the
|
40
|
+
* algorithm is not available for the requested encoding. Otherwise,
|
41
|
+
* returns a pointer to a newly created stemmer for the requested algorithm.
|
42
|
+
* The returned pointer must be deleted by calling sb_stemmer_delete().
|
43
|
+
*
|
44
|
+
* @note NULL will also be returned if an out of memory error occurs.
|
45
|
+
*/
|
46
|
+
struct sb_stemmer * sb_stemmer_new(const char * algorithm, const char * charenc);
|
47
|
+
|
48
|
+
/** Delete a stemmer object.
|
49
|
+
*
|
50
|
+
* This frees all resources allocated for the stemmer. After calling
|
51
|
+
* this function, the supplied stemmer may no longer be used in any way.
|
52
|
+
*
|
53
|
+
* It is safe to pass a null pointer to this function - this will have
|
54
|
+
* no effect.
|
55
|
+
*/
|
56
|
+
void sb_stemmer_delete(struct sb_stemmer * stemmer);
|
57
|
+
|
58
|
+
/** Stem a word.
|
59
|
+
*
|
60
|
+
* The return value is owned by the stemmer - it must not be freed or
|
61
|
+
* modified, and it will become invalid when the stemmer is called again,
|
62
|
+
* or if the stemmer is freed.
|
63
|
+
*
|
64
|
+
* The length of the return value can be obtained using sb_stemmer_length().
|
65
|
+
*
|
66
|
+
* If an out-of-memory error occurs, this will return NULL.
|
67
|
+
*/
|
68
|
+
const sb_symbol * sb_stemmer_stem(struct sb_stemmer * stemmer,
|
69
|
+
const sb_symbol * word, int size);
|
70
|
+
|
71
|
+
/** Get the length of the result of the last stemmed word.
|
72
|
+
* This should not be called before sb_stemmer_stem() has been called.
|
73
|
+
*/
|
74
|
+
int sb_stemmer_length(struct sb_stemmer * stemmer);
|
75
|
+
|
76
|
+
#ifdef __cplusplus
|
77
|
+
}
|
78
|
+
#endif
|
79
|
+
|
@@ -0,0 +1,93 @@
|
|
1
|
+
|
2
|
+
#include <stdlib.h>
|
3
|
+
#include <string.h>
|
4
|
+
#include "../include/libstemmer.h"
|
5
|
+
#include "../runtime/api.h"
|
6
|
+
#include "modules.h"
|
7
|
+
|
8
|
+
struct sb_stemmer {
|
9
|
+
struct SN_env * (*create)(void);
|
10
|
+
void (*close)(struct SN_env *);
|
11
|
+
int (*stem)(struct SN_env *);
|
12
|
+
|
13
|
+
struct SN_env * env;
|
14
|
+
};
|
15
|
+
|
16
|
+
extern const char **
|
17
|
+
sb_stemmer_list(void)
|
18
|
+
{
|
19
|
+
return algorithm_names;
|
20
|
+
}
|
21
|
+
|
22
|
+
static stemmer_encoding_t
|
23
|
+
sb_getenc(const char * charenc)
|
24
|
+
{
|
25
|
+
struct stemmer_encoding * encoding;
|
26
|
+
if (charenc == NULL) return ENC_UTF_8;
|
27
|
+
for (encoding = encodings; encoding->name != 0; encoding++) {
|
28
|
+
if (strcmp(encoding->name, charenc) == 0) break;
|
29
|
+
}
|
30
|
+
if (encoding->name == NULL) return ENC_UNKNOWN;
|
31
|
+
return encoding->enc;
|
32
|
+
}
|
33
|
+
|
34
|
+
extern struct sb_stemmer *
|
35
|
+
sb_stemmer_new(const char * algorithm, const char * charenc)
|
36
|
+
{
|
37
|
+
stemmer_encoding_t enc;
|
38
|
+
struct stemmer_modules * module;
|
39
|
+
struct sb_stemmer * stemmer =
|
40
|
+
(struct sb_stemmer *) malloc(sizeof(struct sb_stemmer));
|
41
|
+
if (stemmer == NULL) return NULL;
|
42
|
+
enc = sb_getenc(charenc);
|
43
|
+
if (enc == ENC_UNKNOWN) return NULL;
|
44
|
+
|
45
|
+
for (module = modules; module->name != 0; module++) {
|
46
|
+
if (strcmp(module->name, algorithm) == 0 && module->enc == enc) break;
|
47
|
+
}
|
48
|
+
if (module->name == NULL) return NULL;
|
49
|
+
|
50
|
+
stemmer->create = module->create;
|
51
|
+
stemmer->close = module->close;
|
52
|
+
stemmer->stem = module->stem;
|
53
|
+
|
54
|
+
stemmer->env = stemmer->create();
|
55
|
+
if (stemmer->env == NULL)
|
56
|
+
{
|
57
|
+
sb_stemmer_delete(stemmer);
|
58
|
+
return NULL;
|
59
|
+
}
|
60
|
+
|
61
|
+
return stemmer;
|
62
|
+
}
|
63
|
+
|
64
|
+
void
|
65
|
+
sb_stemmer_delete(struct sb_stemmer * stemmer)
|
66
|
+
{
|
67
|
+
if (stemmer == 0) return;
|
68
|
+
if (stemmer->close == 0) return;
|
69
|
+
stemmer->close(stemmer->env);
|
70
|
+
stemmer->close = 0;
|
71
|
+
free(stemmer);
|
72
|
+
}
|
73
|
+
|
74
|
+
const sb_symbol *
|
75
|
+
sb_stemmer_stem(struct sb_stemmer * stemmer, const sb_symbol * word, int size)
|
76
|
+
{
|
77
|
+
int ret;
|
78
|
+
if (SN_set_current(stemmer->env, size, (const symbol *)(word)))
|
79
|
+
{
|
80
|
+
stemmer->env->l = 0;
|
81
|
+
return NULL;
|
82
|
+
}
|
83
|
+
ret = stemmer->stem(stemmer->env);
|
84
|
+
if (ret < 0) return NULL;
|
85
|
+
stemmer->env->p[stemmer->env->l] = 0;
|
86
|
+
return (const sb_symbol *)(stemmer->env->p);
|
87
|
+
}
|
88
|
+
|
89
|
+
int
|
90
|
+
sb_stemmer_length(struct sb_stemmer * stemmer)
|
91
|
+
{
|
92
|
+
return stemmer->env->l;
|
93
|
+
}
|
@@ -0,0 +1,93 @@
|
|
1
|
+
|
2
|
+
#include <stdlib.h>
|
3
|
+
#include <string.h>
|
4
|
+
#include "../include/libstemmer.h"
|
5
|
+
#include "../runtime/api.h"
|
6
|
+
#include "modules_utf8.h"
|
7
|
+
|
8
|
+
struct sb_stemmer {
|
9
|
+
struct SN_env * (*create)(void);
|
10
|
+
void (*close)(struct SN_env *);
|
11
|
+
int (*stem)(struct SN_env *);
|
12
|
+
|
13
|
+
struct SN_env * env;
|
14
|
+
};
|
15
|
+
|
16
|
+
extern const char **
|
17
|
+
sb_stemmer_list(void)
|
18
|
+
{
|
19
|
+
return algorithm_names;
|
20
|
+
}
|
21
|
+
|
22
|
+
static stemmer_encoding_t
|
23
|
+
sb_getenc(const char * charenc)
|
24
|
+
{
|
25
|
+
struct stemmer_encoding * encoding;
|
26
|
+
if (charenc == NULL) return ENC_UTF_8;
|
27
|
+
for (encoding = encodings; encoding->name != 0; encoding++) {
|
28
|
+
if (strcmp(encoding->name, charenc) == 0) break;
|
29
|
+
}
|
30
|
+
if (encoding->name == NULL) return ENC_UNKNOWN;
|
31
|
+
return encoding->enc;
|
32
|
+
}
|
33
|
+
|
34
|
+
extern struct sb_stemmer *
|
35
|
+
sb_stemmer_new(const char * algorithm, const char * charenc)
|
36
|
+
{
|
37
|
+
stemmer_encoding_t enc;
|
38
|
+
struct stemmer_modules * module;
|
39
|
+
struct sb_stemmer * stemmer =
|
40
|
+
(struct sb_stemmer *) malloc(sizeof(struct sb_stemmer));
|
41
|
+
if (stemmer == NULL) return NULL;
|
42
|
+
enc = sb_getenc(charenc);
|
43
|
+
if (enc == ENC_UNKNOWN) return NULL;
|
44
|
+
|
45
|
+
for (module = modules; module->name != 0; module++) {
|
46
|
+
if (strcmp(module->name, algorithm) == 0 && module->enc == enc) break;
|
47
|
+
}
|
48
|
+
if (module->name == NULL) return NULL;
|
49
|
+
|
50
|
+
stemmer->create = module->create;
|
51
|
+
stemmer->close = module->close;
|
52
|
+
stemmer->stem = module->stem;
|
53
|
+
|
54
|
+
stemmer->env = stemmer->create();
|
55
|
+
if (stemmer->env == NULL)
|
56
|
+
{
|
57
|
+
sb_stemmer_delete(stemmer);
|
58
|
+
return NULL;
|
59
|
+
}
|
60
|
+
|
61
|
+
return stemmer;
|
62
|
+
}
|
63
|
+
|
64
|
+
void
|
65
|
+
sb_stemmer_delete(struct sb_stemmer * stemmer)
|
66
|
+
{
|
67
|
+
if (stemmer == 0) return;
|
68
|
+
if (stemmer->close == 0) return;
|
69
|
+
stemmer->close(stemmer->env);
|
70
|
+
stemmer->close = 0;
|
71
|
+
free(stemmer);
|
72
|
+
}
|
73
|
+
|
74
|
+
const sb_symbol *
|
75
|
+
sb_stemmer_stem(struct sb_stemmer * stemmer, const sb_symbol * word, int size)
|
76
|
+
{
|
77
|
+
int ret;
|
78
|
+
if (SN_set_current(stemmer->env, size, (const symbol *)(word)))
|
79
|
+
{
|
80
|
+
stemmer->env->l = 0;
|
81
|
+
return NULL;
|
82
|
+
}
|
83
|
+
ret = stemmer->stem(stemmer->env);
|
84
|
+
if (ret < 0) return NULL;
|
85
|
+
stemmer->env->p[stemmer->env->l] = 0;
|
86
|
+
return (const sb_symbol *)(stemmer->env->p);
|
87
|
+
}
|
88
|
+
|
89
|
+
int
|
90
|
+
sb_stemmer_length(struct sb_stemmer * stemmer)
|
91
|
+
{
|
92
|
+
return stemmer->env->l;
|
93
|
+
}
|
@@ -0,0 +1,190 @@
|
|
1
|
+
/* libstemmer/modules.h: List of stemming modules.
|
2
|
+
*
|
3
|
+
* This file is generated by mkmodules.pl from a list of module names.
|
4
|
+
* Do not edit manually.
|
5
|
+
*
|
6
|
+
* Modules included by this file are: danish, dutch, english, finnish, french,
|
7
|
+
* german, hungarian, italian, norwegian, porter, portuguese, romanian,
|
8
|
+
* russian, spanish, swedish, turkish
|
9
|
+
*/
|
10
|
+
|
11
|
+
#include "../src_c/stem_ISO_8859_1_danish.h"
|
12
|
+
#include "../src_c/stem_UTF_8_danish.h"
|
13
|
+
#include "../src_c/stem_ISO_8859_1_dutch.h"
|
14
|
+
#include "../src_c/stem_UTF_8_dutch.h"
|
15
|
+
#include "../src_c/stem_ISO_8859_1_english.h"
|
16
|
+
#include "../src_c/stem_UTF_8_english.h"
|
17
|
+
#include "../src_c/stem_ISO_8859_1_finnish.h"
|
18
|
+
#include "../src_c/stem_UTF_8_finnish.h"
|
19
|
+
#include "../src_c/stem_ISO_8859_1_french.h"
|
20
|
+
#include "../src_c/stem_UTF_8_french.h"
|
21
|
+
#include "../src_c/stem_ISO_8859_1_german.h"
|
22
|
+
#include "../src_c/stem_UTF_8_german.h"
|
23
|
+
#include "../src_c/stem_ISO_8859_1_hungarian.h"
|
24
|
+
#include "../src_c/stem_UTF_8_hungarian.h"
|
25
|
+
#include "../src_c/stem_ISO_8859_1_italian.h"
|
26
|
+
#include "../src_c/stem_UTF_8_italian.h"
|
27
|
+
#include "../src_c/stem_ISO_8859_1_norwegian.h"
|
28
|
+
#include "../src_c/stem_UTF_8_norwegian.h"
|
29
|
+
#include "../src_c/stem_ISO_8859_1_porter.h"
|
30
|
+
#include "../src_c/stem_UTF_8_porter.h"
|
31
|
+
#include "../src_c/stem_ISO_8859_1_portuguese.h"
|
32
|
+
#include "../src_c/stem_UTF_8_portuguese.h"
|
33
|
+
#include "../src_c/stem_ISO_8859_2_romanian.h"
|
34
|
+
#include "../src_c/stem_UTF_8_romanian.h"
|
35
|
+
#include "../src_c/stem_KOI8_R_russian.h"
|
36
|
+
#include "../src_c/stem_UTF_8_russian.h"
|
37
|
+
#include "../src_c/stem_ISO_8859_1_spanish.h"
|
38
|
+
#include "../src_c/stem_UTF_8_spanish.h"
|
39
|
+
#include "../src_c/stem_ISO_8859_1_swedish.h"
|
40
|
+
#include "../src_c/stem_UTF_8_swedish.h"
|
41
|
+
#include "../src_c/stem_UTF_8_turkish.h"
|
42
|
+
|
43
|
+
typedef enum {
|
44
|
+
ENC_UNKNOWN=0,
|
45
|
+
ENC_ISO_8859_1,
|
46
|
+
ENC_ISO_8859_2,
|
47
|
+
ENC_KOI8_R,
|
48
|
+
ENC_UTF_8
|
49
|
+
} stemmer_encoding_t;
|
50
|
+
|
51
|
+
struct stemmer_encoding {
|
52
|
+
const char * name;
|
53
|
+
stemmer_encoding_t enc;
|
54
|
+
};
|
55
|
+
static struct stemmer_encoding encodings[] = {
|
56
|
+
{"ISO_8859_1", ENC_ISO_8859_1},
|
57
|
+
{"ISO_8859_2", ENC_ISO_8859_2},
|
58
|
+
{"KOI8_R", ENC_KOI8_R},
|
59
|
+
{"UTF_8", ENC_UTF_8},
|
60
|
+
{0,ENC_UNKNOWN}
|
61
|
+
};
|
62
|
+
|
63
|
+
struct stemmer_modules {
|
64
|
+
const char * name;
|
65
|
+
stemmer_encoding_t enc;
|
66
|
+
struct SN_env * (*create)(void);
|
67
|
+
void (*close)(struct SN_env *);
|
68
|
+
int (*stem)(struct SN_env *);
|
69
|
+
};
|
70
|
+
static struct stemmer_modules modules[] = {
|
71
|
+
{"da", ENC_ISO_8859_1, danish_ISO_8859_1_create_env, danish_ISO_8859_1_close_env, danish_ISO_8859_1_stem},
|
72
|
+
{"da", ENC_UTF_8, danish_UTF_8_create_env, danish_UTF_8_close_env, danish_UTF_8_stem},
|
73
|
+
{"dan", ENC_ISO_8859_1, danish_ISO_8859_1_create_env, danish_ISO_8859_1_close_env, danish_ISO_8859_1_stem},
|
74
|
+
{"dan", ENC_UTF_8, danish_UTF_8_create_env, danish_UTF_8_close_env, danish_UTF_8_stem},
|
75
|
+
{"danish", ENC_ISO_8859_1, danish_ISO_8859_1_create_env, danish_ISO_8859_1_close_env, danish_ISO_8859_1_stem},
|
76
|
+
{"danish", ENC_UTF_8, danish_UTF_8_create_env, danish_UTF_8_close_env, danish_UTF_8_stem},
|
77
|
+
{"de", ENC_ISO_8859_1, german_ISO_8859_1_create_env, german_ISO_8859_1_close_env, german_ISO_8859_1_stem},
|
78
|
+
{"de", ENC_UTF_8, german_UTF_8_create_env, german_UTF_8_close_env, german_UTF_8_stem},
|
79
|
+
{"deu", ENC_ISO_8859_1, german_ISO_8859_1_create_env, german_ISO_8859_1_close_env, german_ISO_8859_1_stem},
|
80
|
+
{"deu", ENC_UTF_8, german_UTF_8_create_env, german_UTF_8_close_env, german_UTF_8_stem},
|
81
|
+
{"dut", ENC_ISO_8859_1, dutch_ISO_8859_1_create_env, dutch_ISO_8859_1_close_env, dutch_ISO_8859_1_stem},
|
82
|
+
{"dut", ENC_UTF_8, dutch_UTF_8_create_env, dutch_UTF_8_close_env, dutch_UTF_8_stem},
|
83
|
+
{"dutch", ENC_ISO_8859_1, dutch_ISO_8859_1_create_env, dutch_ISO_8859_1_close_env, dutch_ISO_8859_1_stem},
|
84
|
+
{"dutch", ENC_UTF_8, dutch_UTF_8_create_env, dutch_UTF_8_close_env, dutch_UTF_8_stem},
|
85
|
+
{"en", ENC_ISO_8859_1, english_ISO_8859_1_create_env, english_ISO_8859_1_close_env, english_ISO_8859_1_stem},
|
86
|
+
{"en", ENC_UTF_8, english_UTF_8_create_env, english_UTF_8_close_env, english_UTF_8_stem},
|
87
|
+
{"eng", ENC_ISO_8859_1, english_ISO_8859_1_create_env, english_ISO_8859_1_close_env, english_ISO_8859_1_stem},
|
88
|
+
{"eng", ENC_UTF_8, english_UTF_8_create_env, english_UTF_8_close_env, english_UTF_8_stem},
|
89
|
+
{"english", ENC_ISO_8859_1, english_ISO_8859_1_create_env, english_ISO_8859_1_close_env, english_ISO_8859_1_stem},
|
90
|
+
{"english", ENC_UTF_8, english_UTF_8_create_env, english_UTF_8_close_env, english_UTF_8_stem},
|
91
|
+
{"es", ENC_ISO_8859_1, spanish_ISO_8859_1_create_env, spanish_ISO_8859_1_close_env, spanish_ISO_8859_1_stem},
|
92
|
+
{"es", ENC_UTF_8, spanish_UTF_8_create_env, spanish_UTF_8_close_env, spanish_UTF_8_stem},
|
93
|
+
{"esl", ENC_ISO_8859_1, spanish_ISO_8859_1_create_env, spanish_ISO_8859_1_close_env, spanish_ISO_8859_1_stem},
|
94
|
+
{"esl", ENC_UTF_8, spanish_UTF_8_create_env, spanish_UTF_8_close_env, spanish_UTF_8_stem},
|
95
|
+
{"fi", ENC_ISO_8859_1, finnish_ISO_8859_1_create_env, finnish_ISO_8859_1_close_env, finnish_ISO_8859_1_stem},
|
96
|
+
{"fi", ENC_UTF_8, finnish_UTF_8_create_env, finnish_UTF_8_close_env, finnish_UTF_8_stem},
|
97
|
+
{"fin", ENC_ISO_8859_1, finnish_ISO_8859_1_create_env, finnish_ISO_8859_1_close_env, finnish_ISO_8859_1_stem},
|
98
|
+
{"fin", ENC_UTF_8, finnish_UTF_8_create_env, finnish_UTF_8_close_env, finnish_UTF_8_stem},
|
99
|
+
{"finnish", ENC_ISO_8859_1, finnish_ISO_8859_1_create_env, finnish_ISO_8859_1_close_env, finnish_ISO_8859_1_stem},
|
100
|
+
{"finnish", ENC_UTF_8, finnish_UTF_8_create_env, finnish_UTF_8_close_env, finnish_UTF_8_stem},
|
101
|
+
{"fr", ENC_ISO_8859_1, french_ISO_8859_1_create_env, french_ISO_8859_1_close_env, french_ISO_8859_1_stem},
|
102
|
+
{"fr", ENC_UTF_8, french_UTF_8_create_env, french_UTF_8_close_env, french_UTF_8_stem},
|
103
|
+
{"fra", ENC_ISO_8859_1, french_ISO_8859_1_create_env, french_ISO_8859_1_close_env, french_ISO_8859_1_stem},
|
104
|
+
{"fra", ENC_UTF_8, french_UTF_8_create_env, french_UTF_8_close_env, french_UTF_8_stem},
|
105
|
+
{"fre", ENC_ISO_8859_1, french_ISO_8859_1_create_env, french_ISO_8859_1_close_env, french_ISO_8859_1_stem},
|
106
|
+
{"fre", ENC_UTF_8, french_UTF_8_create_env, french_UTF_8_close_env, french_UTF_8_stem},
|
107
|
+
{"french", ENC_ISO_8859_1, french_ISO_8859_1_create_env, french_ISO_8859_1_close_env, french_ISO_8859_1_stem},
|
108
|
+
{"french", ENC_UTF_8, french_UTF_8_create_env, french_UTF_8_close_env, french_UTF_8_stem},
|
109
|
+
{"ger", ENC_ISO_8859_1, german_ISO_8859_1_create_env, german_ISO_8859_1_close_env, german_ISO_8859_1_stem},
|
110
|
+
{"ger", ENC_UTF_8, german_UTF_8_create_env, german_UTF_8_close_env, german_UTF_8_stem},
|
111
|
+
{"german", ENC_ISO_8859_1, german_ISO_8859_1_create_env, german_ISO_8859_1_close_env, german_ISO_8859_1_stem},
|
112
|
+
{"german", ENC_UTF_8, german_UTF_8_create_env, german_UTF_8_close_env, german_UTF_8_stem},
|
113
|
+
{"hu", ENC_ISO_8859_1, hungarian_ISO_8859_1_create_env, hungarian_ISO_8859_1_close_env, hungarian_ISO_8859_1_stem},
|
114
|
+
{"hu", ENC_UTF_8, hungarian_UTF_8_create_env, hungarian_UTF_8_close_env, hungarian_UTF_8_stem},
|
115
|
+
{"hun", ENC_ISO_8859_1, hungarian_ISO_8859_1_create_env, hungarian_ISO_8859_1_close_env, hungarian_ISO_8859_1_stem},
|
116
|
+
{"hun", ENC_UTF_8, hungarian_UTF_8_create_env, hungarian_UTF_8_close_env, hungarian_UTF_8_stem},
|
117
|
+
{"hungarian", ENC_ISO_8859_1, hungarian_ISO_8859_1_create_env, hungarian_ISO_8859_1_close_env, hungarian_ISO_8859_1_stem},
|
118
|
+
{"hungarian", ENC_UTF_8, hungarian_UTF_8_create_env, hungarian_UTF_8_close_env, hungarian_UTF_8_stem},
|
119
|
+
{"it", ENC_ISO_8859_1, italian_ISO_8859_1_create_env, italian_ISO_8859_1_close_env, italian_ISO_8859_1_stem},
|
120
|
+
{"it", ENC_UTF_8, italian_UTF_8_create_env, italian_UTF_8_close_env, italian_UTF_8_stem},
|
121
|
+
{"ita", ENC_ISO_8859_1, italian_ISO_8859_1_create_env, italian_ISO_8859_1_close_env, italian_ISO_8859_1_stem},
|
122
|
+
{"ita", ENC_UTF_8, italian_UTF_8_create_env, italian_UTF_8_close_env, italian_UTF_8_stem},
|
123
|
+
{"italian", ENC_ISO_8859_1, italian_ISO_8859_1_create_env, italian_ISO_8859_1_close_env, italian_ISO_8859_1_stem},
|
124
|
+
{"italian", ENC_UTF_8, italian_UTF_8_create_env, italian_UTF_8_close_env, italian_UTF_8_stem},
|
125
|
+
{"nl", ENC_ISO_8859_1, dutch_ISO_8859_1_create_env, dutch_ISO_8859_1_close_env, dutch_ISO_8859_1_stem},
|
126
|
+
{"nl", ENC_UTF_8, dutch_UTF_8_create_env, dutch_UTF_8_close_env, dutch_UTF_8_stem},
|
127
|
+
{"nld", ENC_ISO_8859_1, dutch_ISO_8859_1_create_env, dutch_ISO_8859_1_close_env, dutch_ISO_8859_1_stem},
|
128
|
+
{"nld", ENC_UTF_8, dutch_UTF_8_create_env, dutch_UTF_8_close_env, dutch_UTF_8_stem},
|
129
|
+
{"no", ENC_ISO_8859_1, norwegian_ISO_8859_1_create_env, norwegian_ISO_8859_1_close_env, norwegian_ISO_8859_1_stem},
|
130
|
+
{"no", ENC_UTF_8, norwegian_UTF_8_create_env, norwegian_UTF_8_close_env, norwegian_UTF_8_stem},
|
131
|
+
{"nor", ENC_ISO_8859_1, norwegian_ISO_8859_1_create_env, norwegian_ISO_8859_1_close_env, norwegian_ISO_8859_1_stem},
|
132
|
+
{"nor", ENC_UTF_8, norwegian_UTF_8_create_env, norwegian_UTF_8_close_env, norwegian_UTF_8_stem},
|
133
|
+
{"norwegian", ENC_ISO_8859_1, norwegian_ISO_8859_1_create_env, norwegian_ISO_8859_1_close_env, norwegian_ISO_8859_1_stem},
|
134
|
+
{"norwegian", ENC_UTF_8, norwegian_UTF_8_create_env, norwegian_UTF_8_close_env, norwegian_UTF_8_stem},
|
135
|
+
{"por", ENC_ISO_8859_1, portuguese_ISO_8859_1_create_env, portuguese_ISO_8859_1_close_env, portuguese_ISO_8859_1_stem},
|
136
|
+
{"por", ENC_UTF_8, portuguese_UTF_8_create_env, portuguese_UTF_8_close_env, portuguese_UTF_8_stem},
|
137
|
+
{"porter", ENC_ISO_8859_1, porter_ISO_8859_1_create_env, porter_ISO_8859_1_close_env, porter_ISO_8859_1_stem},
|
138
|
+
{"porter", ENC_UTF_8, porter_UTF_8_create_env, porter_UTF_8_close_env, porter_UTF_8_stem},
|
139
|
+
{"portuguese", ENC_ISO_8859_1, portuguese_ISO_8859_1_create_env, portuguese_ISO_8859_1_close_env, portuguese_ISO_8859_1_stem},
|
140
|
+
{"portuguese", ENC_UTF_8, portuguese_UTF_8_create_env, portuguese_UTF_8_close_env, portuguese_UTF_8_stem},
|
141
|
+
{"pt", ENC_ISO_8859_1, portuguese_ISO_8859_1_create_env, portuguese_ISO_8859_1_close_env, portuguese_ISO_8859_1_stem},
|
142
|
+
{"pt", ENC_UTF_8, portuguese_UTF_8_create_env, portuguese_UTF_8_close_env, portuguese_UTF_8_stem},
|
143
|
+
{"ro", ENC_ISO_8859_2, romanian_ISO_8859_2_create_env, romanian_ISO_8859_2_close_env, romanian_ISO_8859_2_stem},
|
144
|
+
{"ro", ENC_UTF_8, romanian_UTF_8_create_env, romanian_UTF_8_close_env, romanian_UTF_8_stem},
|
145
|
+
{"romanian", ENC_ISO_8859_2, romanian_ISO_8859_2_create_env, romanian_ISO_8859_2_close_env, romanian_ISO_8859_2_stem},
|
146
|
+
{"romanian", ENC_UTF_8, romanian_UTF_8_create_env, romanian_UTF_8_close_env, romanian_UTF_8_stem},
|
147
|
+
{"ron", ENC_ISO_8859_2, romanian_ISO_8859_2_create_env, romanian_ISO_8859_2_close_env, romanian_ISO_8859_2_stem},
|
148
|
+
{"ron", ENC_UTF_8, romanian_UTF_8_create_env, romanian_UTF_8_close_env, romanian_UTF_8_stem},
|
149
|
+
{"ru", ENC_KOI8_R, russian_KOI8_R_create_env, russian_KOI8_R_close_env, russian_KOI8_R_stem},
|
150
|
+
{"ru", ENC_UTF_8, russian_UTF_8_create_env, russian_UTF_8_close_env, russian_UTF_8_stem},
|
151
|
+
{"rum", ENC_ISO_8859_2, romanian_ISO_8859_2_create_env, romanian_ISO_8859_2_close_env, romanian_ISO_8859_2_stem},
|
152
|
+
{"rum", ENC_UTF_8, romanian_UTF_8_create_env, romanian_UTF_8_close_env, romanian_UTF_8_stem},
|
153
|
+
{"rus", ENC_KOI8_R, russian_KOI8_R_create_env, russian_KOI8_R_close_env, russian_KOI8_R_stem},
|
154
|
+
{"rus", ENC_UTF_8, russian_UTF_8_create_env, russian_UTF_8_close_env, russian_UTF_8_stem},
|
155
|
+
{"russian", ENC_KOI8_R, russian_KOI8_R_create_env, russian_KOI8_R_close_env, russian_KOI8_R_stem},
|
156
|
+
{"russian", ENC_UTF_8, russian_UTF_8_create_env, russian_UTF_8_close_env, russian_UTF_8_stem},
|
157
|
+
{"spa", ENC_ISO_8859_1, spanish_ISO_8859_1_create_env, spanish_ISO_8859_1_close_env, spanish_ISO_8859_1_stem},
|
158
|
+
{"spa", ENC_UTF_8, spanish_UTF_8_create_env, spanish_UTF_8_close_env, spanish_UTF_8_stem},
|
159
|
+
{"spanish", ENC_ISO_8859_1, spanish_ISO_8859_1_create_env, spanish_ISO_8859_1_close_env, spanish_ISO_8859_1_stem},
|
160
|
+
{"spanish", ENC_UTF_8, spanish_UTF_8_create_env, spanish_UTF_8_close_env, spanish_UTF_8_stem},
|
161
|
+
{"sv", ENC_ISO_8859_1, swedish_ISO_8859_1_create_env, swedish_ISO_8859_1_close_env, swedish_ISO_8859_1_stem},
|
162
|
+
{"sv", ENC_UTF_8, swedish_UTF_8_create_env, swedish_UTF_8_close_env, swedish_UTF_8_stem},
|
163
|
+
{"swe", ENC_ISO_8859_1, swedish_ISO_8859_1_create_env, swedish_ISO_8859_1_close_env, swedish_ISO_8859_1_stem},
|
164
|
+
{"swe", ENC_UTF_8, swedish_UTF_8_create_env, swedish_UTF_8_close_env, swedish_UTF_8_stem},
|
165
|
+
{"swedish", ENC_ISO_8859_1, swedish_ISO_8859_1_create_env, swedish_ISO_8859_1_close_env, swedish_ISO_8859_1_stem},
|
166
|
+
{"swedish", ENC_UTF_8, swedish_UTF_8_create_env, swedish_UTF_8_close_env, swedish_UTF_8_stem},
|
167
|
+
{"tr", ENC_UTF_8, turkish_UTF_8_create_env, turkish_UTF_8_close_env, turkish_UTF_8_stem},
|
168
|
+
{"tur", ENC_UTF_8, turkish_UTF_8_create_env, turkish_UTF_8_close_env, turkish_UTF_8_stem},
|
169
|
+
{"turkish", ENC_UTF_8, turkish_UTF_8_create_env, turkish_UTF_8_close_env, turkish_UTF_8_stem},
|
170
|
+
{0,ENC_UNKNOWN,0,0,0}
|
171
|
+
};
|
172
|
+
static const char * algorithm_names[] = {
|
173
|
+
"danish",
|
174
|
+
"dutch",
|
175
|
+
"english",
|
176
|
+
"finnish",
|
177
|
+
"french",
|
178
|
+
"german",
|
179
|
+
"hungarian",
|
180
|
+
"italian",
|
181
|
+
"norwegian",
|
182
|
+
"porter",
|
183
|
+
"portuguese",
|
184
|
+
"romanian",
|
185
|
+
"russian",
|
186
|
+
"spanish",
|
187
|
+
"swedish",
|
188
|
+
"turkish",
|
189
|
+
0
|
190
|
+
};
|