@delali/narsil-react-native 0.3.1-next.155 → 0.3.1-next.156
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/NOTICE +144 -0
- package/NarsilReactNative.podspec +2 -2
- package/android/CMakeLists.txt +1 -1
- package/cpp/HybridNarsilCoreAnalysis.cpp +277 -0
- package/cpp/HybridNarsilCoreImpl.hpp +43 -0
- package/cpp/HybridNarsilCoreKeyword.cpp +253 -0
- package/cpp/HybridNarsilCoreMath.cpp +47 -0
- package/cpp/core/include/narsil_analysis.h +86 -0
- package/cpp/core/include/narsil_core.h +1 -1
- package/cpp/core/include/narsil_keyword.h +110 -0
- package/cpp/core/include/narsil_math.h +36 -0
- package/cpp/core/src/analysis/analyse.c +281 -0
- package/cpp/core/src/analysis/analyser.h +49 -0
- package/cpp/core/src/analysis/languages.c +79 -0
- package/cpp/core/src/analysis/languages.h +42 -0
- package/cpp/core/src/analysis/normalise.c +242 -0
- package/cpp/core/src/analysis/normalise.h +28 -0
- package/cpp/core/src/analysis/split.c +183 -0
- package/cpp/core/src/analysis/split.h +27 -0
- package/cpp/core/src/analysis/stop_words.c +82 -0
- package/cpp/core/src/analysis/stop_words.h +19 -0
- package/cpp/core/src/analysis/term_cache.c +88 -0
- package/cpp/core/src/analysis/term_cache.h +36 -0
- package/cpp/core/src/analysis/text.c +161 -0
- package/cpp/core/src/analysis/text.h +45 -0
- package/cpp/core/src/analysis/transform.c +103 -0
- package/cpp/core/src/analysis/unicode.c +252 -0
- package/cpp/core/src/analysis/unicode.h +54 -0
- package/cpp/core/src/internal.h +4 -0
- package/cpp/core/src/keyword/pass.h +117 -0
- package/cpp/core/src/keyword/prune.c +324 -0
- package/cpp/core/src/keyword/request.c +179 -0
- package/cpp/core/src/keyword/rows.c +83 -0
- package/cpp/core/src/keyword/rows.h +18 -0
- package/cpp/core/src/keyword/score.c +257 -0
- package/cpp/core/src/keyword/score.h +17 -0
- package/cpp/core/src/keyword/scratch.c +226 -0
- package/cpp/core/src/keyword/scratch.h +55 -0
- package/cpp/core/src/keyword/select.c +155 -0
- package/cpp/core/src/keyword/select.h +17 -0
- package/cpp/core/src/languages/bulgarian_stemmer.c +114 -0
- package/cpp/core/src/languages/data/albanian.c +54 -0
- package/cpp/core/src/languages/data/amharic.c +93 -0
- package/cpp/core/src/languages/data/arabic.c +146 -0
- package/cpp/core/src/languages/data/armenian.c +46 -0
- package/cpp/core/src/languages/data/azerbaijani.c +68 -0
- package/cpp/core/src/languages/data/bambara.c +48 -0
- package/cpp/core/src/languages/data/basque.c +62 -0
- package/cpp/core/src/languages/data/belarusian.c +56 -0
- package/cpp/core/src/languages/data/bengali.c +63 -0
- package/cpp/core/src/languages/data/bosnian.c +73 -0
- package/cpp/core/src/languages/data/breton.c +52 -0
- package/cpp/core/src/languages/data/bulgarian.c +144 -0
- package/cpp/core/src/languages/data/burmese.c +48 -0
- package/cpp/core/src/languages/data/catalan.c +84 -0
- package/cpp/core/src/languages/data/chinese.c +70 -0
- package/cpp/core/src/languages/data/croatian.c +72 -0
- package/cpp/core/src/languages/data/czech.c +72 -0
- package/cpp/core/src/languages/data/dagbani.c +88 -0
- package/cpp/core/src/languages/data/danish.c +56 -0
- package/cpp/core/src/languages/data/declarations.h +114 -0
- package/cpp/core/src/languages/data/dutch.c +58 -0
- package/cpp/core/src/languages/data/english.c +172 -0
- package/cpp/core/src/languages/data/esperanto.c +71 -0
- package/cpp/core/src/languages/data/estonian.c +417 -0
- package/cpp/core/src/languages/data/ewe.c +89 -0
- package/cpp/core/src/languages/data/faroese.c +54 -0
- package/cpp/core/src/languages/data/fijian.c +44 -0
- package/cpp/core/src/languages/data/finnish.c +100 -0
- package/cpp/core/src/languages/data/french.c +74 -0
- package/cpp/core/src/languages/data/ga.c +83 -0
- package/cpp/core/src/languages/data/galician.c +70 -0
- package/cpp/core/src/languages/data/georgian.c +57 -0
- package/cpp/core/src/languages/data/german.c +96 -0
- package/cpp/core/src/languages/data/greek.c +202 -0
- package/cpp/core/src/languages/data/guarani.c +110 -0
- package/cpp/core/src/languages/data/gujarati.c +90 -0
- package/cpp/core/src/languages/data/haitian-creole.c +54 -0
- package/cpp/core/src/languages/data/hausa.c +85 -0
- package/cpp/core/src/languages/data/hawaiian.c +50 -0
- package/cpp/core/src/languages/data/hebrew.c +85 -0
- package/cpp/core/src/languages/data/hindi.c +91 -0
- package/cpp/core/src/languages/data/hungarian.c +98 -0
- package/cpp/core/src/languages/data/icelandic.c +73 -0
- package/cpp/core/src/languages/data/igbo.c +90 -0
- package/cpp/core/src/languages/data/indonesian.c +296 -0
- package/cpp/core/src/languages/data/irish.c +60 -0
- package/cpp/core/src/languages/data/italian.c +107 -0
- package/cpp/core/src/languages/data/japanese.c +52 -0
- package/cpp/core/src/languages/data/kannada.c +60 -0
- package/cpp/core/src/languages/data/kazakh.c +57 -0
- package/cpp/core/src/languages/data/khmer.c +71 -0
- package/cpp/core/src/languages/data/kinyarwanda.c +54 -0
- package/cpp/core/src/languages/data/kirundi.c +53 -0
- package/cpp/core/src/languages/data/korean.c +74 -0
- package/cpp/core/src/languages/data/kurmanji.c +43 -0
- package/cpp/core/src/languages/data/kyrgyz.c +65 -0
- package/cpp/core/src/languages/data/lao.c +46 -0
- package/cpp/core/src/languages/data/latin.c +46 -0
- package/cpp/core/src/languages/data/latvian.c +75 -0
- package/cpp/core/src/languages/data/lingala.c +88 -0
- package/cpp/core/src/languages/data/lithuanian.c +65 -0
- package/cpp/core/src/languages/data/luxembourgish.c +80 -0
- package/cpp/core/src/languages/data/macedonian.c +286 -0
- package/cpp/core/src/languages/data/malagasy.c +51 -0
- package/cpp/core/src/languages/data/malay.c +184 -0
- package/cpp/core/src/languages/data/malayalam.c +64 -0
- package/cpp/core/src/languages/data/maltese.c +102 -0
- package/cpp/core/src/languages/data/maori.c +52 -0
- package/cpp/core/src/languages/data/marathi.c +62 -0
- package/cpp/core/src/languages/data/nepali.c +72 -0
- package/cpp/core/src/languages/data/norwegian.c +74 -0
- package/cpp/core/src/languages/data/oromo.c +51 -0
- package/cpp/core/src/languages/data/persian.c +150 -0
- package/cpp/core/src/languages/data/polish.c +111 -0
- package/cpp/core/src/languages/data/portuguese.c +91 -0
- package/cpp/core/src/languages/data/punjabi.c +80 -0
- package/cpp/core/src/languages/data/registry.c +116 -0
- package/cpp/core/src/languages/data/romanian.c +93 -0
- package/cpp/core/src/languages/data/russian.c +76 -0
- package/cpp/core/src/languages/data/samoan.c +47 -0
- package/cpp/core/src/languages/data/sanskrit.c +53 -0
- package/cpp/core/src/languages/data/scottish-gaelic.c +119 -0
- package/cpp/core/src/languages/data/serbian.c +84 -0
- package/cpp/core/src/languages/data/shona.c +55 -0
- package/cpp/core/src/languages/data/sinhala.c +78 -0
- package/cpp/core/src/languages/data/slovak.c +135 -0
- package/cpp/core/src/languages/data/slovenian.c +139 -0
- package/cpp/core/src/languages/data/sorani.c +55 -0
- package/cpp/core/src/languages/data/spanish.c +84 -0
- package/cpp/core/src/languages/data/swahili.c +59 -0
- package/cpp/core/src/languages/data/swedish.c +60 -0
- package/cpp/core/src/languages/data/tagalog.c +75 -0
- package/cpp/core/src/languages/data/tamil.c +69 -0
- package/cpp/core/src/languages/data/tatar.c +548 -0
- package/cpp/core/src/languages/data/telugu.c +54 -0
- package/cpp/core/src/languages/data/thai.c +46 -0
- package/cpp/core/src/languages/data/tibetan.c +55 -0
- package/cpp/core/src/languages/data/tigrinya.c +73 -0
- package/cpp/core/src/languages/data/tongan.c +47 -0
- package/cpp/core/src/languages/data/turkish.c +100 -0
- package/cpp/core/src/languages/data/twi.c +88 -0
- package/cpp/core/src/languages/data/ukrainian.c +99 -0
- package/cpp/core/src/languages/data/urdu.c +174 -0
- package/cpp/core/src/languages/data/vietnamese.c +88 -0
- package/cpp/core/src/languages/data/welsh.c +143 -0
- package/cpp/core/src/languages/data/wolof.c +51 -0
- package/cpp/core/src/languages/data/xhosa.c +56 -0
- package/cpp/core/src/languages/data/yoruba.c +93 -0
- package/cpp/core/src/languages/data/zulu.c +58 -0
- package/cpp/core/src/languages/marks.c +64 -0
- package/cpp/core/src/languages/rules.c +63 -0
- package/cpp/core/src/languages/rules.h +68 -0
- package/cpp/core/src/languages/sanskrit_stemmer.c +50 -0
- package/cpp/core/src/languages/scripts.c +143 -0
- package/cpp/core/src/languages/slovenian_stemmer.c +46 -0
- package/cpp/core/src/languages/snowball/api.c +52 -0
- package/cpp/core/src/languages/snowball/api.h +43 -0
- package/cpp/core/src/languages/snowball/arabic_stemmer.c +1506 -0
- package/cpp/core/src/languages/snowball/arabic_stemmer.h +24 -0
- package/cpp/core/src/languages/snowball/armenian_stemmer.c +523 -0
- package/cpp/core/src/languages/snowball/armenian_stemmer.h +24 -0
- package/cpp/core/src/languages/snowball/basque_stemmer.c +1089 -0
- package/cpp/core/src/languages/snowball/basque_stemmer.h +24 -0
- package/cpp/core/src/languages/snowball/catalan_stemmer.c +1380 -0
- package/cpp/core/src/languages/snowball/catalan_stemmer.h +24 -0
- package/cpp/core/src/languages/snowball/czech_stemmer.c +582 -0
- package/cpp/core/src/languages/snowball/czech_stemmer.h +24 -0
- package/cpp/core/src/languages/snowball/danish_stemmer.c +352 -0
- package/cpp/core/src/languages/snowball/danish_stemmer.h +24 -0
- package/cpp/core/src/languages/snowball/dutch_stemmer.c +1657 -0
- package/cpp/core/src/languages/snowball/dutch_stemmer.h +24 -0
- package/cpp/core/src/languages/snowball/english_stemmer.c +1019 -0
- package/cpp/core/src/languages/snowball/english_stemmer.h +24 -0
- package/cpp/core/src/languages/snowball/esperanto_stemmer.c +716 -0
- package/cpp/core/src/languages/snowball/esperanto_stemmer.h +24 -0
- package/cpp/core/src/languages/snowball/estonian_stemmer.c +1361 -0
- package/cpp/core/src/languages/snowball/estonian_stemmer.h +24 -0
- package/cpp/core/src/languages/snowball/finnish_stemmer.c +800 -0
- package/cpp/core/src/languages/snowball/finnish_stemmer.h +24 -0
- package/cpp/core/src/languages/snowball/french_stemmer.c +1217 -0
- package/cpp/core/src/languages/snowball/french_stemmer.h +24 -0
- package/cpp/core/src/languages/snowball/german_stemmer.c +567 -0
- package/cpp/core/src/languages/snowball/german_stemmer.h +24 -0
- package/cpp/core/src/languages/snowball/greek_stemmer.c +3450 -0
- package/cpp/core/src/languages/snowball/greek_stemmer.h +24 -0
- package/cpp/core/src/languages/snowball/hindi_stemmer.c +325 -0
- package/cpp/core/src/languages/snowball/hindi_stemmer.h +24 -0
- package/cpp/core/src/languages/snowball/hungarian_stemmer.c +742 -0
- package/cpp/core/src/languages/snowball/hungarian_stemmer.h +24 -0
- package/cpp/core/src/languages/snowball/indonesian_stemmer.c +443 -0
- package/cpp/core/src/languages/snowball/indonesian_stemmer.h +24 -0
- package/cpp/core/src/languages/snowball/irish_stemmer.c +428 -0
- package/cpp/core/src/languages/snowball/irish_stemmer.h +24 -0
- package/cpp/core/src/languages/snowball/italian_stemmer.c +983 -0
- package/cpp/core/src/languages/snowball/italian_stemmer.h +24 -0
- package/cpp/core/src/languages/snowball/lithuanian_stemmer.c +794 -0
- package/cpp/core/src/languages/snowball/lithuanian_stemmer.h +24 -0
- package/cpp/core/src/languages/snowball/nepali_stemmer.c +361 -0
- package/cpp/core/src/languages/snowball/nepali_stemmer.h +24 -0
- package/cpp/core/src/languages/snowball/norwegian_stemmer.c +332 -0
- package/cpp/core/src/languages/snowball/norwegian_stemmer.h +24 -0
- package/cpp/core/src/languages/snowball/persian_stemmer.c +538 -0
- package/cpp/core/src/languages/snowball/persian_stemmer.h +24 -0
- package/cpp/core/src/languages/snowball/polish_stemmer.c +501 -0
- package/cpp/core/src/languages/snowball/polish_stemmer.h +24 -0
- package/cpp/core/src/languages/snowball/portuguese_stemmer.c +866 -0
- package/cpp/core/src/languages/snowball/portuguese_stemmer.h +24 -0
- package/cpp/core/src/languages/snowball/romanian_stemmer.c +940 -0
- package/cpp/core/src/languages/snowball/romanian_stemmer.h +24 -0
- package/cpp/core/src/languages/snowball/russian_stemmer.c +598 -0
- package/cpp/core/src/languages/snowball/russian_stemmer.h +24 -0
- package/cpp/core/src/languages/snowball/serbian_stemmer.c +6765 -0
- package/cpp/core/src/languages/snowball/serbian_stemmer.h +24 -0
- package/cpp/core/src/languages/snowball/snowball_runtime.h +122 -0
- package/cpp/core/src/languages/snowball/spanish_stemmer.c +955 -0
- package/cpp/core/src/languages/snowball/spanish_stemmer.h +24 -0
- package/cpp/core/src/languages/snowball/swedish_stemmer.c +364 -0
- package/cpp/core/src/languages/snowball/swedish_stemmer.h +24 -0
- package/cpp/core/src/languages/snowball/tamil_stemmer.c +1305 -0
- package/cpp/core/src/languages/snowball/tamil_stemmer.h +24 -0
- package/cpp/core/src/languages/snowball/turkish_stemmer.c +1700 -0
- package/cpp/core/src/languages/snowball/turkish_stemmer.h +24 -0
- package/cpp/core/src/languages/snowball/utilities.c +716 -0
- package/cpp/core/src/languages/substitutions.c +131 -0
- package/cpp/core/src/languages/swahili_stemmer.c +85 -0
- package/cpp/core/src/languages/ukrainian_stemmer.c +168 -0
- package/cpp/core/src/locks.c +2 -15
- package/cpp/core/src/math/apply.c +44 -0
- package/cpp/core/src/math/fdlibm.h +16 -0
- package/cpp/core/src/math/inverse.c +242 -0
- package/cpp/core/src/math/log.c +90 -0
- package/cpp/core/src/math/reduce.c +325 -0
- package/cpp/core/src/math/trig.c +188 -0
- package/cpp/core/src/math/words.h +45 -0
- package/cpp/core/src/thread_yield.h +18 -0
- package/cpp/core/src/unicode/decompositions.c +310 -0
- package/cpp/core/src/unicode/lowercase.c +172 -0
- package/cpp/core/src/unicode/properties.c +214 -0
- package/cpp/core/src/workspace.c +5 -1
- package/lib/engine/analysis.d.ts +3 -0
- package/lib/engine/analysis.js +76 -0
- package/lib/engine/core.js +5 -0
- package/lib/engine/keyword.d.ts +3 -0
- package/lib/engine/keyword.js +25 -0
- package/lib/specs/core.nitro.d.ts +1 -0
- package/nitrogen/generated/shared/c++/HybridNarsilCoreSpec.cpp +1 -0
- package/nitrogen/generated/shared/c++/HybridNarsilCoreSpec.hpp +1 -0
- package/package.json +4 -3
- package/src/engine/analysis.ts +87 -0
- package/src/engine/core.ts +6 -0
- package/src/engine/keyword.ts +26 -0
- package/src/specs/core.nitro.ts +1 -0
|
@@ -0,0 +1,24 @@
|
|
|
1
|
+
/*
|
|
2
|
+
* Generated from algorithms/hungarian.sbl by the Snowball compiler 3.1.1, with each symbol widened to a
|
|
3
|
+
* UTF-16 code unit so that the stemmer reads the same units as the TypeScript stemmer.
|
|
4
|
+
*
|
|
5
|
+
* Snowball is BSD-3-Clause, copyright (c) 2001, Dr Martin Porter and contributors. The NOTICE file at the root
|
|
6
|
+
* of the repository carries the licence. Source from https://github.com/snowballstem/snowball.git at
|
|
7
|
+
* 6772636350acfd63797e8cd24ff86c70fd2df6fc, verified against https://github.com/snowballstem/snowball-data.git at
|
|
8
|
+
* a0ec0d0a2839ec885878868de20fcb63209d92b0.
|
|
9
|
+
*
|
|
10
|
+
* packages/ts/snowball/build.ts writes this file, so regenerate it with that script.
|
|
11
|
+
*/
|
|
12
|
+
|
|
13
|
+
#ifdef __cplusplus
|
|
14
|
+
extern "C" {
|
|
15
|
+
#endif
|
|
16
|
+
|
|
17
|
+
extern struct SN_env * hungarian_create_env(void);
|
|
18
|
+
#define hungarian_close_env SN_delete_env
|
|
19
|
+
|
|
20
|
+
extern int hungarian_stem(struct SN_env * z);
|
|
21
|
+
|
|
22
|
+
#ifdef __cplusplus
|
|
23
|
+
}
|
|
24
|
+
#endif
|
|
@@ -0,0 +1,443 @@
|
|
|
1
|
+
/*
|
|
2
|
+
* Generated from algorithms/indonesian.sbl by the Snowball compiler 3.1.1, with each symbol widened to a
|
|
3
|
+
* UTF-16 code unit so that the stemmer reads the same units as the TypeScript stemmer.
|
|
4
|
+
*
|
|
5
|
+
* Snowball is BSD-3-Clause, copyright (c) 2001, Dr Martin Porter and contributors. The NOTICE file at the root
|
|
6
|
+
* of the repository carries the licence. Source from https://github.com/snowballstem/snowball.git at
|
|
7
|
+
* 6772636350acfd63797e8cd24ff86c70fd2df6fc, verified against https://github.com/snowballstem/snowball-data.git at
|
|
8
|
+
* a0ec0d0a2839ec885878868de20fcb63209d92b0.
|
|
9
|
+
*
|
|
10
|
+
* packages/ts/snowball/build.ts writes this file, so regenerate it with that script.
|
|
11
|
+
*/
|
|
12
|
+
|
|
13
|
+
#if defined(__GNUC__)
|
|
14
|
+
#pragma GCC diagnostic ignored "-Wcast-align"
|
|
15
|
+
#pragma GCC diagnostic ignored "-Wsign-conversion"
|
|
16
|
+
#endif
|
|
17
|
+
|
|
18
|
+
#include "indonesian_stemmer.h"
|
|
19
|
+
|
|
20
|
+
#include <stddef.h>
|
|
21
|
+
|
|
22
|
+
#include "./snowball_runtime.h"
|
|
23
|
+
|
|
24
|
+
struct SN_local {
|
|
25
|
+
struct SN_env z;
|
|
26
|
+
int i_prefix;
|
|
27
|
+
int i_measure;
|
|
28
|
+
};
|
|
29
|
+
|
|
30
|
+
typedef struct SN_local SN_local;
|
|
31
|
+
|
|
32
|
+
#ifdef __cplusplus
|
|
33
|
+
extern "C" {
|
|
34
|
+
#endif
|
|
35
|
+
extern int indonesian_stem(struct SN_env * z);
|
|
36
|
+
#ifdef __cplusplus
|
|
37
|
+
}
|
|
38
|
+
#endif
|
|
39
|
+
|
|
40
|
+
static int r_remove_suffix(struct SN_env * z);
|
|
41
|
+
static int r_remove_second_order_prefix(struct SN_env * z);
|
|
42
|
+
|
|
43
|
+
static const symbol s_0[] = { 'a', 'j', 'a', 'r' };
|
|
44
|
+
static const symbol s_1[] = { 'a', 'j', 'a', 'r' };
|
|
45
|
+
static const symbol s_2[] = { 'e', 'r' };
|
|
46
|
+
static const symbol s_3[] = { 's' };
|
|
47
|
+
static const symbol s_4[] = { 's' };
|
|
48
|
+
static const symbol s_5[] = { 'p' };
|
|
49
|
+
static const symbol s_6[] = { 'p' };
|
|
50
|
+
|
|
51
|
+
static const symbol s_0_0[3] = { 'k', 'a', 'h' };
|
|
52
|
+
static const symbol s_0_1[3] = { 'l', 'a', 'h' };
|
|
53
|
+
static const symbol s_0_2[3] = { 'p', 'u', 'n' };
|
|
54
|
+
static const struct among a_0[3] = {
|
|
55
|
+
{ 3, s_0_0, 0, 1, 0},
|
|
56
|
+
{ 3, s_0_1, 0, 1, 0},
|
|
57
|
+
{ 3, s_0_2, 0, 1, 0}
|
|
58
|
+
};
|
|
59
|
+
|
|
60
|
+
static const symbol s_1_0[3] = { 'n', 'y', 'a' };
|
|
61
|
+
static const symbol s_1_1[2] = { 'k', 'u' };
|
|
62
|
+
static const symbol s_1_2[2] = { 'm', 'u' };
|
|
63
|
+
static const struct among a_1[3] = {
|
|
64
|
+
{ 3, s_1_0, 0, 1, 0},
|
|
65
|
+
{ 2, s_1_1, 0, 1, 0},
|
|
66
|
+
{ 2, s_1_2, 0, 1, 0}
|
|
67
|
+
};
|
|
68
|
+
|
|
69
|
+
static const symbol s_2_0[1] = { 'i' };
|
|
70
|
+
static const symbol s_2_1[2] = { 'a', 'n' };
|
|
71
|
+
static const struct among a_2[2] = {
|
|
72
|
+
{ 1, s_2_0, 0, 2, 0},
|
|
73
|
+
{ 2, s_2_1, 0, 1, 0}
|
|
74
|
+
};
|
|
75
|
+
|
|
76
|
+
static const symbol s_3_0[2] = { 'd', 'i' };
|
|
77
|
+
static const symbol s_3_1[2] = { 'k', 'e' };
|
|
78
|
+
static const symbol s_3_2[2] = { 'm', 'e' };
|
|
79
|
+
static const symbol s_3_3[3] = { 'm', 'e', 'm' };
|
|
80
|
+
static const symbol s_3_4[3] = { 'm', 'e', 'n' };
|
|
81
|
+
static const symbol s_3_5[4] = { 'm', 'e', 'n', 'g' };
|
|
82
|
+
static const symbol s_3_6[3] = { 'p', 'e', 'm' };
|
|
83
|
+
static const symbol s_3_7[3] = { 'p', 'e', 'n' };
|
|
84
|
+
static const symbol s_3_8[4] = { 'p', 'e', 'n', 'g' };
|
|
85
|
+
static const symbol s_3_9[3] = { 't', 'e', 'r' };
|
|
86
|
+
static const struct among a_3[10] = {
|
|
87
|
+
{ 2, s_3_0, 0, 1, 0},
|
|
88
|
+
{ 2, s_3_1, 0, 3, 0},
|
|
89
|
+
{ 2, s_3_2, 0, 1, 0},
|
|
90
|
+
{ 3, s_3_3, -1, 5, 0},
|
|
91
|
+
{ 3, s_3_4, -2, 2, 0},
|
|
92
|
+
{ 4, s_3_5, -1, 1, 0},
|
|
93
|
+
{ 3, s_3_6, 0, 6, 0},
|
|
94
|
+
{ 3, s_3_7, 0, 4, 0},
|
|
95
|
+
{ 4, s_3_8, -1, 3, 0},
|
|
96
|
+
{ 3, s_3_9, 0, 1, 0}
|
|
97
|
+
};
|
|
98
|
+
|
|
99
|
+
static const symbol s_4_0[2] = { 'b', 'e' };
|
|
100
|
+
static const symbol s_4_1[2] = { 'p', 'e' };
|
|
101
|
+
static const struct among a_4[2] = {
|
|
102
|
+
{ 2, s_4_0, 0, 2, 0},
|
|
103
|
+
{ 2, s_4_1, 0, 1, 0}
|
|
104
|
+
};
|
|
105
|
+
|
|
106
|
+
static const unsigned char g_vowel[] = { 17, 65, 16 };
|
|
107
|
+
|
|
108
|
+
static int r_remove_suffix(struct SN_env * z) {
|
|
109
|
+
int among_var;
|
|
110
|
+
z->ket = z->c;
|
|
111
|
+
if (z->c <= z->lb || (z->p[z->c - 1] != 105 && z->p[z->c - 1] != 110)) return 0;
|
|
112
|
+
among_var = find_among_b(z, a_2, 2, 0);
|
|
113
|
+
if (!among_var) return 0;
|
|
114
|
+
z->bra = z->c;
|
|
115
|
+
switch (among_var) {
|
|
116
|
+
case 1:
|
|
117
|
+
do {
|
|
118
|
+
int v_1 = z->l - z->c;
|
|
119
|
+
if (((SN_local *)z)->i_prefix == 3) goto lab0;
|
|
120
|
+
if (((SN_local *)z)->i_prefix == 2) goto lab0;
|
|
121
|
+
if (z->c <= z->lb || z->p[z->c - 1] != 'k') goto lab0;
|
|
122
|
+
z->c--;
|
|
123
|
+
z->bra = z->c;
|
|
124
|
+
break;
|
|
125
|
+
lab0:
|
|
126
|
+
z->c = z->l - v_1;
|
|
127
|
+
if (((SN_local *)z)->i_prefix == 1) return 0;
|
|
128
|
+
} while (0);
|
|
129
|
+
break;
|
|
130
|
+
case 2:
|
|
131
|
+
if (((SN_local *)z)->i_prefix > 2) return 0;
|
|
132
|
+
if (z->c <= z->lb || z->p[z->c - 1] != 's') goto lab1;
|
|
133
|
+
z->c--;
|
|
134
|
+
return 0;
|
|
135
|
+
lab1:
|
|
136
|
+
break;
|
|
137
|
+
}
|
|
138
|
+
{
|
|
139
|
+
int ret = slice_del(z);
|
|
140
|
+
if (ret < 0) return ret;
|
|
141
|
+
}
|
|
142
|
+
((SN_local *)z)->i_measure -= 1;
|
|
143
|
+
return 1;
|
|
144
|
+
}
|
|
145
|
+
|
|
146
|
+
static int r_remove_second_order_prefix(struct SN_env * z) {
|
|
147
|
+
int among_var;
|
|
148
|
+
z->bra = z->c;
|
|
149
|
+
if (z->c + 1 >= z->l || z->p[z->c + 1] != 101) return 0;
|
|
150
|
+
among_var = find_among(z, a_4, 2, 0);
|
|
151
|
+
if (!among_var) return 0;
|
|
152
|
+
switch (among_var) {
|
|
153
|
+
case 1:
|
|
154
|
+
do {
|
|
155
|
+
int v_1 = z->c;
|
|
156
|
+
if (z->c == z->l || z->p[z->c] != 'r') goto lab0;
|
|
157
|
+
z->c++;
|
|
158
|
+
z->ket = z->c;
|
|
159
|
+
((SN_local *)z)->i_prefix = 2;
|
|
160
|
+
break;
|
|
161
|
+
lab0:
|
|
162
|
+
z->c = v_1;
|
|
163
|
+
if (z->c == z->l || z->p[z->c] != 'l') goto lab1;
|
|
164
|
+
z->c++;
|
|
165
|
+
z->ket = z->c;
|
|
166
|
+
if (!(eq_s(z, 4, s_0))) goto lab1;
|
|
167
|
+
break;
|
|
168
|
+
lab1:
|
|
169
|
+
z->c = v_1;
|
|
170
|
+
z->ket = z->c;
|
|
171
|
+
((SN_local *)z)->i_prefix = 2;
|
|
172
|
+
} while (0);
|
|
173
|
+
break;
|
|
174
|
+
case 2:
|
|
175
|
+
do {
|
|
176
|
+
int v_2 = z->c;
|
|
177
|
+
if (z->c == z->l || z->p[z->c] != 'r') goto lab2;
|
|
178
|
+
z->c++;
|
|
179
|
+
z->ket = z->c;
|
|
180
|
+
break;
|
|
181
|
+
lab2:
|
|
182
|
+
z->c = v_2;
|
|
183
|
+
if (z->c == z->l || z->p[z->c] != 'l') goto lab3;
|
|
184
|
+
z->c++;
|
|
185
|
+
z->ket = z->c;
|
|
186
|
+
if (!(eq_s(z, 4, s_1))) goto lab3;
|
|
187
|
+
break;
|
|
188
|
+
lab3:
|
|
189
|
+
z->c = v_2;
|
|
190
|
+
z->ket = z->c;
|
|
191
|
+
if (out_grouping(z, g_vowel, 97, 117, 0)) return 0;
|
|
192
|
+
if (!(eq_s(z, 2, s_2))) return 0;
|
|
193
|
+
} while (0);
|
|
194
|
+
((SN_local *)z)->i_prefix = 4;
|
|
195
|
+
break;
|
|
196
|
+
}
|
|
197
|
+
((SN_local *)z)->i_measure -= 1;
|
|
198
|
+
{
|
|
199
|
+
int ret = slice_del(z);
|
|
200
|
+
if (ret < 0) return ret;
|
|
201
|
+
}
|
|
202
|
+
return 1;
|
|
203
|
+
}
|
|
204
|
+
|
|
205
|
+
extern int indonesian_stem(struct SN_env * z) {
|
|
206
|
+
int among_var;
|
|
207
|
+
((SN_local *)z)->i_measure = 0;
|
|
208
|
+
{
|
|
209
|
+
int v_1 = z->c;
|
|
210
|
+
while (1) {
|
|
211
|
+
int v_2 = z->c;
|
|
212
|
+
{
|
|
213
|
+
int ret = out_grouping(z, g_vowel, 97, 117, 1);
|
|
214
|
+
if (ret < 0) goto lab1;
|
|
215
|
+
z->c += ret;
|
|
216
|
+
}
|
|
217
|
+
((SN_local *)z)->i_measure += 1;
|
|
218
|
+
continue;
|
|
219
|
+
lab1:
|
|
220
|
+
z->c = v_2;
|
|
221
|
+
break;
|
|
222
|
+
}
|
|
223
|
+
z->c = v_1;
|
|
224
|
+
}
|
|
225
|
+
if (((SN_local *)z)->i_measure < 3) return 0;
|
|
226
|
+
((SN_local *)z)->i_prefix = 0;
|
|
227
|
+
z->lb = z->c; z->c = z->l;
|
|
228
|
+
{
|
|
229
|
+
int v_3 = z->l - z->c;
|
|
230
|
+
z->ket = z->c;
|
|
231
|
+
if (z->c - 2 <= z->lb || (z->p[z->c - 1] != 104 && z->p[z->c - 1] != 110)) goto lab2;
|
|
232
|
+
if (!find_among_b(z, a_0, 3, 0)) goto lab2;
|
|
233
|
+
z->bra = z->c;
|
|
234
|
+
{
|
|
235
|
+
int ret = slice_del(z);
|
|
236
|
+
if (ret < 0) return ret;
|
|
237
|
+
}
|
|
238
|
+
((SN_local *)z)->i_measure -= 1;
|
|
239
|
+
lab2:
|
|
240
|
+
z->c = z->l - v_3;
|
|
241
|
+
}
|
|
242
|
+
if (((SN_local *)z)->i_measure < 3) return 0;
|
|
243
|
+
{
|
|
244
|
+
int v_4 = z->l - z->c;
|
|
245
|
+
z->ket = z->c;
|
|
246
|
+
if (z->c - 1 <= z->lb || (z->p[z->c - 1] != 97 && z->p[z->c - 1] != 117)) goto lab3;
|
|
247
|
+
if (!find_among_b(z, a_1, 3, 0)) goto lab3;
|
|
248
|
+
z->bra = z->c;
|
|
249
|
+
{
|
|
250
|
+
int ret = slice_del(z);
|
|
251
|
+
if (ret < 0) return ret;
|
|
252
|
+
}
|
|
253
|
+
((SN_local *)z)->i_measure -= 1;
|
|
254
|
+
lab3:
|
|
255
|
+
z->c = z->l - v_4;
|
|
256
|
+
}
|
|
257
|
+
z->c = z->lb;
|
|
258
|
+
if (((SN_local *)z)->i_measure < 3) return 0;
|
|
259
|
+
do {
|
|
260
|
+
int v_5 = z->c;
|
|
261
|
+
{
|
|
262
|
+
int v_6 = z->c;
|
|
263
|
+
z->bra = z->c;
|
|
264
|
+
if (z->c + 1 >= z->l || (z->p[z->c + 1] != 105 && z->p[z->c + 1] != 101)) goto lab4;
|
|
265
|
+
among_var = find_among(z, a_3, 10, 0);
|
|
266
|
+
if (!among_var) goto lab4;
|
|
267
|
+
z->ket = z->c;
|
|
268
|
+
switch (among_var) {
|
|
269
|
+
case 1:
|
|
270
|
+
{
|
|
271
|
+
int ret = slice_del(z);
|
|
272
|
+
if (ret < 0) return ret;
|
|
273
|
+
}
|
|
274
|
+
((SN_local *)z)->i_prefix = 1;
|
|
275
|
+
((SN_local *)z)->i_measure -= 1;
|
|
276
|
+
break;
|
|
277
|
+
case 2:
|
|
278
|
+
do {
|
|
279
|
+
int v_7 = z->c;
|
|
280
|
+
if (z->c == z->l || z->p[z->c] != 'y') goto lab5;
|
|
281
|
+
z->c++;
|
|
282
|
+
{
|
|
283
|
+
int v_8 = z->c;
|
|
284
|
+
if (in_grouping(z, g_vowel, 97, 117, 0)) goto lab5;
|
|
285
|
+
z->c = v_8;
|
|
286
|
+
}
|
|
287
|
+
z->ket = z->c;
|
|
288
|
+
{
|
|
289
|
+
int ret = slice_from_s(z, 1, s_3);
|
|
290
|
+
if (ret < 0) return ret;
|
|
291
|
+
}
|
|
292
|
+
((SN_local *)z)->i_prefix = 1;
|
|
293
|
+
((SN_local *)z)->i_measure -= 1;
|
|
294
|
+
break;
|
|
295
|
+
lab5:
|
|
296
|
+
z->c = v_7;
|
|
297
|
+
{
|
|
298
|
+
int ret = slice_del(z);
|
|
299
|
+
if (ret < 0) return ret;
|
|
300
|
+
}
|
|
301
|
+
((SN_local *)z)->i_prefix = 1;
|
|
302
|
+
((SN_local *)z)->i_measure -= 1;
|
|
303
|
+
} while (0);
|
|
304
|
+
break;
|
|
305
|
+
case 3:
|
|
306
|
+
{
|
|
307
|
+
int ret = slice_del(z);
|
|
308
|
+
if (ret < 0) return ret;
|
|
309
|
+
}
|
|
310
|
+
((SN_local *)z)->i_prefix = 3;
|
|
311
|
+
((SN_local *)z)->i_measure -= 1;
|
|
312
|
+
break;
|
|
313
|
+
case 4:
|
|
314
|
+
do {
|
|
315
|
+
int v_9 = z->c;
|
|
316
|
+
if (z->c == z->l || z->p[z->c] != 'y') goto lab6;
|
|
317
|
+
z->c++;
|
|
318
|
+
{
|
|
319
|
+
int v_10 = z->c;
|
|
320
|
+
if (in_grouping(z, g_vowel, 97, 117, 0)) goto lab6;
|
|
321
|
+
z->c = v_10;
|
|
322
|
+
}
|
|
323
|
+
z->ket = z->c;
|
|
324
|
+
{
|
|
325
|
+
int ret = slice_from_s(z, 1, s_4);
|
|
326
|
+
if (ret < 0) return ret;
|
|
327
|
+
}
|
|
328
|
+
((SN_local *)z)->i_prefix = 3;
|
|
329
|
+
((SN_local *)z)->i_measure -= 1;
|
|
330
|
+
break;
|
|
331
|
+
lab6:
|
|
332
|
+
z->c = v_9;
|
|
333
|
+
{
|
|
334
|
+
int ret = slice_del(z);
|
|
335
|
+
if (ret < 0) return ret;
|
|
336
|
+
}
|
|
337
|
+
((SN_local *)z)->i_prefix = 3;
|
|
338
|
+
((SN_local *)z)->i_measure -= 1;
|
|
339
|
+
} while (0);
|
|
340
|
+
break;
|
|
341
|
+
case 5:
|
|
342
|
+
((SN_local *)z)->i_prefix = 1;
|
|
343
|
+
((SN_local *)z)->i_measure -= 1;
|
|
344
|
+
do {
|
|
345
|
+
int v_11 = z->c;
|
|
346
|
+
{
|
|
347
|
+
int v_12 = z->c;
|
|
348
|
+
if (in_grouping(z, g_vowel, 97, 117, 0)) goto lab7;
|
|
349
|
+
z->c = v_12;
|
|
350
|
+
{
|
|
351
|
+
int ret = slice_from_s(z, 1, s_5);
|
|
352
|
+
if (ret < 0) return ret;
|
|
353
|
+
}
|
|
354
|
+
}
|
|
355
|
+
break;
|
|
356
|
+
lab7:
|
|
357
|
+
z->c = v_11;
|
|
358
|
+
{
|
|
359
|
+
int ret = slice_del(z);
|
|
360
|
+
if (ret < 0) return ret;
|
|
361
|
+
}
|
|
362
|
+
} while (0);
|
|
363
|
+
break;
|
|
364
|
+
case 6:
|
|
365
|
+
((SN_local *)z)->i_prefix = 3;
|
|
366
|
+
((SN_local *)z)->i_measure -= 1;
|
|
367
|
+
do {
|
|
368
|
+
int v_13 = z->c;
|
|
369
|
+
{
|
|
370
|
+
int v_14 = z->c;
|
|
371
|
+
if (in_grouping(z, g_vowel, 97, 117, 0)) goto lab8;
|
|
372
|
+
z->c = v_14;
|
|
373
|
+
{
|
|
374
|
+
int ret = slice_from_s(z, 1, s_6);
|
|
375
|
+
if (ret < 0) return ret;
|
|
376
|
+
}
|
|
377
|
+
}
|
|
378
|
+
break;
|
|
379
|
+
lab8:
|
|
380
|
+
z->c = v_13;
|
|
381
|
+
{
|
|
382
|
+
int ret = slice_del(z);
|
|
383
|
+
if (ret < 0) return ret;
|
|
384
|
+
}
|
|
385
|
+
} while (0);
|
|
386
|
+
break;
|
|
387
|
+
}
|
|
388
|
+
{
|
|
389
|
+
int v_15 = z->c;
|
|
390
|
+
{
|
|
391
|
+
int v_16 = z->c;
|
|
392
|
+
if (((SN_local *)z)->i_measure < 3) goto lab9;
|
|
393
|
+
z->lb = z->c; z->c = z->l;
|
|
394
|
+
{
|
|
395
|
+
int ret = r_remove_suffix(z);
|
|
396
|
+
if (ret == 0) goto lab9;
|
|
397
|
+
if (ret < 0) return ret;
|
|
398
|
+
}
|
|
399
|
+
z->c = z->lb;
|
|
400
|
+
z->c = v_16;
|
|
401
|
+
}
|
|
402
|
+
if (((SN_local *)z)->i_measure < 3) goto lab9;
|
|
403
|
+
{
|
|
404
|
+
int ret = r_remove_second_order_prefix(z);
|
|
405
|
+
if (ret == 0) goto lab9;
|
|
406
|
+
if (ret < 0) return ret;
|
|
407
|
+
}
|
|
408
|
+
lab9:
|
|
409
|
+
z->c = v_15;
|
|
410
|
+
}
|
|
411
|
+
z->c = v_6;
|
|
412
|
+
}
|
|
413
|
+
break;
|
|
414
|
+
lab4:
|
|
415
|
+
z->c = v_5;
|
|
416
|
+
{
|
|
417
|
+
int v_17 = z->c;
|
|
418
|
+
{
|
|
419
|
+
int ret = r_remove_second_order_prefix(z);
|
|
420
|
+
if (ret < 0) return ret;
|
|
421
|
+
}
|
|
422
|
+
z->c = v_17;
|
|
423
|
+
}
|
|
424
|
+
{
|
|
425
|
+
int v_18 = z->c;
|
|
426
|
+
if (((SN_local *)z)->i_measure < 3) goto lab10;
|
|
427
|
+
z->lb = z->c; z->c = z->l;
|
|
428
|
+
{
|
|
429
|
+
int ret = r_remove_suffix(z);
|
|
430
|
+
if (ret == 0) goto lab10;
|
|
431
|
+
if (ret < 0) return ret;
|
|
432
|
+
}
|
|
433
|
+
z->c = z->lb;
|
|
434
|
+
lab10:
|
|
435
|
+
z->c = v_18;
|
|
436
|
+
}
|
|
437
|
+
} while (0);
|
|
438
|
+
return 1;
|
|
439
|
+
}
|
|
440
|
+
|
|
441
|
+
extern struct SN_env * indonesian_create_env(void) {
|
|
442
|
+
return SN_new_env(sizeof(SN_local));
|
|
443
|
+
}
|
|
@@ -0,0 +1,24 @@
|
|
|
1
|
+
/*
|
|
2
|
+
* Generated from algorithms/indonesian.sbl by the Snowball compiler 3.1.1, with each symbol widened to a
|
|
3
|
+
* UTF-16 code unit so that the stemmer reads the same units as the TypeScript stemmer.
|
|
4
|
+
*
|
|
5
|
+
* Snowball is BSD-3-Clause, copyright (c) 2001, Dr Martin Porter and contributors. The NOTICE file at the root
|
|
6
|
+
* of the repository carries the licence. Source from https://github.com/snowballstem/snowball.git at
|
|
7
|
+
* 6772636350acfd63797e8cd24ff86c70fd2df6fc, verified against https://github.com/snowballstem/snowball-data.git at
|
|
8
|
+
* a0ec0d0a2839ec885878868de20fcb63209d92b0.
|
|
9
|
+
*
|
|
10
|
+
* packages/ts/snowball/build.ts writes this file, so regenerate it with that script.
|
|
11
|
+
*/
|
|
12
|
+
|
|
13
|
+
#ifdef __cplusplus
|
|
14
|
+
extern "C" {
|
|
15
|
+
#endif
|
|
16
|
+
|
|
17
|
+
extern struct SN_env * indonesian_create_env(void);
|
|
18
|
+
#define indonesian_close_env SN_delete_env
|
|
19
|
+
|
|
20
|
+
extern int indonesian_stem(struct SN_env * z);
|
|
21
|
+
|
|
22
|
+
#ifdef __cplusplus
|
|
23
|
+
}
|
|
24
|
+
#endif
|