@delali/narsil-react-native 0.3.1-next.155 → 0.3.1-next.157
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 +1,69 @@
|
|
|
1
|
+
#include "../../analysis/languages.h"
|
|
2
|
+
#include "declarations.h"
|
|
3
|
+
#include "../snowball/tamil_stemmer.h"
|
|
4
|
+
|
|
5
|
+
#include <stddef.h>
|
|
6
|
+
#include <stdint.h>
|
|
7
|
+
|
|
8
|
+
static const uint16_t stop_word_units[503] = {
|
|
9
|
+
2949, 2969, 3021, 2965, 3009, 2949, 2969, 3021, 2965, 3015, 2949, 2975, 3009, 2980, 3021, 2980, 2949, 2980, 2985,
|
|
10
|
+
3006, 2994, 3021, 2949, 2980, 2985, 3021, 2949, 2980, 2993, 3021, 2965, 3009, 2949, 2980, 3007, 2965, 2949, 2980,
|
|
11
|
+
3007, 2994, 3021, 2949, 2980, 3009, 2949, 2980, 3015, 2949, 2980, 3016, 2949, 2984, 3021, 2980, 2949, 2985, 3021,
|
|
12
|
+
2993, 3009, 2949, 2994, 3021, 2994, 2980, 3009, 2949, 2997, 2985, 3021, 2949, 2997, 2992, 2980, 3009, 2949, 2997,
|
|
13
|
+
2992, 3021, 2949, 2997, 2992, 3021, 2965, 2995, 3021, 2949, 2997, 2995, 3021, 2949, 2997, 3016, 2950, 2965, 3007,
|
|
14
|
+
2991, 2950, 2965, 3007, 2991, 3019, 2992, 3021, 2950, 2965, 3009, 2990, 3021, 2951, 2969, 3021, 2965, 3009, 2951,
|
|
15
|
+
2969, 3021, 2965, 3015, 2951, 2975, 2990, 3021, 2951, 2980, 2985, 3006, 2994, 3021, 2951, 2980, 2985, 3016, 2951,
|
|
16
|
+
2980, 2985, 3021, 2951, 2980, 2993, 3021, 2965, 3009, 2951, 2980, 3007, 2994, 3021, 2951, 2980, 3009, 2951, 2980,
|
|
17
|
+
3016, 2951, 2984, 3021, 2980, 2951, 2985, 3021, 2985, 3009, 2990, 3021, 2951, 2986, 3021, 2986, 3019, 2980, 3009,
|
|
18
|
+
2951, 2992, 3009, 2951, 2992, 3009, 2965, 3021, 2965, 3009, 2990, 3021, 2951, 2992, 3009, 2984, 3021, 2980, 2951,
|
|
19
|
+
2992, 3009, 2984, 3021, 2980, 3009, 2951, 2997, 2992, 3021, 2951, 2997, 3016, 2953, 2985, 3021, 2953, 2995, 3021,
|
|
20
|
+
2995, 2953, 2995, 3021, 2995, 2980, 3009, 2953, 2995, 3021, 2995, 2985, 2958, 2984, 3021, 2980, 2958, 2985, 2958,
|
|
21
|
+
2985, 2997, 3015, 2958, 2985, 3009, 2990, 3021, 2958, 2985, 3021, 2958, 2985, 3021, 2985, 2958, 2985, 3021, 2985,
|
|
22
|
+
3009, 2990, 3021, 2958, 2985, 3021, 2986, 2980, 3009, 2958, 2985, 3021, 2986, 2980, 3016, 2958, 2985, 3021, 2993,
|
|
23
|
+
2958, 2985, 3021, 2993, 3009, 2958, 2985, 3021, 2993, 3009, 2990, 3021, 2958, 2994, 3021, 2994, 3006, 2990, 3021,
|
|
24
|
+
2959, 2985, 3021, 2962, 2992, 3009, 2962, 2992, 3015, 2963, 2992, 3021, 2965, 3018, 2979, 3021, 2975, 2965, 3018,
|
|
25
|
+
2979, 3021, 2975, 3009, 2965, 3018, 2995, 3021, 2995, 2970, 2993, 3021, 2993, 3009, 2970, 3007, 2993, 3009, 2970,
|
|
26
|
+
3007, 2994, 2980, 2985, 2980, 3009, 2980, 2985, 3021, 2980, 2997, 3007, 2992, 2980, 3006, 2985, 3021, 2984, 3006,
|
|
27
|
+
2985, 3021, 2984, 3006, 2990, 3021, 2984, 3008, 2986, 2993, 3021, 2993, 3007, 2986, 2993, 3021, 2993, 3007, 2991,
|
|
28
|
+
2986, 2994, 2986, 2994, 3021, 2997, 3015, 2993, 3009, 2986, 3007, 2985, 3021, 2986, 3007, 2985, 3021, 2985, 2992,
|
|
29
|
+
3021, 2986, 3007, 2993, 2986, 3007, 2993, 2965, 3009, 2986, 3015, 2992, 3021, 2986, 3019, 2980, 3009, 2986, 3019,
|
|
30
|
+
2985, 3021, 2993, 2986, 3019, 2994, 2986, 3019, 2994, 3021, 2990, 2975, 3021, 2975, 3009, 2990, 3015, 2990, 2975,
|
|
31
|
+
3021, 2975, 3009, 2990, 3021, 2990, 2993, 3021, 2993, 2990, 2993, 3021, 2993, 3009, 2990, 3021, 2990, 3007, 2965,
|
|
32
|
+
2990, 3007, 2965, 2997, 3009, 2990, 3021, 2990, 3008, 2980, 3009, 2990, 3009, 2980, 2994, 3021, 2990, 3009, 2993,
|
|
33
|
+
3016, 2990, 3015, 2994, 3009, 2990, 3021, 2990, 3015, 2994, 3021, 2991, 3006, 2992, 3021, 2997, 2984, 3021, 2980,
|
|
34
|
+
2997, 2984, 3021, 2980, 3009, 2997, 2992, 3009, 2990, 3021, 2997, 2992, 3016, 2997, 3007, 2975, 2997, 3015, 2979,
|
|
35
|
+
3021, 2975, 3009, 2990, 3021, 2997, 3015, 2993, 3009,
|
|
36
|
+
};
|
|
37
|
+
|
|
38
|
+
static const uint32_t stop_word_offsets[109] = {
|
|
39
|
+
0, 5, 10, 16, 22, 26, 32, 36, 41, 44, 47, 50, 54, 59, 65, 69, 74, 78, 85, 89, 92, 96, 103, 108, 113, 118, 122, 128,
|
|
40
|
+
132, 136, 142, 147, 150, 153, 157, 164, 171, 174, 183, 189, 196, 200, 203, 206, 210, 216, 221, 225, 227, 231, 236,
|
|
41
|
+
239, 243, 250, 256, 262, 266, 271, 278, 285, 288, 291, 294, 297, 302, 308, 313, 318, 322, 325, 329, 332, 336, 340,
|
|
42
|
+
344, 348, 350, 355, 361, 363, 370, 374, 381, 384, 389, 393, 397, 402, 405, 409, 416, 423, 427, 434, 437, 444, 448,
|
|
43
|
+
453, 457, 463, 467, 471, 475, 480, 485, 488, 491, 499, 503,
|
|
44
|
+
};
|
|
45
|
+
|
|
46
|
+
static const uint32_t keep_ranges[8] = {
|
|
47
|
+
48, 57, 65, 90, 97, 122, 2944, 3071,
|
|
48
|
+
};
|
|
49
|
+
|
|
50
|
+
const narsil_language narsil_language_tamil = {
|
|
51
|
+
.name = "tamil",
|
|
52
|
+
.revision = "f859bf3d19447fbc",
|
|
53
|
+
.stop_word_units = stop_word_units,
|
|
54
|
+
.stop_word_offsets = stop_word_offsets,
|
|
55
|
+
.stop_word_count = 108,
|
|
56
|
+
.ascii_keep = {0x03FF000000000000ULL, 0x07FFFFFE07FFFFFEULL},
|
|
57
|
+
.keep_ranges = keep_ranges,
|
|
58
|
+
.keep_range_count = 4,
|
|
59
|
+
.splits_code_points = 0,
|
|
60
|
+
.strips_possessive = 0,
|
|
61
|
+
.normalises_diacritics = 0,
|
|
62
|
+
.min_token_length = 1,
|
|
63
|
+
.ngram_size = 0,
|
|
64
|
+
.normalise = NULL,
|
|
65
|
+
.stem_by_rules = NULL,
|
|
66
|
+
.create_snowball = tamil_create_env,
|
|
67
|
+
.snowball_stem = tamil_stem,
|
|
68
|
+
.close_snowball = tamil_close_env,
|
|
69
|
+
};
|