@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 +1,93 @@
|
|
|
1
|
+
#include "../../analysis/languages.h"
|
|
2
|
+
#include "declarations.h"
|
|
3
|
+
#include "../rules.h"
|
|
4
|
+
#include "../snowball/romanian_stemmer.h"
|
|
5
|
+
|
|
6
|
+
#include <stddef.h>
|
|
7
|
+
#include <stdint.h>
|
|
8
|
+
|
|
9
|
+
static const uint16_t stop_word_units[1050] = {
|
|
10
|
+
97, 99, 101, 97, 97, 99, 101, 97, 115, 116, 97, 97, 99, 101, 97, 115, 116, 259, 97, 99, 101, 101, 97, 97, 99, 101,
|
|
11
|
+
105, 97, 99, 101, 105, 97, 97, 99, 101, 108, 97, 99, 101, 108, 97, 97, 99, 101, 108, 101, 97, 99, 101, 108, 101, 97,
|
|
12
|
+
97, 99, 101, 115, 116, 97, 99, 101, 115, 116, 97, 97, 99, 101, 115, 116, 101, 97, 99, 101, 115, 116, 101, 97, 97, 99,
|
|
13
|
+
101, 351, 116, 105, 97, 99, 101, 351, 116, 105, 97, 97, 99, 101, 537, 116, 105, 97, 99, 101, 537, 116, 105, 97, 97,
|
|
14
|
+
99, 111, 108, 111, 97, 99, 117, 109, 97, 105, 97, 105, 99, 105, 97, 108, 97, 108, 101, 97, 108, 116, 99, 101, 118, 97,
|
|
15
|
+
97, 108, 116, 99, 105, 110, 101, 118, 97, 97, 109, 97, 114, 97, 114, 101, 97, 115, 101, 109, 101, 110, 101, 97, 97,
|
|
16
|
+
115, 116, 97, 97, 115, 116, 101, 97, 97, 115, 116, 259, 122, 105, 97, 115, 117, 112, 114, 97, 97, 117, 97, 118, 101,
|
|
17
|
+
97, 97, 118, 101, 109, 97, 118, 101, 355, 105, 97, 118, 101, 539, 105, 97, 122, 105, 97, 351, 97, 351, 97, 100, 97,
|
|
18
|
+
114, 97, 355, 105, 97, 537, 97, 537, 97, 100, 97, 114, 97, 539, 105, 98, 105, 110, 101, 99, 97, 99, 97, 114, 101, 99,
|
|
19
|
+
101, 99, 101, 108, 99, 101, 118, 97, 99, 104, 105, 97, 114, 99, 105, 110, 101, 99, 105, 110, 101, 118, 97, 99, 111,
|
|
20
|
+
110, 116, 114, 97, 99, 117, 99, 117, 109, 99, 117, 109, 118, 97, 99, 226, 110, 100, 99, 226, 116, 99, 226, 116, 101,
|
|
21
|
+
99, 226, 355, 105, 99, 226, 539, 105, 99, 259, 99, 259, 99, 105, 99, 259, 114, 101, 105, 99, 259, 114, 111, 114, 99,
|
|
22
|
+
259, 114, 117, 105, 99, 259, 116, 114, 101, 100, 97, 100, 97, 99, 259, 100, 97, 114, 100, 97, 116, 111, 114, 105, 116,
|
|
23
|
+
259, 100, 97, 116, 259, 100, 101, 100, 101, 99, 105, 100, 101, 106, 97, 100, 101, 111, 97, 114, 101, 99, 101, 100,
|
|
24
|
+
101, 112, 97, 114, 116, 101, 100, 101, 351, 105, 100, 101, 537, 105, 100, 105, 110, 100, 105, 110, 116, 114, 101, 100,
|
|
25
|
+
111, 105, 100, 111, 117, 259, 100, 114, 101, 112, 116, 100, 117, 112, 259, 101, 97, 101, 105, 101, 108, 101, 108, 101,
|
|
26
|
+
101, 114, 97, 109, 101, 115, 116, 101, 101, 117, 101, 351, 116, 105, 101, 537, 116, 105, 102, 97, 99, 101, 102, 105,
|
|
27
|
+
102, 105, 101, 102, 105, 101, 99, 97, 114, 101, 102, 105, 105, 102, 105, 109, 102, 105, 117, 102, 105, 355, 105, 102,
|
|
28
|
+
105, 539, 105, 102, 259, 114, 259, 105, 97, 114, 105, 101, 114, 105, 108, 97, 108, 101, 108, 105, 108, 111, 114, 108,
|
|
29
|
+
117, 105, 108, 226, 110, 103, 259, 109, 97, 105, 109, 101, 97, 109, 101, 105, 109, 101, 108, 101, 109, 101, 114, 101,
|
|
30
|
+
117, 109, 101, 117, 109, 105, 109, 105, 101, 109, 105, 110, 101, 109, 117, 108, 116, 109, 117, 108, 116, 259, 109,
|
|
31
|
+
117, 108, 355, 105, 109, 117, 108, 539, 105, 109, 226, 105, 110, 101, 109, 259, 110, 101, 110, 101, 118, 111, 105,
|
|
32
|
+
101, 110, 105, 99, 105, 110, 105, 99, 259, 105, 101, 114, 105, 110, 105, 109, 101, 110, 105, 110, 105, 109, 105, 99,
|
|
33
|
+
110, 105, 351, 116, 101, 110, 105, 537, 116, 101, 110, 111, 97, 115, 116, 114, 101, 110, 111, 97, 115, 116, 114, 259,
|
|
34
|
+
110, 111, 105, 110, 111, 115, 116, 114, 117, 110, 111, 117, 259, 110, 111, 351, 116, 114, 105, 110, 111, 537, 116,
|
|
35
|
+
114, 105, 110, 117, 111, 114, 105, 111, 114, 105, 99, 97, 114, 101, 111, 114, 105, 99, 101, 111, 114, 105, 99, 105,
|
|
36
|
+
110, 101, 111, 114, 105, 99, 117, 109, 111, 114, 105, 99, 226, 110, 100, 111, 114, 105, 99, 226, 116, 111, 114, 105,
|
|
37
|
+
117, 110, 100, 101, 112, 101, 112, 101, 110, 116, 114, 117, 112, 101, 115, 116, 101, 112, 111, 97, 116, 101, 112, 111,
|
|
38
|
+
116, 112, 114, 101, 97, 112, 114, 105, 109, 97, 112, 114, 105, 109, 117, 108, 112, 114, 105, 110, 112, 117, 355, 105,
|
|
39
|
+
110, 112, 117, 539, 105, 110, 112, 226, 110, 259, 115, 97, 115, 97, 108, 101, 115, 97, 117, 115, 101, 115, 112, 114,
|
|
40
|
+
101, 115, 117, 98, 115, 117, 110, 116, 115, 117, 110, 116, 101, 109, 115, 117, 110, 116, 101, 355, 105, 115, 117, 110,
|
|
41
|
+
116, 101, 539, 105, 115, 259, 115, 259, 105, 115, 259, 117, 116, 97, 116, 97, 108, 101, 116, 101, 116, 105, 109, 112,
|
|
42
|
+
116, 105, 110, 101, 116, 111, 97, 116, 101, 116, 111, 97, 116, 259, 116, 111, 116, 116, 111, 116, 117, 351, 105, 116,
|
|
43
|
+
111, 116, 117, 537, 105, 116, 111, 355, 105, 116, 111, 539, 105, 116, 114, 101, 105, 116, 117, 116, 259, 105, 116,
|
|
44
|
+
259, 117, 117, 110, 117, 110, 97, 117, 110, 100, 101, 117, 110, 100, 101, 118, 97, 117, 110, 101, 105, 117, 110, 101,
|
|
45
|
+
108, 101, 117, 110, 105, 105, 117, 110, 111, 114, 117, 110, 117, 117, 110, 117, 105, 117, 110, 117, 108, 118, 105,
|
|
46
|
+
118, 111, 97, 115, 116, 114, 101, 118, 111, 97, 115, 116, 114, 259, 118, 111, 105, 118, 111, 115, 116, 114, 117, 118,
|
|
47
|
+
111, 117, 259, 118, 111, 351, 116, 114, 105, 118, 111, 537, 116, 114, 105, 118, 114, 101, 109, 101, 118, 114, 101,
|
|
48
|
+
111, 118, 114, 101, 117, 110, 118, 259, 122, 101, 99, 101, 122, 105, 238, 105, 238, 108, 238, 109, 105, 238, 109, 112,
|
|
49
|
+
111, 116, 114, 105, 118, 97, 238, 110, 238, 110, 97, 105, 110, 116, 101, 238, 110, 97, 105, 110, 116, 101, 97, 238,
|
|
50
|
+
110, 99, 111, 116, 114, 111, 238, 110, 99, 226, 116, 238, 110, 116, 114, 101, 238, 110, 116, 114, 117, 99, 226, 116,
|
|
51
|
+
238, 355, 105, 238, 539, 105, 259, 108, 97, 259, 108, 101, 97, 259, 115, 116, 97, 259, 115, 116, 101, 97, 351, 97,
|
|
52
|
+
112, 116, 101, 351, 97, 115, 101, 351, 105, 355, 105, 355, 105, 101, 537, 97, 112, 116, 101, 537, 97, 115, 101, 537,
|
|
53
|
+
105, 539, 105, 539, 105, 101,
|
|
54
|
+
};
|
|
55
|
+
|
|
56
|
+
static const uint32_t stop_word_offsets[248] = {
|
|
57
|
+
0, 4, 11, 18, 23, 27, 32, 36, 41, 46, 52, 57, 63, 69, 76, 82, 89, 95, 102, 107, 111, 113, 117, 119, 122, 129, 138,
|
|
58
|
+
140, 142, 145, 153, 157, 162, 168, 174, 176, 180, 184, 189, 194, 197, 199, 205, 208, 210, 216, 219, 223, 225, 229,
|
|
59
|
+
231, 234, 238, 243, 247, 253, 259, 261, 264, 269, 273, 276, 280, 284, 288, 290, 294, 299, 304, 309, 314, 316, 320,
|
|
60
|
+
323, 331, 335, 337, 341, 345, 353, 360, 364, 368, 371, 377, 380, 384, 389, 393, 395, 397, 399, 402, 406, 410, 412,
|
|
61
|
+
416, 420, 424, 426, 429, 436, 439, 442, 445, 449, 453, 457, 460, 464, 466, 468, 470, 473, 476, 481, 484, 487, 490,
|
|
62
|
+
494, 499, 502, 504, 507, 511, 515, 520, 525, 530, 535, 537, 539, 545, 549, 557, 563, 568, 573, 578, 585, 592, 595,
|
|
63
|
+
601, 605, 611, 617, 619, 622, 629, 634, 641, 647, 654, 660, 667, 669, 675, 680, 685, 688, 692, 697, 703, 707, 712,
|
|
64
|
+
717, 721, 723, 727, 730, 732, 736, 739, 743, 749, 756, 763, 765, 768, 771, 773, 777, 779, 783, 787, 792, 797, 800,
|
|
65
|
+
806, 812, 816, 820, 824, 826, 829, 832, 834, 837, 841, 847, 851, 856, 860, 864, 867, 871, 875, 877, 884, 891, 894,
|
|
66
|
+
900, 904, 910, 916, 921, 925, 930, 932, 936, 938, 940, 942, 945, 954, 956, 963, 971, 978, 983, 988, 996, 999, 1002,
|
|
67
|
+
1005, 1009, 1013, 1018, 1023, 1027, 1029, 1031, 1034, 1039, 1043, 1045, 1047, 1050,
|
|
68
|
+
};
|
|
69
|
+
|
|
70
|
+
static const uint32_t keep_ranges[22] = {
|
|
71
|
+
48, 57, 65, 90, 97, 122, 194, 194, 206, 206, 226, 226, 238, 238, 258, 259, 350, 351, 354, 355, 536, 539,
|
|
72
|
+
};
|
|
73
|
+
|
|
74
|
+
const narsil_language narsil_language_romanian = {
|
|
75
|
+
.name = "romanian",
|
|
76
|
+
.revision = "daf797aac5310bd4",
|
|
77
|
+
.stop_word_units = stop_word_units,
|
|
78
|
+
.stop_word_offsets = stop_word_offsets,
|
|
79
|
+
.stop_word_count = 247,
|
|
80
|
+
.ascii_keep = {0x03FF000000000000ULL, 0x07FFFFFE07FFFFFEULL},
|
|
81
|
+
.keep_ranges = keep_ranges,
|
|
82
|
+
.keep_range_count = 11,
|
|
83
|
+
.splits_code_points = 0,
|
|
84
|
+
.strips_possessive = 0,
|
|
85
|
+
.normalises_diacritics = 0,
|
|
86
|
+
.min_token_length = 1,
|
|
87
|
+
.ngram_size = 0,
|
|
88
|
+
.normalise = narsil_romanian_normalise,
|
|
89
|
+
.stem_by_rules = NULL,
|
|
90
|
+
.create_snowball = romanian_create_env,
|
|
91
|
+
.snowball_stem = romanian_stem,
|
|
92
|
+
.close_snowball = romanian_close_env,
|
|
93
|
+
};
|
|
@@ -0,0 +1,76 @@
|
|
|
1
|
+
#include "../../analysis/languages.h"
|
|
2
|
+
#include "declarations.h"
|
|
3
|
+
#include "../snowball/russian_stemmer.h"
|
|
4
|
+
|
|
5
|
+
#include <stddef.h>
|
|
6
|
+
#include <stdint.h>
|
|
7
|
+
|
|
8
|
+
static const uint16_t stop_word_units[595] = {
|
|
9
|
+
1072, 1073, 1077, 1079, 1073, 1086, 1083, 1077, 1077, 1073, 1086, 1083, 1100, 1096, 1077, 1073, 1091, 1076, 1077,
|
|
10
|
+
1090, 1073, 1091, 1076, 1090, 1086, 1073, 1099, 1073, 1099, 1083, 1073, 1099, 1083, 1072, 1073, 1099, 1083, 1080,
|
|
11
|
+
1073, 1099, 1083, 1086, 1073, 1099, 1090, 1100, 1074, 1074, 1072, 1084, 1074, 1072, 1089, 1074, 1076, 1088, 1091,
|
|
12
|
+
1075, 1074, 1077, 1076, 1100, 1074, 1086, 1074, 1086, 1090, 1074, 1087, 1088, 1086, 1095, 1077, 1084, 1074, 1089,
|
|
13
|
+
1077, 1074, 1089, 1077, 1075, 1076, 1072, 1074, 1089, 1077, 1075, 1086, 1074, 1089, 1077, 1093, 1074, 1089, 1102,
|
|
14
|
+
1074, 1099, 1075, 1076, 1077, 1075, 1086, 1074, 1086, 1088, 1080, 1083, 1076, 1072, 1076, 1072, 1078, 1077, 1076,
|
|
15
|
+
1074, 1072, 1076, 1083, 1103, 1076, 1086, 1076, 1088, 1091, 1075, 1086, 1081, 1077, 1075, 1086, 1077, 1077, 1077,
|
|
16
|
+
1081, 1077, 1084, 1091, 1077, 1089, 1083, 1080, 1077, 1089, 1090, 1100, 1077, 1097, 1077, 1078, 1078, 1077, 1078,
|
|
17
|
+
1080, 1079, 1085, 1100, 1079, 1072, 1079, 1072, 1095, 1077, 1084, 1079, 1076, 1077, 1089, 1100, 1080, 1080, 1079,
|
|
18
|
+
1080, 1083, 1080, 1080, 1084, 1080, 1085, 1086, 1075, 1076, 1072, 1080, 1093, 1082, 1082, 1072, 1078, 1077, 1090,
|
|
19
|
+
1089, 1103, 1082, 1072, 1082, 1082, 1072, 1082, 1072, 1103, 1082, 1072, 1082, 1086, 1081, 1082, 1086, 1075, 1076,
|
|
20
|
+
1072, 1082, 1086, 1085, 1077, 1095, 1085, 1086, 1082, 1090, 1086, 1082, 1091, 1076, 1072, 1083, 1080, 1083, 1091,
|
|
21
|
+
1095, 1096, 1077, 1084, 1077, 1078, 1076, 1091, 1084, 1077, 1085, 1103, 1084, 1085, 1077, 1084, 1085, 1086, 1075,
|
|
22
|
+
1086, 1084, 1086, 1078, 1077, 1090, 1084, 1086, 1078, 1085, 1086, 1084, 1086, 1081, 1084, 1086, 1103, 1084, 1099,
|
|
23
|
+
1085, 1072, 1085, 1072, 1076, 1085, 1072, 1076, 1086, 1085, 1072, 1082, 1086, 1085, 1077, 1094, 1085, 1072, 1089,
|
|
24
|
+
1085, 1077, 1085, 1077, 1075, 1086, 1085, 1077, 1077, 1085, 1077, 1081, 1085, 1077, 1083, 1100, 1079, 1103, 1085,
|
|
25
|
+
1077, 1090, 1085, 1080, 1085, 1080, 1073, 1091, 1076, 1100, 1085, 1080, 1082, 1086, 1075, 1076, 1072, 1085, 1080,
|
|
26
|
+
1084, 1085, 1080, 1093, 1085, 1080, 1095, 1077, 1075, 1086, 1085, 1086, 1085, 1091, 1086, 1086, 1073, 1086, 1076,
|
|
27
|
+
1080, 1085, 1086, 1085, 1086, 1085, 1072, 1086, 1085, 1080, 1086, 1087, 1103, 1090, 1100, 1086, 1090, 1087, 1077,
|
|
28
|
+
1088, 1077, 1076, 1087, 1086, 1087, 1086, 1076, 1087, 1086, 1089, 1083, 1077, 1087, 1086, 1090, 1086, 1084, 1087,
|
|
29
|
+
1086, 1090, 1086, 1084, 1091, 1087, 1086, 1095, 1090, 1080, 1087, 1088, 1080, 1087, 1088, 1086, 1088, 1072, 1079,
|
|
30
|
+
1088, 1072, 1079, 1074, 1077, 1089, 1089, 1072, 1084, 1089, 1074, 1086, 1102, 1089, 1077, 1073, 1077, 1089, 1077,
|
|
31
|
+
1073, 1103, 1089, 1077, 1075, 1086, 1076, 1085, 1103, 1089, 1077, 1081, 1095, 1072, 1089, 1089, 1082, 1072, 1079,
|
|
32
|
+
1072, 1083, 1089, 1082, 1072, 1079, 1072, 1083, 1072, 1089, 1082, 1072, 1079, 1072, 1090, 1100, 1089, 1086, 1089,
|
|
33
|
+
1086, 1074, 1089, 1077, 1084, 1090, 1072, 1082, 1090, 1072, 1082, 1086, 1081, 1090, 1072, 1084, 1090, 1077, 1073,
|
|
34
|
+
1103, 1090, 1077, 1084, 1090, 1077, 1087, 1077, 1088, 1100, 1090, 1086, 1090, 1086, 1075, 1076, 1072, 1090, 1086,
|
|
35
|
+
1075, 1086, 1090, 1086, 1078, 1077, 1090, 1086, 1083, 1100, 1082, 1086, 1090, 1086, 1084, 1090, 1086, 1090, 1090,
|
|
36
|
+
1088, 1080, 1090, 1091, 1090, 1090, 1099, 1091, 1091, 1078, 1091, 1078, 1077, 1093, 1086, 1088, 1086, 1096, 1086,
|
|
37
|
+
1093, 1086, 1090, 1100, 1095, 1077, 1075, 1086, 1095, 1077, 1083, 1086, 1074, 1077, 1082, 1095, 1077, 1084, 1095,
|
|
38
|
+
1077, 1088, 1077, 1079, 1095, 1090, 1086, 1095, 1090, 1086, 1073, 1095, 1090, 1086, 1073, 1099, 1095, 1091, 1090,
|
|
39
|
+
1100, 1101, 1090, 1080, 1101, 1090, 1086, 1075, 1086, 1101, 1090, 1086, 1081, 1101, 1090, 1086, 1084, 1101, 1090,
|
|
40
|
+
1086, 1090, 1101, 1090, 1091, 1103,
|
|
41
|
+
};
|
|
42
|
+
|
|
43
|
+
static const uint32_t stop_word_offsets[160] = {
|
|
44
|
+
0, 1, 4, 9, 15, 20, 25, 27, 30, 34, 38, 42, 46, 47, 50, 53, 58, 62, 64, 67, 74, 77, 83, 88, 92, 95, 97, 100, 107, 109,
|
|
45
|
+
113, 116, 119, 121, 127, 130, 132, 134, 137, 141, 145, 148, 149, 151, 156, 158, 163, 168, 169, 171, 174, 176, 182,
|
|
46
|
+
184, 185, 192, 195, 200, 205, 210, 217, 220, 224, 226, 231, 236, 240, 243, 248, 253, 258, 261, 264, 266, 268, 271,
|
|
47
|
+
275, 282, 285, 287, 291, 294, 297, 303, 306, 308, 314, 321, 324, 327, 333, 335, 337, 338, 340, 344, 346, 349, 352,
|
|
48
|
+
357, 359, 364, 366, 369, 374, 379, 385, 390, 393, 396, 399, 404, 405, 408, 412, 416, 420, 427, 433, 439, 446, 453,
|
|
49
|
+
455, 461, 464, 469, 472, 476, 479, 485, 487, 492, 496, 500, 506, 509, 512, 515, 518, 520, 521, 523, 526, 532, 536,
|
|
50
|
+
540, 547, 550, 555, 558, 562, 567, 571, 574, 579, 583, 587, 591, 594, 595,
|
|
51
|
+
};
|
|
52
|
+
|
|
53
|
+
static const uint32_t keep_ranges[22] = {
|
|
54
|
+
45, 45, 48, 57, 65, 90, 97, 122, 1025, 1025, 1030, 1030, 1040, 1103, 1105, 1105, 1110, 1110, 1122, 1123, 7296, 7303,
|
|
55
|
+
};
|
|
56
|
+
|
|
57
|
+
const narsil_language narsil_language_russian = {
|
|
58
|
+
.name = "russian",
|
|
59
|
+
.revision = "87767119dd36215c",
|
|
60
|
+
.stop_word_units = stop_word_units,
|
|
61
|
+
.stop_word_offsets = stop_word_offsets,
|
|
62
|
+
.stop_word_count = 159,
|
|
63
|
+
.ascii_keep = {0x03FF200000000000ULL, 0x07FFFFFE07FFFFFEULL},
|
|
64
|
+
.keep_ranges = keep_ranges,
|
|
65
|
+
.keep_range_count = 11,
|
|
66
|
+
.splits_code_points = 0,
|
|
67
|
+
.strips_possessive = 0,
|
|
68
|
+
.normalises_diacritics = 0,
|
|
69
|
+
.min_token_length = 1,
|
|
70
|
+
.ngram_size = 0,
|
|
71
|
+
.normalise = NULL,
|
|
72
|
+
.stem_by_rules = NULL,
|
|
73
|
+
.create_snowball = russian_create_env,
|
|
74
|
+
.snowball_stem = russian_stem,
|
|
75
|
+
.close_snowball = russian_close_env,
|
|
76
|
+
};
|
|
@@ -0,0 +1,47 @@
|
|
|
1
|
+
#include "../../analysis/languages.h"
|
|
2
|
+
#include "declarations.h"
|
|
3
|
+
|
|
4
|
+
#include <stddef.h>
|
|
5
|
+
#include <stdint.h>
|
|
6
|
+
|
|
7
|
+
static const uint16_t stop_word_units[149] = {
|
|
8
|
+
39, 111, 97, 97, 39, 101, 97, 39, 117, 97, 101, 97, 102, 97, 105, 97, 105, 97, 116, 117, 97, 117, 97, 101, 102, 111,
|
|
9
|
+
39, 105, 105, 105, 97, 105, 102, 111, 105, 110, 97, 105, 115, 105, 108, 97, 110, 97, 108, 97, 116, 111, 117, 108, 97,
|
|
10
|
+
117, 97, 108, 97, 118, 97, 108, 101, 108, 101, 97, 108, 101, 110, 97, 108, 101, 110, 101, 105, 108, 111, 39, 117, 108,
|
|
11
|
+
111, 110, 97, 108, 111, 117, 108, 275, 109, 97, 109, 97, 105, 109, 97, 116, 111, 117, 109, 97, 117, 97, 109, 111, 110,
|
|
12
|
+
97, 110, 101, 105, 110, 105, 110, 105, 115, 105, 111, 111, 101, 111, 110, 97, 112, 101, 97, 112, 111, 39, 111, 115,
|
|
13
|
+
97, 115, 101, 115, 257, 116, 97, 116, 111, 117, 116, 97, 117, 97, 116, 101, 116, 111, 101, 117, 97, 117, 109, 97,
|
|
14
|
+
};
|
|
15
|
+
|
|
16
|
+
static const uint32_t stop_word_offsets[52] = {
|
|
17
|
+
0, 2, 3, 6, 9, 11, 15, 17, 20, 23, 24, 28, 29, 31, 34, 37, 40, 44, 49, 53, 57, 59, 62, 66, 71, 75, 79, 82, 84, 86, 89,
|
|
18
|
+
94, 98, 100, 102, 105, 107, 111, 112, 114, 117, 120, 124, 126, 128, 130, 135, 139, 141, 144, 146, 149,
|
|
19
|
+
};
|
|
20
|
+
|
|
21
|
+
static const uint32_t keep_ranges[76] = {
|
|
22
|
+
39, 39, 48, 57, 65, 90, 97, 122, 170, 170, 186, 186, 192, 214, 216, 246, 248, 696, 736, 740, 7424, 7461, 7468, 7516,
|
|
23
|
+
7522, 7525, 7531, 7543, 7545, 7614, 7680, 7935, 8305, 8305, 8319, 8319, 8336, 8348, 8490, 8491, 8498, 8498, 8526,
|
|
24
|
+
8526, 8544, 8584, 11360, 11391, 42786, 42887, 42891, 42972, 42993, 43007, 43824, 43866, 43868, 43876, 43878, 43881,
|
|
25
|
+
64256, 64262, 65313, 65338, 65345, 65370, 67456, 67461, 67463, 67504, 67506, 67514, 122624, 122654, 122661, 122666,
|
|
26
|
+
};
|
|
27
|
+
|
|
28
|
+
const narsil_language narsil_language_samoan = {
|
|
29
|
+
.name = "samoan",
|
|
30
|
+
.revision = "1",
|
|
31
|
+
.stop_word_units = stop_word_units,
|
|
32
|
+
.stop_word_offsets = stop_word_offsets,
|
|
33
|
+
.stop_word_count = 51,
|
|
34
|
+
.ascii_keep = {0x03FF008000000000ULL, 0x07FFFFFE07FFFFFEULL},
|
|
35
|
+
.keep_ranges = keep_ranges,
|
|
36
|
+
.keep_range_count = 38,
|
|
37
|
+
.splits_code_points = 1,
|
|
38
|
+
.strips_possessive = 0,
|
|
39
|
+
.normalises_diacritics = 0,
|
|
40
|
+
.min_token_length = 1,
|
|
41
|
+
.ngram_size = 0,
|
|
42
|
+
.normalise = NULL,
|
|
43
|
+
.stem_by_rules = NULL,
|
|
44
|
+
.create_snowball = NULL,
|
|
45
|
+
.snowball_stem = NULL,
|
|
46
|
+
.close_snowball = NULL,
|
|
47
|
+
};
|
|
@@ -0,0 +1,53 @@
|
|
|
1
|
+
#include "../../analysis/languages.h"
|
|
2
|
+
#include "declarations.h"
|
|
3
|
+
#include "../rules.h"
|
|
4
|
+
|
|
5
|
+
#include <stddef.h>
|
|
6
|
+
#include <stdint.h>
|
|
7
|
+
|
|
8
|
+
static const uint16_t stop_word_units[244] = {
|
|
9
|
+
2309, 2340, 2309, 2340, 2307, 2309, 2340, 2381, 2352, 2309, 2341, 2309, 2341, 2357, 2366, 2309, 2343, 2367, 2309,
|
|
10
|
+
2344, 2369, 2309, 2344, 2381, 2351, 2340, 2381, 2309, 2344, 2381, 2351, 2340, 2381, 2352, 2309, 2346, 2352, 2350,
|
|
11
|
+
2381, 2309, 2346, 2367, 2309, 2349, 2367, 2309, 2351, 2350, 2381, 2309, 2357, 2309, 2360, 2381, 2340, 2367, 2309,
|
|
12
|
+
2360, 2381, 2351, 2309, 2360, 2381, 2351, 2366, 2307, 2310, 2311, 2340, 2307, 2311, 2340, 2367, 2311, 2342, 2350,
|
|
13
|
+
2381, 2311, 2351, 2350, 2381, 2311, 2357, 2313, 2340, 2313, 2346, 2319, 2340, 2340, 2381, 2319, 2357, 2319, 2357,
|
|
14
|
+
2350, 2381, 2319, 2359, 2307, 2319, 2359, 2366, 2325, 2341, 2350, 2381, 2325, 2342, 2366, 2325, 2367, 2334, 2381,
|
|
15
|
+
2330, 2367, 2340, 2381, 2325, 2367, 2350, 2381, 2325, 2369, 2340, 2307, 2325, 2369, 2340, 2381, 2352, 2326, 2354,
|
|
16
|
+
2369, 2330, 2340, 2340, 2307, 2340, 2340, 2381, 2340, 2340, 2381, 2352, 2340, 2341, 2366, 2340, 2352, 2381, 2361,
|
|
17
|
+
2367, 2340, 2357, 2340, 2360, 2381, 2351, 2340, 2360, 2381, 2351, 2366, 2307, 2340, 2369, 2344, 2344, 2366, 2350,
|
|
18
|
+
2344, 2367, 2307, 2344, 2369, 2346, 2352, 2344, 2381, 2340, 2369, 2346, 2352, 2367, 2346, 2369, 2344, 2307, 2346,
|
|
19
|
+
2381, 2352, 2340, 2367, 2349, 2357, 2340, 2367, 2350, 2350, 2351, 2340, 2307, 2351, 2340, 2381, 2351, 2340, 2381,
|
|
20
|
+
2352, 2351, 2341, 2366, 2351, 2342, 2367, 2357, 2366, 2357, 2367, 2344, 2366, 2360, 2307, 2360, 2352, 2381, 2357,
|
|
21
|
+
2340, 2381, 2352, 2360, 2352, 2381, 2357, 2350, 2381, 2360, 2361, 2360, 2366, 2361, 2361, 2367,
|
|
22
|
+
};
|
|
23
|
+
|
|
24
|
+
static const uint32_t stop_word_offsets[72] = {
|
|
25
|
+
0, 2, 5, 9, 11, 15, 18, 21, 27, 34, 39, 42, 45, 49, 51, 56, 60, 66, 67, 70, 73, 77, 81, 83, 85, 87, 91, 93, 97, 100,
|
|
26
|
+
103, 107, 110, 118, 122, 126, 131, 134, 135, 138, 141, 145, 148, 153, 155, 159, 165, 167, 168, 171, 174, 176, 182,
|
|
27
|
+
185, 189, 194, 198, 200, 203, 206, 210, 213, 216, 218, 222, 224, 231, 237, 239, 241, 242, 244,
|
|
28
|
+
};
|
|
29
|
+
|
|
30
|
+
static const uint32_t keep_ranges[14] = {
|
|
31
|
+
48, 57, 65, 90, 97, 122, 2304, 2403, 2406, 2431, 7746, 7747, 7770, 7771,
|
|
32
|
+
};
|
|
33
|
+
|
|
34
|
+
const narsil_language narsil_language_sanskrit = {
|
|
35
|
+
.name = "sanskrit",
|
|
36
|
+
.revision = "1",
|
|
37
|
+
.stop_word_units = stop_word_units,
|
|
38
|
+
.stop_word_offsets = stop_word_offsets,
|
|
39
|
+
.stop_word_count = 71,
|
|
40
|
+
.ascii_keep = {0x03FF000000000000ULL, 0x07FFFFFE07FFFFFEULL},
|
|
41
|
+
.keep_ranges = keep_ranges,
|
|
42
|
+
.keep_range_count = 7,
|
|
43
|
+
.splits_code_points = 0,
|
|
44
|
+
.strips_possessive = 0,
|
|
45
|
+
.normalises_diacritics = 0,
|
|
46
|
+
.min_token_length = 1,
|
|
47
|
+
.ngram_size = 0,
|
|
48
|
+
.normalise = NULL,
|
|
49
|
+
.stem_by_rules = narsil_sanskrit_stem,
|
|
50
|
+
.create_snowball = NULL,
|
|
51
|
+
.snowball_stem = NULL,
|
|
52
|
+
.close_snowball = NULL,
|
|
53
|
+
};
|
|
@@ -0,0 +1,119 @@
|
|
|
1
|
+
#include "../../analysis/languages.h"
|
|
2
|
+
#include "declarations.h"
|
|
3
|
+
|
|
4
|
+
#include <stddef.h>
|
|
5
|
+
#include <stdint.h>
|
|
6
|
+
|
|
7
|
+
static const uint16_t stop_word_units[1603] = {
|
|
8
|
+
39, 97, 100, 39, 97, 114, 39, 100, 39, 103, 39, 103, 97, 39, 103, 97, 109, 39, 103, 97, 110, 39, 103, 97, 114, 39,
|
|
9
|
+
103, 117, 114, 39, 109, 39, 110, 39, 110, 97, 39, 110, 97, 100, 39, 110, 97, 109, 39, 110, 97, 110, 39, 110, 97, 114,
|
|
10
|
+
39, 110, 117, 97, 105, 114, 39, 110, 117, 114, 39, 115, 39, 115, 97, 39, 115, 97, 110, 39, 115, 97, 110, 110, 39, 115,
|
|
11
|
+
101, 39, 115, 110, 97, 97, 97, 39, 97, 39, 100, 97, 39, 109, 97, 45, 99, 104, 232, 105, 108, 101, 97, 45, 115, 101,
|
|
12
|
+
111, 97, 45, 115, 105, 110, 97, 45, 115, 105, 117, 100, 97, 99, 39, 97, 99, 97, 97, 99, 97, 45, 115, 97, 110, 97, 99,
|
|
13
|
+
97, 115, 97, 110, 97, 99, 104, 97, 103, 97, 103, 97, 100, 97, 103, 97, 100, 45, 115, 97, 97, 103, 97, 100, 115, 39,
|
|
14
|
+
97, 103, 97, 100, 115, 97, 97, 103, 97, 105, 98, 104, 97, 103, 97, 105, 98, 104, 115, 101, 97, 103, 97, 105, 110, 110,
|
|
15
|
+
97, 103, 97, 105, 110, 110, 101, 97, 103, 97, 109, 97, 103, 97, 109, 45, 115, 97, 97, 103, 97, 109, 115, 39, 97, 103,
|
|
16
|
+
97, 109, 115, 97, 97, 103, 117, 115, 97, 105, 99, 101, 97, 105, 99, 101, 45, 115, 101, 97, 105, 99, 101, 115, 101, 97,
|
|
17
|
+
105, 103, 97, 105, 103, 39, 97, 105, 103, 101, 97, 105, 103, 101, 45, 115, 97, 110, 97, 105, 103, 101, 115, 97, 110,
|
|
18
|
+
97, 105, 114, 97, 105, 114, 45, 115, 97, 110, 97, 105, 114, 115, 97, 110, 97, 109, 97, 110, 97, 110, 110, 97, 110,
|
|
19
|
+
110, 97, 100, 97, 110, 110, 97, 109, 97, 110, 110, 97, 109, 45, 115, 39, 97, 110, 110, 97, 109, 115, 97, 97, 110, 110,
|
|
20
|
+
115, 97, 110, 110, 116, 97, 97, 111, 110, 97, 114, 97, 115, 97, 115, 97, 100, 97, 115, 100, 97, 97, 115, 116, 97, 98,
|
|
21
|
+
39, 98, 104, 111, 98, 104, 111, 110, 98, 104, 117, 97, 105, 100, 104, 101, 98, 104, 117, 97, 105, 110, 110, 98, 104,
|
|
22
|
+
117, 97, 105, 112, 101, 98, 104, 117, 97, 105, 116, 104, 101, 98, 104, 117, 97, 112, 97, 98, 104, 117, 114, 98, 114,
|
|
23
|
+
105, 116, 104, 98, 114, 236, 98, 117, 99, 39, 224, 99, 97, 114, 99, 97, 114, 115, 111, 110, 99, 104, 97, 99, 104, 97,
|
|
24
|
+
110, 99, 104, 105, 111, 110, 110, 99, 104, 111, 105, 114, 99, 104, 111, 110, 99, 104, 117, 110, 99, 104, 232, 105,
|
|
25
|
+
108, 101, 99, 104, 233, 105, 108, 101, 99, 104, 242, 105, 114, 99, 105, 97, 99, 105, 97, 109, 97, 114, 99, 111, 45,
|
|
26
|
+
100, 104, 105, 117, 98, 104, 99, 117, 105, 100, 101, 99, 117, 105, 110, 99, 117, 105, 110, 39, 99, 117, 105, 110, 101,
|
|
27
|
+
99, 224, 99, 224, 39, 99, 224, 105, 108, 99, 224, 105, 116, 99, 224, 105, 116, 39, 99, 224, 105, 116, 101, 99, 242,
|
|
28
|
+
99, 243, 100, 39, 100, 97, 100, 101, 100, 104, 39, 100, 104, 97, 100, 104, 97, 105, 98, 104, 100, 104, 97, 105, 98,
|
|
29
|
+
104, 45, 115, 97, 110, 100, 104, 97, 105, 98, 104, 115, 97, 110, 100, 104, 97, 110, 100, 104, 97, 115, 97, 110, 100,
|
|
30
|
+
104, 101, 100, 104, 101, 110, 100, 104, 101, 116, 104, 100, 104, 105, 100, 104, 105, 111, 109, 100, 104, 105, 111,
|
|
31
|
+
116, 100, 104, 105, 116, 104, 100, 104, 105, 117, 98, 104, 100, 104, 111, 109, 104, 100, 104, 111, 109, 104, 45, 115,
|
|
32
|
+
39, 100, 104, 111, 109, 104, 115, 97, 100, 104, 117, 39, 115, 97, 100, 104, 117, 105, 98, 104, 100, 104, 117, 105, 98,
|
|
33
|
+
104, 115, 101, 100, 104, 117, 105, 110, 110, 100, 104, 117, 105, 110, 110, 101, 100, 104, 117, 105, 116, 100, 104,
|
|
34
|
+
117, 116, 100, 104, 117, 116, 45, 115, 97, 100, 104, 117, 116, 115, 97, 100, 104, 224, 100, 104, 224, 45, 115, 97,
|
|
35
|
+
110, 100, 104, 224, 115, 97, 110, 100, 104, 242, 109, 104, 115, 97, 100, 105, 117, 98, 104, 100, 111, 100, 111, 99,
|
|
36
|
+
104, 97, 100, 111, 110, 100, 224, 100, 232, 100, 233, 100, 242, 99, 104, 39, 100, 242, 99, 104, 97, 101, 101, 97, 100,
|
|
37
|
+
97, 114, 101, 97, 116, 97, 114, 114, 97, 101, 97, 116, 111, 114, 114, 97, 101, 105, 108, 101, 101, 115, 97, 110, 102,
|
|
38
|
+
97, 102, 97, 114, 102, 101, 117, 100, 102, 104, 97, 100, 102, 104, 101, 97, 114, 114, 102, 104, 101, 105, 110, 102,
|
|
39
|
+
104, 101, 117, 100, 97, 114, 102, 104, 101, 224, 114, 114, 102, 104, 232, 105, 110, 102, 104, 233, 105, 110, 102, 104,
|
|
40
|
+
236, 110, 102, 111, 102, 111, 100, 104, 97, 102, 111, 100, 104, 97, 105, 110, 110, 102, 111, 105, 112, 101, 102, 111,
|
|
41
|
+
110, 102, 232, 105, 110, 103, 97, 103, 97, 99, 104, 103, 97, 109, 103, 97, 110, 103, 101, 103, 101, 100, 103, 117,
|
|
42
|
+
103, 117, 110, 103, 117, 114, 103, 117, 115, 104, 45, 117, 105, 108, 101, 105, 105, 97, 100, 105, 97, 100, 115, 97,
|
|
43
|
+
110, 105, 110, 110, 116, 101, 105, 115, 105, 115, 101, 108, 101, 108, 101, 97, 109, 108, 101, 97, 109, 45, 115, 97,
|
|
44
|
+
108, 101, 97, 109, 115, 97, 108, 101, 97, 116, 108, 101, 97, 116, 45, 115, 97, 108, 101, 97, 116, 104, 97, 108, 101,
|
|
45
|
+
97, 116, 115, 97, 108, 101, 105, 98, 104, 108, 101, 105, 115, 108, 101, 105, 115, 45, 115, 97, 110, 108, 101, 111,
|
|
46
|
+
116, 104, 39, 108, 101, 111, 116, 104, 97, 108, 101, 111, 116, 104, 97, 45, 115, 97, 110, 108, 105, 110, 110, 109, 39,
|
|
47
|
+
109, 39, 97, 109, 97, 109, 97, 99, 109, 97, 110, 109, 97, 114, 109, 97, 115, 109, 97, 116, 104, 97, 105, 100, 109,
|
|
48
|
+
104, 101, 117, 100, 109, 105, 109, 105, 115, 39, 109, 105, 115, 101, 109, 111, 109, 117, 109, 117, 110, 109, 117, 114,
|
|
49
|
+
109, 117, 114, 97, 109, 117, 115, 110, 97, 110, 97, 99, 104, 110, 97, 100, 110, 97, 109, 110, 97, 110, 110, 97, 114,
|
|
50
|
+
110, 97, 115, 110, 101, 111, 110, 111, 110, 117, 97, 105, 114, 111, 111, 39, 110, 111, 105, 114, 111, 105, 114, 98,
|
|
51
|
+
104, 111, 105, 114, 98, 104, 45, 115, 101, 111, 105, 114, 110, 110, 111, 105, 114, 110, 110, 101, 111, 105, 114, 114,
|
|
52
|
+
101, 111, 110, 111, 114, 109, 111, 114, 109, 45, 115, 97, 111, 114, 109, 115, 97, 111, 114, 114, 97, 111, 114, 114,
|
|
53
|
+
97, 45, 115, 97, 110, 111, 114, 114, 97, 115, 97, 110, 111, 114, 116, 111, 115, 114, 39, 114, 105, 114, 105, 98, 104,
|
|
54
|
+
114, 105, 110, 110, 114, 105, 115, 114, 105, 116, 104, 101, 114, 105, 116, 104, 101, 45, 115, 101, 114, 105, 117, 109,
|
|
55
|
+
114, 105, 117, 109, 45, 115, 97, 114, 105, 117, 109, 115, 39, 114, 105, 117, 109, 115, 97, 114, 105, 117, 116, 114,
|
|
56
|
+
105, 117, 116, 104, 39, 114, 105, 117, 116, 104, 97, 114, 105, 117, 116, 104, 97, 115, 97, 110, 114, 111, 114, 111,
|
|
57
|
+
39, 110, 114, 111, 105, 109, 104, 114, 111, 105, 109, 104, 101, 114, 111, 109, 104, 97, 105, 110, 110, 114, 111, 109,
|
|
58
|
+
104, 97, 109, 114, 111, 109, 104, 112, 97, 114, 111, 110, 114, 117, 105, 98, 104, 114, 117, 105, 103, 101, 114, 117,
|
|
59
|
+
105, 110, 110, 114, 117, 105, 110, 110, 101, 115, 97, 115, 97, 110, 115, 97, 110, 110, 115, 101, 115, 101, 97, 99,
|
|
60
|
+
104, 115, 101, 111, 115, 101, 111, 116, 104, 97, 99, 104, 115, 104, 105, 110, 115, 105, 98, 104, 115, 105, 98, 104,
|
|
61
|
+
45, 115, 101, 115, 105, 98, 104, 115, 101, 115, 105, 110, 115, 105, 110, 101, 97, 99, 104, 115, 105, 110, 110, 115,
|
|
62
|
+
105, 110, 110, 101, 115, 105, 111, 100, 115, 105, 111, 100, 97, 99, 104, 115, 105, 117, 100, 115, 105, 117, 100, 97,
|
|
63
|
+
99, 104, 115, 110, 97, 115, 111, 110, 115, 232, 116, 39, 116, 97, 114, 115, 97, 105, 110, 103, 116, 97, 114, 115, 97,
|
|
64
|
+
105, 110, 110, 116, 97, 114, 115, 117, 105, 110, 110, 116, 104, 97, 114, 116, 104, 111, 105, 103, 104, 116, 104, 114,
|
|
65
|
+
111, 116, 104, 117, 116, 104, 117, 99, 39, 116, 104, 117, 99, 97, 116, 104, 117, 103, 97, 100, 116, 104, 117, 103, 97,
|
|
66
|
+
105, 98, 104, 116, 104, 117, 103, 97, 105, 110, 110, 116, 104, 117, 103, 97, 109, 116, 104, 117, 103, 97, 109, 115,
|
|
67
|
+
97, 116, 104, 117, 105, 99, 101, 116, 104, 117, 105, 103, 101, 116, 104, 117, 115, 39, 116, 104, 117, 115, 97, 116,
|
|
68
|
+
105, 109, 99, 104, 101, 97, 108, 108, 116, 111, 105, 103, 104, 116, 111, 105, 108, 116, 114, 111, 116, 114, 111, 39,
|
|
69
|
+
116, 114, 111, 105, 109, 104, 116, 114, 111, 105, 109, 104, 101, 116, 114, 111, 110, 116, 117, 116, 117, 115, 97, 117,
|
|
70
|
+
97, 105, 114, 117, 100, 117, 103, 97, 105, 98, 104, 117, 103, 97, 109, 45, 115, 39, 117, 103, 97, 109, 45, 115, 97,
|
|
71
|
+
117, 105, 99, 101, 117, 105, 103, 101, 117, 105, 103, 101, 45, 115, 97, 110, 117, 109, 97, 100, 117, 110, 110, 116,
|
|
72
|
+
97, 117, 114, 117, 114, 114, 97, 105, 110, 110, 224, 224, 115, 224, 115, 97, 110, 225, 225, 115, 232, 236, 242, 243,
|
|
73
|
+
};
|
|
74
|
+
|
|
75
|
+
static const uint32_t stop_word_offsets[359] = {
|
|
76
|
+
0, 3, 6, 8, 10, 13, 17, 21, 25, 29, 31, 33, 36, 40, 44, 48, 52, 58, 62, 64, 67, 71, 76, 79, 83, 84, 86, 89, 92, 100,
|
|
77
|
+
105, 110, 116, 119, 122, 129, 135, 138, 140, 144, 151, 157, 163, 169, 177, 183, 190, 194, 201, 207, 213, 217, 221,
|
|
78
|
+
228, 234, 237, 241, 245, 253, 260, 263, 270, 276, 278, 280, 283, 288, 293, 301, 308, 312, 317, 320, 322, 324, 328,
|
|
79
|
+
332, 336, 338, 341, 345, 353, 360, 367, 375, 381, 385, 390, 393, 395, 398, 401, 407, 410, 414, 420, 425, 429, 433,
|
|
80
|
+
439, 445, 450, 453, 459, 468, 473, 477, 482, 487, 489, 492, 496, 500, 505, 510, 512, 514, 516, 518, 520, 523, 526,
|
|
81
|
+
532, 542, 551, 555, 561, 564, 568, 573, 576, 581, 586, 591, 597, 602, 610, 617, 623, 629, 637, 643, 650, 655, 659,
|
|
82
|
+
666, 672, 675, 682, 688, 695, 700, 702, 707, 710, 712, 714, 716, 721, 726, 727, 732, 739, 746, 750, 754, 756, 759,
|
|
83
|
+
763, 767, 773, 778, 785, 791, 796, 801, 805, 807, 812, 820, 825, 828, 832, 834, 838, 841, 844, 846, 849, 851, 854,
|
|
84
|
+
857, 860, 866, 867, 870, 876, 881, 883, 886, 888, 892, 899, 905, 909, 916, 922, 928, 933, 937, 945, 951, 957, 967,
|
|
85
|
+
971, 973, 976, 978, 981, 984, 987, 990, 997, 1002, 1004, 1008, 1012, 1014, 1016, 1019, 1022, 1026, 1029, 1031, 1035,
|
|
86
|
+
1038, 1041, 1044, 1047, 1050, 1053, 1055, 1060, 1061, 1064, 1067, 1072, 1080, 1085, 1091, 1096, 1098, 1101, 1107,
|
|
87
|
+
1112, 1116, 1124, 1131, 1134, 1136, 1138, 1140, 1144, 1148, 1151, 1156, 1164, 1168, 1175, 1181, 1187, 1191, 1197,
|
|
88
|
+
1203, 1212, 1214, 1218, 1223, 1229, 1237, 1243, 1249, 1252, 1257, 1262, 1267, 1273, 1275, 1278, 1282, 1284, 1289,
|
|
89
|
+
1292, 1300, 1304, 1308, 1315, 1321, 1324, 1331, 1335, 1340, 1344, 1351, 1355, 1362, 1365, 1368, 1370, 1372, 1380,
|
|
90
|
+
1388, 1396, 1400, 1406, 1410, 1413, 1418, 1423, 1429, 1437, 1445, 1451, 1459, 1465, 1471, 1476, 1481, 1490, 1495,
|
|
91
|
+
1499, 1502, 1506, 1512, 1519, 1523, 1525, 1529, 1533, 1535, 1541, 1548, 1555, 1559, 1563, 1571, 1575, 1580, 1582,
|
|
92
|
+
1589, 1590, 1592, 1596, 1597, 1599, 1600, 1601, 1602, 1603,
|
|
93
|
+
};
|
|
94
|
+
|
|
95
|
+
static const uint32_t keep_ranges[30] = {
|
|
96
|
+
39, 39, 45, 45, 48, 57, 65, 90, 97, 122, 192, 193, 200, 201, 204, 204, 210, 211, 217, 217, 224, 225, 232, 233, 236,
|
|
97
|
+
236, 242, 243, 249, 249,
|
|
98
|
+
};
|
|
99
|
+
|
|
100
|
+
const narsil_language narsil_language_scottish_gaelic = {
|
|
101
|
+
.name = "scottish-gaelic",
|
|
102
|
+
.revision = "1",
|
|
103
|
+
.stop_word_units = stop_word_units,
|
|
104
|
+
.stop_word_offsets = stop_word_offsets,
|
|
105
|
+
.stop_word_count = 358,
|
|
106
|
+
.ascii_keep = {0x03FF208000000000ULL, 0x07FFFFFE07FFFFFEULL},
|
|
107
|
+
.keep_ranges = keep_ranges,
|
|
108
|
+
.keep_range_count = 15,
|
|
109
|
+
.splits_code_points = 0,
|
|
110
|
+
.strips_possessive = 0,
|
|
111
|
+
.normalises_diacritics = 0,
|
|
112
|
+
.min_token_length = 1,
|
|
113
|
+
.ngram_size = 0,
|
|
114
|
+
.normalise = NULL,
|
|
115
|
+
.stem_by_rules = NULL,
|
|
116
|
+
.create_snowball = NULL,
|
|
117
|
+
.snowball_stem = NULL,
|
|
118
|
+
.close_snowball = NULL,
|
|
119
|
+
};
|
|
@@ -0,0 +1,84 @@
|
|
|
1
|
+
#include "../../analysis/languages.h"
|
|
2
|
+
#include "declarations.h"
|
|
3
|
+
#include "../rules.h"
|
|
4
|
+
#include "../snowball/serbian_stemmer.h"
|
|
5
|
+
|
|
6
|
+
#include <stddef.h>
|
|
7
|
+
#include <stdint.h>
|
|
8
|
+
|
|
9
|
+
static const uint16_t stop_word_units[749] = {
|
|
10
|
+
97, 97, 107, 111, 97, 108, 105, 98, 97, 353, 98, 101, 122, 98, 105, 98, 105, 104, 98, 105, 108, 97, 98, 105, 108, 105,
|
|
11
|
+
98, 105, 108, 111, 98, 105, 111, 98, 105, 115, 109, 111, 98, 105, 115, 116, 101, 98, 105, 116, 105, 98, 117, 100, 101,
|
|
12
|
+
100, 97, 100, 111, 100, 111, 107, 103, 100, 101, 105, 105, 97, 107, 111, 105, 108, 105, 105, 115, 112, 111, 100, 105,
|
|
13
|
+
122, 105, 122, 109, 101, 273, 117, 105, 122, 110, 97, 100, 106, 97, 106, 101, 106, 101, 114, 106, 111, 353, 107, 97,
|
|
14
|
+
107, 97, 100, 107, 97, 107, 111, 107, 111, 107, 111, 106, 97, 107, 111, 106, 101, 107, 111, 106, 105, 107, 111, 108,
|
|
15
|
+
105, 107, 111, 107, 114, 111, 122, 108, 105, 109, 97, 100, 97, 109, 97, 108, 111, 109, 105, 109, 110, 111, 103, 111,
|
|
16
|
+
109, 111, 106, 110, 97, 110, 97, 353, 110, 101, 110, 101, 103, 111, 110, 101, 107, 111, 110, 101, 353, 116, 111, 110,
|
|
17
|
+
105, 110, 105, 107, 97, 100, 110, 105, 107, 111, 110, 105, 116, 105, 110, 105, 353, 116, 97, 110, 106, 101, 103, 111,
|
|
18
|
+
118, 110, 106, 101, 110, 110, 106, 105, 104, 111, 118, 111, 100, 111, 110, 111, 110, 97, 111, 110, 97, 106, 111, 110,
|
|
19
|
+
97, 107, 111, 111, 110, 100, 97, 111, 110, 101, 111, 110, 105, 111, 110, 111, 111, 115, 105, 109, 111, 118, 97, 106,
|
|
20
|
+
111, 118, 97, 107, 111, 111, 118, 100, 101, 111, 118, 111, 112, 111, 112, 111, 114, 101, 100, 112, 111, 115, 108, 101,
|
|
21
|
+
112, 114, 101, 112, 114, 101, 109, 97, 112, 114, 105, 115, 97, 115, 97, 100, 97, 115, 97, 109, 115, 97, 109, 111, 115,
|
|
22
|
+
101, 98, 101, 115, 101, 98, 105, 115, 105, 115, 109, 111, 115, 116, 101, 115, 117, 115, 118, 101, 115, 118, 105, 116,
|
|
23
|
+
97, 106, 116, 97, 107, 111, 116, 97, 109, 111, 116, 105, 116, 111, 116, 118, 111, 106, 117, 117, 118, 101, 107, 118,
|
|
24
|
+
97, 353, 118, 101, 263, 118, 105, 122, 97, 122, 97, 353, 116, 111, 122, 98, 111, 103, 263, 101, 263, 101, 109, 111,
|
|
25
|
+
263, 101, 116, 101, 263, 101, 353, 263, 117, 353, 116, 97, 1072, 1072, 1082, 1086, 1072, 1083, 1080, 1073, 1072, 1096,
|
|
26
|
+
1073, 1077, 1079, 1073, 1080, 1073, 1080, 1083, 1072, 1073, 1080, 1083, 1080, 1073, 1080, 1083, 1086, 1073, 1080,
|
|
27
|
+
1086, 1073, 1080, 1089, 1084, 1086, 1073, 1080, 1089, 1090, 1077, 1073, 1080, 1090, 1080, 1073, 1080, 1093, 1073,
|
|
28
|
+
1091, 1076, 1077, 1074, 1072, 1096, 1074, 1077, 1115, 1074, 1080, 1075, 1076, 1077, 1076, 1072, 1076, 1086, 1076,
|
|
29
|
+
1086, 1082, 1079, 1072, 1079, 1072, 1096, 1090, 1086, 1079, 1073, 1086, 1075, 1080, 1080, 1072, 1082, 1086, 1080,
|
|
30
|
+
1079, 1080, 1079, 1084, 1077, 1106, 1091, 1080, 1079, 1085, 1072, 1076, 1080, 1083, 1080, 1080, 1089, 1087, 1086,
|
|
31
|
+
1076, 1082, 1072, 1082, 1072, 1076, 1082, 1072, 1082, 1086, 1082, 1086, 1082, 1086, 1083, 1080, 1082, 1086, 1082,
|
|
32
|
+
1086, 1112, 1072, 1082, 1086, 1112, 1077, 1082, 1086, 1112, 1080, 1082, 1088, 1086, 1079, 1083, 1080, 1084, 1072,
|
|
33
|
+
1076, 1072, 1084, 1072, 1083, 1086, 1084, 1080, 1084, 1085, 1086, 1075, 1086, 1084, 1086, 1112, 1085, 1072, 1085,
|
|
34
|
+
1072, 1096, 1085, 1077, 1085, 1077, 1075, 1086, 1085, 1077, 1082, 1086, 1085, 1077, 1096, 1090, 1086, 1085, 1080,
|
|
35
|
+
1085, 1080, 1082, 1072, 1076, 1085, 1080, 1082, 1086, 1085, 1080, 1090, 1080, 1085, 1080, 1096, 1090, 1072, 1086,
|
|
36
|
+
1074, 1072, 1082, 1086, 1086, 1074, 1072, 1112, 1086, 1074, 1076, 1077, 1086, 1074, 1086, 1086, 1076, 1086, 1085,
|
|
37
|
+
1086, 1085, 1072, 1086, 1085, 1072, 1082, 1086, 1086, 1085, 1072, 1112, 1086, 1085, 1076, 1072, 1086, 1085, 1077,
|
|
38
|
+
1086, 1085, 1080, 1086, 1085, 1086, 1086, 1089, 1080, 1084, 1087, 1086, 1087, 1086, 1088, 1077, 1076, 1087, 1086,
|
|
39
|
+
1089, 1083, 1077, 1087, 1088, 1077, 1087, 1088, 1077, 1084, 1072, 1087, 1088, 1080, 1089, 1072, 1089, 1072, 1076,
|
|
40
|
+
1072, 1089, 1072, 1084, 1089, 1072, 1084, 1086, 1089, 1074, 1077, 1089, 1074, 1080, 1089, 1077, 1073, 1077, 1089,
|
|
41
|
+
1077, 1073, 1080, 1089, 1080, 1089, 1084, 1086, 1089, 1090, 1077, 1089, 1091, 1090, 1072, 1082, 1086, 1090, 1072,
|
|
42
|
+
1084, 1086, 1090, 1072, 1112, 1090, 1074, 1086, 1112, 1090, 1080, 1090, 1086, 1091, 1091, 1074, 1077, 1082, 1096,
|
|
43
|
+
1090, 1072, 1112, 1072, 1112, 1077, 1112, 1077, 1088, 1112, 1086, 1096, 1114, 1077, 1075, 1086, 1074, 1114, 1077,
|
|
44
|
+
1085, 1114, 1080, 1093, 1086, 1074, 1115, 1077, 1115, 1077, 1084, 1086, 1115, 1077, 1090, 1077, 1115, 1077, 1096,
|
|
45
|
+
1115, 1091,
|
|
46
|
+
};
|
|
47
|
+
|
|
48
|
+
static const uint32_t stop_word_offsets[223] = {
|
|
49
|
+
0, 1, 4, 7, 10, 13, 15, 18, 22, 26, 30, 33, 38, 43, 47, 51, 53, 55, 58, 61, 62, 66, 69, 74, 76, 82, 87, 89, 91, 94,
|
|
50
|
+
97, 99, 102, 106, 108, 112, 116, 120, 126, 130, 132, 136, 140, 142, 147, 150, 152, 155, 157, 161, 165, 170, 172, 177,
|
|
51
|
+
181, 185, 190, 196, 200, 206, 208, 210, 213, 217, 222, 226, 229, 232, 235, 239, 243, 248, 252, 255, 257, 262, 267,
|
|
52
|
+
270, 275, 278, 280, 284, 287, 291, 295, 299, 301, 304, 307, 309, 312, 315, 318, 322, 326, 328, 330, 334, 335, 339,
|
|
53
|
+
342, 345, 347, 349, 354, 358, 360, 364, 368, 371, 373, 376, 377, 380, 383, 386, 389, 391, 395, 399, 403, 406, 411,
|
|
54
|
+
416, 420, 423, 427, 430, 433, 435, 438, 440, 442, 445, 447, 452, 456, 457, 461, 463, 469, 474, 477, 482, 484, 487,
|
|
55
|
+
491, 493, 499, 503, 507, 511, 515, 517, 521, 525, 527, 532, 535, 537, 540, 542, 546, 550, 555, 557, 562, 566, 570,
|
|
56
|
+
575, 580, 584, 588, 591, 593, 595, 598, 603, 607, 611, 614, 617, 620, 624, 626, 631, 636, 639, 644, 647, 649, 653,
|
|
57
|
+
656, 660, 663, 666, 670, 674, 676, 679, 682, 684, 688, 692, 695, 699, 701, 703, 704, 708, 711, 713, 715, 718, 721,
|
|
58
|
+
726, 729, 734, 736, 740, 744, 747, 749,
|
|
59
|
+
};
|
|
60
|
+
|
|
61
|
+
static const uint32_t keep_ranges[20] = {
|
|
62
|
+
48, 57, 65, 90, 97, 122, 262, 263, 268, 269, 272, 273, 352, 353, 381, 382, 1024, 1279, 7296, 7303,
|
|
63
|
+
};
|
|
64
|
+
|
|
65
|
+
const narsil_language narsil_language_serbian = {
|
|
66
|
+
.name = "serbian",
|
|
67
|
+
.revision = "86c7d873d1fd8723",
|
|
68
|
+
.stop_word_units = stop_word_units,
|
|
69
|
+
.stop_word_offsets = stop_word_offsets,
|
|
70
|
+
.stop_word_count = 222,
|
|
71
|
+
.ascii_keep = {0x03FF000000000000ULL, 0x07FFFFFE07FFFFFEULL},
|
|
72
|
+
.keep_ranges = keep_ranges,
|
|
73
|
+
.keep_range_count = 10,
|
|
74
|
+
.splits_code_points = 0,
|
|
75
|
+
.strips_possessive = 0,
|
|
76
|
+
.normalises_diacritics = 0,
|
|
77
|
+
.min_token_length = 1,
|
|
78
|
+
.ngram_size = 0,
|
|
79
|
+
.normalise = narsil_serbian_normalise,
|
|
80
|
+
.stem_by_rules = NULL,
|
|
81
|
+
.create_snowball = serbian_create_env,
|
|
82
|
+
.snowball_stem = serbian_stem,
|
|
83
|
+
.close_snowball = serbian_close_env,
|
|
84
|
+
};
|
|
@@ -0,0 +1,55 @@
|
|
|
1
|
+
#include "../../analysis/languages.h"
|
|
2
|
+
#include "declarations.h"
|
|
3
|
+
|
|
4
|
+
#include <stddef.h>
|
|
5
|
+
#include <stdint.h>
|
|
6
|
+
|
|
7
|
+
static const uint16_t stop_word_units[339] = {
|
|
8
|
+
97, 105, 118, 97, 97, 107, 97, 110, 103, 97, 97, 110, 103, 97, 97, 112, 97, 97, 112, 111, 97, 114, 105, 97, 115, 105,
|
|
9
|
+
97, 118, 97, 97, 118, 111, 99, 104, 97, 107, 101, 99, 104, 97, 110, 103, 117, 99, 104, 105, 109, 119, 101, 99, 104,
|
|
10
|
+
105, 114, 105, 100, 122, 97, 107, 101, 100, 122, 105, 109, 119, 101, 100, 122, 111, 115, 101, 104, 101, 114, 101, 105,
|
|
11
|
+
99, 104, 105, 105, 99, 104, 111, 105, 100, 122, 105, 105, 100, 122, 111, 105, 109, 105, 105, 109, 119, 101, 105, 110,
|
|
12
|
+
105, 105, 112, 97, 112, 111, 105, 114, 105, 105, 114, 111, 105, 115, 117, 105, 118, 111, 105, 119, 101, 105, 121, 101,
|
|
13
|
+
105, 121, 105, 105, 121, 111, 105, 122, 118, 105, 105, 122, 118, 111, 107, 97, 110, 97, 107, 117, 110, 101, 107, 117,
|
|
14
|
+
110, 122, 101, 107, 117, 116, 105, 109, 117, 107, 97, 116, 105, 109, 117, 109, 119, 101, 109, 117, 110, 101, 109, 117,
|
|
15
|
+
115, 104, 117, 114, 101, 110, 101, 107, 117, 116, 105, 110, 111, 107, 117, 116, 105, 111, 115, 101, 112, 97, 107, 97,
|
|
16
|
+
116, 105, 112, 97, 109, 117, 115, 111, 114, 111, 112, 97, 110, 101, 112, 97, 115, 105, 114, 105, 109, 119, 101, 115,
|
|
17
|
+
97, 107, 97, 115, 101, 105, 115, 104, 117, 114, 101, 117, 107, 111, 117, 121, 101, 117, 121, 111, 117, 121, 117, 118,
|
|
18
|
+
97, 109, 119, 101, 118, 97, 114, 105, 118, 111, 115, 101, 119, 97, 107, 101, 119, 97, 107, 111, 119, 97, 110, 103,
|
|
19
|
+
117, 119, 97, 118, 111, 119, 101, 100, 117, 119, 101, 110, 121, 117, 121, 97, 107, 101, 121, 97, 107, 111, 121, 97,
|
|
20
|
+
110, 103, 117, 121, 97, 118, 111, 121, 101, 100, 117, 121, 101, 110, 121, 117, 122, 118, 97, 107, 101, 122, 118, 97,
|
|
21
|
+
110, 103, 117, 122, 118, 105, 109, 119, 101, 122, 118, 105, 110, 111, 122, 118, 105, 114, 105, 122, 118, 111, 115,
|
|
22
|
+
101,
|
|
23
|
+
};
|
|
24
|
+
|
|
25
|
+
static const uint32_t stop_word_offsets[80] = {
|
|
26
|
+
0, 4, 10, 14, 17, 20, 23, 26, 29, 32, 37, 43, 49, 54, 59, 65, 70, 74, 78, 82, 86, 90, 93, 97, 100, 105, 108, 111, 114,
|
|
27
|
+
117, 120, 123, 126, 129, 133, 137, 141, 145, 150, 154, 160, 165, 169, 176, 182, 188, 191, 197, 205, 209, 213, 218,
|
|
28
|
+
222, 225, 230, 233, 236, 239, 242, 247, 251, 255, 259, 263, 268, 272, 276, 281, 285, 289, 294, 298, 302, 307, 312,
|
|
29
|
+
318, 324, 329, 334, 339,
|
|
30
|
+
};
|
|
31
|
+
|
|
32
|
+
static const uint32_t keep_ranges[6] = {
|
|
33
|
+
48, 57, 65, 90, 97, 122,
|
|
34
|
+
};
|
|
35
|
+
|
|
36
|
+
const narsil_language narsil_language_shona = {
|
|
37
|
+
.name = "shona",
|
|
38
|
+
.revision = "1",
|
|
39
|
+
.stop_word_units = stop_word_units,
|
|
40
|
+
.stop_word_offsets = stop_word_offsets,
|
|
41
|
+
.stop_word_count = 79,
|
|
42
|
+
.ascii_keep = {0x03FF000000000000ULL, 0x07FFFFFE07FFFFFEULL},
|
|
43
|
+
.keep_ranges = keep_ranges,
|
|
44
|
+
.keep_range_count = 3,
|
|
45
|
+
.splits_code_points = 0,
|
|
46
|
+
.strips_possessive = 0,
|
|
47
|
+
.normalises_diacritics = 0,
|
|
48
|
+
.min_token_length = 1,
|
|
49
|
+
.ngram_size = 0,
|
|
50
|
+
.normalise = NULL,
|
|
51
|
+
.stem_by_rules = NULL,
|
|
52
|
+
.create_snowball = NULL,
|
|
53
|
+
.snowball_stem = NULL,
|
|
54
|
+
.close_snowball = NULL,
|
|
55
|
+
};
|
|
@@ -0,0 +1,78 @@
|
|
|
1
|
+
#include "../../analysis/languages.h"
|
|
2
|
+
#include "declarations.h"
|
|
3
|
+
#include "../rules.h"
|
|
4
|
+
|
|
5
|
+
#include <stddef.h>
|
|
6
|
+
#include <stdint.h>
|
|
7
|
+
|
|
8
|
+
static const uint16_t stop_word_units[632] = {
|
|
9
|
+
3461, 3501, 3515, 3461, 3505, 3530, 3505, 3461, 3505, 3540, 3520, 3461, 3505, 3546, 3461, 3507, 3549, 3461, 3508,
|
|
10
|
+
3548, 3514, 3538, 3461, 3508, 3549, 3461, 3514, 3538, 3514, 3549, 3461, 3514, 3540, 3515, 3538, 3505, 3530, 3461,
|
|
11
|
+
3514, 3540, 3515, 3540, 3461, 3515, 3510, 3514, 3535, 3461, 3524, 3535, 3462, 3462, 3514, 3538, 3462, 3524, 3530,
|
|
12
|
+
3463, 3501, 3538, 3463, 3501, 3540, 3517, 3540, 3463, 3501, 3540, 3525, 3540, 3463, 3514, 3538, 3465, 3482, 3530,
|
|
13
|
+
3510, 3538, 3501, 3538, 3465, 3501, 3535, 3465, 3501, 3538, 3505, 3530, 3465, 3507, 3540, 3515, 3535, 3467, 3503,
|
|
14
|
+
3545, 3523, 3535, 3468, 3514, 3538, 3473, 3473, 3505, 3512, 3540, 3501, 3530, 3473, 3505, 3538, 3523, 3535, 3473,
|
|
15
|
+
3510, 3536, 3520, 3538, 3505, 3530, 3473, 3512, 3473, 3512, 3530, 3510, 3517, 3473, 3512, 3530, 3510, 3535, 3473,
|
|
16
|
+
3524, 3545, 3501, 3530, 3473, 3524, 3545, 3505, 3530, 3473, 3524, 3546, 3474, 3476, 3505, 3530, 3505, 3477, 3524,
|
|
17
|
+
3549, 3482, 3482, 3517, 3539, 3482, 3530, 3482, 3538, 3512, 3482, 3538, 3512, 3530, 3484, 3535, 3505, 3545, 3484,
|
|
18
|
+
3536, 3505, 3484, 3546, 3488, 3538, 3482, 3530, 3488, 3538, 3524, 3530, 3488, 3539, 3495, 3501, 3520, 3501, 3520,
|
|
19
|
+
3501, 3530, 3501, 3535, 3482, 3530, 3501, 3540, 3515, 3535, 3501, 3540, 3515, 3535, 3520, 3495, 3501, 3540, 3515,
|
|
20
|
+
3540, 3501, 3540, 3517, 3538, 3505, 3530, 3501, 3540, 3520, 3482, 3530, 3501, 3540, 3525, 3501, 3545, 3482, 3530,
|
|
21
|
+
3503, 3503, 3482, 3530, 3520, 3535, 3503, 3536, 3505, 3495, 3503, 3536, 3505, 3530, 3503, 3539, 3503, 3549, 3503,
|
|
22
|
+
3549, 3524, 3549, 3505, 3512, 3530, 3505, 3512, 3540, 3501, 3530, 3505, 3520, 3505, 3537, 3505, 3538, 3501, 3515,
|
|
23
|
+
3505, 3538, 3501, 3538, 3505, 3538, 3501, 3538, 3501, 3530, 3505, 3538, 3501, 3548, 3515, 3505, 3538, 3523, 3535,
|
|
24
|
+
3505, 3538, 3523, 3535, 3520, 3545, 3505, 3530, 3505, 3548, 3524, 3548, 3501, 3530, 3508, 3495, 3505, 3530, 3508,
|
|
25
|
+
3501, 3535, 3508, 3512, 3499, 3508, 3512, 3499, 3538, 3505, 3530, 3508, 3512, 3505, 3508, 3515, 3538, 3503, 3538,
|
|
26
|
+
3508, 3520, 3535, 3508, 3535, 3523, 3535, 3508, 3538, 3499, 3538, 3523, 3508, 3538, 3525, 3538, 3510, 3507, 3508,
|
|
27
|
+
3538, 3525, 3538, 3510, 3507, 3520, 3508, 3540, 3505, 3510, 3507, 3540, 3510, 3520, 3510, 3520, 3495, 3510, 3520,
|
|
28
|
+
3501, 3530, 3510, 3520, 3503, 3510, 3520, 3545, 3505, 3530, 3510, 3536, 3520, 3538, 3505, 3530, 3510, 3548, 3517,
|
|
29
|
+
3510, 3548, 3524, 3549, 3512, 3512, 3484, 3538, 3505, 3530, 3512, 3501, 3512, 3505, 3530, 3503, 3512, 3524, 3512,
|
|
30
|
+
3524, 3535, 3512, 3538, 3523, 3512, 3540, 3501, 3530, 3512, 3545, 3501, 3545, 3482, 3530, 3512, 3545, 3505, 3530,
|
|
31
|
+
3512, 3545, 3505, 3530, 3505, 3512, 3545, 3512, 3512, 3545, 3514, 3538, 3505, 3530, 3512, 3545, 3517, 3545, 3523,
|
|
32
|
+
3512, 3545, 3523, 3546, 3512, 3545, 3524, 3538, 3512, 3545, 3524, 3545, 3505, 3530, 3512, 3545, 3524, 3546, 3512,
|
|
33
|
+
3546, 3512, 3546, 3501, 3535, 3482, 3530, 3514, 3514, 3505, 3514, 3517, 3538, 3517, 3545, 3523, 3520, 3497, 3535,
|
|
34
|
+
3520, 3497, 3535, 3501, 3530, 3512, 3520, 3501, 3530, 3520, 3505, 3520, 3505, 3530, 3520, 3505, 3535, 3524, 3538,
|
|
35
|
+
3520, 3521, 3514, 3545, 3505, 3530, 3520, 3523, 3530, 3520, 3524, 3535, 3520, 3536, 3497, 3538, 3520, 3536, 3505,
|
|
36
|
+
3538, 3520, 3538, 3495, 3520, 3538, 3495, 3538, 3505, 3530, 3520, 3538, 3505, 3535, 3520, 3538, 3521, 3546, 3522,
|
|
37
|
+
3520, 3538, 3523, 3538, 3505, 3530, 3520, 3545, 3501, 3520, 3545, 3501, 3495, 3520, 3545, 3501, 3538, 3505, 3530,
|
|
38
|
+
3520, 3545, 3505, 3520, 3545, 3505, 3540, 3520, 3495, 3520, 3545, 3505, 3540, 3520, 3545, 3505, 3530, 3523, 3507,
|
|
39
|
+
3524, 3535, 3523, 3512, 3484, 3523, 3512, 3487, 3523, 3524, 3523, 3535, 3523, 3536, 3505, 3538, 3505, 3530, 3523,
|
|
40
|
+
3538, 3495, 3523, 3538, 3495, 3505, 3530, 3523, 3538, 3503, 3540, 3523, 3545, 3503, 3523, 3546, 3523, 3546, 3482,
|
|
41
|
+
3523, 3546, 3482, 3530, 3524, 3505, 3538, 3482, 3524, 3535, 3524, 3536, 3515, 3524, 3545, 3514, 3538, 3505, 3530,
|
|
42
|
+
3524, 3545, 3520, 3501, 3530,
|
|
43
|
+
};
|
|
44
|
+
|
|
45
|
+
static const uint32_t stop_word_offsets[164] = {
|
|
46
|
+
0, 3, 7, 11, 14, 17, 22, 25, 30, 37, 42, 47, 50, 51, 54, 57, 60, 65, 70, 73, 80, 83, 88, 93, 98, 101, 102, 108, 113,
|
|
47
|
+
120, 122, 127, 132, 137, 142, 145, 146, 150, 153, 154, 157, 159, 162, 166, 170, 173, 175, 179, 183, 185, 186, 188,
|
|
48
|
+
192, 196, 200, 206, 210, 216, 221, 224, 228, 229, 234, 238, 242, 244, 246, 250, 253, 258, 260, 262, 266, 270, 276,
|
|
49
|
+
281, 285, 293, 299, 303, 306, 309, 315, 318, 323, 326, 330, 335, 341, 348, 351, 354, 356, 359, 363, 366, 371, 377,
|
|
50
|
+
380, 384, 385, 390, 392, 396, 398, 401, 404, 408, 414, 418, 423, 426, 432, 437, 441, 445, 451, 455, 457, 463, 464,
|
|
51
|
+
466, 469, 472, 475, 481, 484, 486, 489, 494, 500, 503, 506, 510, 514, 517, 523, 527, 532, 538, 541, 545, 551, 554,
|
|
52
|
+
560, 568, 572, 575, 578, 580, 582, 588, 591, 596, 600, 603, 605, 608, 612, 616, 618, 621, 627, 632,
|
|
53
|
+
};
|
|
54
|
+
|
|
55
|
+
static const uint32_t keep_ranges[10] = {
|
|
56
|
+
48, 57, 65, 90, 97, 122, 3456, 3583, 8205, 8205,
|
|
57
|
+
};
|
|
58
|
+
|
|
59
|
+
const narsil_language narsil_language_sinhala = {
|
|
60
|
+
.name = "sinhala",
|
|
61
|
+
.revision = "1",
|
|
62
|
+
.stop_word_units = stop_word_units,
|
|
63
|
+
.stop_word_offsets = stop_word_offsets,
|
|
64
|
+
.stop_word_count = 163,
|
|
65
|
+
.ascii_keep = {0x03FF000000000000ULL, 0x07FFFFFE07FFFFFEULL},
|
|
66
|
+
.keep_ranges = keep_ranges,
|
|
67
|
+
.keep_range_count = 5,
|
|
68
|
+
.splits_code_points = 0,
|
|
69
|
+
.strips_possessive = 0,
|
|
70
|
+
.normalises_diacritics = 0,
|
|
71
|
+
.min_token_length = 1,
|
|
72
|
+
.ngram_size = 0,
|
|
73
|
+
.normalise = narsil_sinhala_normalise,
|
|
74
|
+
.stem_by_rules = NULL,
|
|
75
|
+
.create_snowball = NULL,
|
|
76
|
+
.snowball_stem = NULL,
|
|
77
|
+
.close_snowball = NULL,
|
|
78
|
+
};
|