@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 +1,68 @@
|
|
|
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[601] = {
|
|
8
|
+
97, 109, 109, 97, 97, 114, 97, 115, 305, 110, 100, 97, 97, 114, 116, 305, 113, 97, 121, 97, 122, 98, 97, 120, 98, 101,
|
|
9
|
+
108, 601, 98, 101, 351, 98, 105, 108, 601, 114, 98, 105, 114, 98, 105, 114, 97, 122, 98, 105, 114, 105, 98, 105, 114,
|
|
10
|
+
351, 101, 121, 98, 105, 122, 98, 105, 122, 105, 109, 98, 105, 122, 108, 601, 114, 98, 117, 98, 117, 110, 97, 98, 117,
|
|
11
|
+
110, 100, 97, 110, 98, 117, 110, 108, 97, 114, 305, 110, 98, 117, 110, 117, 98, 117, 110, 117, 110, 98, 117, 114, 97,
|
|
12
|
+
100, 97, 110, 98, 252, 116, 252, 110, 98, 601, 108, 105, 98, 601, 108, 107, 601, 98, 601, 121, 98, 601, 122, 105, 98,
|
|
13
|
+
601, 122, 601, 110, 100, 97, 104, 97, 100, 101, 100, 105, 100, 101, 121, 105, 108, 100, 105, 114, 100, 252, 122, 100,
|
|
14
|
+
601, 100, 601, 107, 100, 601, 110, 100, 601, 113, 105, 113, 601, 101, 100, 105, 114, 101, 100, 601, 110, 101, 108,
|
|
15
|
+
601, 101, 116, 101, 116, 100, 105, 101, 116, 109, 601, 101, 116, 109, 601, 107, 102, 97, 105, 122, 103, 105, 108, 601,
|
|
16
|
+
103, 246, 114, 601, 104, 97, 104, 97, 113, 113, 305, 110, 100, 97, 104, 97, 114, 97, 100, 97, 104, 101, 231, 104, 601,
|
|
17
|
+
104, 601, 109, 104, 601, 109, 105, 110, 104, 601, 109, 105, 351, 601, 104, 601, 114, 105, 100, 105, 105, 108, 105,
|
|
18
|
+
108, 100, 601, 105, 108, 107, 105, 108, 601, 105, 110, 105, 110, 100, 105, 105, 115, 116, 105, 102, 97, 100, 601, 105,
|
|
19
|
+
115, 601, 107, 105, 107, 105, 109, 107, 105, 109, 105, 107, 105, 109, 601, 108, 97, 107, 105, 110, 108, 97, 112, 109,
|
|
20
|
+
105, 114, 351, 101, 121, 109, 601, 104, 122, 109, 601, 110, 109, 601, 110, 601, 110, 105, 121, 601, 110, 601, 110,
|
|
21
|
+
601, 104, 97, 121, 601, 116, 111, 111, 98, 105, 114, 105, 115, 105, 111, 102, 111, 108, 97, 110, 111, 108, 97, 114,
|
|
22
|
+
111, 108, 97, 114, 97, 113, 111, 108, 100, 117, 111, 108, 100, 117, 287, 117, 111, 108, 109, 97, 100, 305, 111, 108,
|
|
23
|
+
109, 97, 122, 111, 108, 109, 117, 351, 100, 117, 114, 111, 108, 115, 117, 110, 111, 108, 117, 114, 111, 110, 111, 110,
|
|
24
|
+
97, 111, 110, 100, 97, 110, 111, 110, 108, 97, 114, 111, 110, 108, 97, 114, 100, 97, 110, 111, 110, 108, 97, 114, 305,
|
|
25
|
+
110, 111, 110, 115, 117, 122, 100, 97, 111, 110, 117, 111, 110, 117, 110, 111, 114, 97, 100, 97, 110, 113, 97, 114,
|
|
26
|
+
351, 305, 113, 601, 100, 601, 114, 115, 97, 97, 116, 115, 97, 100, 601, 99, 601, 115, 97, 110, 105, 121, 601, 115,
|
|
27
|
+
105, 122, 115, 105, 122, 105, 110, 115, 105, 122, 108, 601, 114, 115, 111, 110, 114, 97, 115, 601, 104, 118, 115, 601,
|
|
28
|
+
110, 115, 601, 110, 105, 110, 115, 601, 110, 601, 116, 601, 601, 115, 115, 252, 102, 118, 97, 114, 118, 601, 120, 97,
|
|
29
|
+
110, 120, 97, 110, 305, 109, 120, 101, 121, 114, 121, 97, 121, 97, 108, 110, 305, 122, 121, 97, 120, 351, 305, 121,
|
|
30
|
+
101, 100, 100, 105, 121, 101, 110, 601, 121, 111, 120, 121, 111, 120, 100, 117, 114, 121, 111, 120, 115, 97, 121, 601,
|
|
31
|
+
110, 105, 122, 97, 109, 97, 110, 231, 111, 120, 231, 252, 110, 107, 105, 246, 122, 246, 122, 252, 252, 231, 252, 110,
|
|
32
|
+
601, 103, 601, 114, 601, 108, 98, 601, 116, 116, 601, 601, 110, 601, 115, 108, 105, 110, 100, 601,
|
|
33
|
+
};
|
|
34
|
+
|
|
35
|
+
static const uint32_t stop_word_offsets[141] = {
|
|
36
|
+
0, 4, 12, 17, 19, 21, 24, 28, 31, 36, 39, 44, 48, 54, 57, 62, 68, 70, 74, 80, 88, 92, 97, 104, 109, 113, 118, 121,
|
|
37
|
+
125, 130, 134, 138, 143, 146, 149, 151, 154, 157, 163, 167, 171, 174, 176, 180, 184, 189, 193, 197, 201, 203, 211,
|
|
38
|
+
217, 220, 222, 225, 230, 236, 239, 242, 244, 248, 251, 254, 256, 260, 268, 271, 273, 276, 280, 284, 289, 292, 298,
|
|
39
|
+
302, 305, 309, 313, 315, 322, 323, 330, 332, 336, 340, 346, 350, 356, 362, 367, 375, 380, 384, 386, 389, 394, 399,
|
|
40
|
+
407, 414, 421, 424, 428, 434, 439, 444, 448, 454, 460, 463, 468, 474, 479, 483, 486, 491, 495, 502, 505, 507, 510,
|
|
41
|
+
515, 519, 521, 527, 532, 537, 541, 544, 550, 555, 559, 564, 567, 572, 574, 577, 581, 585, 592, 594, 601,
|
|
42
|
+
};
|
|
43
|
+
|
|
44
|
+
static const uint32_t keep_ranges[28] = {
|
|
45
|
+
48, 57, 65, 90, 97, 122, 199, 199, 214, 214, 220, 220, 231, 231, 246, 246, 252, 252, 286, 287, 305, 305, 350, 351,
|
|
46
|
+
399, 399, 601, 601,
|
|
47
|
+
};
|
|
48
|
+
|
|
49
|
+
const narsil_language narsil_language_azerbaijani = {
|
|
50
|
+
.name = "azerbaijani",
|
|
51
|
+
.revision = "1",
|
|
52
|
+
.stop_word_units = stop_word_units,
|
|
53
|
+
.stop_word_offsets = stop_word_offsets,
|
|
54
|
+
.stop_word_count = 140,
|
|
55
|
+
.ascii_keep = {0x03FF000000000000ULL, 0x07FFFFFE07FFFFFEULL},
|
|
56
|
+
.keep_ranges = keep_ranges,
|
|
57
|
+
.keep_range_count = 14,
|
|
58
|
+
.splits_code_points = 0,
|
|
59
|
+
.strips_possessive = 0,
|
|
60
|
+
.normalises_diacritics = 0,
|
|
61
|
+
.min_token_length = 1,
|
|
62
|
+
.ngram_size = 0,
|
|
63
|
+
.normalise = NULL,
|
|
64
|
+
.stem_by_rules = NULL,
|
|
65
|
+
.create_snowball = NULL,
|
|
66
|
+
.snowball_stem = NULL,
|
|
67
|
+
.close_snowball = NULL,
|
|
68
|
+
};
|
|
@@ -0,0 +1,48 @@
|
|
|
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[151] = {
|
|
8
|
+
97, 97, 98, 97, 100, 97, 97, 108, 101, 97, 110, 97, 110, 105, 97, 119, 98, 603, 98, 603, 603, 99, 97, 109, 97, 110,
|
|
9
|
+
100, 101, 100, 111, 110, 100, 596, 100, 596, 114, 596, 110, 100, 596, 596, 110, 105, 102, 97, 110, 97, 102, 111, 102,
|
|
10
|
+
603, 104, 97, 108, 105, 105, 107, 97, 107, 97, 98, 105, 110, 105, 107, 111, 107, 111, 115, 603, 98, 603, 107, 596,
|
|
11
|
+
107, 596, 102, 603, 107, 596, 110, 596, 108, 97, 109, 105, 110, 109, 105, 110, 110, 117, 110, 97, 110, 101, 110, 105,
|
|
12
|
+
110, 105, 110, 110, 105, 110, 110, 117, 110, 107, 97, 111, 111, 108, 117, 115, 97, 110, 102, 603, 115, 105, 115, 97,
|
|
13
|
+
110, 116, 117, 103, 117, 110, 116, 603, 117, 119, 97, 119, 97, 108, 105, 109, 97, 121, 97, 110, 121, 101, 121, 101,
|
|
14
|
+
110, 121, 603, 114, 603,
|
|
15
|
+
};
|
|
16
|
+
|
|
17
|
+
static const uint32_t stop_word_offsets[49] = {
|
|
18
|
+
0, 1, 6, 9, 11, 14, 16, 18, 21, 26, 28, 31, 33, 38, 43, 47, 49, 51, 55, 56, 58, 64, 66, 72, 74, 78, 82, 84, 87, 92,
|
|
19
|
+
94, 96, 98, 101, 106, 109, 110, 113, 118, 123, 128, 130, 131, 133, 139, 142, 144, 147, 151,
|
|
20
|
+
};
|
|
21
|
+
|
|
22
|
+
static const uint32_t keep_ranges[74] = {
|
|
23
|
+
48, 57, 65, 90, 97, 122, 170, 170, 186, 186, 192, 214, 216, 246, 248, 696, 736, 740, 7424, 7461, 7468, 7516, 7522,
|
|
24
|
+
7525, 7531, 7543, 7545, 7614, 7680, 7935, 8305, 8305, 8319, 8319, 8336, 8348, 8490, 8491, 8498, 8498, 8526, 8526,
|
|
25
|
+
8544, 8584, 11360, 11391, 42786, 42887, 42891, 42972, 42993, 43007, 43824, 43866, 43868, 43876, 43878, 43881, 64256,
|
|
26
|
+
64262, 65313, 65338, 65345, 65370, 67456, 67461, 67463, 67504, 67506, 67514, 122624, 122654, 122661, 122666,
|
|
27
|
+
};
|
|
28
|
+
|
|
29
|
+
const narsil_language narsil_language_bambara = {
|
|
30
|
+
.name = "bambara",
|
|
31
|
+
.revision = "1",
|
|
32
|
+
.stop_word_units = stop_word_units,
|
|
33
|
+
.stop_word_offsets = stop_word_offsets,
|
|
34
|
+
.stop_word_count = 48,
|
|
35
|
+
.ascii_keep = {0x03FF000000000000ULL, 0x07FFFFFE07FFFFFEULL},
|
|
36
|
+
.keep_ranges = keep_ranges,
|
|
37
|
+
.keep_range_count = 37,
|
|
38
|
+
.splits_code_points = 1,
|
|
39
|
+
.strips_possessive = 0,
|
|
40
|
+
.normalises_diacritics = 0,
|
|
41
|
+
.min_token_length = 1,
|
|
42
|
+
.ngram_size = 0,
|
|
43
|
+
.normalise = NULL,
|
|
44
|
+
.stem_by_rules = NULL,
|
|
45
|
+
.create_snowball = NULL,
|
|
46
|
+
.snowball_stem = NULL,
|
|
47
|
+
.close_snowball = NULL,
|
|
48
|
+
};
|
|
@@ -0,0 +1,62 @@
|
|
|
1
|
+
#include "../../analysis/languages.h"
|
|
2
|
+
#include "declarations.h"
|
|
3
|
+
#include "../snowball/basque_stemmer.h"
|
|
4
|
+
|
|
5
|
+
#include <stddef.h>
|
|
6
|
+
#include <stdint.h>
|
|
7
|
+
|
|
8
|
+
static const uint16_t stop_word_units[478] = {
|
|
9
|
+
97, 108, 97, 110, 105, 116, 122, 97, 114, 97, 98, 101, 114, 97, 97, 115, 107, 111, 98, 97, 105, 110, 97, 98, 97, 116,
|
|
10
|
+
98, 97, 116, 101, 97, 110, 98, 97, 116, 101, 107, 98, 97, 116, 105, 98, 97, 116, 122, 117, 101, 105, 98, 97, 116, 122,
|
|
11
|
+
117, 101, 107, 98, 97, 116, 122, 117, 101, 116, 97, 110, 98, 97, 116, 122, 117, 107, 98, 101, 114, 97, 98, 101, 114,
|
|
12
|
+
97, 105, 101, 107, 98, 101, 114, 97, 117, 98, 101, 114, 97, 117, 101, 107, 98, 101, 114, 101, 98, 101, 114, 111, 114,
|
|
13
|
+
105, 98, 101, 114, 111, 114, 105, 101, 107, 98, 101, 115, 116, 101, 98, 101, 122, 97, 108, 97, 100, 97, 100, 97, 103,
|
|
14
|
+
111, 100, 105, 114, 97, 100, 105, 116, 117, 100, 117, 100, 117, 116, 101, 101, 100, 111, 101, 103, 105, 110, 101, 114,
|
|
15
|
+
101, 101, 116, 97, 101, 117, 114, 97, 107, 101, 122, 103, 97, 105, 110, 101, 114, 97, 103, 117, 103, 117, 116, 120,
|
|
16
|
+
105, 103, 117, 122, 116, 105, 104, 97, 105, 101, 105, 104, 97, 105, 101, 107, 104, 97, 105, 101, 116, 97, 110, 104,
|
|
17
|
+
97, 105, 110, 98, 101, 115, 116, 101, 104, 97, 108, 97, 104, 97, 110, 104, 97, 110, 100, 105, 107, 104, 97, 110, 103,
|
|
18
|
+
111, 104, 97, 114, 97, 104, 97, 114, 105, 104, 97, 114, 107, 104, 97, 114, 116, 97, 110, 104, 97, 117, 104, 97, 117,
|
|
19
|
+
101, 105, 104, 97, 117, 101, 107, 104, 97, 117, 101, 116, 97, 110, 104, 101, 109, 101, 110, 104, 101, 109, 101, 110,
|
|
20
|
+
100, 105, 107, 104, 101, 109, 101, 110, 103, 111, 104, 105, 104, 111, 110, 97, 104, 111, 110, 101, 107, 104, 111, 110,
|
|
21
|
+
101, 108, 97, 104, 111, 110, 101, 116, 97, 110, 104, 111, 110, 105, 104, 111, 114, 104, 111, 114, 105, 104, 111, 114,
|
|
22
|
+
105, 101, 105, 104, 111, 114, 105, 101, 107, 104, 111, 114, 105, 101, 116, 97, 110, 104, 111, 114, 107, 111, 104, 111,
|
|
23
|
+
114, 114, 97, 104, 111, 114, 114, 101, 107, 104, 111, 114, 114, 101, 108, 97, 104, 111, 114, 114, 101, 116, 97, 110,
|
|
24
|
+
104, 111, 114, 114, 105, 104, 111, 114, 116, 105, 107, 104, 117, 114, 97, 105, 122, 97, 110, 110, 105, 110, 111, 105,
|
|
25
|
+
122, 110, 111, 108, 97, 110, 111, 110, 110, 111, 110, 100, 105, 107, 110, 111, 110, 103, 111, 110, 111, 114, 110, 111,
|
|
26
|
+
114, 97, 122, 101, 122, 101, 105, 110, 122, 101, 110, 122, 101, 110, 98, 97, 105, 116, 122, 101, 110, 98, 97, 116,
|
|
27
|
+
122, 101, 114, 122, 101, 114, 103, 97, 116, 105, 107, 122, 105, 114, 101, 110, 122, 105, 116, 117, 101, 110, 122, 117,
|
|
28
|
+
122, 117, 101, 107, 122, 117, 101, 110, 122, 117, 116, 101, 110,
|
|
29
|
+
};
|
|
30
|
+
|
|
31
|
+
static const uint32_t stop_word_offsets[99] = {
|
|
32
|
+
0, 2, 7, 14, 18, 23, 26, 32, 37, 41, 48, 55, 64, 70, 74, 81, 86, 93, 97, 103, 111, 116, 122, 124, 128, 132, 136, 138,
|
|
33
|
+
142, 145, 149, 152, 155, 160, 162, 169, 171, 176, 181, 186, 191, 198, 207, 211, 214, 220, 225, 229, 233, 237, 243,
|
|
34
|
+
246, 251, 256, 263, 268, 276, 283, 285, 289, 294, 300, 307, 311, 314, 318, 324, 330, 338, 343, 348, 354, 361, 369,
|
|
35
|
+
374, 380, 384, 388, 390, 394, 398, 401, 407, 412, 415, 419, 421, 425, 428, 435, 441, 444, 452, 457, 463, 465, 469,
|
|
36
|
+
473, 478,
|
|
37
|
+
};
|
|
38
|
+
|
|
39
|
+
static const uint32_t keep_ranges[10] = {
|
|
40
|
+
48, 57, 65, 90, 97, 122, 209, 209, 241, 241,
|
|
41
|
+
};
|
|
42
|
+
|
|
43
|
+
const narsil_language narsil_language_basque = {
|
|
44
|
+
.name = "basque",
|
|
45
|
+
.revision = "287e5c13460de4c3",
|
|
46
|
+
.stop_word_units = stop_word_units,
|
|
47
|
+
.stop_word_offsets = stop_word_offsets,
|
|
48
|
+
.stop_word_count = 98,
|
|
49
|
+
.ascii_keep = {0x03FF000000000000ULL, 0x07FFFFFE07FFFFFEULL},
|
|
50
|
+
.keep_ranges = keep_ranges,
|
|
51
|
+
.keep_range_count = 5,
|
|
52
|
+
.splits_code_points = 0,
|
|
53
|
+
.strips_possessive = 0,
|
|
54
|
+
.normalises_diacritics = 0,
|
|
55
|
+
.min_token_length = 1,
|
|
56
|
+
.ngram_size = 0,
|
|
57
|
+
.normalise = NULL,
|
|
58
|
+
.stem_by_rules = NULL,
|
|
59
|
+
.create_snowball = basque_create_env,
|
|
60
|
+
.snowball_stem = basque_stem,
|
|
61
|
+
.close_snowball = basque_close_env,
|
|
62
|
+
};
|
|
@@ -0,0 +1,56 @@
|
|
|
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[287] = {
|
|
8
|
+
1072, 1072, 1073, 1072, 1073, 1086, 1072, 1076, 1072, 1076, 1079, 1110, 1085, 1072, 1083, 1077, 1073, 1077, 1079,
|
|
9
|
+
1073, 1086, 1073, 1099, 1083, 1072, 1073, 1099, 1083, 1110, 1073, 1099, 1094, 1100, 1073, 1099, 1118, 1074, 1072,
|
|
10
|
+
1089, 1074, 1072, 1096, 1074, 1077, 1083, 1100, 1084, 1110, 1074, 1099, 1075, 1101, 1090, 1072, 1075, 1101, 1090,
|
|
11
|
+
1072, 1103, 1075, 1101, 1090, 1099, 1075, 1101, 1090, 1099, 1103, 1076, 1072, 1076, 1074, 1072, 1076, 1079, 1077,
|
|
12
|
+
1076, 1083, 1103, 1076, 1099, 1079, 1079, 39, 1103, 1118, 1083, 1103, 1077, 1094, 1094, 1072, 1079, 1072, 1082, 1072,
|
|
13
|
+
1073, 1082, 1072, 1083, 1103, 1082, 1072, 1083, 1110, 1082, 1086, 1078, 1085, 1099, 1082, 1091, 1076, 1099, 1084,
|
|
14
|
+
1086, 1081, 1084, 1099, 1084, 1103, 1085, 1077, 1085, 1072, 1085, 1072, 1076, 1085, 1072, 1089, 1085, 1072, 1096,
|
|
15
|
+
1085, 1077, 1085, 1103, 1085, 1103, 1084, 1072, 1087, 1072, 1087, 1072, 1076, 1087, 1072, 1084, 1110, 1078, 1087,
|
|
16
|
+
1072, 1089, 1083, 1103, 1087, 1077, 1088, 1072, 1076, 1087, 1088, 1072, 1087, 1088, 1072, 1079, 1087, 1088, 1099,
|
|
17
|
+
1089, 1072, 1089, 1072, 1084, 1089, 1074, 1072, 1077, 1089, 1074, 1072, 1102, 1089, 1074, 1086, 1081, 1089, 1103,
|
|
18
|
+
1088, 1086, 1076, 1090, 1072, 1082, 1090, 1072, 1082, 1089, 1072, 1084, 1072, 1090, 1072, 1084, 1091, 1090, 1072,
|
|
19
|
+
1103, 1090, 1074, 1086, 1081, 1090, 1086, 1081, 1090, 1086, 1083, 1100, 1082, 1110, 1090, 1099, 1090, 1099, 1103,
|
|
20
|
+
1091, 1091, 1078, 1086, 1091, 1089, 1077, 1091, 1089, 1105, 1094, 1103, 1073, 1077, 1094, 1110, 1095, 1072, 1084,
|
|
21
|
+
1091, 1096, 1090, 1086, 1103, 1103, 1075, 1086, 1103, 1077, 1103, 1082, 1103, 1085, 1072, 1103, 1085, 1086, 1103,
|
|
22
|
+
1085, 1099, 1103, 1096, 1095, 1101, 1105, 1085, 1105, 1089, 1094, 1100, 1110, 1110, 1085, 1096, 1099, 1110, 1093,
|
|
23
|
+
1118,
|
|
24
|
+
};
|
|
25
|
+
|
|
26
|
+
static const uint32_t stop_word_offsets[89] = {
|
|
27
|
+
0, 1, 3, 6, 8, 13, 16, 19, 21, 25, 29, 33, 36, 39, 42, 48, 50, 54, 59, 63, 68, 70, 73, 76, 79, 81, 82, 92, 94, 97,
|
|
28
|
+
101, 105, 110, 114, 117, 119, 123, 125, 128, 131, 134, 136, 138, 142, 144, 147, 152, 157, 162, 165, 169, 172, 174,
|
|
29
|
+
177, 181, 185, 189, 194, 197, 204, 208, 211, 215, 218, 224, 226, 229, 230, 233, 236, 239, 243, 245, 249, 252, 253,
|
|
30
|
+
256, 258, 260, 263, 266, 269, 273, 275, 279, 280, 284, 286, 287,
|
|
31
|
+
};
|
|
32
|
+
|
|
33
|
+
static const uint32_t keep_ranges[8] = {
|
|
34
|
+
39, 39, 48, 57, 1024, 1279, 7296, 7303,
|
|
35
|
+
};
|
|
36
|
+
|
|
37
|
+
const narsil_language narsil_language_belarusian = {
|
|
38
|
+
.name = "belarusian",
|
|
39
|
+
.revision = "0c1d3db2f7d0be62",
|
|
40
|
+
.stop_word_units = stop_word_units,
|
|
41
|
+
.stop_word_offsets = stop_word_offsets,
|
|
42
|
+
.stop_word_count = 88,
|
|
43
|
+
.ascii_keep = {0x03FF008000000000ULL, 0x0000000000000000ULL},
|
|
44
|
+
.keep_ranges = keep_ranges,
|
|
45
|
+
.keep_range_count = 4,
|
|
46
|
+
.splits_code_points = 0,
|
|
47
|
+
.strips_possessive = 0,
|
|
48
|
+
.normalises_diacritics = 0,
|
|
49
|
+
.min_token_length = 1,
|
|
50
|
+
.ngram_size = 0,
|
|
51
|
+
.normalise = NULL,
|
|
52
|
+
.stem_by_rules = NULL,
|
|
53
|
+
.create_snowball = NULL,
|
|
54
|
+
.snowball_stem = NULL,
|
|
55
|
+
.close_snowball = NULL,
|
|
56
|
+
};
|
|
@@ -0,0 +1,63 @@
|
|
|
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[408] = {
|
|
8
|
+
2437, 2472, 2503, 2453, 2437, 2472, 2509, 2479, 2437, 2476, 2486, 2509, 2479, 2438, 2455, 2503, 2438, 2459, 2503,
|
|
9
|
+
2438, 2460, 2438, 2476, 2494, 2480, 2438, 2478, 2480, 2494, 2438, 2478, 2494, 2470, 2503, 2480, 2438, 2480, 2439,
|
|
10
|
+
2441, 2468, 2509, 2468, 2480, 2441, 2474, 2480, 2441, 2474, 2480, 2503, 2447, 2447, 2439, 2447, 2453, 2509, 2447,
|
|
11
|
+
2454, 2472, 2447, 2468, 2447, 2476, 2447, 2478, 2472, 2447, 2478, 2472, 2495, 2447, 2480, 2447, 2488, 2447, 2488,
|
|
12
|
+
2503, 2451, 2451, 2439, 2453, 2478, 2472, 2503, 2453, 2479, 2492, 2503, 2453, 2453, 2480, 2494, 2453, 2480, 2503,
|
|
13
|
+
2453, 2494, 2459, 2503, 2453, 2494, 2460, 2453, 2494, 2460, 2503, 2453, 2494, 2480, 2467, 2453, 2495, 2453, 2495,
|
|
14
|
+
2459, 2497, 2453, 2503, 2453, 2503, 2441, 2453, 2503, 2454, 2494, 2453, 2503, 2472, 2453, 2507, 2463, 2495, 2453,
|
|
15
|
+
2507, 2472, 2507, 2454, 2497, 2476, 2455, 2495, 2479, 2492, 2503, 2455, 2503, 2482, 2458, 2494, 2480, 2458, 2494,
|
|
16
|
+
2482, 2497, 2458, 2503, 2487, 2509, 2463, 2494, 2459, 2495, 2482, 2460, 2494, 2472, 2494, 2460, 2509, 2472, 2460,
|
|
17
|
+
2472, 2463, 2495, 2468, 2454, 2472, 2468, 2476, 2503, 2468, 2494, 2468, 2494, 2439, 2468, 2507, 2469, 2494, 2453,
|
|
18
|
+
2494, 2469, 2503, 2453, 2503, 2470, 2495, 2472, 2470, 2497, 2470, 2497, 2439, 2470, 2503, 2451, 2479, 2492, 2494,
|
|
19
|
+
2471, 2494, 2478, 2494, 2480, 2472, 2468, 2497, 2472, 2472, 2479, 2492, 2472, 2494, 2472, 2494, 2455, 2494, 2470,
|
|
20
|
+
2472, 2495, 2479, 2492, 2503, 2472, 2503, 2451, 2479, 2492, 2494, 2474, 2480, 2474, 2480, 2503, 2474, 2494, 2458,
|
|
21
|
+
2474, 2495, 2474, 2503, 2479, 2492, 2509, 2480, 2509, 2474, 2509, 2480, 2468, 2495, 2474, 2509, 2480, 2469, 2478,
|
|
22
|
+
2474, 2509, 2480, 2479, 2472, 2509, 2468, 2474, 2509, 2480, 2494, 2469, 2478, 2495, 2453, 2474, 2509, 2480, 2494,
|
|
23
|
+
2479, 2492, 2476, 2453, 2509, 2468, 2476, 2509, 2479, 2476, 2472, 2476, 2482, 2494, 2476, 2482, 2503, 2476, 2482,
|
|
24
|
+
2503, 2472, 2476, 2489, 2497, 2476, 2494, 2476, 2495, 2476, 2495, 2477, 2495, 2472, 2509, 2472, 2476, 2503, 2486,
|
|
25
|
+
2476, 2503, 2486, 2495, 2478, 2468, 2507, 2478, 2471, 2509, 2479, 2503, 2478, 2472, 2503, 2479, 2454, 2472, 2479,
|
|
26
|
+
2470, 2495, 2479, 2494, 2479, 2494, 2451, 2479, 2492, 2494, 2479, 2503, 2480, 2480, 2453, 2478, 2482, 2453, 2509,
|
|
27
|
+
2487, 2486, 2497, 2471, 2497, 2486, 2497, 2480, 2497, 2488, 2457, 2509, 2455, 2503, 2488, 2476, 2488, 2489, 2488,
|
|
28
|
+
2494, 2471, 2494, 2480, 2467, 2488, 2494, 2478, 2472, 2503, 2488, 2495, 2488, 2503, 2488, 2503, 2439, 2489, 2468,
|
|
29
|
+
2503, 2489, 2479, 2492, 2489, 2494, 2460, 2494, 2480,
|
|
30
|
+
};
|
|
31
|
+
|
|
32
|
+
static const uint32_t stop_word_offsets[117] = {
|
|
33
|
+
0, 4, 8, 13, 16, 19, 21, 25, 29, 35, 37, 38, 43, 46, 50, 51, 53, 56, 59, 61, 63, 66, 70, 72, 74, 77, 78, 80, 84, 89,
|
|
34
|
+
92, 95, 99, 102, 106, 110, 112, 116, 118, 121, 125, 128, 132, 136, 139, 144, 147, 150, 154, 160, 163, 167, 172, 174,
|
|
35
|
+
177, 180, 182, 185, 187, 191, 195, 198, 200, 203, 209, 214, 218, 221, 223, 228, 233, 239, 241, 244, 247, 249, 256,
|
|
36
|
+
261, 266, 273, 281, 287, 294, 296, 299, 302, 306, 309, 311, 313, 320, 323, 327, 330, 335, 338, 341, 344, 346, 352,
|
|
37
|
+
354, 355, 358, 362, 366, 370, 375, 377, 379, 385, 390, 392, 394, 397, 400, 403, 408,
|
|
38
|
+
};
|
|
39
|
+
|
|
40
|
+
static const uint32_t keep_ranges[8] = {
|
|
41
|
+
48, 57, 65, 90, 97, 122, 2432, 2559,
|
|
42
|
+
};
|
|
43
|
+
|
|
44
|
+
const narsil_language narsil_language_bengali = {
|
|
45
|
+
.name = "bengali",
|
|
46
|
+
.revision = "1",
|
|
47
|
+
.stop_word_units = stop_word_units,
|
|
48
|
+
.stop_word_offsets = stop_word_offsets,
|
|
49
|
+
.stop_word_count = 116,
|
|
50
|
+
.ascii_keep = {0x03FF000000000000ULL, 0x07FFFFFE07FFFFFEULL},
|
|
51
|
+
.keep_ranges = keep_ranges,
|
|
52
|
+
.keep_range_count = 4,
|
|
53
|
+
.splits_code_points = 0,
|
|
54
|
+
.strips_possessive = 0,
|
|
55
|
+
.normalises_diacritics = 0,
|
|
56
|
+
.min_token_length = 1,
|
|
57
|
+
.ngram_size = 0,
|
|
58
|
+
.normalise = NULL,
|
|
59
|
+
.stem_by_rules = NULL,
|
|
60
|
+
.create_snowball = NULL,
|
|
61
|
+
.snowball_stem = NULL,
|
|
62
|
+
.close_snowball = NULL,
|
|
63
|
+
};
|
|
@@ -0,0 +1,73 @@
|
|
|
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[688] = {
|
|
8
|
+
97, 97, 107, 111, 97, 108, 105, 98, 105, 98, 105, 104, 98, 105, 108, 97, 98, 105, 108, 105, 98, 105, 108, 111, 98,
|
|
9
|
+
105, 111, 98, 105, 115, 109, 111, 98, 105, 115, 116, 101, 98, 105, 116, 105, 98, 117, 109, 111, 100, 97, 100, 111,
|
|
10
|
+
100, 117, 382, 103, 97, 104, 111, 263, 101, 104, 111, 263, 101, 109, 111, 104, 111, 263, 101, 116, 101, 104, 111, 263,
|
|
11
|
+
101, 353, 104, 111, 263, 117, 105, 105, 97, 107, 111, 105, 104, 105, 108, 105, 105, 122, 106, 97, 106, 101, 106, 101,
|
|
12
|
+
100, 110, 97, 106, 101, 100, 110, 101, 106, 101, 100, 110, 111, 106, 101, 114, 106, 101, 115, 97, 109, 106, 101, 115,
|
|
13
|
+
105, 106, 101, 115, 109, 111, 106, 101, 115, 116, 106, 101, 115, 116, 101, 106, 101, 115, 117, 106, 105, 109, 106,
|
|
14
|
+
111, 106, 106, 111, 353, 106, 117, 107, 97, 100, 97, 107, 97, 107, 111, 107, 97, 111, 107, 111, 107, 111, 106, 97,
|
|
15
|
+
107, 111, 106, 101, 107, 111, 106, 105, 107, 111, 106, 105, 109, 97, 107, 111, 106, 117, 107, 114, 111, 122, 108, 105,
|
|
16
|
+
109, 101, 109, 101, 110, 101, 109, 101, 110, 105, 109, 105, 109, 105, 109, 111, 109, 111, 106, 109, 111, 106, 97, 109,
|
|
17
|
+
111, 106, 101, 109, 117, 110, 97, 110, 97, 100, 110, 97, 107, 111, 110, 110, 97, 109, 110, 97, 109, 97, 110, 97, 115,
|
|
18
|
+
110, 97, 353, 110, 97, 353, 97, 110, 97, 353, 101, 110, 97, 353, 101, 103, 110, 101, 110, 101, 103, 111, 110, 101,
|
|
19
|
+
107, 97, 110, 101, 107, 105, 110, 101, 107, 111, 110, 101, 107, 111, 103, 110, 101, 107, 117, 110, 101, 109, 97, 110,
|
|
20
|
+
101, 116, 107, 111, 110, 101, 263, 101, 110, 101, 263, 101, 109, 111, 110, 101, 263, 101, 116, 101, 110, 101, 263,
|
|
21
|
+
101, 353, 110, 101, 263, 117, 110, 101, 353, 116, 111, 110, 105, 110, 105, 106, 101, 110, 105, 107, 111, 110, 105,
|
|
22
|
+
107, 111, 103, 97, 110, 105, 107, 111, 106, 101, 110, 105, 107, 111, 106, 117, 110, 105, 115, 97, 109, 110, 105, 115,
|
|
23
|
+
105, 110, 105, 115, 109, 111, 110, 105, 115, 116, 101, 110, 105, 115, 117, 110, 106, 101, 103, 97, 110, 106, 101, 103,
|
|
24
|
+
111, 118, 110, 106, 101, 103, 111, 118, 97, 110, 106, 101, 103, 111, 118, 111, 110, 106, 101, 109, 117, 110, 106, 101,
|
|
25
|
+
122, 105, 110, 110, 106, 101, 122, 105, 110, 97, 110, 106, 101, 122, 105, 110, 111, 110, 106, 105, 104, 110, 106, 105,
|
|
26
|
+
104, 111, 118, 110, 106, 105, 104, 111, 118, 97, 110, 106, 105, 104, 111, 118, 111, 110, 106, 105, 109, 110, 106, 105,
|
|
27
|
+
109, 97, 110, 106, 111, 106, 110, 106, 117, 110, 111, 111, 111, 100, 111, 100, 109, 97, 104, 111, 110, 111, 110, 97,
|
|
28
|
+
111, 110, 105, 111, 110, 111, 111, 118, 97, 112, 97, 112, 97, 107, 112, 111, 112, 111, 100, 112, 111, 114, 101, 100,
|
|
29
|
+
112, 114, 105, 106, 101, 115, 115, 97, 115, 97, 109, 115, 97, 109, 111, 115, 101, 115, 101, 98, 101, 115, 101, 98,
|
|
30
|
+
105, 115, 105, 115, 109, 111, 115, 116, 101, 115, 117, 115, 118, 101, 115, 118, 105, 115, 118, 111, 103, 115, 118,
|
|
31
|
+
111, 106, 115, 118, 111, 106, 97, 115, 118, 111, 106, 101, 115, 118, 111, 109, 116, 97, 116, 97, 100, 97, 116, 97,
|
|
32
|
+
106, 116, 97, 107, 111, 116, 97, 107, 111, 273, 101, 116, 97, 107, 111, 273, 101, 114, 116, 101, 116, 101, 98, 101,
|
|
33
|
+
116, 101, 98, 105, 116, 105, 116, 111, 116, 111, 106, 116, 111, 109, 101, 116, 117, 116, 118, 111, 106, 116, 118, 111,
|
|
34
|
+
106, 97, 116, 118, 111, 106, 101, 117, 117, 122, 118, 97, 109, 118, 97, 109, 97, 118, 97, 115, 118, 97, 353, 118, 97,
|
|
35
|
+
353, 97, 118, 97, 353, 101, 118, 101, 263, 118, 105, 118, 114, 108, 111, 122, 97, 122, 97, 114, 263, 101, 263, 101,
|
|
36
|
+
109, 111, 263, 101, 116, 101, 263, 101, 353, 263, 117, 353, 116, 97, 353, 116, 111,
|
|
37
|
+
};
|
|
38
|
+
|
|
39
|
+
static const uint32_t stop_word_offsets[186] = {
|
|
40
|
+
0, 1, 4, 7, 9, 12, 16, 20, 24, 27, 32, 37, 41, 45, 47, 49, 52, 54, 58, 64, 70, 75, 79, 80, 84, 86, 89, 91, 93, 95,
|
|
41
|
+
100, 105, 110, 113, 118, 122, 127, 131, 136, 140, 143, 146, 149, 151, 155, 159, 162, 164, 168, 172, 176, 182, 186,
|
|
42
|
+
190, 192, 194, 198, 202, 204, 208, 211, 215, 219, 221, 223, 226, 231, 234, 238, 241, 244, 248, 252, 257, 259, 263,
|
|
43
|
+
267, 271, 275, 280, 284, 288, 293, 297, 303, 309, 314, 318, 323, 325, 329, 333, 339, 345, 351, 356, 360, 365, 370,
|
|
44
|
+
374, 379, 385, 392, 399, 404, 410, 417, 424, 428, 434, 441, 448, 452, 457, 461, 464, 466, 467, 469, 474, 476, 479,
|
|
45
|
+
482, 485, 488, 490, 493, 495, 498, 503, 508, 509, 511, 514, 518, 520, 524, 528, 530, 533, 536, 538, 541, 544, 548,
|
|
46
|
+
552, 557, 562, 566, 568, 572, 575, 579, 585, 592, 594, 598, 602, 604, 606, 609, 613, 615, 619, 624, 629, 630, 632,
|
|
47
|
+
635, 639, 642, 645, 649, 653, 656, 658, 662, 664, 667, 669, 673, 677, 680, 682, 685, 688,
|
|
48
|
+
};
|
|
49
|
+
|
|
50
|
+
static const uint32_t keep_ranges[16] = {
|
|
51
|
+
48, 57, 65, 90, 97, 122, 262, 263, 268, 269, 272, 273, 352, 353, 381, 382,
|
|
52
|
+
};
|
|
53
|
+
|
|
54
|
+
const narsil_language narsil_language_bosnian = {
|
|
55
|
+
.name = "bosnian",
|
|
56
|
+
.revision = "1",
|
|
57
|
+
.stop_word_units = stop_word_units,
|
|
58
|
+
.stop_word_offsets = stop_word_offsets,
|
|
59
|
+
.stop_word_count = 185,
|
|
60
|
+
.ascii_keep = {0x03FF000000000000ULL, 0x07FFFFFE07FFFFFEULL},
|
|
61
|
+
.keep_ranges = keep_ranges,
|
|
62
|
+
.keep_range_count = 8,
|
|
63
|
+
.splits_code_points = 0,
|
|
64
|
+
.strips_possessive = 0,
|
|
65
|
+
.normalises_diacritics = 0,
|
|
66
|
+
.min_token_length = 1,
|
|
67
|
+
.ngram_size = 0,
|
|
68
|
+
.normalise = NULL,
|
|
69
|
+
.stem_by_rules = NULL,
|
|
70
|
+
.create_snowball = NULL,
|
|
71
|
+
.snowball_stem = NULL,
|
|
72
|
+
.close_snowball = NULL,
|
|
73
|
+
};
|
|
@@ -0,0 +1,52 @@
|
|
|
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[278] = {
|
|
8
|
+
97, 97, 108, 97, 108, 108, 97, 109, 97, 241, 97, 110, 97, 114, 97, 114, 97, 108, 108, 97, 116, 97, 118, 97, 118, 97,
|
|
9
|
+
116, 97, 122, 101, 98, 101, 116, 98, 101, 116, 101, 107, 98, 114, 101, 109, 97, 241, 99, 39, 104, 111, 97, 122, 104,
|
|
10
|
+
100, 97, 100, 105, 110, 100, 97, 110, 100, 105, 111, 117, 122, 104, 100, 105, 119, 97, 114, 100, 114, 101, 101, 101,
|
|
11
|
+
103, 105, 108, 101, 101, 110, 101, 110, 111, 101, 111, 101, 114, 101, 116, 114, 101, 101, 117, 115, 101, 118, 101,
|
|
12
|
+
108, 101, 118, 105, 116, 101, 122, 101, 241, 103, 97, 110, 116, 103, 111, 117, 100, 101, 103, 119, 97, 108, 108, 104,
|
|
13
|
+
97, 104, 97, 103, 104, 101, 104, 101, 112, 104, 105, 104, 111, 104, 111, 103, 101, 110, 104, 111, 108, 108, 104, 111,
|
|
14
|
+
110, 104, 111, 114, 105, 110, 116, 105, 118, 101, 122, 107, 97, 101, 114, 107, 101, 108, 107, 101, 109, 101, 110, 116,
|
|
15
|
+
107, 101, 110, 107, 101, 110, 116, 114, 101, 107, 101, 116, 109, 97, 109, 97, 114, 109, 101, 109, 101, 116, 110, 97,
|
|
16
|
+
103, 110, 101, 110, 101, 109, 101, 116, 110, 101, 117, 122, 101, 110, 105, 111, 111, 97, 111, 117, 122, 104, 112, 97,
|
|
17
|
+
112, 101, 112, 101, 103, 111, 117, 108, 122, 112, 101, 110, 97, 111, 115, 112, 101, 112, 112, 101, 114, 97, 107, 112,
|
|
18
|
+
101, 116, 114, 97, 112, 105, 118, 114, 97, 107, 114, 101, 115, 101, 115, 101, 117, 108, 116, 101, 116, 114, 97, 117,
|
|
19
|
+
108, 117, 110, 117, 114, 118, 97, 119, 97, 114, 122, 111,
|
|
20
|
+
};
|
|
21
|
+
|
|
22
|
+
static const uint32_t stop_word_offsets[85] = {
|
|
23
|
+
0, 1, 3, 6, 10, 12, 14, 19, 23, 27, 30, 33, 38, 44, 51, 53, 59, 65, 70, 73, 74, 79, 81, 84, 86, 88, 92, 95, 99, 103,
|
|
24
|
+
105, 107, 111, 116, 121, 123, 126, 128, 131, 133, 135, 140, 144, 147, 150, 153, 157, 161, 164, 170, 173, 179, 182,
|
|
25
|
+
184, 187, 189, 192, 195, 197, 202, 207, 209, 210, 212, 216, 218, 220, 227, 233, 236, 241, 246, 249, 252, 254, 256,
|
|
26
|
+
260, 262, 265, 267, 269, 271, 273, 276, 278,
|
|
27
|
+
};
|
|
28
|
+
|
|
29
|
+
static const uint32_t keep_ranges[20] = {
|
|
30
|
+
39, 39, 48, 57, 65, 90, 97, 122, 202, 202, 209, 209, 217, 217, 234, 234, 241, 241, 249, 249,
|
|
31
|
+
};
|
|
32
|
+
|
|
33
|
+
const narsil_language narsil_language_breton = {
|
|
34
|
+
.name = "breton",
|
|
35
|
+
.revision = "1",
|
|
36
|
+
.stop_word_units = stop_word_units,
|
|
37
|
+
.stop_word_offsets = stop_word_offsets,
|
|
38
|
+
.stop_word_count = 84,
|
|
39
|
+
.ascii_keep = {0x03FF008000000000ULL, 0x07FFFFFE07FFFFFEULL},
|
|
40
|
+
.keep_ranges = keep_ranges,
|
|
41
|
+
.keep_range_count = 10,
|
|
42
|
+
.splits_code_points = 0,
|
|
43
|
+
.strips_possessive = 0,
|
|
44
|
+
.normalises_diacritics = 0,
|
|
45
|
+
.min_token_length = 1,
|
|
46
|
+
.ngram_size = 0,
|
|
47
|
+
.normalise = NULL,
|
|
48
|
+
.stem_by_rules = NULL,
|
|
49
|
+
.create_snowball = NULL,
|
|
50
|
+
.snowball_stem = NULL,
|
|
51
|
+
.close_snowball = NULL,
|
|
52
|
+
};
|
|
@@ -0,0 +1,144 @@
|
|
|
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[1072] = {
|
|
9
|
+
1072, 1072, 1074, 1090, 1077, 1085, 1090, 1080, 1095, 1077, 1085, 1072, 1079, 1072, 1082, 1086, 1072, 1083, 1072,
|
|
10
|
+
1073, 1077, 1073, 1077, 1079, 1073, 1077, 1096, 1077, 1073, 1080, 1073, 1080, 1074, 1096, 1073, 1080, 1074, 1096,
|
|
11
|
+
1072, 1073, 1080, 1074, 1096, 1086, 1073, 1080, 1083, 1073, 1080, 1083, 1072, 1073, 1080, 1083, 1080, 1073, 1080,
|
|
12
|
+
1083, 1086, 1073, 1083, 1072, 1075, 1086, 1076, 1072, 1088, 1103, 1073, 1083, 1080, 1079, 1086, 1073, 1098, 1076,
|
|
13
|
+
1072, 1090, 1073, 1098, 1076, 1077, 1073, 1103, 1093, 1072, 1074, 1074, 1072, 1089, 1074, 1072, 1096, 1074, 1072,
|
|
14
|
+
1096, 1072, 1074, 1077, 1088, 1086, 1103, 1090, 1085, 1086, 1074, 1077, 1095, 1077, 1074, 1079, 1077, 1084, 1072,
|
|
15
|
+
1074, 1080, 1074, 1080, 1077, 1074, 1080, 1085, 1072, 1075, 1080, 1074, 1085, 1080, 1084, 1072, 1074, 1072, 1074,
|
|
16
|
+
1088, 1077, 1084, 1077, 1074, 1089, 1077, 1074, 1089, 1077, 1082, 1080, 1074, 1089, 1080, 1095, 1082, 1080, 1074,
|
|
17
|
+
1089, 1080, 1095, 1082, 1086, 1074, 1089, 1103, 1082, 1072, 1074, 1098, 1074, 1074, 1098, 1087, 1088, 1077, 1082,
|
|
18
|
+
1080, 1074, 1098, 1088, 1093, 1091, 1075, 1075, 1080, 1075, 1083, 1072, 1074, 1077, 1085, 1075, 1083, 1072, 1074,
|
|
19
|
+
1085, 1072, 1075, 1083, 1072, 1074, 1085, 1086, 1075, 1083, 1072, 1089, 1075, 1086, 1075, 1086, 1076, 1080, 1085,
|
|
20
|
+
1072, 1075, 1086, 1076, 1080, 1085, 1080, 1075, 1086, 1076, 1080, 1096, 1077, 1085, 1076, 1076, 1072, 1076, 1072,
|
|
21
|
+
1083, 1080, 1076, 1074, 1072, 1076, 1074, 1072, 1084, 1072, 1076, 1074, 1072, 1084, 1072, 1090, 1072, 1076, 1074,
|
|
22
|
+
1077, 1076, 1074, 1077, 1090, 1077, 1076, 1077, 1085, 1076, 1085, 1077, 1089, 1076, 1085, 1080, 1076, 1086, 1076,
|
|
23
|
+
1086, 1073, 1088, 1072, 1076, 1086, 1073, 1088, 1077, 1076, 1086, 1073, 1088, 1086, 1076, 1086, 1073, 1098, 1088,
|
|
24
|
+
1076, 1086, 1082, 1072, 1090, 1086, 1076, 1086, 1082, 1086, 1075, 1072, 1076, 1086, 1088, 1080, 1076, 1086, 1089,
|
|
25
|
+
1077, 1075, 1072, 1076, 1086, 1089, 1090, 1072, 1076, 1088, 1091, 1075, 1076, 1088, 1091, 1075, 1072, 1076, 1088,
|
|
26
|
+
1091, 1075, 1080, 1077, 1077, 1074, 1090, 1080, 1085, 1077, 1076, 1074, 1072, 1077, 1076, 1080, 1085, 1077, 1076,
|
|
27
|
+
1085, 1072, 1077, 1076, 1085, 1072, 1082, 1074, 1072, 1077, 1076, 1085, 1072, 1082, 1074, 1080, 1077, 1076, 1085,
|
|
28
|
+
1072, 1082, 1098, 1074, 1077, 1076, 1085, 1086, 1077, 1082, 1080, 1087, 1077, 1090, 1086, 1078, 1080, 1074, 1086,
|
|
29
|
+
1090, 1079, 1072, 1079, 1072, 1073, 1072, 1074, 1103, 1084, 1079, 1072, 1076, 1079, 1072, 1077, 1076, 1085, 1086,
|
|
30
|
+
1079, 1072, 1088, 1072, 1076, 1080, 1079, 1072, 1089, 1077, 1075, 1072, 1079, 1072, 1089, 1087, 1072, 1083, 1079,
|
|
31
|
+
1072, 1090, 1086, 1074, 1072, 1079, 1072, 1097, 1086, 1079, 1072, 1097, 1086, 1090, 1086, 1080, 1080, 1079, 1080,
|
|
32
|
+
1083, 1080, 1080, 1084, 1080, 1084, 1072, 1080, 1084, 1072, 1090, 1080, 1089, 1082, 1072, 1081, 1082, 1072, 1079,
|
|
33
|
+
1072, 1082, 1072, 1082, 1082, 1072, 1082, 1074, 1072, 1082, 1072, 1082, 1074, 1086, 1082, 1072, 1082, 1090, 1086,
|
|
34
|
+
1082, 1072, 1082, 1098, 1074, 1082, 1072, 1090, 1086, 1082, 1086, 1075, 1072, 1082, 1086, 1075, 1072, 1090, 1086,
|
|
35
|
+
1082, 1086, 1077, 1090, 1086, 1082, 1086, 1080, 1090, 1086, 1082, 1086, 1081, 1082, 1086, 1081, 1090, 1086, 1082,
|
|
36
|
+
1086, 1083, 1082, 1086, 1082, 1086, 1103, 1090, 1086, 1082, 1098, 1076, 1077, 1082, 1098, 1076, 1077, 1090, 1086,
|
|
37
|
+
1082, 1098, 1084, 1083, 1077, 1089, 1077, 1085, 1083, 1077, 1089, 1085, 1086, 1083, 1080, 1083, 1086, 1096, 1084,
|
|
38
|
+
1084, 1072, 1081, 1084, 1072, 1083, 1082, 1086, 1084, 1077, 1084, 1077, 1078, 1076, 1091, 1084, 1077, 1082, 1084,
|
|
39
|
+
1077, 1085, 1084, 1077, 1089, 1077, 1094, 1084, 1080, 1084, 1085, 1086, 1075, 1086, 1084, 1085, 1086, 1079, 1080,
|
|
40
|
+
1085, 1072, 1084, 1086, 1075, 1072, 1084, 1086, 1075, 1072, 1090, 1084, 1086, 1078, 1077, 1084, 1086, 1082, 1098,
|
|
41
|
+
1088, 1084, 1086, 1083, 1103, 1084, 1086, 1084, 1077, 1085, 1090, 1072, 1084, 1091, 1085, 1085, 1072, 1085, 1072,
|
|
42
|
+
1076, 1085, 1072, 1079, 1072, 1076, 1085, 1072, 1081, 1085, 1072, 1087, 1088, 1072, 1074, 1080, 1085, 1072, 1087,
|
|
43
|
+
1088, 1077, 1076, 1085, 1072, 1087, 1088, 1080, 1084, 1077, 1088, 1085, 1072, 1089, 1085, 1077, 1085, 1077, 1075,
|
|
44
|
+
1086, 1085, 1077, 1097, 1086, 1085, 1077, 1103, 1085, 1080, 1085, 1080, 1077, 1085, 1080, 1082, 1086, 1081, 1085,
|
|
45
|
+
1080, 1090, 1086, 1085, 1080, 1097, 1086, 1085, 1086, 1085, 1086, 1074, 1085, 1086, 1074, 1072, 1085, 1086, 1074,
|
|
46
|
+
1080, 1085, 1086, 1074, 1080, 1085, 1072, 1085, 1103, 1082, 1086, 1080, 1085, 1103, 1082, 1086, 1081, 1085, 1103,
|
|
47
|
+
1082, 1086, 1083, 1082, 1086, 1085, 1103, 1084, 1072, 1086, 1073, 1072, 1095, 1077, 1086, 1082, 1086, 1083, 1086,
|
|
48
|
+
1086, 1089, 1074, 1077, 1085, 1086, 1089, 1086, 1073, 1077, 1085, 1086, 1086, 1090, 1086, 1090, 1075, 1086, 1088,
|
|
49
|
+
1077, 1086, 1090, 1085, 1086, 1074, 1086, 1086, 1097, 1077, 1087, 1072, 1082, 1087, 1086, 1087, 1086, 1074, 1077,
|
|
50
|
+
1095, 1077, 1087, 1086, 1074, 1077, 1095, 1077, 1090, 1086, 1087, 1086, 1076, 1087, 1086, 1085, 1077, 1087, 1086,
|
|
51
|
+
1088, 1072, 1076, 1080, 1087, 1086, 1089, 1083, 1077, 1087, 1086, 1095, 1090, 1080, 1087, 1088, 1072, 1074, 1080,
|
|
52
|
+
1087, 1088, 1077, 1076, 1087, 1088, 1077, 1076, 1080, 1087, 1088, 1077, 1079, 1087, 1088, 1080, 1087, 1098, 1082,
|
|
53
|
+
1087, 1098, 1088, 1074, 1072, 1090, 1072, 1087, 1098, 1088, 1074, 1080, 1087, 1098, 1088, 1074, 1086, 1087, 1098,
|
|
54
|
+
1090, 1080, 1088, 1072, 1074, 1077, 1085, 1088, 1072, 1074, 1085, 1072, 1089, 1089, 1072, 1089, 1072, 1084, 1089,
|
|
55
|
+
1072, 1084, 1086, 1089, 1077, 1089, 1077, 1075, 1072, 1089, 1080, 1089, 1080, 1085, 1089, 1082, 1086, 1088, 1086,
|
|
56
|
+
1089, 1083, 1077, 1076, 1089, 1083, 1077, 1076, 1074, 1072, 1097, 1089, 1084, 1077, 1089, 1084, 1103, 1093, 1089,
|
|
57
|
+
1087, 1086, 1088, 1077, 1076, 1089, 1088, 1077, 1076, 1089, 1088, 1077, 1097, 1091, 1089, 1090, 1077, 1089, 1098,
|
|
58
|
+
1084, 1089, 1098, 1089, 1089, 1098, 1097, 1086, 1090, 1090, 1072, 1079, 1080, 1090, 1072, 1082, 1072, 1090, 1072,
|
|
59
|
+
1082, 1080, 1074, 1072, 1090, 1072, 1082, 1098, 1074, 1090, 1072, 1084, 1090, 1074, 1086, 1081, 1090, 1077, 1090,
|
|
60
|
+
1077, 1079, 1080, 1090, 1080, 1090, 1086, 1090, 1086, 1074, 1072, 1090, 1086, 1075, 1072, 1074, 1072, 1090, 1086,
|
|
61
|
+
1079, 1080, 1090, 1086, 1081, 1090, 1086, 1083, 1082, 1086, 1074, 1072, 1090, 1086, 1095, 1085, 1086, 1090, 1088,
|
|
62
|
+
1080, 1090, 1088, 1103, 1073, 1074, 1072, 1090, 1091, 1082, 1090, 1098, 1081, 1090, 1103, 1090, 1103, 1093, 1091,
|
|
63
|
+
1091, 1090, 1088, 1077, 1093, 1072, 1088, 1077, 1089, 1074, 1072, 1093, 1080, 1083, 1103, 1076, 1080, 1095, 1095,
|
|
64
|
+
1072, 1089, 1072, 1095, 1077, 1095, 1077, 1089, 1090, 1086, 1095, 1088, 1077, 1079, 1097, 1077, 1097, 1086, 1084,
|
|
65
|
+
1102, 1084, 1088, 1091, 1082, 1103, 1103, 1082,
|
|
66
|
+
};
|
|
67
|
+
|
|
68
|
+
static const uint32_t stop_word_offsets[259] = {
|
|
69
|
+
0, 1, 11, 13, 16, 19, 21, 24, 28, 30, 34, 39, 44, 47, 51, 55, 59, 68, 73, 78, 82, 86, 87, 90, 93, 97, 105, 109, 114,
|
|
70
|
+
116, 119, 125, 132, 137, 140, 145, 151, 157, 162, 165, 172, 177, 178, 180, 186, 192, 198, 202, 204, 210, 216, 223,
|
|
71
|
+
224, 226, 230, 233, 238, 245, 248, 253, 256, 260, 263, 265, 270, 275, 280, 285, 291, 297, 301, 307, 312, 316, 321,
|
|
72
|
+
326, 327, 332, 336, 340, 344, 351, 358, 365, 369, 373, 376, 381, 383, 390, 393, 399, 405, 411, 417, 423, 427, 433,
|
|
73
|
+
434, 436, 439, 441, 444, 448, 452, 453, 457, 460, 465, 470, 475, 480, 484, 488, 494, 499, 504, 507, 512, 517, 522,
|
|
74
|
+
526, 532, 535, 540, 545, 547, 550, 551, 554, 559, 561, 566, 569, 572, 577, 579, 584, 591, 595, 600, 604, 609, 613,
|
|
75
|
+
620, 622, 623, 625, 628, 633, 636, 643, 649, 657, 660, 662, 666, 670, 673, 675, 678, 683, 687, 691, 693, 696, 700,
|
|
76
|
+
704, 710, 715, 720, 727, 731, 736, 741, 746, 753, 755, 761, 767, 770, 773, 775, 781, 789, 792, 796, 802, 807, 812,
|
|
77
|
+
817, 821, 826, 830, 833, 836, 843, 848, 853, 857, 862, 867, 868, 870, 873, 877, 879, 883, 885, 888, 893, 897, 904,
|
|
78
|
+
907, 911, 917, 921, 926, 929, 932, 935, 939, 940, 944, 948, 954, 959, 962, 966, 968, 972, 974, 976, 980, 986, 990,
|
|
79
|
+
993, 1000, 1005, 1008, 1014, 1017, 1020, 1022, 1025, 1026, 1030, 1037, 1043, 1044, 1048, 1050, 1055, 1059, 1061, 1064,
|
|
80
|
+
1069, 1070, 1072,
|
|
81
|
+
};
|
|
82
|
+
|
|
83
|
+
static const uint32_t keep_ranges[678] = {
|
|
84
|
+
48, 57, 65, 90, 97, 122, 383, 383, 768, 879, 921, 921, 953, 953, 1037, 1037, 1040, 1103, 1117, 1117, 1155, 1161, 1425,
|
|
85
|
+
1469, 1471, 1471, 1473, 1474, 1476, 1477, 1479, 1479, 1552, 1562, 1611, 1631, 1648, 1648, 1750, 1756, 1759, 1764,
|
|
86
|
+
1767, 1768, 1770, 1773, 1809, 1809, 1840, 1866, 1958, 1968, 2027, 2035, 2045, 2045, 2070, 2073, 2075, 2083, 2085,
|
|
87
|
+
2087, 2089, 2093, 2137, 2139, 2199, 2207, 2250, 2273, 2275, 2307, 2362, 2364, 2366, 2383, 2385, 2391, 2402, 2403,
|
|
88
|
+
2433, 2435, 2492, 2492, 2494, 2500, 2503, 2504, 2507, 2509, 2519, 2519, 2530, 2531, 2558, 2558, 2561, 2563, 2620,
|
|
89
|
+
2620, 2622, 2626, 2631, 2632, 2635, 2637, 2641, 2641, 2672, 2673, 2677, 2677, 2689, 2691, 2748, 2748, 2750, 2757,
|
|
90
|
+
2759, 2761, 2763, 2765, 2786, 2787, 2810, 2815, 2817, 2819, 2876, 2876, 2878, 2884, 2887, 2888, 2891, 2893, 2901,
|
|
91
|
+
2903, 2914, 2915, 2946, 2946, 3006, 3010, 3014, 3016, 3018, 3021, 3031, 3031, 3072, 3076, 3132, 3132, 3134, 3140,
|
|
92
|
+
3142, 3144, 3146, 3149, 3157, 3158, 3170, 3171, 3201, 3203, 3260, 3260, 3262, 3268, 3270, 3272, 3274, 3277, 3285,
|
|
93
|
+
3286, 3298, 3299, 3315, 3315, 3328, 3331, 3387, 3388, 3390, 3396, 3398, 3400, 3402, 3405, 3415, 3415, 3426, 3427,
|
|
94
|
+
3457, 3459, 3530, 3530, 3535, 3540, 3542, 3542, 3544, 3551, 3570, 3571, 3633, 3633, 3636, 3642, 3655, 3662, 3761,
|
|
95
|
+
3761, 3764, 3772, 3784, 3790, 3864, 3865, 3893, 3893, 3895, 3895, 3897, 3897, 3902, 3903, 3953, 3972, 3974, 3975,
|
|
96
|
+
3981, 3991, 3993, 4028, 4038, 4038, 4139, 4158, 4182, 4185, 4190, 4192, 4194, 4196, 4199, 4205, 4209, 4212, 4226,
|
|
97
|
+
4237, 4239, 4239, 4250, 4253, 4957, 4959, 5906, 5909, 5938, 5940, 5970, 5971, 6002, 6003, 6068, 6099, 6109, 6109,
|
|
98
|
+
6155, 6157, 6159, 6159, 6277, 6278, 6313, 6313, 6432, 6443, 6448, 6459, 6679, 6683, 6741, 6750, 6752, 6780, 6783,
|
|
99
|
+
6783, 6832, 6877, 6880, 6891, 6912, 6916, 6964, 6980, 7019, 7027, 7040, 7042, 7073, 7085, 7142, 7155, 7204, 7223,
|
|
100
|
+
7296, 7302, 7376, 7378, 7380, 7400, 7405, 7405, 7412, 7412, 7415, 7417, 7616, 7679, 8126, 8126, 8400, 8432, 8490,
|
|
101
|
+
8490, 11503, 11505, 11647, 11647, 11744, 11775, 12330, 12335, 12441, 12442, 42607, 42610, 42612, 42621, 42654, 42655,
|
|
102
|
+
42736, 42737, 43010, 43010, 43014, 43014, 43019, 43019, 43043, 43047, 43052, 43052, 43136, 43137, 43188, 43205, 43232,
|
|
103
|
+
43249, 43263, 43263, 43302, 43309, 43335, 43347, 43392, 43395, 43443, 43456, 43493, 43493, 43561, 43574, 43587, 43587,
|
|
104
|
+
43596, 43597, 43643, 43645, 43696, 43696, 43698, 43700, 43703, 43704, 43710, 43711, 43713, 43713, 43755, 43759, 43765,
|
|
105
|
+
43766, 44003, 44010, 44012, 44013, 64286, 64286, 65024, 65039, 65056, 65071, 66045, 66045, 66272, 66272, 66422, 66426,
|
|
106
|
+
68097, 68099, 68101, 68102, 68108, 68111, 68152, 68154, 68159, 68159, 68325, 68326, 68900, 68903, 68969, 68973, 69291,
|
|
107
|
+
69292, 69370, 69375, 69446, 69456, 69506, 69509, 69632, 69634, 69688, 69702, 69744, 69744, 69747, 69748, 69759, 69762,
|
|
108
|
+
69808, 69818, 69826, 69826, 69888, 69890, 69927, 69940, 69957, 69958, 70003, 70003, 70016, 70018, 70067, 70080, 70089,
|
|
109
|
+
70092, 70094, 70095, 70188, 70199, 70206, 70206, 70209, 70209, 70367, 70378, 70400, 70403, 70459, 70460, 70462, 70468,
|
|
110
|
+
70471, 70472, 70475, 70477, 70487, 70487, 70498, 70499, 70502, 70508, 70512, 70516, 70584, 70592, 70594, 70594, 70597,
|
|
111
|
+
70597, 70599, 70602, 70604, 70608, 70610, 70610, 70625, 70626, 70709, 70726, 70750, 70750, 70832, 70851, 71087, 71093,
|
|
112
|
+
71096, 71104, 71132, 71133, 71216, 71232, 71339, 71351, 71453, 71467, 71724, 71738, 71984, 71989, 71991, 71992, 71995,
|
|
113
|
+
71998, 72000, 72000, 72002, 72003, 72145, 72151, 72154, 72160, 72164, 72164, 72193, 72202, 72243, 72249, 72251, 72254,
|
|
114
|
+
72263, 72263, 72273, 72283, 72330, 72345, 72544, 72551, 72751, 72758, 72760, 72767, 72850, 72871, 72873, 72886, 73009,
|
|
115
|
+
73014, 73018, 73018, 73020, 73021, 73023, 73029, 73031, 73031, 73098, 73102, 73104, 73105, 73107, 73111, 73459, 73462,
|
|
116
|
+
73472, 73473, 73475, 73475, 73524, 73530, 73534, 73538, 73562, 73562, 78912, 78912, 78919, 78933, 90398, 90415, 92912,
|
|
117
|
+
92916, 92976, 92982, 94031, 94031, 94033, 94087, 94095, 94098, 94180, 94180, 94192, 94193, 113821, 113822, 118528,
|
|
118
|
+
118573, 118576, 118598, 119141, 119145, 119149, 119154, 119163, 119170, 119173, 119179, 119210, 119213, 119362,
|
|
119
|
+
119364, 121344, 121398, 121403, 121452, 121461, 121461, 121476, 121476, 121499, 121503, 121505, 121519, 122880,
|
|
120
|
+
122886, 122888, 122904, 122907, 122913, 122915, 122916, 122918, 122922, 123023, 123023, 123184, 123190, 123566,
|
|
121
|
+
123566, 123628, 123631, 124140, 124143, 124398, 124399, 124643, 124643, 124646, 124646, 124654, 124655, 124661,
|
|
122
|
+
124661, 125136, 125142, 125252, 125258, 917760, 917999,
|
|
123
|
+
};
|
|
124
|
+
|
|
125
|
+
const narsil_language narsil_language_bulgarian = {
|
|
126
|
+
.name = "bulgarian",
|
|
127
|
+
.revision = "1",
|
|
128
|
+
.stop_word_units = stop_word_units,
|
|
129
|
+
.stop_word_offsets = stop_word_offsets,
|
|
130
|
+
.stop_word_count = 258,
|
|
131
|
+
.ascii_keep = {0x03FF000000000000ULL, 0x07FFFFFE07FFFFFEULL},
|
|
132
|
+
.keep_ranges = keep_ranges,
|
|
133
|
+
.keep_range_count = 339,
|
|
134
|
+
.splits_code_points = 1,
|
|
135
|
+
.strips_possessive = 0,
|
|
136
|
+
.normalises_diacritics = 0,
|
|
137
|
+
.min_token_length = 1,
|
|
138
|
+
.ngram_size = 0,
|
|
139
|
+
.normalise = narsil_bulgarian_normalise,
|
|
140
|
+
.stem_by_rules = narsil_bulgarian_stem,
|
|
141
|
+
.create_snowball = NULL,
|
|
142
|
+
.snowball_stem = NULL,
|
|
143
|
+
.close_snowball = NULL,
|
|
144
|
+
};
|
|
@@ -0,0 +1,48 @@
|
|
|
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[175] = {
|
|
8
|
+
4096, 4146, 4151, 4126, 4141, 4143, 4151, 4096, 4155, 4116, 4145, 4140, 4154, 4096, 4155, 4121, 4096, 4155, 4143,
|
|
9
|
+
4117, 4154, 4100, 4139, 4100, 4139, 4096, 4141, 4143, 4100, 4139, 4112, 4141, 4143, 4151, 4112, 4146, 4151, 4113,
|
|
10
|
+
4141, 4143, 4113, 4141, 4143, 4127, 4140, 4114, 4139, 4114, 4142, 4127, 4140, 4116, 4158, 4100, 4151, 4154, 4117,
|
|
11
|
+
4139, 4096, 4117, 4145, 4121, 4146, 4151, 4120, 4140, 4124, 4146, 4121, 4100, 4154, 4152, 4121, 4106, 4151, 4154,
|
|
12
|
+
4124, 4143, 4150, 4152, 4124, 4143, 4150, 4152, 4124, 4155, 4158, 4100, 4154, 4126, 4100, 4154, 4126, 4106, 4151,
|
|
13
|
+
4154, 4126, 4144, 4126, 4144, 4112, 4141, 4143, 4151, 4126, 4144, 4121, 4126, 4144, 4127, 4140, 4126, 4144, 4151,
|
|
14
|
+
4096, 4141, 4143, 4126, 4144, 4151, 4127, 4140, 4126, 4144, 4175, 4127, 4141, 4143, 4127, 4140, 4127, 4145, 4140,
|
|
15
|
+
4114, 4139, 4127, 4145, 4140, 4114, 4142, 4129, 4097, 4155, 4141, 4143, 4151, 4129, 4117, 4145, 4139, 4154, 4129,
|
|
16
|
+
4140, 4152, 4124, 4143, 4150, 4152, 4129, 4146, 4114, 4139, 4132, 4121, 4155, 4158, 4164, 4100, 4154, 4152, 4173,
|
|
17
|
+
4174, 4100, 4154, 4152,
|
|
18
|
+
};
|
|
19
|
+
|
|
20
|
+
static const uint32_t stop_word_offsets[41] = {
|
|
21
|
+
0, 7, 13, 16, 21, 23, 28, 34, 37, 40, 45, 47, 51, 56, 59, 64, 68, 72, 76, 84, 89, 92, 96, 98, 104, 107, 111, 117, 122,
|
|
22
|
+
125, 130, 135, 140, 146, 151, 158, 162, 166, 170, 171, 175,
|
|
23
|
+
};
|
|
24
|
+
|
|
25
|
+
static const uint32_t keep_ranges[14] = {
|
|
26
|
+
48, 57, 65, 90, 97, 122, 383, 383, 4096, 4169, 4172, 4255, 8490, 8490,
|
|
27
|
+
};
|
|
28
|
+
|
|
29
|
+
const narsil_language narsil_language_burmese = {
|
|
30
|
+
.name = "burmese",
|
|
31
|
+
.revision = "1",
|
|
32
|
+
.stop_word_units = stop_word_units,
|
|
33
|
+
.stop_word_offsets = stop_word_offsets,
|
|
34
|
+
.stop_word_count = 40,
|
|
35
|
+
.ascii_keep = {0x03FF000000000000ULL, 0x07FFFFFE07FFFFFEULL},
|
|
36
|
+
.keep_ranges = keep_ranges,
|
|
37
|
+
.keep_range_count = 7,
|
|
38
|
+
.splits_code_points = 1,
|
|
39
|
+
.strips_possessive = 0,
|
|
40
|
+
.normalises_diacritics = 0,
|
|
41
|
+
.min_token_length = 1,
|
|
42
|
+
.ngram_size = 2,
|
|
43
|
+
.normalise = NULL,
|
|
44
|
+
.stem_by_rules = NULL,
|
|
45
|
+
.create_snowball = NULL,
|
|
46
|
+
.snowball_stem = NULL,
|
|
47
|
+
.close_snowball = NULL,
|
|
48
|
+
};
|