@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 +1,46 @@
|
|
|
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[137] = {
|
|
8
|
+
3585, 3633, 3609, 3585, 3633, 3610, 3585, 3655, 3586, 3629, 3586, 3638, 3657, 3609, 3588, 3591, 3588, 3639, 3629,
|
|
9
|
+
3592, 3632, 3592, 3633, 3604, 3592, 3638, 3591, 3595, 3638, 3656, 3591, 3604, 3633, 3591, 3605, 3633, 3657, 3591,
|
|
10
|
+
3605, 3656, 3629, 3606, 3638, 3591, 3606, 3641, 3585, 3606, 3657, 3634, 3607, 3633, 3657, 3591, 3607, 3635, 3607,
|
|
11
|
+
3637, 3656, 3607, 3640, 3585, 3607, 3661, 3634, 3609, 3633, 3585, 3609, 3633, 3657, 3609, 3609, 3635, 3609, 3637,
|
|
12
|
+
3657, 3609, 3656, 3634, 3609, 3661, 3634, 3612, 3621, 3614, 3610, 3617, 3634, 3617, 3637, 3618, 3633, 3591, 3619,
|
|
13
|
+
3633, 3610, 3621, 3591, 3623, 3633, 3609, 3623, 3656, 3634, 3626, 3640, 3604, 3626, 3656, 3591, 3629, 3618, 3641,
|
|
14
|
+
3656, 3629, 3637, 3585, 3649, 3605, 3656, 3651, 3609, 3651, 3627, 3657, 3652, 3604, 3657, 3652, 3611, 3652, 3617,
|
|
15
|
+
3656, 3652, 3623, 3657,
|
|
16
|
+
};
|
|
17
|
+
|
|
18
|
+
static const uint32_t stop_word_offsets[49] = {
|
|
19
|
+
0, 3, 6, 8, 10, 14, 16, 19, 21, 24, 27, 31, 34, 38, 41, 44, 47, 50, 54, 56, 59, 62, 65, 68, 72, 74, 77, 80, 83, 85,
|
|
20
|
+
87, 89, 91, 94, 97, 99, 102, 105, 108, 111, 115, 118, 121, 123, 126, 129, 131, 134, 137,
|
|
21
|
+
};
|
|
22
|
+
|
|
23
|
+
static const uint32_t keep_ranges[16] = {
|
|
24
|
+
48, 57, 65, 90, 97, 122, 383, 383, 3585, 3646, 3648, 3662, 3664, 3673, 8490, 8490,
|
|
25
|
+
};
|
|
26
|
+
|
|
27
|
+
const narsil_language narsil_language_thai = {
|
|
28
|
+
.name = "thai",
|
|
29
|
+
.revision = "1",
|
|
30
|
+
.stop_word_units = stop_word_units,
|
|
31
|
+
.stop_word_offsets = stop_word_offsets,
|
|
32
|
+
.stop_word_count = 48,
|
|
33
|
+
.ascii_keep = {0x03FF000000000000ULL, 0x07FFFFFE07FFFFFEULL},
|
|
34
|
+
.keep_ranges = keep_ranges,
|
|
35
|
+
.keep_range_count = 8,
|
|
36
|
+
.splits_code_points = 1,
|
|
37
|
+
.strips_possessive = 0,
|
|
38
|
+
.normalises_diacritics = 0,
|
|
39
|
+
.min_token_length = 1,
|
|
40
|
+
.ngram_size = 2,
|
|
41
|
+
.normalise = NULL,
|
|
42
|
+
.stem_by_rules = NULL,
|
|
43
|
+
.create_snowball = NULL,
|
|
44
|
+
.snowball_stem = NULL,
|
|
45
|
+
.close_snowball = NULL,
|
|
46
|
+
};
|
|
@@ -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[253] = {
|
|
8
|
+
3904, 3904, 3954, 3904, 3956, 3923, 3904, 4017, 3908, 3904, 4017, 3954, 3904, 4017, 3954, 3923, 3904, 4017, 3954,
|
|
9
|
+
3942, 3905, 3906, 3906, 3928, 3906, 3934, 3923, 3906, 3954, 3906, 3954, 3908, 3906, 3954, 3923, 3906, 3954, 3942,
|
|
10
|
+
3906, 3962, 3906, 3964, 3906, 4017, 3954, 3906, 4017, 3954, 3923, 3906, 4017, 3954, 3942, 3906, 4018, 3908, 3908,
|
|
11
|
+
3928, 3908, 3962, 3908, 3964, 3909, 3908, 3909, 3954, 3906, 3909, 3954, 3908, 3909, 3964, 3906, 3913, 3954, 3921,
|
|
12
|
+
3919, 3928, 3919, 3956, 3919, 3962, 3919, 3964, 3921, 3906, 3921, 3908, 3921, 3928, 3921, 3956, 3921, 3962, 3921,
|
|
13
|
+
3964, 3923, 3923, 3908, 3923, 3928, 3923, 3954, 3923, 3964, 3924, 3942, 3924, 3964, 3925, 4017, 3954, 3926, 3909,
|
|
14
|
+
3942, 3926, 3928, 3926, 3938, 3926, 3942, 3926, 3964, 3928, 3928, 3928, 3928, 3954, 3928, 3954, 3923, 3928, 3964,
|
|
15
|
+
3928, 3966, 3929, 3928, 3930, 3930, 3956, 3930, 3964, 3934, 3954, 3906, 3934, 3954, 3908, 3936, 3906, 3936, 3906,
|
|
16
|
+
3936, 3936, 3908, 3936, 3921, 3954, 3936, 3925, 4018, 3939, 3936, 3926, 3936, 3936, 3928, 3936, 3954, 3936, 3964,
|
|
17
|
+
3937, 3937, 3908, 3937, 3954, 3937, 3954, 3923, 3937, 3954, 3942, 3937, 3964, 3908, 3942, 3938, 3938, 3908, 3938,
|
|
18
|
+
3926, 3938, 3928, 3938, 3954, 3939, 3938, 3956, 3938, 3956, 3908, 3938, 3962, 3938, 3964, 3938, 4003, 3928, 3942,
|
|
19
|
+
3939, 3939, 3906, 3942, 3939, 3928, 3939, 3942, 3939, 3964, 3940, 3942, 3940, 3954, 3906, 3940, 3954, 3908, 3942,
|
|
20
|
+
3942, 3928, 3942, 3956, 3942, 3964, 3942, 3964, 3906, 3942, 3942, 3993, 3962, 3921, 3942, 3999, 3962, 3942, 4004,
|
|
21
|
+
4017, 3954, 3944, 3908, 3944, 3962,
|
|
22
|
+
};
|
|
23
|
+
|
|
24
|
+
static const uint32_t stop_word_offsets[108] = {
|
|
25
|
+
0, 1, 3, 6, 9, 12, 16, 20, 21, 22, 24, 27, 29, 32, 35, 38, 40, 42, 45, 49, 53, 56, 58, 60, 62, 64, 67, 70, 73, 76, 78,
|
|
26
|
+
80, 82, 84, 86, 88, 90, 92, 94, 96, 97, 99, 101, 103, 105, 107, 109, 112, 115, 117, 119, 121, 123, 124, 126, 128, 131,
|
|
27
|
+
133, 135, 137, 138, 140, 142, 145, 148, 150, 153, 155, 158, 162, 165, 167, 169, 171, 172, 174, 176, 179, 182, 186,
|
|
28
|
+
187, 189, 191, 193, 196, 198, 201, 203, 205, 209, 210, 213, 215, 217, 219, 221, 224, 227, 228, 230, 232, 234, 238,
|
|
29
|
+
242, 245, 249, 251, 253,
|
|
30
|
+
};
|
|
31
|
+
|
|
32
|
+
static const uint32_t keep_ranges[10] = {
|
|
33
|
+
48, 57, 65, 90, 97, 122, 3872, 3891, 3904, 4028,
|
|
34
|
+
};
|
|
35
|
+
|
|
36
|
+
const narsil_language narsil_language_tibetan = {
|
|
37
|
+
.name = "tibetan",
|
|
38
|
+
.revision = "1",
|
|
39
|
+
.stop_word_units = stop_word_units,
|
|
40
|
+
.stop_word_offsets = stop_word_offsets,
|
|
41
|
+
.stop_word_count = 107,
|
|
42
|
+
.ascii_keep = {0x03FF000000000000ULL, 0x07FFFFFE07FFFFFEULL},
|
|
43
|
+
.keep_ranges = keep_ranges,
|
|
44
|
+
.keep_range_count = 5,
|
|
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 +1,73 @@
|
|
|
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[539] = {
|
|
8
|
+
4621, 4821, 4618, 4626, 4825, 4626, 4827, 4629, 4866, 4632, 4704, 4621, 4632, 4757, 4632, 4757, 4878, 4632, 4896,
|
|
9
|
+
4757, 4635, 4616, 4725, 4637, 4661, 4637, 4704, 4653, 4637, 4707, 4621, 4637, 4773, 4757, 4722, 4637, 4797, 4757,
|
|
10
|
+
4843, 4721, 4637, 4797, 4757, 4843, 4725, 4637, 4803, 4753, 4637, 4803, 4757, 4755, 4637, 4803, 4758, 4637, 4638,
|
|
11
|
+
4661, 4616, 4661, 4616, 4826, 4661, 4616, 4829, 4704, 4619, 4669, 4817, 4677, 4853, 4634, 4704, 4616, 4704, 4722,
|
|
12
|
+
4704, 4826, 4709, 4637, 4707, 4621, 4709, 4720, 4808, 4659, 4794, 4709, 4792, 4632, 4845, 4709, 4824, 4845, 4709,
|
|
13
|
+
4824, 4845, 4779, 4709, 4825, 4629, 4709, 4827, 4821, 4707, 4709, 4941, 4619, 4845, 4720, 4707, 4610, 4617, 4722,
|
|
14
|
+
4723, 4752, 4704, 4648, 4752, 4722, 4752, 4723, 4752, 4726, 4637, 4752, 4826, 4752, 4845, 4649, 4752, 4872, 4651,
|
|
15
|
+
4725, 4752, 4872, 4653, 4755, 4709, 4755, 4709, 4722, 4755, 4725, 4777, 4637, 4755, 4725, 4778, 4755, 4725, 4779,
|
|
16
|
+
4755, 4725, 4781, 4757, 4755, 4845, 4755, 4845, 4722, 4757, 4629, 4755, 4757, 4657, 4757, 4659, 4757, 4659, 4726,
|
|
17
|
+
4637, 4757, 4661, 4794, 4757, 4661, 4795, 4757, 4661, 4795, 4725, 4777, 4637, 4757, 4661, 4795, 4725, 4781, 4757,
|
|
18
|
+
4757, 4819, 4845, 4770, 4616, 4770, 4617, 4770, 4619, 4770, 4621, 4779, 4770, 4622, 4637, 4770, 4755, 4770, 4795,
|
|
19
|
+
4770, 4841, 4771, 4616, 4793, 4771, 4616, 4809, 4771, 4616, 4814, 4771, 4619, 4771, 4622, 4771, 4709, 4771, 4709,
|
|
20
|
+
4722, 4771, 4709, 4723, 4771, 4709, 4769, 4771, 4709, 4826, 4771, 4752, 4771, 4829, 4841, 4771, 4845, 4782, 4752,
|
|
21
|
+
4757, 4771, 4845, 4784, 4752, 4757, 4773, 4637, 4704, 4653, 4773, 4638, 4773, 4720, 4757, 4773, 4722, 4773, 4723,
|
|
22
|
+
4773, 4726, 4637, 4773, 4757, 4720, 4773, 4757, 4720, 4622, 4773, 4757, 4720, 4798, 4752, 4773, 4757, 4723, 4845,
|
|
23
|
+
4773, 4757, 4776, 4622, 4773, 4787, 4773, 4811, 4757, 4773, 4813, 4757, 4773, 4826, 4773, 4827, 4773, 4830, 4637,
|
|
24
|
+
4773, 4840, 4773, 4840, 4757, 4773, 4841, 4773, 4843, 4773, 4846, 4637, 4776, 4622, 4776, 4632, 4845, 4776, 4637,
|
|
25
|
+
4776, 4637, 4722, 4776, 4637, 4769, 4776, 4637, 4824, 4622, 4776, 4637, 4826, 4776, 4771, 4777, 4617, 4779, 4621,
|
|
26
|
+
4773, 4779, 4621, 4774, 4725, 4779, 4709, 4779, 4709, 4722, 4779, 4709, 4726, 4637, 4781, 4659, 4709, 4781, 4659,
|
|
27
|
+
4821, 4781, 4709, 4621, 4781, 4757, 4848, 4845, 4781, 4757, 4850, 4781, 4792, 4813, 4757, 4782, 4845, 4753, 4784,
|
|
28
|
+
4845, 4753, 4787, 4789, 4617, 4792, 4637, 4792, 4771, 4808, 4845, 4811, 4619, 4813, 4757, 4824, 4616, 4755, 4824,
|
|
29
|
+
4616, 4809, 4824, 4616, 4811, 4824, 4616, 4814, 4824, 4616, 4814, 4637, 4824, 4619, 4824, 4622, 4824, 4845, 4709,
|
|
30
|
+
4617, 4826, 4829, 4653, 4776, 4709, 4829, 4704, 4611, 4621, 4829, 4704, 4616, 4829, 4709, 4621, 4829, 4720, 4707,
|
|
31
|
+
4613, 4616, 4829, 4720, 4795, 4840, 4848, 4829, 4720, 4936, 4619, 4616, 4840, 4829, 4720, 4936, 4619, 4616, 4841,
|
|
32
|
+
4829, 4752, 4704, 4648, 4829, 4752, 4704, 4648, 4725, 4829, 4752, 4704, 4649, 4829, 4779, 4840, 4853, 4829, 4792,
|
|
33
|
+
4813, 4757, 4829, 4797, 4773, 4621, 4829, 4798, 4752, 4829, 4800, 4752, 4840, 4840, 4616, 4757, 4841, 4843, 4845,
|
|
34
|
+
4693, 4648, 4709, 4845, 4709, 4621, 4845, 4792, 4813, 4757, 4845, 4793, 4757, 4845, 4797, 4773, 4621, 4846, 4637,
|
|
35
|
+
4846, 4757, 4848, 4771, 4853, 4629, 4650, 4853, 4635, 4872, 4616, 4872, 4618, 4921, 4872, 4755, 4872, 4845, 4649,
|
|
36
|
+
4877, 4755, 4877, 4757, 4901, 4651, 4845,
|
|
37
|
+
};
|
|
38
|
+
|
|
39
|
+
static const uint32_t stop_word_offsets[183] = {
|
|
40
|
+
0, 3, 5, 7, 9, 12, 14, 17, 20, 23, 25, 28, 31, 35, 40, 45, 48, 52, 56, 57, 59, 62, 67, 69, 72, 74, 76, 78, 82, 87, 91,
|
|
41
|
+
94, 98, 101, 105, 109, 113, 114, 115, 118, 120, 122, 125, 127, 130, 134, 137, 139, 142, 146, 149, 152, 156, 158, 161,
|
|
42
|
+
164, 166, 168, 172, 175, 178, 184, 190, 193, 195, 197, 199, 202, 205, 207, 209, 211, 214, 217, 220, 222, 224, 226,
|
|
43
|
+
229, 232, 235, 238, 240, 243, 248, 253, 257, 259, 262, 264, 266, 269, 272, 276, 281, 285, 289, 291, 294, 297, 299,
|
|
44
|
+
301, 304, 306, 309, 311, 313, 316, 318, 321, 323, 326, 329, 333, 336, 338, 340, 343, 347, 349, 352, 356, 359, 362,
|
|
45
|
+
365, 369, 372, 376, 379, 382, 383, 385, 387, 389, 391, 393, 395, 398, 401, 404, 407, 411, 413, 415, 419, 420, 424,
|
|
46
|
+
428, 431, 434, 439, 444, 450, 456, 460, 465, 469, 473, 477, 481, 484, 487, 488, 491, 492, 493, 497, 500, 504, 507,
|
|
47
|
+
511, 513, 515, 517, 520, 522, 524, 527, 529, 532, 534, 536, 539,
|
|
48
|
+
};
|
|
49
|
+
|
|
50
|
+
static const uint32_t keep_ranges[10] = {
|
|
51
|
+
48, 57, 65, 90, 97, 122, 4608, 4959, 4969, 4988,
|
|
52
|
+
};
|
|
53
|
+
|
|
54
|
+
const narsil_language narsil_language_tigrinya = {
|
|
55
|
+
.name = "tigrinya",
|
|
56
|
+
.revision = "1",
|
|
57
|
+
.stop_word_units = stop_word_units,
|
|
58
|
+
.stop_word_offsets = stop_word_offsets,
|
|
59
|
+
.stop_word_count = 182,
|
|
60
|
+
.ascii_keep = {0x03FF000000000000ULL, 0x07FFFFFE07FFFFFEULL},
|
|
61
|
+
.keep_ranges = keep_ranges,
|
|
62
|
+
.keep_range_count = 5,
|
|
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 +1,47 @@
|
|
|
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[140] = {
|
|
8
|
+
39, 97, 39, 101, 39, 101, 110, 97, 39, 101, 110, 101, 39, 101, 110, 105, 39, 105, 39, 105, 107, 97, 105, 39, 111, 39,
|
|
9
|
+
111, 107, 117, 39, 111, 115, 105, 97, 105, 97, 110, 103, 101, 97, 116, 117, 97, 117, 101, 102, 111, 107, 105, 104, 97,
|
|
10
|
+
104, 97, 107, 101, 104, 101, 104, 105, 102, 111, 104, 111, 104, 111, 107, 117, 104, 111, 110, 111, 105, 97, 107, 97,
|
|
11
|
+
107, 97, 101, 107, 97, 112, 97, 117, 107, 101, 107, 105, 107, 111, 107, 111, 101, 107, 111, 101, 39, 117, 104, 105,
|
|
12
|
+
107, 117, 111, 107, 257, 116, 111, 97, 108, 111, 108, 111, 116, 111, 110, 103, 97, 109, 97, 105, 109, 101, 105, 109,
|
|
13
|
+
111, 110, 97, 39, 97, 110, 97, 39, 101, 110, 105, 112, 101, 112, 275, 116, 101,
|
|
14
|
+
};
|
|
15
|
+
|
|
16
|
+
static const uint32_t stop_word_offsets[45] = {
|
|
17
|
+
0, 2, 4, 8, 12, 16, 18, 23, 25, 29, 33, 35, 39, 42, 44, 45, 49, 51, 55, 57, 61, 63, 67, 71, 73, 75, 78, 83, 85, 87,
|
|
18
|
+
89, 92, 99, 102, 107, 116, 119, 122, 124, 128, 132, 134, 136, 138, 140,
|
|
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_tongan = {
|
|
29
|
+
.name = "tongan",
|
|
30
|
+
.revision = "1",
|
|
31
|
+
.stop_word_units = stop_word_units,
|
|
32
|
+
.stop_word_offsets = stop_word_offsets,
|
|
33
|
+
.stop_word_count = 44,
|
|
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 +1,100 @@
|
|
|
1
|
+
#include "../../analysis/languages.h"
|
|
2
|
+
#include "declarations.h"
|
|
3
|
+
#include "../snowball/turkish_stemmer.h"
|
|
4
|
+
|
|
5
|
+
#include <stddef.h>
|
|
6
|
+
#include <stdint.h>
|
|
7
|
+
|
|
8
|
+
static const uint16_t stop_word_units[1222] = {
|
|
9
|
+
97, 99, 97, 98, 97, 97, 99, 101, 112, 97, 100, 101, 116, 97, 97, 108, 116, 109, 305, 351, 97, 108, 116, 305, 97, 109,
|
|
10
|
+
97, 97, 110, 99, 97, 107, 97, 114, 97, 100, 97, 97, 115, 108, 305, 110, 100, 97, 97, 121, 110, 101, 110, 97, 121, 114,
|
|
11
|
+
305, 99, 97, 97, 122, 98, 97, 110, 97, 98, 97, 114, 105, 98, 97, 122, 101, 110, 98, 97, 122, 305, 98, 101, 108, 107,
|
|
12
|
+
105, 98, 101, 110, 98, 101, 110, 100, 101, 110, 98, 101, 110, 105, 98, 101, 110, 105, 109, 98, 101, 114, 105, 98, 101,
|
|
13
|
+
351, 98, 105, 108, 101, 98, 105, 110, 98, 105, 114, 98, 105, 114, 97, 122, 98, 105, 114, 105, 98, 105, 114, 107, 97,
|
|
14
|
+
231, 98, 105, 114, 107, 101, 122, 98, 105, 114, 231, 111, 107, 98, 105, 114, 351, 101, 121, 98, 105, 114, 351, 101,
|
|
15
|
+
121, 105, 98, 105, 122, 98, 105, 122, 100, 101, 110, 98, 105, 122, 101, 98, 105, 122, 105, 98, 105, 122, 105, 109, 98,
|
|
16
|
+
117, 98, 117, 110, 97, 98, 117, 110, 100, 97, 98, 117, 110, 100, 97, 110, 98, 117, 110, 108, 97, 114, 98, 117, 110,
|
|
17
|
+
108, 97, 114, 305, 98, 117, 110, 108, 97, 114, 305, 110, 98, 117, 110, 117, 98, 117, 110, 117, 110, 98, 117, 114, 97,
|
|
18
|
+
100, 97, 98, 246, 121, 108, 101, 98, 246, 121, 108, 101, 99, 101, 98, 252, 116, 252, 110, 100, 97, 100, 97, 104, 97,
|
|
19
|
+
100, 97, 104, 105, 100, 97, 104, 105, 108, 100, 97, 105, 109, 97, 100, 97, 105, 114, 100, 97, 121, 97, 110, 97, 114,
|
|
20
|
+
97, 107, 100, 101, 100, 101, 102, 97, 100, 101, 287, 105, 108, 100, 105, 121, 101, 100, 105, 287, 101, 114, 100, 111,
|
|
21
|
+
107, 115, 97, 110, 100, 111, 107, 117, 122, 100, 111, 108, 97, 121, 305, 100, 111, 108, 97, 121, 305, 115, 305, 121,
|
|
22
|
+
108, 97, 100, 246, 114, 116, 101, 100, 101, 99, 101, 107, 101, 100, 101, 110, 101, 100, 101, 114, 101, 107, 101, 100,
|
|
23
|
+
105, 108, 101, 99, 101, 107, 101, 100, 105, 108, 105, 121, 111, 114, 101, 100, 105, 108, 109, 101, 115, 105, 101, 100,
|
|
24
|
+
105, 121, 111, 114, 101, 108, 108, 105, 101, 110, 101, 116, 109, 101, 115, 105, 101, 116, 116, 105, 101, 116, 116,
|
|
25
|
+
105, 287, 105, 101, 116, 116, 105, 287, 105, 110, 105, 101, 287, 101, 114, 102, 97, 107, 97, 116, 103, 105, 98, 105,
|
|
26
|
+
103, 246, 114, 101, 104, 97, 108, 98, 117, 107, 105, 104, 97, 108, 101, 110, 104, 97, 110, 103, 105, 104, 97, 110,
|
|
27
|
+
105, 104, 97, 114, 105, 231, 104, 97, 116, 116, 97, 104, 101, 108, 101, 104, 101, 109, 104, 101, 110, 252, 122, 104,
|
|
28
|
+
101, 112, 104, 101, 112, 115, 105, 104, 101, 114, 104, 101, 114, 104, 97, 110, 103, 105, 104, 101, 114, 107, 101, 115,
|
|
29
|
+
104, 101, 114, 107, 101, 115, 105, 110, 104, 105, 231, 104, 105, 231, 98, 105, 114, 105, 107, 101, 110, 105, 107, 105,
|
|
30
|
+
105, 108, 97, 105, 108, 101, 105, 108, 103, 105, 108, 105, 105, 108, 107, 105, 108, 108, 97, 105, 115, 101, 105, 116,
|
|
31
|
+
105, 98, 97, 114, 101, 110, 105, 116, 105, 98, 97, 114, 105, 121, 108, 101, 105, 121, 105, 105, 121, 105, 99, 101,
|
|
32
|
+
105, 231, 105, 110, 105, 351, 116, 101, 107, 97, 100, 97, 114, 107, 97, 114, 351, 305, 110, 107, 97, 116, 114, 105,
|
|
33
|
+
108, 121, 111, 110, 107, 101, 110, 100, 105, 107, 101, 110, 100, 105, 108, 101, 114, 105, 110, 101, 107, 101, 110,
|
|
34
|
+
100, 105, 110, 105, 107, 101, 110, 100, 105, 115, 105, 107, 101, 110, 100, 105, 115, 105, 110, 101, 107, 101, 110,
|
|
35
|
+
100, 105, 115, 105, 110, 105, 107, 101, 114, 101, 107, 101, 122, 107, 105, 107, 105, 109, 107, 105, 109, 100, 101,
|
|
36
|
+
110, 107, 105, 109, 101, 107, 105, 109, 105, 107, 105, 109, 115, 101, 107, 305, 114, 107, 108, 97, 107, 105, 110, 109,
|
|
37
|
+
97, 100, 101, 109, 109, 105, 108, 121, 97, 114, 109, 105, 108, 121, 111, 110, 109, 117, 109, 252, 109, 305, 110, 97,
|
|
38
|
+
115, 305, 108, 110, 101, 110, 101, 100, 101, 110, 110, 101, 100, 101, 110, 108, 101, 110, 101, 114, 100, 101, 110,
|
|
39
|
+
101, 114, 101, 110, 101, 114, 101, 100, 101, 110, 101, 114, 101, 121, 101, 110, 105, 116, 101, 107, 105, 109, 110,
|
|
40
|
+
105, 121, 101, 110, 105, 231, 105, 110, 111, 111, 108, 97, 110, 111, 108, 97, 114, 97, 107, 111, 108, 100, 117, 111,
|
|
41
|
+
108, 100, 117, 107, 108, 97, 114, 305, 110, 305, 111, 108, 100, 117, 287, 117, 111, 108, 100, 117, 287, 117, 110, 117,
|
|
42
|
+
111, 108, 109, 97, 100, 305, 111, 108, 109, 97, 100, 305, 287, 305, 111, 108, 109, 97, 107, 111, 108, 109, 97, 115,
|
|
43
|
+
305, 111, 108, 109, 97, 121, 97, 110, 111, 108, 109, 97, 122, 111, 108, 115, 97, 111, 108, 115, 117, 110, 111, 108,
|
|
44
|
+
117, 112, 111, 108, 117, 114, 111, 108, 117, 114, 115, 97, 111, 108, 117, 121, 111, 114, 111, 110, 111, 110, 97, 111,
|
|
45
|
+
110, 100, 97, 110, 111, 110, 108, 97, 114, 111, 110, 108, 97, 114, 100, 97, 110, 111, 110, 108, 97, 114, 105, 111,
|
|
46
|
+
110, 108, 97, 114, 305, 111, 110, 108, 97, 114, 305, 110, 111, 110, 117, 111, 110, 117, 110, 111, 116, 117, 122, 111,
|
|
47
|
+
121, 115, 97, 112, 101, 107, 114, 97, 287, 109, 101, 110, 115, 97, 100, 101, 99, 101, 115, 97, 110, 107, 105, 115,
|
|
48
|
+
101, 107, 105, 122, 115, 101, 107, 115, 101, 110, 115, 101, 110, 115, 101, 110, 100, 101, 110, 115, 101, 110, 105,
|
|
49
|
+
115, 101, 110, 105, 110, 115, 105, 122, 115, 105, 122, 100, 101, 110, 115, 105, 122, 105, 115, 105, 122, 105, 110,
|
|
50
|
+
115, 111, 110, 114, 97, 116, 97, 114, 97, 102, 305, 110, 100, 97, 110, 116, 114, 105, 108, 121, 111, 110, 116, 252,
|
|
51
|
+
109, 118, 97, 114, 118, 97, 114, 100, 305, 118, 101, 118, 101, 121, 97, 118, 101, 121, 97, 104, 117, 116, 121, 97,
|
|
52
|
+
121, 97, 104, 117, 116, 121, 97, 110, 105, 121, 97, 112, 97, 99, 97, 107, 121, 97, 112, 109, 97, 107, 121, 97, 112,
|
|
53
|
+
116, 305, 121, 97, 112, 116, 305, 107, 108, 97, 114, 305, 121, 97, 112, 116, 305, 287, 305, 121, 97, 112, 116, 305,
|
|
54
|
+
287, 305, 110, 305, 121, 97, 112, 305, 108, 97, 110, 121, 97, 112, 305, 108, 109, 97, 115, 305, 121, 97, 112, 305,
|
|
55
|
+
121, 111, 114, 121, 101, 100, 105, 121, 101, 114, 105, 110, 101, 121, 101, 116, 109, 105, 351, 121, 105, 110, 101,
|
|
56
|
+
121, 105, 114, 109, 105, 121, 111, 107, 115, 97, 121, 252, 122, 122, 97, 116, 101, 110, 231, 111, 107, 231, 252, 110,
|
|
57
|
+
107, 252, 246, 121, 108, 101, 252, 122, 101, 114, 101, 252, 231, 351, 101, 121, 351, 101, 121, 100, 101, 110, 351,
|
|
58
|
+
101, 121, 105, 351, 101, 121, 108, 101, 114, 351, 117, 351, 117, 110, 97, 351, 117, 110, 100, 97, 351, 117, 110, 100,
|
|
59
|
+
97, 110, 351, 117, 110, 108, 97, 114, 305, 351, 117, 110, 117, 351, 246, 121, 108, 101,
|
|
60
|
+
};
|
|
61
|
+
|
|
62
|
+
static const uint32_t stop_word_offsets[243] = {
|
|
63
|
+
0, 5, 9, 14, 20, 24, 27, 32, 37, 44, 49, 55, 57, 61, 65, 70, 74, 79, 82, 88, 92, 97, 101, 104, 108, 111, 114, 119,
|
|
64
|
+
123, 129, 135, 141, 147, 154, 157, 163, 167, 171, 176, 178, 182, 187, 193, 199, 206, 214, 218, 223, 229, 234, 241,
|
|
65
|
+
246, 248, 252, 256, 261, 266, 270, 279, 281, 285, 290, 294, 299, 305, 310, 316, 327, 331, 337, 341, 347, 355, 363,
|
|
66
|
+
371, 377, 381, 383, 389, 393, 399, 407, 411, 416, 420, 424, 431, 436, 441, 445, 450, 455, 459, 462, 467, 470, 475,
|
|
67
|
+
478, 486, 492, 500, 503, 509, 513, 516, 519, 522, 528, 531, 535, 538, 546, 556, 559, 564, 568, 572, 577, 583, 592,
|
|
68
|
+
597, 608, 615, 622, 631, 640, 644, 647, 649, 652, 658, 662, 666, 671, 675, 680, 685, 691, 697, 699, 701, 703, 708,
|
|
69
|
+
710, 715, 722, 727, 731, 737, 743, 750, 754, 759, 760, 764, 770, 774, 785, 791, 799, 805, 813, 818, 824, 831, 836,
|
|
70
|
+
840, 845, 849, 853, 859, 865, 867, 870, 875, 880, 888, 894, 900, 907, 910, 914, 918, 922, 925, 931, 937, 942, 947,
|
|
71
|
+
953, 956, 962, 966, 971, 974, 980, 984, 989, 994, 1004, 1011, 1014, 1017, 1022, 1024, 1028, 1035, 1037, 1042, 1046,
|
|
72
|
+
1053, 1059, 1064, 1074, 1081, 1090, 1097, 1106, 1113, 1117, 1123, 1129, 1133, 1138, 1143, 1146, 1151, 1154, 1159,
|
|
73
|
+
1163, 1168, 1170, 1173, 1179, 1183, 1189, 1191, 1195, 1200, 1206, 1213, 1217, 1222,
|
|
74
|
+
};
|
|
75
|
+
|
|
76
|
+
static const uint32_t keep_ranges[32] = {
|
|
77
|
+
48, 57, 65, 90, 97, 122, 194, 194, 199, 199, 206, 206, 214, 214, 220, 220, 226, 226, 231, 231, 238, 238, 246, 246,
|
|
78
|
+
252, 252, 286, 287, 305, 305, 350, 351,
|
|
79
|
+
};
|
|
80
|
+
|
|
81
|
+
const narsil_language narsil_language_turkish = {
|
|
82
|
+
.name = "turkish",
|
|
83
|
+
.revision = "c537dbdae6350a3c",
|
|
84
|
+
.stop_word_units = stop_word_units,
|
|
85
|
+
.stop_word_offsets = stop_word_offsets,
|
|
86
|
+
.stop_word_count = 242,
|
|
87
|
+
.ascii_keep = {0x03FF000000000000ULL, 0x07FFFFFE07FFFFFEULL},
|
|
88
|
+
.keep_ranges = keep_ranges,
|
|
89
|
+
.keep_range_count = 16,
|
|
90
|
+
.splits_code_points = 0,
|
|
91
|
+
.strips_possessive = 0,
|
|
92
|
+
.normalises_diacritics = 0,
|
|
93
|
+
.min_token_length = 1,
|
|
94
|
+
.ngram_size = 0,
|
|
95
|
+
.normalise = NULL,
|
|
96
|
+
.stem_by_rules = NULL,
|
|
97
|
+
.create_snowball = turkish_create_env,
|
|
98
|
+
.snowball_stem = turkish_stem,
|
|
99
|
+
.close_snowball = turkish_close_env,
|
|
100
|
+
};
|
|
@@ -0,0 +1,88 @@
|
|
|
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[216] = {
|
|
9
|
+
97, 97, 107, 121, 105, 114, 101, 97, 110, 97, 97, 97, 110, 107, 97, 115, 97, 97, 110, 115, 97, 97, 114, 97, 97, 115,
|
|
10
|
+
101, 97, 115, 603, 98, 97, 98, 101, 114, 101, 98, 105, 98, 603, 100, 97, 100, 101, 101, 104, 111, 101, 109, 117, 101,
|
|
11
|
+
110, 111, 101, 110, 116, 105, 101, 115, 111, 101, 121, 105, 102, 97, 102, 105, 103, 121, 101, 104, 111, 104, 119, 101,
|
|
12
|
+
104, 119, 101, 101, 107, 97, 107, 97, 107, 114, 97, 107, 111, 114, 596, 107, 121, 101, 114, 603, 107, 121, 603, 110,
|
|
13
|
+
107, 596, 109, 97, 109, 101, 109, 109, 111, 109, 109, 111, 109, 112, 111, 109, 117, 110, 97, 110, 97, 110, 115, 111,
|
|
14
|
+
110, 101, 110, 107, 111, 114, 97, 97, 110, 110, 121, 101, 110, 111, 110, 115, 111, 110, 116, 105, 110, 121, 101, 110,
|
|
15
|
+
121, 105, 110, 97, 97, 111, 98, 105, 111, 98, 105, 97, 114, 97, 111, 111, 112, 97, 97, 114, 101, 115, 97, 110, 115,
|
|
16
|
+
101, 115, 111, 115, 603, 115, 603, 100, 101, 603, 116, 101, 119, 101, 105, 119, 111, 119, 596, 110, 121, 105, 121,
|
|
17
|
+
603, 121, 603, 110, 596, 110, 111, 603, 102, 105, 114, 105, 603, 110, 111, 603, 115, 105, 97, 110, 101,
|
|
18
|
+
};
|
|
19
|
+
|
|
20
|
+
static const uint32_t stop_word_offsets[70] = {
|
|
21
|
+
0, 1, 7, 11, 17, 21, 24, 27, 30, 32, 36, 38, 40, 42, 44, 47, 50, 53, 57, 60, 63, 65, 67, 70, 72, 75, 79, 81, 86, 90,
|
|
22
|
+
95, 99, 101, 103, 105, 109, 111, 114, 116, 118, 123, 125, 131, 135, 137, 140, 143, 146, 152, 155, 161, 163, 166, 168,
|
|
23
|
+
171, 173, 175, 177, 182, 184, 187, 189, 192, 194, 196, 199, 202, 207, 210, 216,
|
|
24
|
+
};
|
|
25
|
+
|
|
26
|
+
static const uint32_t keep_ranges[696] = {
|
|
27
|
+
39, 39, 48, 57, 65, 90, 97, 122, 330, 331, 383, 383, 390, 390, 400, 400, 596, 596, 603, 603, 768, 879, 891, 891, 917,
|
|
28
|
+
917, 921, 921, 949, 949, 953, 953, 1013, 1013, 1021, 1021, 1155, 1161, 1425, 1469, 1471, 1471, 1473, 1474, 1476, 1477,
|
|
29
|
+
1479, 1479, 1499, 1499, 1552, 1562, 1611, 1631, 1648, 1648, 1750, 1756, 1759, 1764, 1767, 1768, 1770, 1773, 1809,
|
|
30
|
+
1809, 1840, 1866, 1958, 1968, 2027, 2035, 2045, 2045, 2070, 2073, 2075, 2083, 2085, 2087, 2089, 2093, 2137, 2139,
|
|
31
|
+
2199, 2207, 2250, 2273, 2275, 2307, 2362, 2364, 2366, 2383, 2385, 2391, 2402, 2403, 2433, 2435, 2492, 2492, 2494,
|
|
32
|
+
2500, 2503, 2504, 2507, 2509, 2519, 2519, 2530, 2531, 2558, 2558, 2561, 2563, 2620, 2620, 2622, 2626, 2631, 2632,
|
|
33
|
+
2635, 2637, 2641, 2641, 2672, 2673, 2677, 2677, 2689, 2691, 2748, 2748, 2750, 2757, 2759, 2761, 2763, 2765, 2786,
|
|
34
|
+
2787, 2810, 2815, 2817, 2819, 2876, 2876, 2878, 2884, 2887, 2888, 2891, 2893, 2901, 2903, 2914, 2915, 2946, 2946,
|
|
35
|
+
3006, 3010, 3014, 3016, 3018, 3021, 3031, 3031, 3072, 3076, 3132, 3132, 3134, 3140, 3142, 3144, 3146, 3149, 3157,
|
|
36
|
+
3158, 3170, 3171, 3201, 3203, 3260, 3260, 3262, 3268, 3270, 3272, 3274, 3277, 3285, 3286, 3298, 3299, 3315, 3315,
|
|
37
|
+
3328, 3331, 3387, 3388, 3390, 3396, 3398, 3400, 3402, 3405, 3415, 3415, 3426, 3427, 3457, 3459, 3530, 3530, 3535,
|
|
38
|
+
3540, 3542, 3542, 3544, 3551, 3570, 3571, 3633, 3633, 3636, 3642, 3655, 3662, 3761, 3761, 3764, 3772, 3784, 3790,
|
|
39
|
+
3864, 3865, 3893, 3893, 3895, 3895, 3897, 3897, 3902, 3903, 3953, 3972, 3974, 3975, 3981, 3991, 3993, 4028, 4038,
|
|
40
|
+
4038, 4139, 4158, 4182, 4185, 4190, 4192, 4194, 4196, 4199, 4205, 4209, 4212, 4226, 4237, 4239, 4239, 4250, 4253,
|
|
41
|
+
4957, 4959, 5906, 5909, 5938, 5940, 5970, 5971, 6002, 6003, 6068, 6099, 6109, 6109, 6155, 6157, 6159, 6159, 6277,
|
|
42
|
+
6278, 6313, 6313, 6432, 6443, 6448, 6459, 6679, 6683, 6741, 6750, 6752, 6780, 6783, 6783, 6832, 6877, 6880, 6891,
|
|
43
|
+
6912, 6916, 6964, 6980, 7019, 7027, 7040, 7042, 7073, 7085, 7142, 7155, 7204, 7223, 7376, 7378, 7380, 7400, 7405,
|
|
44
|
+
7405, 7412, 7412, 7415, 7417, 7616, 7679, 8126, 8126, 8400, 8432, 8490, 8490, 8579, 8580, 11503, 11505, 11647, 11647,
|
|
45
|
+
11744, 11775, 12330, 12335, 12441, 12442, 42607, 42610, 42612, 42621, 42654, 42655, 42736, 42737, 43010, 43010, 43014,
|
|
46
|
+
43014, 43019, 43019, 43043, 43047, 43052, 43052, 43136, 43137, 43188, 43205, 43232, 43249, 43263, 43263, 43302, 43309,
|
|
47
|
+
43335, 43347, 43392, 43395, 43443, 43456, 43493, 43493, 43561, 43574, 43587, 43587, 43596, 43597, 43643, 43645, 43696,
|
|
48
|
+
43696, 43698, 43700, 43703, 43704, 43710, 43711, 43713, 43713, 43755, 43759, 43765, 43766, 44003, 44010, 44012, 44013,
|
|
49
|
+
64286, 64286, 65024, 65039, 65056, 65071, 66045, 66045, 66272, 66272, 66422, 66426, 68097, 68099, 68101, 68102, 68108,
|
|
50
|
+
68111, 68152, 68154, 68159, 68159, 68325, 68326, 68900, 68903, 68969, 68973, 69291, 69292, 69370, 69375, 69446, 69456,
|
|
51
|
+
69506, 69509, 69632, 69634, 69688, 69702, 69744, 69744, 69747, 69748, 69759, 69762, 69808, 69818, 69826, 69826, 69888,
|
|
52
|
+
69890, 69927, 69940, 69957, 69958, 70003, 70003, 70016, 70018, 70067, 70080, 70089, 70092, 70094, 70095, 70188, 70199,
|
|
53
|
+
70206, 70206, 70209, 70209, 70367, 70378, 70400, 70403, 70459, 70460, 70462, 70468, 70471, 70472, 70475, 70477, 70487,
|
|
54
|
+
70487, 70498, 70499, 70502, 70508, 70512, 70516, 70584, 70592, 70594, 70594, 70597, 70597, 70599, 70602, 70604, 70608,
|
|
55
|
+
70610, 70610, 70625, 70626, 70709, 70726, 70750, 70750, 70832, 70851, 71087, 71093, 71096, 71104, 71132, 71133, 71216,
|
|
56
|
+
71232, 71339, 71351, 71453, 71467, 71724, 71738, 71984, 71989, 71991, 71992, 71995, 71998, 72000, 72000, 72002, 72003,
|
|
57
|
+
72145, 72151, 72154, 72160, 72164, 72164, 72193, 72202, 72243, 72249, 72251, 72254, 72263, 72263, 72273, 72283, 72330,
|
|
58
|
+
72345, 72544, 72551, 72751, 72758, 72760, 72767, 72850, 72871, 72873, 72886, 73009, 73014, 73018, 73018, 73020, 73021,
|
|
59
|
+
73023, 73029, 73031, 73031, 73098, 73102, 73104, 73105, 73107, 73111, 73459, 73462, 73472, 73473, 73475, 73475, 73524,
|
|
60
|
+
73530, 73534, 73538, 73562, 73562, 78912, 78912, 78919, 78933, 90398, 90415, 92912, 92916, 92976, 92982, 94031, 94031,
|
|
61
|
+
94033, 94087, 94095, 94098, 94180, 94180, 94192, 94193, 113821, 113822, 118528, 118573, 118576, 118598, 119141,
|
|
62
|
+
119145, 119149, 119154, 119163, 119170, 119173, 119179, 119210, 119213, 119362, 119364, 121344, 121398, 121403,
|
|
63
|
+
121452, 121461, 121461, 121476, 121476, 121499, 121503, 121505, 121519, 122880, 122886, 122888, 122904, 122907,
|
|
64
|
+
122913, 122915, 122916, 122918, 122922, 123023, 123023, 123184, 123190, 123566, 123566, 123628, 123631, 124140,
|
|
65
|
+
124143, 124398, 124399, 124643, 124643, 124646, 124646, 124654, 124655, 124661, 124661, 125136, 125142, 125252,
|
|
66
|
+
125258, 917760, 917999,
|
|
67
|
+
};
|
|
68
|
+
|
|
69
|
+
const narsil_language narsil_language_twi = {
|
|
70
|
+
.name = "twi",
|
|
71
|
+
.revision = "1",
|
|
72
|
+
.stop_word_units = stop_word_units,
|
|
73
|
+
.stop_word_offsets = stop_word_offsets,
|
|
74
|
+
.stop_word_count = 69,
|
|
75
|
+
.ascii_keep = {0x03FF008000000000ULL, 0x07FFFFFE07FFFFFEULL},
|
|
76
|
+
.keep_ranges = keep_ranges,
|
|
77
|
+
.keep_range_count = 348,
|
|
78
|
+
.splits_code_points = 1,
|
|
79
|
+
.strips_possessive = 0,
|
|
80
|
+
.normalises_diacritics = 0,
|
|
81
|
+
.min_token_length = 1,
|
|
82
|
+
.ngram_size = 0,
|
|
83
|
+
.normalise = narsil_twi_normalise,
|
|
84
|
+
.stem_by_rules = NULL,
|
|
85
|
+
.create_snowball = NULL,
|
|
86
|
+
.snowball_stem = NULL,
|
|
87
|
+
.close_snowball = NULL,
|
|
88
|
+
};
|
|
@@ -0,0 +1,99 @@
|
|
|
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[964] = {
|
|
9
|
+
1072, 1072, 1073, 1080, 1072, 1073, 1086, 1072, 1074, 1078, 1077, 1078, 1072, 1076, 1078, 1077, 1072, 1083, 1077,
|
|
10
|
+
1072, 1085, 1110, 1073, 1073, 1072, 1075, 1072, 1090, 1086, 1073, 1077, 1079, 1073, 1080, 1073, 1086, 1073, 1091,
|
|
11
|
+
1074, 1073, 1091, 1076, 1077, 1073, 1091, 1076, 1091, 1090, 1100, 1073, 1091, 1083, 1072, 1073, 1091, 1083, 1080,
|
|
12
|
+
1073, 1091, 1083, 1086, 1073, 1091, 1090, 1080, 1073, 1110, 1083, 1100, 1096, 1073, 1110, 1083, 1103, 1074, 1074,
|
|
13
|
+
1072, 1084, 1074, 1072, 1084, 1080, 1074, 1072, 1089, 1074, 1072, 1096, 1074, 1072, 1096, 1072, 1074, 1072, 1096,
|
|
14
|
+
1077, 1074, 1072, 1096, 1110, 1074, 1077, 1089, 1100, 1074, 1078, 1077, 1074, 1079, 1076, 1086, 1074, 1078, 1074,
|
|
15
|
+
1080, 1074, 1085, 1080, 1079, 1074, 1085, 1080, 1079, 1091, 1074, 1086, 1085, 1072, 1074, 1086, 1085, 1080, 1074,
|
|
16
|
+
1086, 1085, 1086, 1074, 1089, 1077, 1074, 1089, 1077, 1088, 1077, 1076, 1080, 1085, 1110, 1074, 1089, 1102, 1074,
|
|
17
|
+
1089, 1103, 1074, 1089, 1110, 1074, 1089, 1110, 1093, 1074, 1110, 1076, 1074, 1110, 1085, 1076, 1072, 1076, 1072,
|
|
18
|
+
1074, 1072, 1081, 1076, 1072, 1074, 1072, 1090, 1080, 1076, 1077, 1076, 1077, 1082, 1110, 1083, 1100, 1082, 1072,
|
|
19
|
+
1076, 1077, 1097, 1086, 1076, 1077, 1103, 1082, 1110, 1076, 1083, 1103, 1076, 1086, 1078, 1078, 1077, 1079, 1079,
|
|
20
|
+
1072, 1079, 1072, 1074, 1078, 1076, 1080, 1079, 1072, 1084, 1110, 1089, 1090, 1100, 1079, 1072, 1088, 1072, 1079,
|
|
21
|
+
1079, 1072, 1090, 1077, 1079, 1074, 1110, 1076, 1082, 1080, 1079, 1074, 1110, 1076, 1089, 1080, 1079, 1074, 1110,
|
|
22
|
+
1076, 1090, 1080, 1079, 1110, 1081, 1081, 1086, 1075, 1086, 1081, 1086, 1084, 1091, 1082, 1086, 1075, 1086, 1082,
|
|
23
|
+
1086, 1078, 1077, 1085, 1082, 1086, 1078, 1085, 1072, 1082, 1086, 1078, 1085, 1077, 1082, 1086, 1083, 1080, 1082,
|
|
24
|
+
1086, 1083, 1086, 1082, 1086, 1084, 1091, 1082, 1086, 1090, 1088, 1072, 1082, 1086, 1090, 1088, 1077, 1082, 1086,
|
|
25
|
+
1090, 1088, 1080, 1081, 1082, 1086, 1090, 1088, 1110, 1082, 1088, 1110, 1084, 1082, 1091, 1076, 1080, 1082, 1110,
|
|
26
|
+
1083, 1100, 1082, 1072, 1083, 1077, 1076, 1074, 1077, 1083, 1080, 1096, 1077, 1084, 1072, 1081, 1078, 1077, 1084,
|
|
27
|
+
1072, 1083, 1086, 1084, 1077, 1085, 1077, 1084, 1077, 1085, 1110, 1084, 1080, 1084, 1085, 1086, 1102, 1084, 1086,
|
|
28
|
+
1074, 1084, 1086, 1075, 1086, 1084, 1086, 1078, 1077, 1084, 1086, 1078, 1085, 1072, 1084, 1086, 1102, 1084, 1086,
|
|
29
|
+
1103, 1084, 1086, 1108, 1084, 1086, 1108, 1084, 1091, 1084, 1086, 1111, 1084, 1086, 1111, 1081, 1084, 1086, 1111,
|
|
30
|
+
1093, 1084, 1110, 1078, 1084, 1110, 1081, 1085, 1072, 1085, 1072, 1074, 1082, 1086, 1083, 1086, 1085, 1072, 1074,
|
|
31
|
+
1110, 1090, 1100, 1085, 1072, 1076, 1085, 1072, 1084, 1085, 1072, 1084, 1080, 1085, 1072, 1089, 1085, 1072, 1095,
|
|
32
|
+
1077, 1085, 1072, 1096, 1085, 1072, 1096, 1072, 1085, 1072, 1096, 1077, 1085, 1072, 1096, 1080, 1093, 1085, 1072,
|
|
33
|
+
1096, 1110, 1085, 1077, 1085, 1077, 1084, 1086, 1074, 1085, 1077, 1102, 1085, 1077, 1111, 1085, 1080, 1084, 1085,
|
|
34
|
+
1080, 1084, 1080, 1085, 1080, 1093, 1085, 1100, 1086, 1075, 1086, 1085, 1100, 1086, 1084, 1091, 1085, 1110, 1085,
|
|
35
|
+
1110, 1073, 1080, 1085, 1110, 1078, 1085, 1110, 1082, 1086, 1075, 1086, 1085, 1110, 1082, 1086, 1083, 1080, 1085,
|
|
36
|
+
1110, 1093, 1090, 1086, 1085, 1110, 1095, 1086, 1075, 1086, 1085, 1110, 1097, 1086, 1086, 1076, 1085, 1072, 1082,
|
|
37
|
+
1086, 1082, 1088, 1110, 1084, 1086, 1090, 1086, 1090, 1078, 1077, 1086, 1090, 1086, 1078, 1087, 1077, 1088, 1077,
|
|
38
|
+
1076, 1087, 1086, 1087, 1086, 1079, 1072, 1087, 1088, 1080, 1087, 1088, 1086, 1087, 1088, 1086, 1090, 1077, 1087,
|
|
39
|
+
1088, 1086, 1090, 1080, 1087, 1110, 1076, 1087, 1110, 1089, 1083, 1103, 1089, 1072, 1084, 1089, 1072, 1084, 1072,
|
|
40
|
+
1089, 1072, 1084, 1077, 1089, 1072, 1084, 1110, 1089, 1074, 1086, 1075, 1086, 1089, 1074, 1086, 1102, 1089, 1074,
|
|
41
|
+
1086, 1103, 1089, 1074, 1086, 1108, 1089, 1074, 1086, 1111, 1089, 1074, 1110, 1081, 1089, 1077, 1073, 1077, 1089,
|
|
42
|
+
1077, 1088, 1077, 1076, 1089, 1082, 1110, 1083, 1100, 1082, 1080, 1089, 1086, 1073, 1086, 1102, 1089, 1086, 1073,
|
|
43
|
+
1110, 1089, 1102, 1076, 1080, 1090, 1072, 1090, 1072, 1082, 1090, 1072, 1082, 1080, 1081, 1090, 1072, 1082, 1080,
|
|
44
|
+
1093, 1090, 1072, 1082, 1086, 1078, 1090, 1072, 1084, 1090, 1077, 1090, 1077, 1073, 1077, 1090, 1077, 1078, 1090,
|
|
45
|
+
1077, 1087, 1077, 1088, 1090, 1080, 1090, 1080, 1084, 1090, 1080, 1093, 1090, 1086, 1090, 1086, 1073, 1086, 1102,
|
|
46
|
+
1090, 1086, 1073, 1090, 1086, 1090, 1086, 1073, 1110, 1090, 1086, 1075, 1086, 1090, 1086, 1076, 1110, 1090, 1086,
|
|
47
|
+
1078, 1090, 1086, 1081, 1090, 1086, 1084, 1091, 1090, 1086, 1097, 1086, 1090, 1091, 1090, 1091, 1076, 1080, 1090,
|
|
48
|
+
1091, 1090, 1090, 1110, 1090, 1110, 1081, 1090, 1110, 1083, 1100, 1082, 1080, 1091, 1091, 1074, 1077, 1089, 1100,
|
|
49
|
+
1091, 1078, 1077, 1091, 1089, 1077, 1091, 1089, 1102, 1091, 1089, 1103, 1091, 1089, 1110, 1093, 1072, 1081, 1093,
|
|
50
|
+
1086, 1095, 1093, 1086, 1095, 1072, 1093, 1090, 1086, 1093, 1090, 1086, 1089, 1100, 1094, 1077, 1094, 1077, 1081,
|
|
51
|
+
1094, 1080, 1084, 1094, 1080, 1093, 1094, 1100, 1086, 1075, 1086, 1094, 1100, 1086, 1084, 1091, 1094, 1102, 1094,
|
|
52
|
+
1103, 1094, 1110, 1094, 1110, 1081, 1095, 1072, 1089, 1090, 1086, 1095, 1077, 1088, 1077, 1079, 1095, 1080, 1095,
|
|
53
|
+
1080, 1081, 1095, 1080, 1103, 1095, 1080, 1108, 1095, 1080, 1111, 1095, 1086, 1075, 1086, 1095, 1086, 1084, 1091,
|
|
54
|
+
1097, 1077, 1097, 1086, 1097, 1086, 1073, 1097, 1086, 1073, 1080, 1097, 1086, 1076, 1086, 1097, 1086, 1089, 1100,
|
|
55
|
+
1103, 1103, 1082, 1103, 1082, 1072, 1103, 1082, 1073, 1080, 1103, 1082, 1077, 1103, 1082, 1080, 1081, 1103, 1082,
|
|
56
|
+
1086, 1111, 1103, 1082, 1097, 1086, 1103, 1082, 1110, 1108, 1110, 1110, 1079, 1110, 1085, 1082, 1086, 1083, 1080,
|
|
57
|
+
1110, 1085, 1086, 1076, 1110, 1110, 1085, 1096, 1072, 1110, 1085, 1096, 1077, 1110, 1085, 1096, 1080, 1081, 1110,
|
|
58
|
+
1085, 1096, 1080, 1093, 1110, 1085, 1096, 1110, 1111, 1081, 1111, 1084, 1111, 1093, 1111, 1093, 1085, 1103, 1111,
|
|
59
|
+
1093, 1085, 1108, 1111, 1093, 1085, 1110, 1111, 1093, 1085, 1110, 1081, 1111, 1111,
|
|
60
|
+
};
|
|
61
|
+
|
|
62
|
+
static const uint32_t stop_word_offsets[259] = {
|
|
63
|
+
0, 1, 4, 7, 12, 16, 19, 22, 23, 29, 32, 34, 36, 39, 43, 49, 53, 57, 61, 65, 70, 74, 75, 78, 82, 85, 88, 92, 96, 100,
|
|
64
|
+
104, 107, 113, 115, 119, 124, 128, 132, 136, 139, 148, 151, 154, 157, 161, 164, 167, 169, 174, 180, 182, 190, 194,
|
|
65
|
+
199, 202, 204, 205, 207, 208, 210, 216, 223, 228, 232, 238, 244, 250, 252, 253, 257, 261, 265, 270, 275, 280, 284,
|
|
66
|
+
288, 292, 297, 302, 308, 313, 317, 321, 327, 332, 336, 341, 345, 349, 353, 355, 359, 362, 366, 370, 375, 378, 381,
|
|
67
|
+
384, 389, 392, 396, 400, 403, 406, 408, 415, 421, 424, 427, 431, 434, 438, 441, 445, 449, 454, 458, 460, 465, 468,
|
|
68
|
+
471, 474, 478, 481, 486, 491, 493, 497, 500, 506, 512, 517, 523, 527, 532, 537, 539, 543, 547, 552, 554, 558, 561,
|
|
69
|
+
564, 569, 574, 577, 582, 585, 589, 593, 597, 602, 606, 610, 614, 618, 622, 626, 631, 638, 643, 647, 651, 653, 656,
|
|
70
|
+
661, 666, 671, 674, 676, 680, 683, 688, 690, 693, 696, 698, 703, 708, 712, 716, 720, 723, 726, 730, 734, 736, 740,
|
|
71
|
+
743, 745, 748, 754, 755, 760, 763, 766, 769, 772, 775, 778, 781, 785, 788, 793, 795, 798, 801, 804, 809, 814, 816,
|
|
72
|
+
818, 820, 823, 828, 833, 835, 838, 841, 844, 847, 851, 855, 857, 859, 862, 866, 870, 874, 875, 877, 880, 884, 887,
|
|
73
|
+
891, 895, 899, 902, 903, 904, 906, 912, 917, 921, 925, 930, 935, 939, 941, 943, 945, 949, 953, 957, 962, 964,
|
|
74
|
+
};
|
|
75
|
+
|
|
76
|
+
static const uint32_t keep_ranges[22] = {
|
|
77
|
+
39, 39, 48, 57, 65, 90, 97, 122, 1028, 1028, 1030, 1031, 1040, 1103, 1108, 1108, 1110, 1111, 1168, 1169, 7296, 7302,
|
|
78
|
+
};
|
|
79
|
+
|
|
80
|
+
const narsil_language narsil_language_ukrainian = {
|
|
81
|
+
.name = "ukrainian",
|
|
82
|
+
.revision = "1",
|
|
83
|
+
.stop_word_units = stop_word_units,
|
|
84
|
+
.stop_word_offsets = stop_word_offsets,
|
|
85
|
+
.stop_word_count = 258,
|
|
86
|
+
.ascii_keep = {0x03FF008000000000ULL, 0x07FFFFFE07FFFFFEULL},
|
|
87
|
+
.keep_ranges = keep_ranges,
|
|
88
|
+
.keep_range_count = 11,
|
|
89
|
+
.splits_code_points = 0,
|
|
90
|
+
.strips_possessive = 0,
|
|
91
|
+
.normalises_diacritics = 0,
|
|
92
|
+
.min_token_length = 1,
|
|
93
|
+
.ngram_size = 0,
|
|
94
|
+
.normalise = NULL,
|
|
95
|
+
.stem_by_rules = narsil_ukrainian_stem,
|
|
96
|
+
.create_snowball = NULL,
|
|
97
|
+
.snowball_stem = NULL,
|
|
98
|
+
.close_snowball = NULL,
|
|
99
|
+
};
|