@delali/narsil 0.1.14 → 0.2.0
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 +54 -0
- package/README.md +38 -972
- package/dist/adapters/filesystem.mjs +1 -1
- package/dist/adapters/indexeddb.mjs +1 -1
- package/dist/{chunk-WISGW24C.mjs → chunk-AUMKFYTG.mjs} +1 -1
- package/dist/{chunk-HRABJLPB.mjs → chunk-BKAI2WXI.mjs} +1 -0
- package/dist/chunk-GHJWC2E5.mjs +9 -0
- package/dist/chunk-KH3YGMVT.mjs +15 -0
- package/dist/{chunk-ALB35PRQ.mjs → chunk-M2EDG55Y.mjs} +8 -0
- package/dist/{chunk-HMO6OMZ3.mjs → chunk-S6IAXOBA.mjs} +1073 -506
- package/dist/chunk-TGP5M5R7.mjs +880 -0
- package/dist/chunk-U2BKYEC7.mjs +254 -0
- package/dist/{chunk-MEW7HD75.mjs → chunk-XAKNSWIK.mjs} +2543 -1154
- package/dist/{config-CojEPOgS.d.ts → config-P27AtSn4.d.ts} +28 -7
- package/dist/direct-executor-2FV3MEDD.mjs +5 -0
- package/dist/distribution/coordinator/etcd.d.ts +2 -2
- package/dist/distribution/coordinator/etcd.mjs +2 -2
- package/dist/distribution/coordinator/in-memory.d.ts +2 -2
- package/dist/distribution.d.ts +10 -4
- package/dist/distribution.mjs +1875 -1772
- package/dist/embeddings/openai.mjs +1 -1
- package/dist/index.browser.mjs +4566 -1783
- package/dist/index.d.ts +14 -5
- package/dist/index.mjs +6 -5
- package/dist/{language-ONQ3iVDf.d.ts → language-CmJBHf1j.d.ts} +4 -1
- package/dist/languages/albanian.d.ts +5 -0
- package/dist/languages/albanian.mjs +103 -0
- package/dist/languages/amharic.d.ts +5 -0
- package/dist/languages/amharic.mjs +240 -0
- package/dist/languages/arabic.d.ts +1 -1
- package/dist/languages/arabic.mjs +939 -154
- package/dist/languages/armenian.d.ts +1 -1
- package/dist/languages/armenian.mjs +275 -243
- package/dist/languages/azerbaijani.d.ts +5 -0
- package/dist/languages/azerbaijani.mjs +152 -0
- package/dist/languages/bambara.d.ts +5 -0
- package/dist/languages/bambara.mjs +60 -0
- package/dist/languages/basque.d.ts +5 -0
- package/dist/languages/basque.mjs +726 -0
- package/dist/languages/belarusian.d.ts +5 -0
- package/dist/languages/belarusian.mjs +100 -0
- package/dist/languages/bengali.d.ts +5 -0
- package/dist/languages/bengali.mjs +128 -0
- package/dist/languages/bosnian.d.ts +5 -0
- package/dist/languages/bosnian.mjs +197 -0
- package/dist/languages/breton.d.ts +5 -0
- package/dist/languages/breton.mjs +96 -0
- package/dist/languages/bulgarian.d.ts +1 -1
- package/dist/languages/bulgarian.mjs +11 -2
- package/dist/languages/burmese.d.ts +5 -0
- package/dist/languages/burmese.mjs +57 -0
- package/dist/languages/catalan.d.ts +5 -0
- package/dist/languages/catalan.mjs +987 -0
- package/dist/languages/chinese.d.ts +1 -1
- package/dist/languages/chinese.mjs +4 -3
- package/dist/languages/croatian.d.ts +5 -0
- package/dist/languages/croatian.mjs +191 -0
- package/dist/languages/czech.d.ts +5 -0
- package/dist/languages/czech.mjs +628 -0
- package/dist/languages/dagbani.d.ts +1 -1
- package/dist/languages/dagbani.mjs +11 -2
- package/dist/languages/danish.d.ts +1 -1
- package/dist/languages/danish.mjs +215 -110
- package/dist/languages/dutch.d.ts +1 -1
- package/dist/languages/dutch.mjs +1248 -163
- package/dist/languages/english.d.ts +1 -1
- package/dist/languages/english.mjs +2 -1
- package/dist/languages/esperanto.d.ts +5 -0
- package/dist/languages/esperanto.mjs +689 -0
- package/dist/languages/estonian.d.ts +5 -0
- package/dist/languages/estonian.mjs +2160 -0
- package/dist/languages/ewe.d.ts +1 -1
- package/dist/languages/ewe.mjs +11 -2
- package/dist/languages/faroese.d.ts +5 -0
- package/dist/languages/faroese.mjs +93 -0
- package/dist/languages/fijian.d.ts +5 -0
- package/dist/languages/fijian.mjs +55 -0
- package/dist/languages/finnish.d.ts +1 -1
- package/dist/languages/finnish.mjs +528 -263
- package/dist/languages/french.d.ts +1 -1
- package/dist/languages/french.mjs +950 -437
- package/dist/languages/ga.d.ts +1 -1
- package/dist/languages/ga.mjs +2 -1
- package/dist/languages/galician.d.ts +5 -0
- package/dist/languages/galician.mjs +172 -0
- package/dist/languages/georgian.d.ts +5 -0
- package/dist/languages/georgian.mjs +91 -0
- package/dist/languages/german.d.ts +1 -1
- package/dist/languages/german.mjs +432 -162
- package/dist/languages/greek.d.ts +1 -1
- package/dist/languages/greek.mjs +1970 -281
- package/dist/languages/guarani.d.ts +5 -0
- package/dist/languages/guarani.mjs +132 -0
- package/dist/languages/gujarati.d.ts +5 -0
- package/dist/languages/gujarati.mjs +234 -0
- package/dist/languages/haitian-creole.d.ts +5 -0
- package/dist/languages/haitian-creole.mjs +106 -0
- package/dist/languages/hausa.d.ts +1 -1
- package/dist/languages/hausa.mjs +2 -1
- package/dist/languages/hawaiian.d.ts +5 -0
- package/dist/languages/hawaiian.mjs +69 -0
- package/dist/languages/hebrew.d.ts +5 -0
- package/dist/languages/hebrew.mjs +213 -0
- package/dist/languages/hindi.d.ts +1 -1
- package/dist/languages/hindi.mjs +175 -97
- package/dist/languages/hungarian.d.ts +1 -1
- package/dist/languages/hungarian.mjs +412 -326
- package/dist/languages/icelandic.d.ts +5 -0
- package/dist/languages/icelandic.mjs +161 -0
- package/dist/languages/igbo.d.ts +1 -1
- package/dist/languages/igbo.mjs +2 -1
- package/dist/languages/indonesian.d.ts +1 -1
- package/dist/languages/indonesian.mjs +329 -137
- package/dist/languages/irish.d.ts +1 -1
- package/dist/languages/irish.mjs +223 -139
- package/dist/languages/italian.d.ts +1 -1
- package/dist/languages/italian.mjs +703 -406
- package/dist/languages/japanese.d.ts +1 -1
- package/dist/languages/japanese.mjs +5 -108
- package/dist/languages/kannada.d.ts +5 -0
- package/dist/languages/kannada.mjs +94 -0
- package/dist/languages/kazakh.d.ts +5 -0
- package/dist/languages/kazakh.mjs +90 -0
- package/dist/languages/khmer.d.ts +5 -0
- package/dist/languages/khmer.mjs +181 -0
- package/dist/languages/kinyarwanda.d.ts +5 -0
- package/dist/languages/kinyarwanda.mjs +99 -0
- package/dist/languages/kirundi.d.ts +5 -0
- package/dist/languages/kirundi.mjs +92 -0
- package/dist/languages/korean.d.ts +5 -0
- package/dist/languages/korean.mjs +252 -0
- package/dist/languages/kurmanji.d.ts +5 -0
- package/dist/languages/kurmanji.mjs +50 -0
- package/dist/languages/kyrgyz.d.ts +5 -0
- package/dist/languages/kyrgyz.mjs +108 -0
- package/dist/languages/lao.d.ts +5 -0
- package/dist/languages/lao.mjs +63 -0
- package/dist/languages/latin.d.ts +5 -0
- package/dist/languages/latin.mjs +69 -0
- package/dist/languages/latvian.d.ts +5 -0
- package/dist/languages/latvian.mjs +173 -0
- package/dist/languages/lingala.d.ts +5 -0
- package/dist/languages/lingala.mjs +61 -0
- package/dist/languages/lithuanian.d.ts +5 -0
- package/dist/languages/lithuanian.mjs +571 -0
- package/dist/languages/luxembourgish.d.ts +5 -0
- package/dist/languages/luxembourgish.mjs +214 -0
- package/dist/languages/macedonian.d.ts +5 -0
- package/dist/languages/macedonian.mjs +808 -0
- package/dist/languages/malagasy.d.ts +5 -0
- package/dist/languages/malagasy.mjs +67 -0
- package/dist/languages/malay.d.ts +5 -0
- package/dist/languages/malay.mjs +482 -0
- package/dist/languages/malayalam.d.ts +5 -0
- package/dist/languages/malayalam.mjs +122 -0
- package/dist/languages/maltese.d.ts +5 -0
- package/dist/languages/maltese.mjs +139 -0
- package/dist/languages/maori.d.ts +5 -0
- package/dist/languages/maori.mjs +79 -0
- package/dist/languages/marathi.d.ts +5 -0
- package/dist/languages/marathi.mjs +111 -0
- package/dist/languages/nepali.d.ts +1 -1
- package/dist/languages/nepali.mjs +209 -99
- package/dist/languages/norwegian.d.ts +1 -1
- package/dist/languages/norwegian.mjs +218 -85
- package/dist/languages/oromo.d.ts +5 -0
- package/dist/languages/oromo.mjs +78 -0
- package/dist/languages/persian.d.ts +5 -0
- package/dist/languages/persian.mjs +710 -0
- package/dist/languages/polish.d.ts +5 -0
- package/dist/languages/polish.mjs +614 -0
- package/dist/languages/portuguese.d.ts +1 -1
- package/dist/languages/portuguese.mjs +613 -380
- package/dist/languages/punjabi.d.ts +5 -0
- package/dist/languages/punjabi.mjs +197 -0
- package/dist/languages/romanian.d.ts +1 -1
- package/dist/languages/romanian.mjs +561 -382
- package/dist/languages/russian.d.ts +1 -1
- package/dist/languages/russian.mjs +373 -244
- package/dist/languages/samoan.d.ts +5 -0
- package/dist/languages/samoan.mjs +63 -0
- package/dist/languages/sanskrit.d.ts +1 -1
- package/dist/languages/sanskrit.mjs +2 -2
- package/dist/languages/scottish-gaelic.d.ts +5 -0
- package/dist/languages/scottish-gaelic.mjs +370 -0
- package/dist/languages/serbian.d.ts +1 -1
- package/dist/languages/serbian.mjs +3635 -155
- package/dist/languages/shona.d.ts +5 -0
- package/dist/languages/shona.mjs +91 -0
- package/dist/languages/sinhala.d.ts +5 -0
- package/dist/languages/sinhala.mjs +182 -0
- package/dist/languages/slovak.d.ts +5 -0
- package/dist/languages/slovak.mjs +430 -0
- package/dist/languages/slovenian.d.ts +1 -1
- package/dist/languages/slovenian.mjs +11 -2
- package/dist/languages/sorani.d.ts +5 -0
- package/dist/languages/sorani.mjs +156 -0
- package/dist/languages/spanish.d.ts +1 -1
- package/dist/languages/spanish.mjs +676 -374
- package/dist/languages/swahili.d.ts +1 -1
- package/dist/languages/swahili.mjs +2 -1
- package/dist/languages/swedish.d.ts +1 -1
- package/dist/languages/swedish.mjs +237 -95
- package/dist/languages/tagalog.d.ts +5 -0
- package/dist/languages/tagalog.mjs +159 -0
- package/dist/languages/tamil.d.ts +1 -1
- package/dist/languages/tamil.mjs +836 -64
- package/dist/languages/tatar.d.ts +5 -0
- package/dist/languages/tatar.mjs +1017 -0
- package/dist/languages/telugu.d.ts +5 -0
- package/dist/languages/telugu.mjs +62 -0
- package/dist/languages/thai.d.ts +5 -0
- package/dist/languages/thai.mjs +65 -0
- package/dist/languages/tibetan.d.ts +5 -0
- package/dist/languages/tibetan.mjs +119 -0
- package/dist/languages/tigrinya.d.ts +5 -0
- package/dist/languages/tigrinya.mjs +194 -0
- package/dist/languages/tongan.d.ts +5 -0
- package/dist/languages/tongan.mjs +56 -0
- package/dist/languages/turkish.d.ts +1 -1
- package/dist/languages/turkish.mjs +1246 -393
- package/dist/languages/twi.d.ts +1 -1
- package/dist/languages/twi.mjs +13 -2
- package/dist/languages/ukrainian.d.ts +1 -1
- package/dist/languages/ukrainian.mjs +125 -47
- package/dist/languages/urdu.d.ts +5 -0
- package/dist/languages/urdu.mjs +528 -0
- package/dist/languages/vietnamese.d.ts +5 -0
- package/dist/languages/vietnamese.mjs +277 -0
- package/dist/languages/welsh.d.ts +5 -0
- package/dist/languages/welsh.mjs +446 -0
- package/dist/languages/wolof.d.ts +5 -0
- package/dist/languages/wolof.mjs +79 -0
- package/dist/languages/xhosa.d.ts +5 -0
- package/dist/languages/xhosa.mjs +95 -0
- package/dist/languages/yoruba.d.ts +1 -1
- package/dist/languages/yoruba.mjs +1 -0
- package/dist/languages/zulu.d.ts +1 -1
- package/dist/languages/zulu.mjs +2 -1
- package/dist/narsil-CMZRPDlQ.d.ts +159 -0
- package/dist/persistence/durability/checkpoint-worker.d.ts +13 -0
- package/dist/persistence/durability/checkpoint-worker.mjs +6 -5
- package/dist/{schema-DCgkO1j4.d.ts → schema-BnvLvjRp.d.ts} +5 -3
- package/dist/server.d.ts +12 -3
- package/dist/server.mjs +183 -64
- package/dist/{types-DHX6zGeD.d.ts → types-BRTcBgAp.d.ts} +1 -1
- package/dist/workers/entry.mjs +2 -2
- package/package.json +10 -3
- package/dist/chunk-TW6BXTOL.mjs +0 -338
- package/dist/direct-executor-E52FFGDE.mjs +0 -4
- package/dist/narsil-D-84Nvm4.d.ts +0 -72
|
@@ -0,0 +1,132 @@
|
|
|
1
|
+
import { withNormalisedSpellings } from '../chunk-GHJWC2E5.mjs';
|
|
2
|
+
|
|
3
|
+
// src/languages/guarani.ts
|
|
4
|
+
var PUSO = "'";
|
|
5
|
+
var SALTILLO = "\uA78C";
|
|
6
|
+
var SALTILLO_PATTERN = /\u{a78c}/gu;
|
|
7
|
+
function bothPusoSpellings(words) {
|
|
8
|
+
return words.flatMap((word) => word.includes(PUSO) ? [word, word.split(PUSO).join(SALTILLO)] : [word]);
|
|
9
|
+
}
|
|
10
|
+
var stopWords = new Set(
|
|
11
|
+
bothPusoSpellings([
|
|
12
|
+
"aja",
|
|
13
|
+
"akue",
|
|
14
|
+
"ambue",
|
|
15
|
+
"ambu\xE9va",
|
|
16
|
+
"amo",
|
|
17
|
+
"ani",
|
|
18
|
+
"apyt\xE9pe",
|
|
19
|
+
"araka'e",
|
|
20
|
+
"\xE1ri",
|
|
21
|
+
"avave",
|
|
22
|
+
"ave",
|
|
23
|
+
"avei",
|
|
24
|
+
"che",
|
|
25
|
+
"ch\xE9ve",
|
|
26
|
+
"chugui",
|
|
27
|
+
"chupe",
|
|
28
|
+
"gotyo",
|
|
29
|
+
"gui",
|
|
30
|
+
"guive",
|
|
31
|
+
"gu\xFDpe",
|
|
32
|
+
"ha",
|
|
33
|
+
"ha'e",
|
|
34
|
+
"ha'eku\xE9ra",
|
|
35
|
+
"ha'\xE9va",
|
|
36
|
+
"hag\u0303ua",
|
|
37
|
+
"h\xE1katu",
|
|
38
|
+
"he\u1EBD",
|
|
39
|
+
"hendive",
|
|
40
|
+
"hese",
|
|
41
|
+
"heta",
|
|
42
|
+
"hiku\xE1i",
|
|
43
|
+
"h\xEDna",
|
|
44
|
+
"\xEDna",
|
|
45
|
+
"jare",
|
|
46
|
+
"jave",
|
|
47
|
+
"jepe",
|
|
48
|
+
"katu",
|
|
49
|
+
"ko",
|
|
50
|
+
"ko'\xE3",
|
|
51
|
+
"ko'\xE1g\u0303a",
|
|
52
|
+
"ko'\xE3ga",
|
|
53
|
+
"ko'\xE1pe",
|
|
54
|
+
"ko'\xE3va",
|
|
55
|
+
"k\xF3va",
|
|
56
|
+
"ku",
|
|
57
|
+
"ku\xE9ra",
|
|
58
|
+
"kuri",
|
|
59
|
+
"m\xE1va",
|
|
60
|
+
"m\xE1vapa",
|
|
61
|
+
"mayma",
|
|
62
|
+
"maym\xE1va",
|
|
63
|
+
"mba'e",
|
|
64
|
+
"mba'\xE9icha",
|
|
65
|
+
"mba'\xE9ichapa",
|
|
66
|
+
"mba'\xE9re",
|
|
67
|
+
"mbovy",
|
|
68
|
+
"mboy",
|
|
69
|
+
"mboyve",
|
|
70
|
+
"mbyt\xE9pe",
|
|
71
|
+
"me",
|
|
72
|
+
"mo\xF5",
|
|
73
|
+
"nah\xE1niri",
|
|
74
|
+
"\xF1ande",
|
|
75
|
+
"\xF1ane",
|
|
76
|
+
"ndaje",
|
|
77
|
+
"nde",
|
|
78
|
+
"nd\xE9ve",
|
|
79
|
+
"ndive",
|
|
80
|
+
"ne",
|
|
81
|
+
"ngu\xE9ra",
|
|
82
|
+
"niko",
|
|
83
|
+
"nko",
|
|
84
|
+
"o\xF1ondive",
|
|
85
|
+
"opa",
|
|
86
|
+
"opaite",
|
|
87
|
+
"ore",
|
|
88
|
+
"pe",
|
|
89
|
+
"pe\u1EBD",
|
|
90
|
+
"pegua",
|
|
91
|
+
"p\xE9icha",
|
|
92
|
+
"pende",
|
|
93
|
+
"pete\u0129",
|
|
94
|
+
"p\xE9va",
|
|
95
|
+
"peve",
|
|
96
|
+
"piko",
|
|
97
|
+
"ramo",
|
|
98
|
+
"rehe",
|
|
99
|
+
"rehegua",
|
|
100
|
+
"reta",
|
|
101
|
+
"reve",
|
|
102
|
+
"rire",
|
|
103
|
+
"rupi",
|
|
104
|
+
"rupive",
|
|
105
|
+
"t\xE9r\xE3",
|
|
106
|
+
"umi",
|
|
107
|
+
"um\xEDva",
|
|
108
|
+
"upe",
|
|
109
|
+
"up\xE9i",
|
|
110
|
+
"up\xE9icha",
|
|
111
|
+
"up\xE9pe",
|
|
112
|
+
"up\xE9r\xF5",
|
|
113
|
+
"up\xE9va",
|
|
114
|
+
"up\xE9vare",
|
|
115
|
+
"va'ekue",
|
|
116
|
+
"va'er\xE3",
|
|
117
|
+
"voi"
|
|
118
|
+
])
|
|
119
|
+
);
|
|
120
|
+
function normalize(token) {
|
|
121
|
+
return token.replace(SALTILLO_PATTERN, PUSO);
|
|
122
|
+
}
|
|
123
|
+
var guarani = {
|
|
124
|
+
name: "guarani",
|
|
125
|
+
revision: "1",
|
|
126
|
+
stemmer: null,
|
|
127
|
+
stopWords: withNormalisedSpellings(stopWords, normalize),
|
|
128
|
+
normalizer: normalize,
|
|
129
|
+
tokenizer: { splitPattern: /[^\p{Script=Latin}\p{M}0-9']+/giu }
|
|
130
|
+
};
|
|
131
|
+
|
|
132
|
+
export { guarani };
|
|
@@ -0,0 +1,234 @@
|
|
|
1
|
+
// src/languages/gujarati.ts
|
|
2
|
+
var stopWords = /* @__PURE__ */ new Set([
|
|
3
|
+
"\u0A85\u0A82\u0A97\u0AC7",
|
|
4
|
+
"\u0A85\u0AA5\u0AB5\u0ABE",
|
|
5
|
+
"\u0A85\u0A82\u0AA6\u0AB0",
|
|
6
|
+
"\u0A85\u0AA8\u0AC7",
|
|
7
|
+
"\u0A85\u0AAE\u0AA8\u0AC7",
|
|
8
|
+
"\u0A85\u0AAE\u0ABE\u0AB0\u0AC1\u0A82",
|
|
9
|
+
"\u0A85\u0AAE\u0AC7",
|
|
10
|
+
"\u0A85\u0AB9\u0AC0\u0A82",
|
|
11
|
+
"\u0A86",
|
|
12
|
+
"\u0A86\u0A97\u0AB3",
|
|
13
|
+
"\u0A86\u0AA5\u0AC0",
|
|
14
|
+
"\u0A86\u0AA8\u0AC1\u0A82",
|
|
15
|
+
"\u0A86\u0AA8\u0AC7",
|
|
16
|
+
"\u0A86\u0AAA\u0AA3\u0AA8\u0AC7",
|
|
17
|
+
"\u0A86\u0AAA\u0AA3\u0AC1\u0A82",
|
|
18
|
+
"\u0A86\u0AAA\u0AA3\u0AC7",
|
|
19
|
+
"\u0A86\u0AAA\u0AC0",
|
|
20
|
+
"\u0A86\u0AB0",
|
|
21
|
+
"\u0A86\u0AB5\u0AC0",
|
|
22
|
+
"\u0A86\u0AB5\u0AC7",
|
|
23
|
+
"\u0A89\u0AAA\u0AB0",
|
|
24
|
+
"\u0A89\u0AAD\u0ABE",
|
|
25
|
+
"\u0A8A\u0A82\u0A9A\u0AC7",
|
|
26
|
+
"\u0A8A\u0AAD\u0AC1\u0A82",
|
|
27
|
+
"\u0A8F",
|
|
28
|
+
"\u0A8F\u0A95",
|
|
29
|
+
"\u0A8F\u0AA8",
|
|
30
|
+
"\u0A8F\u0AA8\u0ABE",
|
|
31
|
+
"\u0A8F\u0AA8\u0ABE\u0A82",
|
|
32
|
+
"\u0A8F\u0AA8\u0AC0",
|
|
33
|
+
"\u0A8F\u0AA8\u0AC1\u0A82",
|
|
34
|
+
"\u0A8F\u0AA8\u0AC7",
|
|
35
|
+
"\u0A8F\u0AA8\u0ACB",
|
|
36
|
+
"\u0A8F\u0AAE",
|
|
37
|
+
"\u0A8F\u0AB5\u0ABE",
|
|
38
|
+
"\u0A8F\u0AB5\u0ABE\u0A82",
|
|
39
|
+
"\u0A8F\u0AB5\u0AC0",
|
|
40
|
+
"\u0A8F\u0AB5\u0AC1\u0A82",
|
|
41
|
+
"\u0A8F\u0AB5\u0ACB",
|
|
42
|
+
"\u0A93\u0A9B\u0AC1\u0A82",
|
|
43
|
+
"\u0A95\u0A88",
|
|
44
|
+
"\u0A95\u0A82\u0A88\u0A95",
|
|
45
|
+
"\u0A95\u0AAF\u0AC1\u0A82",
|
|
46
|
+
"\u0A95\u0AAF\u0ACB",
|
|
47
|
+
"\u0A95\u0AB0\u0AA4\u0ABE\u0A82",
|
|
48
|
+
"\u0A95\u0AB0\u0AB5\u0AC1\u0A82",
|
|
49
|
+
"\u0A95\u0AB0\u0AC0",
|
|
50
|
+
"\u0A95\u0AB0\u0AC0\u0A8F",
|
|
51
|
+
"\u0A95\u0AB0\u0AC1\u0A82",
|
|
52
|
+
"\u0A95\u0AB0\u0AC7",
|
|
53
|
+
"\u0A95\u0AB0\u0AC7\u0AB2\u0AC1\u0A82",
|
|
54
|
+
"\u0A95\u0AB0\u0ACD\u0AAF\u0ABE",
|
|
55
|
+
"\u0A95\u0AB0\u0ACD\u0AAF\u0ABE\u0A82",
|
|
56
|
+
"\u0A95\u0AB0\u0ACD\u0AAF\u0AC1\u0A82",
|
|
57
|
+
"\u0A95\u0AB0\u0ACD\u0AAF\u0ACB",
|
|
58
|
+
"\u0A95\u0ABE\u0A82\u0A88",
|
|
59
|
+
"\u0A95\u0AC7",
|
|
60
|
+
"\u0A95\u0AC7\u0A9F\u0AB2\u0AC1\u0A82",
|
|
61
|
+
"\u0A95\u0AC7\u0AAE",
|
|
62
|
+
"\u0A95\u0AC7\u0AB5\u0AC0",
|
|
63
|
+
"\u0A95\u0AC7\u0AB5\u0AC1\u0A82",
|
|
64
|
+
"\u0A95\u0ACB\u0A88",
|
|
65
|
+
"\u0A95\u0ACB\u0A88\u0A95",
|
|
66
|
+
"\u0A95\u0ACB\u0AA3",
|
|
67
|
+
"\u0A95\u0ACB\u0AA3\u0AC7",
|
|
68
|
+
"\u0A95\u0ACB\u0AA8\u0AC7",
|
|
69
|
+
"\u0A95\u0ACD\u0AAF\u0ABE\u0A82",
|
|
70
|
+
"\u0A95\u0ACD\u0AAF\u0ABE\u0AB0\u0AC7",
|
|
71
|
+
"\u0A96\u0AC2\u0AAC",
|
|
72
|
+
"\u0A97\u0A88",
|
|
73
|
+
"\u0A97\u0AAF\u0ABE",
|
|
74
|
+
"\u0A97\u0AAF\u0ABE\u0A82",
|
|
75
|
+
"\u0A97\u0AAF\u0AC1\u0A82",
|
|
76
|
+
"\u0A97\u0AAF\u0ACB",
|
|
77
|
+
"\u0A98\u0AA3\u0AC1\u0A82",
|
|
78
|
+
"\u0A9B",
|
|
79
|
+
"\u0A9B\u0AA4\u0ABE\u0A82",
|
|
80
|
+
"\u0A9B\u0AC0\u0A8F",
|
|
81
|
+
"\u0A9B\u0AC1\u0A82",
|
|
82
|
+
"\u0A9B\u0AC7",
|
|
83
|
+
"\u0A9B\u0AC7\u0A95",
|
|
84
|
+
"\u0A9B\u0ACB",
|
|
85
|
+
"\u0A9C",
|
|
86
|
+
"\u0A9C\u0ABE\u0AAF",
|
|
87
|
+
"\u0A9C\u0AC0",
|
|
88
|
+
"\u0A9C\u0AC7",
|
|
89
|
+
"\u0A9C\u0AC7\u0A9F\u0AB2\u0AC1\u0A82",
|
|
90
|
+
"\u0A9C\u0AC7\u0AA8\u0AC7",
|
|
91
|
+
"\u0A9C\u0AC7\u0AAE",
|
|
92
|
+
"\u0A9C\u0AC7\u0AB5\u0AC0",
|
|
93
|
+
"\u0A9C\u0AC7\u0AB5\u0AC1\u0A82",
|
|
94
|
+
"\u0A9C\u0AC7\u0AB5\u0ACB",
|
|
95
|
+
"\u0A9C\u0ACB",
|
|
96
|
+
"\u0A9C\u0ACB\u0A88\u0A8F",
|
|
97
|
+
"\u0A9C\u0ACD\u0AAF\u0ABE\u0A82",
|
|
98
|
+
"\u0A9C\u0ACD\u0AAF\u0ABE\u0AB0\u0AC7",
|
|
99
|
+
"\u0A9D\u0ABE\u0A9D\u0AC1\u0A82",
|
|
100
|
+
"\u0AA4\u0AA8\u0AC7",
|
|
101
|
+
"\u0AA4\u0AAE\u0AA8\u0AC7",
|
|
102
|
+
"\u0AA4\u0AAE\u0ABE\u0AB0\u0AC1\u0A82",
|
|
103
|
+
"\u0AA4\u0AAE\u0AC7",
|
|
104
|
+
"\u0AA4\u0ABE",
|
|
105
|
+
"\u0AA4\u0ABE\u0AB0\u0ABE\u0AA5\u0AC0",
|
|
106
|
+
"\u0AA4\u0ABE\u0AB0\u0ABE\u0AAE\u0ABE\u0A82",
|
|
107
|
+
"\u0AA4\u0ABE\u0AB0\u0AC1\u0A82",
|
|
108
|
+
"\u0AA4\u0AC1\u0A82",
|
|
109
|
+
"\u0AA4\u0AC7",
|
|
110
|
+
"\u0AA4\u0AC7\u0A82",
|
|
111
|
+
"\u0AA4\u0AC7\u0A93",
|
|
112
|
+
"\u0AA4\u0AC7\u0AA3\u0AC7",
|
|
113
|
+
"\u0AA4\u0AC7\u0AA5\u0AC0",
|
|
114
|
+
"\u0AA4\u0AC7\u0AA8\u0ABE",
|
|
115
|
+
"\u0AA4\u0AC7\u0AA8\u0AC0",
|
|
116
|
+
"\u0AA4\u0AC7\u0AA8\u0AC1\u0A82",
|
|
117
|
+
"\u0AA4\u0AC7\u0AA8\u0AC7",
|
|
118
|
+
"\u0AA4\u0AC7\u0AAE",
|
|
119
|
+
"\u0AA4\u0AC7\u0AAE\u0AA8\u0AC1\u0A82",
|
|
120
|
+
"\u0AA4\u0AC7\u0AAE\u0AA8\u0AC7",
|
|
121
|
+
"\u0AA4\u0AC7\u0AB5\u0AC0",
|
|
122
|
+
"\u0AA4\u0AC7\u0AB5\u0AC1\u0A82",
|
|
123
|
+
"\u0AA4\u0ACB",
|
|
124
|
+
"\u0AA4\u0ACD\u0AAF\u0ABE\u0A82",
|
|
125
|
+
"\u0AA4\u0ACD\u0AAF\u0ABE\u0AB0\u0AC7",
|
|
126
|
+
"\u0AA5\u0A87",
|
|
127
|
+
"\u0AA5\u0A88",
|
|
128
|
+
"\u0AA5\u0A88\u0A8F",
|
|
129
|
+
"\u0AA5\u0AA4\u0ABE",
|
|
130
|
+
"\u0AA5\u0AA4\u0ABE\u0A82",
|
|
131
|
+
"\u0AA5\u0AA4\u0AC0",
|
|
132
|
+
"\u0AA5\u0AA4\u0AC1\u0A82",
|
|
133
|
+
"\u0AA5\u0AA4\u0ACB",
|
|
134
|
+
"\u0AA5\u0AAF\u0ABE",
|
|
135
|
+
"\u0AA5\u0AAF\u0ABE\u0A82",
|
|
136
|
+
"\u0AA5\u0AAF\u0AC1\u0A82",
|
|
137
|
+
"\u0AA5\u0AAF\u0AC7\u0AB2\u0AC1\u0A82",
|
|
138
|
+
"\u0AA5\u0AAF\u0ACB",
|
|
139
|
+
"\u0AA5\u0AB5\u0AC1\u0A82",
|
|
140
|
+
"\u0AA5\u0ABE\u0A89\u0A82",
|
|
141
|
+
"\u0AA5\u0ABE\u0A93",
|
|
142
|
+
"\u0AA5\u0ABE\u0AAF",
|
|
143
|
+
"\u0AA5\u0AC0",
|
|
144
|
+
"\u0AA5\u0ACB\u0AA1\u0AC1\u0A82",
|
|
145
|
+
"\u0AA6\u0AB0\u0AC7\u0A95",
|
|
146
|
+
"\u0AA8",
|
|
147
|
+
"\u0AA8\u0A82",
|
|
148
|
+
"\u0AA8\u0AA5\u0AC0",
|
|
149
|
+
"\u0AA8\u0AB9\u0ABF",
|
|
150
|
+
"\u0AA8\u0AB9\u0AC0",
|
|
151
|
+
"\u0AA8\u0AB9\u0AC0\u0A82",
|
|
152
|
+
"\u0AA8\u0ABE",
|
|
153
|
+
"\u0AA8\u0AC0",
|
|
154
|
+
"\u0AA8\u0AC0\u0A9A\u0AC7",
|
|
155
|
+
"\u0AA8\u0AC1\u0A82",
|
|
156
|
+
"\u0AA8\u0AC7",
|
|
157
|
+
"\u0AA8\u0ACB",
|
|
158
|
+
"\u0AAA\u0A9B\u0AC0",
|
|
159
|
+
"\u0AAA\u0AA3",
|
|
160
|
+
"\u0AAA\u0AB0",
|
|
161
|
+
"\u0AAA\u0AB0\u0A82\u0AA4\u0AC1",
|
|
162
|
+
"\u0AAA\u0AB9\u0AC7\u0AB2\u0ABE\u0A82",
|
|
163
|
+
"\u0AAA\u0ABE\u0A9B\u0AB3",
|
|
164
|
+
"\u0AAA\u0ABE\u0AB8\u0AC7",
|
|
165
|
+
"\u0AAA\u0ACB\u0AA4\u0ABE\u0AA8\u0AC1\u0A82",
|
|
166
|
+
"\u0AAA\u0ACD\u0AB0\u0AA4\u0ACD\u0AAF\u0AC7\u0A95",
|
|
167
|
+
"\u0AAB\u0A95\u0ACD\u0AA4",
|
|
168
|
+
"\u0AAB\u0AB0\u0AC0",
|
|
169
|
+
"\u0AAB\u0AB0\u0AC0\u0AA5\u0AC0",
|
|
170
|
+
"\u0AAC\u0AA7\u0ABE",
|
|
171
|
+
"\u0AAC\u0AA7\u0AC1\u0A82",
|
|
172
|
+
"\u0AAC\u0AA8\u0AC0",
|
|
173
|
+
"\u0AAC\u0A82\u0AA8\u0AC7",
|
|
174
|
+
"\u0AAC\u0AB9\u0ABE\u0AB0",
|
|
175
|
+
"\u0AAC\u0AB9\u0AC1",
|
|
176
|
+
"\u0AAC\u0ABE\u0AA6",
|
|
177
|
+
"\u0AAC\u0AC7",
|
|
178
|
+
"\u0AAE\u0AA8\u0AC7",
|
|
179
|
+
"\u0AAE\u0ABE",
|
|
180
|
+
"\u0AAE\u0ABE\u0A82",
|
|
181
|
+
"\u0AAE\u0ABE\u0A9F\u0AC7",
|
|
182
|
+
"\u0AAE\u0ABE\u0AA4\u0ACD\u0AB0",
|
|
183
|
+
"\u0AAE\u0ABE\u0AB0\u0AC1\u0A82",
|
|
184
|
+
"\u0AAE\u0AC0",
|
|
185
|
+
"\u0AAE\u0AC2\u0A95\u0AB5\u0AC1\u0A82",
|
|
186
|
+
"\u0AAE\u0AC2\u0A95\u0AC0",
|
|
187
|
+
"\u0AAE\u0AC2\u0A95\u0ACD\u0AAF\u0ABE",
|
|
188
|
+
"\u0AAE\u0AC2\u0A95\u0ACD\u0AAF\u0ABE\u0A82",
|
|
189
|
+
"\u0AAE\u0AC2\u0A95\u0ACD\u0AAF\u0AC1\u0A82",
|
|
190
|
+
"\u0AAE\u0AC7\u0A82",
|
|
191
|
+
"\u0AB0\u0AB9\u0AC0",
|
|
192
|
+
"\u0AB0\u0AB9\u0AC7",
|
|
193
|
+
"\u0AB0\u0AB9\u0AC7\u0AB5\u0AC1\u0A82",
|
|
194
|
+
"\u0AB0\u0AB9\u0ACD\u0AAF\u0ABE",
|
|
195
|
+
"\u0AB0\u0AB9\u0ACD\u0AAF\u0ABE\u0A82",
|
|
196
|
+
"\u0AB0\u0AB9\u0ACD\u0AAF\u0ACB",
|
|
197
|
+
"\u0AB0\u0AC0\u0AA4\u0AC7",
|
|
198
|
+
"\u0AB0\u0AC2\u0ABE",
|
|
199
|
+
"\u0AB2\u0AC7\u0AA4\u0ABE",
|
|
200
|
+
"\u0AB2\u0AC7\u0AA4\u0AC1\u0A82",
|
|
201
|
+
"\u0AB2\u0AC7\u0AB5\u0ABE",
|
|
202
|
+
"\u0AB5\u0A97\u0AC7\u0AB0\u0AC7",
|
|
203
|
+
"\u0AB5\u0AA7\u0AC1",
|
|
204
|
+
"\u0AB6\u0A95\u0AC7",
|
|
205
|
+
"\u0AB6\u0ABE",
|
|
206
|
+
"\u0AB6\u0AC1\u0A82",
|
|
207
|
+
"\u0AB8\u0AB0\u0A96\u0AC1\u0A82",
|
|
208
|
+
"\u0AB8\u0ABE\u0AAE\u0AC7",
|
|
209
|
+
"\u0AB8\u0AC1\u0AA7\u0AC0",
|
|
210
|
+
"\u0AB9\u0AA4\u0ABE",
|
|
211
|
+
"\u0AB9\u0AA4\u0ABE\u0A82",
|
|
212
|
+
"\u0AB9\u0AA4\u0AC0",
|
|
213
|
+
"\u0AB9\u0AA4\u0AC1\u0A82",
|
|
214
|
+
"\u0AB9\u0AB5\u0AC7",
|
|
215
|
+
"\u0AB9\u0AB6\u0AC7",
|
|
216
|
+
"\u0AB9\u0AB6\u0ACB",
|
|
217
|
+
"\u0AB9\u0ABE",
|
|
218
|
+
"\u0AB9\u0AC1\u0A82",
|
|
219
|
+
"\u0AB9\u0ACB",
|
|
220
|
+
"\u0AB9\u0ACB\u0A88",
|
|
221
|
+
"\u0AB9\u0ACB\u0A88\u0AB6",
|
|
222
|
+
"\u0AB9\u0ACB\u0A88\u0AB6\u0AC1\u0A82",
|
|
223
|
+
"\u0AB9\u0ACB\u0AAF",
|
|
224
|
+
"\u0AB9\u0ACB\u0AB5\u0ABE"
|
|
225
|
+
]);
|
|
226
|
+
var gujarati = {
|
|
227
|
+
name: "gujarati",
|
|
228
|
+
revision: "1",
|
|
229
|
+
stemmer: null,
|
|
230
|
+
stopWords,
|
|
231
|
+
tokenizer: { splitPattern: /[^\u0a80-\u0affa-z0-9]+/gi }
|
|
232
|
+
};
|
|
233
|
+
|
|
234
|
+
export { gujarati };
|
|
@@ -0,0 +1,106 @@
|
|
|
1
|
+
// src/languages/haitian-creole.ts
|
|
2
|
+
var stopWords = /* @__PURE__ */ new Set([
|
|
3
|
+
"a",
|
|
4
|
+
"ak",
|
|
5
|
+
"an",
|
|
6
|
+
"ank\xF2",
|
|
7
|
+
"ant",
|
|
8
|
+
"ap",
|
|
9
|
+
"apre",
|
|
10
|
+
"at\xF2",
|
|
11
|
+
"avan",
|
|
12
|
+
"avanl\xE8",
|
|
13
|
+
"b\xF2",
|
|
14
|
+
"byen",
|
|
15
|
+
"byenke",
|
|
16
|
+
"chak",
|
|
17
|
+
"de",
|
|
18
|
+
"deja",
|
|
19
|
+
"depi",
|
|
20
|
+
"e",
|
|
21
|
+
"eh",
|
|
22
|
+
"en",
|
|
23
|
+
"epi",
|
|
24
|
+
"\xE8ske",
|
|
25
|
+
"f\xF2",
|
|
26
|
+
"f\xF2k",
|
|
27
|
+
"gen",
|
|
28
|
+
"genyen",
|
|
29
|
+
"ki",
|
|
30
|
+
"kil\xE8s",
|
|
31
|
+
"kisa",
|
|
32
|
+
"konbyen",
|
|
33
|
+
"konn",
|
|
34
|
+
"konnen",
|
|
35
|
+
"konsa",
|
|
36
|
+
"kote",
|
|
37
|
+
"koukou",
|
|
38
|
+
"kouman",
|
|
39
|
+
"kounye",
|
|
40
|
+
"l",
|
|
41
|
+
"la",
|
|
42
|
+
"laa",
|
|
43
|
+
"le",
|
|
44
|
+
"l\xE8",
|
|
45
|
+
"li",
|
|
46
|
+
"l\xF2",
|
|
47
|
+
"lye",
|
|
48
|
+
"m",
|
|
49
|
+
"men",
|
|
50
|
+
"m\xE8si",
|
|
51
|
+
"mwen",
|
|
52
|
+
"nan",
|
|
53
|
+
"nap",
|
|
54
|
+
"non",
|
|
55
|
+
"nou",
|
|
56
|
+
"o",
|
|
57
|
+
"oh",
|
|
58
|
+
"osinon",
|
|
59
|
+
"oswa",
|
|
60
|
+
"ou",
|
|
61
|
+
"oumenm",
|
|
62
|
+
"pa",
|
|
63
|
+
"pami",
|
|
64
|
+
"pandan",
|
|
65
|
+
"paske",
|
|
66
|
+
"pito",
|
|
67
|
+
"pou",
|
|
68
|
+
"pral",
|
|
69
|
+
"preske",
|
|
70
|
+
"pwiske",
|
|
71
|
+
"sa",
|
|
72
|
+
"san",
|
|
73
|
+
"se",
|
|
74
|
+
"selman",
|
|
75
|
+
"si",
|
|
76
|
+
"s\xF2t",
|
|
77
|
+
"sou",
|
|
78
|
+
"swa",
|
|
79
|
+
"ta",
|
|
80
|
+
"tan",
|
|
81
|
+
"tankou",
|
|
82
|
+
"tap",
|
|
83
|
+
"te",
|
|
84
|
+
"t\xE8l",
|
|
85
|
+
"tou",
|
|
86
|
+
"toujou",
|
|
87
|
+
"tout",
|
|
88
|
+
"toutotan",
|
|
89
|
+
"tw\xF2p",
|
|
90
|
+
"w",
|
|
91
|
+
"w\xE8",
|
|
92
|
+
"wi",
|
|
93
|
+
"y",
|
|
94
|
+
"yo",
|
|
95
|
+
"yon",
|
|
96
|
+
"yonn"
|
|
97
|
+
]);
|
|
98
|
+
var haitianCreole = {
|
|
99
|
+
name: "haitian-creole",
|
|
100
|
+
revision: "1",
|
|
101
|
+
stemmer: null,
|
|
102
|
+
stopWords,
|
|
103
|
+
tokenizer: { splitPattern: /[^a-z0-9èò]+/gi }
|
|
104
|
+
};
|
|
105
|
+
|
|
106
|
+
export { haitianCreole };
|
package/dist/languages/hausa.mjs
CHANGED
|
@@ -231,10 +231,11 @@ var stopWords = /* @__PURE__ */ new Set([
|
|
|
231
231
|
]);
|
|
232
232
|
var hausa = {
|
|
233
233
|
name: "hausa",
|
|
234
|
+
revision: "1",
|
|
234
235
|
stemmer: null,
|
|
235
236
|
stopWords,
|
|
236
237
|
tokenizer: {
|
|
237
|
-
splitPattern: /[^a-zA-Zɓɗƙƴʼ0-9]+/gi,
|
|
238
|
+
splitPattern: /[^a-zA-Zɓɗƙƴʼ'0-9]+/gi,
|
|
238
239
|
normalizeDiacritics: false,
|
|
239
240
|
minTokenLength: 1
|
|
240
241
|
}
|
|
@@ -0,0 +1,69 @@
|
|
|
1
|
+
// src/languages/hawaiian.ts
|
|
2
|
+
var stopWords = /* @__PURE__ */ new Set([
|
|
3
|
+
"'a'ohe",
|
|
4
|
+
"'a'ole",
|
|
5
|
+
"'o",
|
|
6
|
+
"'oe",
|
|
7
|
+
"'olua",
|
|
8
|
+
"'oukou",
|
|
9
|
+
"a",
|
|
10
|
+
"a'e",
|
|
11
|
+
"ak\u0101",
|
|
12
|
+
"aku",
|
|
13
|
+
"ana",
|
|
14
|
+
"au",
|
|
15
|
+
"e",
|
|
16
|
+
"he",
|
|
17
|
+
"ho'i",
|
|
18
|
+
"i",
|
|
19
|
+
"ia",
|
|
20
|
+
"i\u0101",
|
|
21
|
+
"iho",
|
|
22
|
+
"in\u0101",
|
|
23
|
+
"ka",
|
|
24
|
+
"k\u0101",
|
|
25
|
+
"ka'u",
|
|
26
|
+
"k\u0101kou",
|
|
27
|
+
"k\u0101na",
|
|
28
|
+
"k\u0101u",
|
|
29
|
+
"k\u0101ua",
|
|
30
|
+
"ke",
|
|
31
|
+
"k\u0113ia",
|
|
32
|
+
"kekahi",
|
|
33
|
+
"k\u0113l\u0101",
|
|
34
|
+
"k\u0113n\u0101",
|
|
35
|
+
"ko",
|
|
36
|
+
"ko'u",
|
|
37
|
+
"kona",
|
|
38
|
+
"kou",
|
|
39
|
+
"l\u0101kou",
|
|
40
|
+
"l\u0101ua",
|
|
41
|
+
"loa",
|
|
42
|
+
"ma",
|
|
43
|
+
"mai",
|
|
44
|
+
"m\u0101kou",
|
|
45
|
+
"mau",
|
|
46
|
+
"m\u0101ua",
|
|
47
|
+
"me",
|
|
48
|
+
"na",
|
|
49
|
+
"n\u0101",
|
|
50
|
+
"na'e",
|
|
51
|
+
"nei",
|
|
52
|
+
"no",
|
|
53
|
+
"n\u014D",
|
|
54
|
+
"o",
|
|
55
|
+
"paha",
|
|
56
|
+
"p\u016B",
|
|
57
|
+
"ua",
|
|
58
|
+
"wale",
|
|
59
|
+
"wau"
|
|
60
|
+
]);
|
|
61
|
+
var hawaiian = {
|
|
62
|
+
name: "hawaiian",
|
|
63
|
+
revision: "1",
|
|
64
|
+
stemmer: null,
|
|
65
|
+
stopWords,
|
|
66
|
+
tokenizer: { splitPattern: /[^\p{Script=Latin}0-9']+/giu }
|
|
67
|
+
};
|
|
68
|
+
|
|
69
|
+
export { hawaiian };
|