@delali/narsil 0.1.15 → 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 -982
- 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 +4 -3
- package/dist/server.mjs +13 -2
- 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 +1,62 @@
|
|
|
1
|
+
import { withNormalisedSpellings } from '../chunk-GHJWC2E5.mjs';
|
|
2
|
+
|
|
3
|
+
// src/languages/telugu.ts
|
|
4
|
+
var stopWords = /* @__PURE__ */ new Set([
|
|
5
|
+
"\u0C05\u0C21\u0C17\u0C21\u0C02",
|
|
6
|
+
"\u0C05\u0C21\u0C17\u0C02\u0C21\u0C3F",
|
|
7
|
+
"\u0C05\u0C21\u0C4D\u0C21\u0C02\u0C17\u0C3E",
|
|
8
|
+
"\u0C05\u0C02\u0C26\u0C30\u0C42",
|
|
9
|
+
"\u0C05\u0C02\u0C26\u0C41\u0C2C\u0C3E\u0C1F\u0C41\u0C32\u0C4B",
|
|
10
|
+
"\u0C05\u0C28\u0C41\u0C17\u0C41\u0C23\u0C02\u0C17\u0C3E",
|
|
11
|
+
"\u0C05\u0C28\u0C41\u0C2E\u0C24\u0C3F\u0C02\u0C1A\u0C41",
|
|
12
|
+
"\u0C05\u0C28\u0C41\u0C2E\u0C24\u0C3F\u0C38\u0C4D\u0C24\u0C41\u0C02\u0C26\u0C3F",
|
|
13
|
+
"\u0C05\u0C2F\u0C3F\u0C24\u0C47",
|
|
14
|
+
"\u0C07\u0C2A\u0C4D\u0C2A\u0C1F\u0C3F\u0C15\u0C47",
|
|
15
|
+
"\u0C09\u0C28\u0C4D\u0C28\u0C3E\u0C30\u0C41",
|
|
16
|
+
"\u0C0E\u0C15\u0C4D\u0C15\u0C21\u0C48\u0C28\u0C3E",
|
|
17
|
+
"\u0C0E\u0C2A\u0C4D\u0C2A\u0C41\u0C21\u0C41",
|
|
18
|
+
"\u0C0E\u0C35\u0C30\u0C48\u0C28\u0C3E",
|
|
19
|
+
"\u0C0F",
|
|
20
|
+
"\u0C0F\u0C26\u0C48\u0C28\u0C3E",
|
|
21
|
+
"\u0C0F\u0C2E\u0C48\u0C28\u0C2A\u0C4D\u0C2A\u0C1F\u0C3F\u0C15\u0C3F",
|
|
22
|
+
"\u0C12\u0C15",
|
|
23
|
+
"\u0C15\u0C28\u0C3F\u0C2A\u0C3F\u0C38\u0C4D\u0C24\u0C3E\u0C2F\u0C3F",
|
|
24
|
+
"\u0C15\u0C3E\u0C26\u0C41",
|
|
25
|
+
"\u0C15\u0C42\u0C21\u0C3E",
|
|
26
|
+
"\u0C17\u0C3E",
|
|
27
|
+
"\u0C17\u0C41\u0C30\u0C3F\u0C02\u0C1A\u0C3F",
|
|
28
|
+
"\u0C1A\u0C41\u0C1F\u0C4D\u0C1F\u0C42",
|
|
29
|
+
"\u0C1A\u0C47\u0C2F\u0C17\u0C32\u0C3F\u0C17\u0C3F\u0C02\u0C26\u0C3F",
|
|
30
|
+
"\u0C24\u0C17\u0C3F\u0C28",
|
|
31
|
+
"\u0C24\u0C30\u0C4D\u0C35\u0C3E\u0C24",
|
|
32
|
+
"\u0C26\u0C3E\u0C26\u0C3E\u0C2A\u0C41",
|
|
33
|
+
"\u0C26\u0C42\u0C30\u0C02\u0C17\u0C3E",
|
|
34
|
+
"\u0C28\u0C3F\u0C1C\u0C02\u0C17\u0C3E",
|
|
35
|
+
"\u0C2A\u0C48",
|
|
36
|
+
"\u0C2A\u0C4D\u0C30\u0C15\u0C3E\u0C30\u0C02",
|
|
37
|
+
"\u0C2E\u0C27\u0C4D\u0C2F",
|
|
38
|
+
"\u0C2E\u0C30\u0C3F\u0C2F\u0C41",
|
|
39
|
+
"\u0C2E\u0C30\u0C4A\u0C15",
|
|
40
|
+
"\u0C2E\u0C33\u0C4D\u0C33\u0C40",
|
|
41
|
+
"\u0C2E\u0C3E\u0C24\u0C4D\u0C30\u0C2E\u0C47",
|
|
42
|
+
"\u0C2E\u0C46\u0C1A\u0C4D\u0C1A\u0C41\u0C15\u0C4B",
|
|
43
|
+
"\u0C35\u0C26\u0C4D\u0C26",
|
|
44
|
+
"\u0C35\u0C46\u0C02\u0C1F",
|
|
45
|
+
"\u0C35\u0C47\u0C30\u0C41\u0C17\u0C3E",
|
|
46
|
+
"\u0C35\u0C4D\u0C2F\u0C24\u0C3F\u0C30\u0C47\u0C15\u0C02\u0C17\u0C3E",
|
|
47
|
+
"\u0C38\u0C02\u0C2C\u0C02\u0C27\u0C02"
|
|
48
|
+
]);
|
|
49
|
+
var ZERO_WIDTH_NON_JOINER = /\u200c/g;
|
|
50
|
+
function normalize(token) {
|
|
51
|
+
return token.replace(ZERO_WIDTH_NON_JOINER, "");
|
|
52
|
+
}
|
|
53
|
+
var telugu = {
|
|
54
|
+
name: "telugu",
|
|
55
|
+
revision: "1",
|
|
56
|
+
stemmer: null,
|
|
57
|
+
stopWords: withNormalisedSpellings(stopWords, normalize),
|
|
58
|
+
normalizer: normalize,
|
|
59
|
+
tokenizer: { splitPattern: /[^\u0c00-\u0c7f\u200ca-z0-9]+/gi }
|
|
60
|
+
};
|
|
61
|
+
|
|
62
|
+
export { telugu };
|
|
@@ -0,0 +1,65 @@
|
|
|
1
|
+
// src/languages/thai.ts
|
|
2
|
+
var stopWords = /* @__PURE__ */ new Set([
|
|
3
|
+
"\u0E01\u0E47",
|
|
4
|
+
"\u0E01\u0E31\u0E19",
|
|
5
|
+
"\u0E01\u0E31\u0E1A",
|
|
6
|
+
"\u0E02\u0E2D",
|
|
7
|
+
"\u0E02\u0E36\u0E49\u0E19",
|
|
8
|
+
"\u0E04\u0E07",
|
|
9
|
+
"\u0E04\u0E37\u0E2D",
|
|
10
|
+
"\u0E08\u0E30",
|
|
11
|
+
"\u0E08\u0E31\u0E14",
|
|
12
|
+
"\u0E08\u0E36\u0E07",
|
|
13
|
+
"\u0E0B\u0E36\u0E48\u0E07",
|
|
14
|
+
"\u0E14\u0E31\u0E07",
|
|
15
|
+
"\u0E44\u0E14\u0E49",
|
|
16
|
+
"\u0E15\u0E48\u0E2D",
|
|
17
|
+
"\u0E15\u0E31\u0E49\u0E07",
|
|
18
|
+
"\u0E41\u0E15\u0E48",
|
|
19
|
+
"\u0E16\u0E49\u0E32",
|
|
20
|
+
"\u0E16\u0E36\u0E07",
|
|
21
|
+
"\u0E16\u0E39\u0E01",
|
|
22
|
+
"\u0E17\u0E31\u0E49\u0E07",
|
|
23
|
+
"\u0E17\u0E4D\u0E32",
|
|
24
|
+
"\u0E17\u0E33",
|
|
25
|
+
"\u0E17\u0E35\u0E48",
|
|
26
|
+
"\u0E17\u0E38\u0E01",
|
|
27
|
+
"\u0E19\u0E31\u0E01",
|
|
28
|
+
"\u0E19\u0E31\u0E49\u0E19",
|
|
29
|
+
"\u0E19\u0E48\u0E32",
|
|
30
|
+
"\u0E19\u0E4D\u0E32",
|
|
31
|
+
"\u0E19\u0E33",
|
|
32
|
+
"\u0E19\u0E35\u0E49",
|
|
33
|
+
"\u0E43\u0E19",
|
|
34
|
+
"\u0E44\u0E1B",
|
|
35
|
+
"\u0E1C\u0E25",
|
|
36
|
+
"\u0E1E\u0E1A",
|
|
37
|
+
"\u0E21\u0E32",
|
|
38
|
+
"\u0E21\u0E35",
|
|
39
|
+
"\u0E44\u0E21\u0E48",
|
|
40
|
+
"\u0E22\u0E31\u0E07",
|
|
41
|
+
"\u0E23\u0E31\u0E1A",
|
|
42
|
+
"\u0E25\u0E07",
|
|
43
|
+
"\u0E27\u0E31\u0E19",
|
|
44
|
+
"\u0E27\u0E48\u0E32",
|
|
45
|
+
"\u0E44\u0E27\u0E49",
|
|
46
|
+
"\u0E2A\u0E48\u0E07",
|
|
47
|
+
"\u0E2A\u0E38\u0E14",
|
|
48
|
+
"\u0E43\u0E2B\u0E49",
|
|
49
|
+
"\u0E2D\u0E22\u0E39\u0E48",
|
|
50
|
+
"\u0E2D\u0E35\u0E01"
|
|
51
|
+
]);
|
|
52
|
+
var thai = {
|
|
53
|
+
name: "thai",
|
|
54
|
+
revision: "1",
|
|
55
|
+
stemmer: null,
|
|
56
|
+
stopWords,
|
|
57
|
+
tokenizer: {
|
|
58
|
+
splitPattern: /[^ก-เ-๎๐-๙a-z0-9]+/giu,
|
|
59
|
+
normalizeDiacritics: false,
|
|
60
|
+
minTokenLength: 1,
|
|
61
|
+
ngramSize: 2
|
|
62
|
+
}
|
|
63
|
+
};
|
|
64
|
+
|
|
65
|
+
export { thai };
|
|
@@ -0,0 +1,119 @@
|
|
|
1
|
+
// src/languages/tibetan.ts
|
|
2
|
+
var stopWords = /* @__PURE__ */ new Set([
|
|
3
|
+
"\u0F40",
|
|
4
|
+
"\u0F40\u0FB1\u0F44",
|
|
5
|
+
"\u0F40\u0FB1\u0F72",
|
|
6
|
+
"\u0F40\u0FB1\u0F72\u0F53",
|
|
7
|
+
"\u0F40\u0FB1\u0F72\u0F66",
|
|
8
|
+
"\u0F40\u0F72",
|
|
9
|
+
"\u0F40\u0F74\u0F53",
|
|
10
|
+
"\u0F41",
|
|
11
|
+
"\u0F42",
|
|
12
|
+
"\u0F42\u0F58",
|
|
13
|
+
"\u0F42\u0F5E\u0F53",
|
|
14
|
+
"\u0F42\u0FB1\u0F72",
|
|
15
|
+
"\u0F42\u0FB1\u0F72\u0F53",
|
|
16
|
+
"\u0F42\u0FB1\u0F72\u0F66",
|
|
17
|
+
"\u0F42\u0FB2\u0F44",
|
|
18
|
+
"\u0F42\u0F72",
|
|
19
|
+
"\u0F42\u0F72\u0F44",
|
|
20
|
+
"\u0F42\u0F72\u0F53",
|
|
21
|
+
"\u0F42\u0F72\u0F66",
|
|
22
|
+
"\u0F42\u0F7A",
|
|
23
|
+
"\u0F42\u0F7C",
|
|
24
|
+
"\u0F44\u0F58",
|
|
25
|
+
"\u0F44\u0F7A",
|
|
26
|
+
"\u0F44\u0F7C",
|
|
27
|
+
"\u0F45\u0F44",
|
|
28
|
+
"\u0F45\u0F72\u0F42",
|
|
29
|
+
"\u0F45\u0F72\u0F44",
|
|
30
|
+
"\u0F45\u0F7C\u0F42",
|
|
31
|
+
"\u0F49\u0F72\u0F51",
|
|
32
|
+
"\u0F4F\u0F58",
|
|
33
|
+
"\u0F4F\u0F74",
|
|
34
|
+
"\u0F4F\u0F7A",
|
|
35
|
+
"\u0F4F\u0F7C",
|
|
36
|
+
"\u0F51\u0F42",
|
|
37
|
+
"\u0F51\u0F44",
|
|
38
|
+
"\u0F51\u0F58",
|
|
39
|
+
"\u0F51\u0F74",
|
|
40
|
+
"\u0F51\u0F7A",
|
|
41
|
+
"\u0F51\u0F7C",
|
|
42
|
+
"\u0F53",
|
|
43
|
+
"\u0F53\u0F44",
|
|
44
|
+
"\u0F53\u0F58",
|
|
45
|
+
"\u0F53\u0F72",
|
|
46
|
+
"\u0F53\u0F7C",
|
|
47
|
+
"\u0F54\u0F66",
|
|
48
|
+
"\u0F54\u0F7C",
|
|
49
|
+
"\u0F55\u0FB1\u0F72",
|
|
50
|
+
"\u0F56\u0F45\u0F66",
|
|
51
|
+
"\u0F56\u0F58",
|
|
52
|
+
"\u0F56\u0F62",
|
|
53
|
+
"\u0F56\u0F66",
|
|
54
|
+
"\u0F56\u0F7C",
|
|
55
|
+
"\u0F58",
|
|
56
|
+
"\u0F58\u0F7E",
|
|
57
|
+
"\u0F58\u0F58",
|
|
58
|
+
"\u0F58\u0F72",
|
|
59
|
+
"\u0F58\u0F72\u0F53",
|
|
60
|
+
"\u0F58\u0F7C",
|
|
61
|
+
"\u0F59\u0F58",
|
|
62
|
+
"\u0F5A",
|
|
63
|
+
"\u0F5A\u0F74",
|
|
64
|
+
"\u0F5A\u0F7C",
|
|
65
|
+
"\u0F5E\u0F72\u0F42",
|
|
66
|
+
"\u0F5E\u0F72\u0F44",
|
|
67
|
+
"\u0F60\u0F42",
|
|
68
|
+
"\u0F60\u0F42\u0F60",
|
|
69
|
+
"\u0F60\u0F44",
|
|
70
|
+
"\u0F60\u0F51\u0F72",
|
|
71
|
+
"\u0F60\u0F55\u0FB2\u0F63",
|
|
72
|
+
"\u0F60\u0F56\u0F60",
|
|
73
|
+
"\u0F60\u0F58",
|
|
74
|
+
"\u0F60\u0F72",
|
|
75
|
+
"\u0F60\u0F7C",
|
|
76
|
+
"\u0F61",
|
|
77
|
+
"\u0F61\u0F44",
|
|
78
|
+
"\u0F61\u0F72",
|
|
79
|
+
"\u0F61\u0F72\u0F53",
|
|
80
|
+
"\u0F61\u0F72\u0F66",
|
|
81
|
+
"\u0F61\u0F7C\u0F44\u0F66",
|
|
82
|
+
"\u0F62",
|
|
83
|
+
"\u0F62\u0F44",
|
|
84
|
+
"\u0F62\u0FA3\u0F58\u0F66",
|
|
85
|
+
"\u0F62\u0F56",
|
|
86
|
+
"\u0F62\u0F58",
|
|
87
|
+
"\u0F62\u0F72\u0F63",
|
|
88
|
+
"\u0F62\u0F74",
|
|
89
|
+
"\u0F62\u0F74\u0F44",
|
|
90
|
+
"\u0F62\u0F7A",
|
|
91
|
+
"\u0F62\u0F7C",
|
|
92
|
+
"\u0F63",
|
|
93
|
+
"\u0F63\u0F42\u0F66",
|
|
94
|
+
"\u0F63\u0F58",
|
|
95
|
+
"\u0F63\u0F66",
|
|
96
|
+
"\u0F63\u0F7C",
|
|
97
|
+
"\u0F64\u0F66",
|
|
98
|
+
"\u0F64\u0F72\u0F42",
|
|
99
|
+
"\u0F64\u0F72\u0F44",
|
|
100
|
+
"\u0F66",
|
|
101
|
+
"\u0F66\u0F99\u0F7A\u0F51",
|
|
102
|
+
"\u0F66\u0F9F\u0F7A",
|
|
103
|
+
"\u0F66\u0FA4\u0FB1\u0F72",
|
|
104
|
+
"\u0F66\u0F58",
|
|
105
|
+
"\u0F66\u0F74",
|
|
106
|
+
"\u0F66\u0F7C",
|
|
107
|
+
"\u0F66\u0F7C\u0F42\u0F66",
|
|
108
|
+
"\u0F68\u0F44",
|
|
109
|
+
"\u0F68\u0F7A"
|
|
110
|
+
]);
|
|
111
|
+
var tibetan = {
|
|
112
|
+
name: "tibetan",
|
|
113
|
+
revision: "1",
|
|
114
|
+
stemmer: null,
|
|
115
|
+
stopWords,
|
|
116
|
+
tokenizer: { splitPattern: /[^\u0f20-\u0f33\u0f40-\u0fbca-z0-9]+/gi }
|
|
117
|
+
};
|
|
118
|
+
|
|
119
|
+
export { tibetan };
|
|
@@ -0,0 +1,194 @@
|
|
|
1
|
+
// src/languages/tigrinya.ts
|
|
2
|
+
var stopWords = /* @__PURE__ */ new Set([
|
|
3
|
+
"\u120D\u12D5\u120A",
|
|
4
|
+
"\u1212\u12D9",
|
|
5
|
+
"\u1212\u12DB",
|
|
6
|
+
"\u1215\u1302",
|
|
7
|
+
"\u1218\u1260\u120D",
|
|
8
|
+
"\u1218\u1295",
|
|
9
|
+
"\u1218\u1295\u130E",
|
|
10
|
+
"\u1218\u1320\u1295",
|
|
11
|
+
"\u121B\u1208\u1275",
|
|
12
|
+
"\u121D\u1235",
|
|
13
|
+
"\u121D\u1260\u122D",
|
|
14
|
+
"\u121D\u1263\u120D",
|
|
15
|
+
"\u121D\u12A5\u1295\u1272",
|
|
16
|
+
"\u121D\u12BD\u1295\u12EB\u1271",
|
|
17
|
+
"\u121D\u12BD\u1295\u12EB\u1275",
|
|
18
|
+
"\u121D\u12C3\u1291",
|
|
19
|
+
"\u121D\u12C3\u1295\u1293",
|
|
20
|
+
"\u121D\u12C3\u1296\u121D",
|
|
21
|
+
"\u121E",
|
|
22
|
+
"\u1235\u1208",
|
|
23
|
+
"\u1235\u1208\u12DA",
|
|
24
|
+
"\u1235\u1208\u12DD\u1260\u120B",
|
|
25
|
+
"\u123D\u12D1",
|
|
26
|
+
"\u1245\u12F5\u121A",
|
|
27
|
+
"\u1260\u1208",
|
|
28
|
+
"\u1260\u1272",
|
|
29
|
+
"\u1260\u12DA",
|
|
30
|
+
"\u1265\u121D\u1263\u120D",
|
|
31
|
+
"\u1265\u1270\u12C8\u1233\u12BA",
|
|
32
|
+
"\u1265\u12B8\u1218\u12ED",
|
|
33
|
+
"\u1265\u12D8\u12ED",
|
|
34
|
+
"\u1265\u12D8\u12ED\u12AB",
|
|
35
|
+
"\u1265\u12D9\u1215",
|
|
36
|
+
"\u1265\u12DB\u12D5\u1263",
|
|
37
|
+
"\u1265\u134D\u120B\u12ED",
|
|
38
|
+
"\u1270\u1263\u1202\u1209",
|
|
39
|
+
"\u1272",
|
|
40
|
+
"\u1273",
|
|
41
|
+
"\u1290\u1260\u1228",
|
|
42
|
+
"\u1290\u1272",
|
|
43
|
+
"\u1290\u1273",
|
|
44
|
+
"\u1290\u1276\u121D",
|
|
45
|
+
"\u1290\u12DA",
|
|
46
|
+
"\u1290\u12ED\u1229",
|
|
47
|
+
"\u1290\u1308\u122B\u1275",
|
|
48
|
+
"\u1290\u1308\u122D",
|
|
49
|
+
"\u1293\u1265",
|
|
50
|
+
"\u1293\u1265\u1272",
|
|
51
|
+
"\u1293\u1275\u12A9\u121D",
|
|
52
|
+
"\u1293\u1275\u12AA",
|
|
53
|
+
"\u1293\u1275\u12AB",
|
|
54
|
+
"\u1293\u1275\u12AD\u1295",
|
|
55
|
+
"\u1293\u12ED",
|
|
56
|
+
"\u1293\u12ED\u1272",
|
|
57
|
+
"\u1295\u1215\u1293",
|
|
58
|
+
"\u1295\u1231",
|
|
59
|
+
"\u1295\u1233",
|
|
60
|
+
"\u1295\u1233\u1276\u121D",
|
|
61
|
+
"\u1295\u1235\u12BA",
|
|
62
|
+
"\u1295\u1235\u12BB",
|
|
63
|
+
"\u1295\u1235\u12BB\u1275\u12A9\u121D",
|
|
64
|
+
"\u1295\u1235\u12BB\u1275\u12AD\u1295",
|
|
65
|
+
"\u1295\u12D3\u12ED",
|
|
66
|
+
"\u12A2\u1208",
|
|
67
|
+
"\u12A2\u1209",
|
|
68
|
+
"\u12A2\u120B",
|
|
69
|
+
"\u12A2\u120D\u12AB",
|
|
70
|
+
"\u12A2\u120E\u121D",
|
|
71
|
+
"\u12A2\u1293",
|
|
72
|
+
"\u12A2\u12BB",
|
|
73
|
+
"\u12A2\u12E9",
|
|
74
|
+
"\u12A3\u1208\u12B9",
|
|
75
|
+
"\u12A3\u1208\u12C9",
|
|
76
|
+
"\u12A3\u1208\u12CE",
|
|
77
|
+
"\u12A3\u120B",
|
|
78
|
+
"\u12A3\u120E",
|
|
79
|
+
"\u12A3\u1265",
|
|
80
|
+
"\u12A3\u1265\u1272",
|
|
81
|
+
"\u12A3\u1265\u1273",
|
|
82
|
+
"\u12A3\u1265\u12A1",
|
|
83
|
+
"\u12A3\u1265\u12DA",
|
|
84
|
+
"\u12A3\u1290",
|
|
85
|
+
"\u12A3\u12DD\u12E9",
|
|
86
|
+
"\u12A3\u12ED\u12AE\u1290\u1295",
|
|
87
|
+
"\u12A3\u12ED\u12B0\u1290\u1295",
|
|
88
|
+
"\u12A5\u121D\u1260\u122D",
|
|
89
|
+
"\u12A5\u121E",
|
|
90
|
+
"\u12A5\u1270\u1295",
|
|
91
|
+
"\u12A5\u1272",
|
|
92
|
+
"\u12A5\u1273",
|
|
93
|
+
"\u12A5\u1276\u121D",
|
|
94
|
+
"\u12A5\u1295\u1270",
|
|
95
|
+
"\u12A5\u1295\u1270\u120E",
|
|
96
|
+
"\u12A5\u1295\u1270\u12BE\u1290",
|
|
97
|
+
"\u12A5\u1295\u1273\u12ED",
|
|
98
|
+
"\u12A5\u1295\u12A8\u120E",
|
|
99
|
+
"\u12A5\u12B3",
|
|
100
|
+
"\u12A5\u12CB\u1295",
|
|
101
|
+
"\u12A5\u12CD\u1295",
|
|
102
|
+
"\u12A5\u12DA",
|
|
103
|
+
"\u12A5\u12DB",
|
|
104
|
+
"\u12A5\u12DE\u121D",
|
|
105
|
+
"\u12A5\u12E8",
|
|
106
|
+
"\u12A5\u12E8\u1295",
|
|
107
|
+
"\u12A5\u12E9",
|
|
108
|
+
"\u12A5\u12EB",
|
|
109
|
+
"\u12A5\u12EE\u121D",
|
|
110
|
+
"\u12A8\u120E",
|
|
111
|
+
"\u12A8\u1218\u12ED",
|
|
112
|
+
"\u12A8\u121D",
|
|
113
|
+
"\u12A8\u121D\u1272",
|
|
114
|
+
"\u12A8\u121D\u12A1",
|
|
115
|
+
"\u12A8\u121D\u12D8\u120E",
|
|
116
|
+
"\u12A8\u121D\u12DA",
|
|
117
|
+
"\u12A8\u12A3",
|
|
118
|
+
"\u12A9\u1209",
|
|
119
|
+
"\u12AB\u120D\u12A5",
|
|
120
|
+
"\u12AB\u120D\u12A6\u1275",
|
|
121
|
+
"\u12AB\u1265",
|
|
122
|
+
"\u12AB\u1265\u1272",
|
|
123
|
+
"\u12AB\u1265\u1276\u121D",
|
|
124
|
+
"\u12AD\u1233\u1265",
|
|
125
|
+
"\u12AD\u1233\u12D5",
|
|
126
|
+
"\u12AD\u1265\u120D",
|
|
127
|
+
"\u12AD\u1295\u12F0\u12ED",
|
|
128
|
+
"\u12AD\u1295\u12F2",
|
|
129
|
+
"\u12AD\u12B8\u12CD\u1295",
|
|
130
|
+
"\u12AE\u12ED\u1291",
|
|
131
|
+
"\u12B0\u12ED\u1291",
|
|
132
|
+
"\u12B3",
|
|
133
|
+
"\u12B5\u1209",
|
|
134
|
+
"\u12B8\u121D",
|
|
135
|
+
"\u12B8\u12A3",
|
|
136
|
+
"\u12C8\u12ED",
|
|
137
|
+
"\u12CB\u120B",
|
|
138
|
+
"\u12CD\u1295",
|
|
139
|
+
"\u12D8\u1208\u1293",
|
|
140
|
+
"\u12D8\u1208\u12C9",
|
|
141
|
+
"\u12D8\u1208\u12CB",
|
|
142
|
+
"\u12D8\u1208\u12CE",
|
|
143
|
+
"\u12D8\u1208\u12CE\u121D",
|
|
144
|
+
"\u12D8\u120B",
|
|
145
|
+
"\u12D8\u120E",
|
|
146
|
+
"\u12D8\u12ED\u1265\u1209",
|
|
147
|
+
"\u12DA",
|
|
148
|
+
"\u12DD\u122D\u12A8\u1265",
|
|
149
|
+
"\u12DD\u1260\u1203\u120D",
|
|
150
|
+
"\u12DD\u1260\u1208",
|
|
151
|
+
"\u12DD\u1265\u120D",
|
|
152
|
+
"\u12DD\u1270\u1263\u1205\u1208",
|
|
153
|
+
"\u12DD\u1270\u12BB\u12E8\u12F0",
|
|
154
|
+
"\u12DD\u1270\u1348\u120B\u1208\u12E8",
|
|
155
|
+
"\u12DD\u1270\u1348\u120B\u1208\u12E9",
|
|
156
|
+
"\u12DD\u1290\u1260\u1228",
|
|
157
|
+
"\u12DD\u1290\u1260\u1228\u1275",
|
|
158
|
+
"\u12DD\u1290\u1260\u1229",
|
|
159
|
+
"\u12DD\u12AB\u12E8\u12F5",
|
|
160
|
+
"\u12DD\u12B8\u12CD\u1295",
|
|
161
|
+
"\u12DD\u12BD\u12A5\u120D",
|
|
162
|
+
"\u12DD\u12BE\u1290",
|
|
163
|
+
"\u12DD\u12C0\u1290",
|
|
164
|
+
"\u12E8",
|
|
165
|
+
"\u12E8\u1208\u1295",
|
|
166
|
+
"\u12E9",
|
|
167
|
+
"\u12EB",
|
|
168
|
+
"\u12ED\u1255\u1228\u1265",
|
|
169
|
+
"\u12ED\u1265\u120D",
|
|
170
|
+
"\u12ED\u12B8\u12CD\u1295",
|
|
171
|
+
"\u12ED\u12B9\u1295",
|
|
172
|
+
"\u12ED\u12BD\u12A5\u120D",
|
|
173
|
+
"\u12EE\u121D",
|
|
174
|
+
"\u12EE\u1295",
|
|
175
|
+
"\u12F0\u12A3",
|
|
176
|
+
"\u12F5\u1215\u122A",
|
|
177
|
+
"\u12F5\u121B",
|
|
178
|
+
"\u1308\u1208",
|
|
179
|
+
"\u1308\u120A\u1339",
|
|
180
|
+
"\u1308\u1293",
|
|
181
|
+
"\u1308\u12ED\u1229",
|
|
182
|
+
"\u130D\u1293",
|
|
183
|
+
"\u130D\u1295",
|
|
184
|
+
"\u1325\u122B\u12ED"
|
|
185
|
+
]);
|
|
186
|
+
var tigrinya = {
|
|
187
|
+
name: "tigrinya",
|
|
188
|
+
revision: "1",
|
|
189
|
+
stemmer: null,
|
|
190
|
+
stopWords,
|
|
191
|
+
tokenizer: { splitPattern: /[^\u1200-\u135f\u1369-\u137ca-z0-9]+/gi }
|
|
192
|
+
};
|
|
193
|
+
|
|
194
|
+
export { tigrinya };
|
|
@@ -0,0 +1,56 @@
|
|
|
1
|
+
// src/languages/tongan.ts
|
|
2
|
+
var stopWords = /* @__PURE__ */ new Set([
|
|
3
|
+
"'a",
|
|
4
|
+
"'e",
|
|
5
|
+
"'ena",
|
|
6
|
+
"'ene",
|
|
7
|
+
"'eni",
|
|
8
|
+
"'i",
|
|
9
|
+
"'ikai",
|
|
10
|
+
"'o",
|
|
11
|
+
"'oku",
|
|
12
|
+
"'osi",
|
|
13
|
+
"ai",
|
|
14
|
+
"ange",
|
|
15
|
+
"atu",
|
|
16
|
+
"au",
|
|
17
|
+
"e",
|
|
18
|
+
"foki",
|
|
19
|
+
"ha",
|
|
20
|
+
"hake",
|
|
21
|
+
"he",
|
|
22
|
+
"hifo",
|
|
23
|
+
"ho",
|
|
24
|
+
"hoku",
|
|
25
|
+
"hono",
|
|
26
|
+
"ia",
|
|
27
|
+
"ka",
|
|
28
|
+
"kae",
|
|
29
|
+
"kapau",
|
|
30
|
+
"k\u0101toa",
|
|
31
|
+
"ke",
|
|
32
|
+
"ki",
|
|
33
|
+
"ko",
|
|
34
|
+
"koe",
|
|
35
|
+
"koe'uhi",
|
|
36
|
+
"kuo",
|
|
37
|
+
"lolotonga",
|
|
38
|
+
"mai",
|
|
39
|
+
"mei",
|
|
40
|
+
"mo",
|
|
41
|
+
"na'a",
|
|
42
|
+
"na'e",
|
|
43
|
+
"ni",
|
|
44
|
+
"pe",
|
|
45
|
+
"p\u0113",
|
|
46
|
+
"te"
|
|
47
|
+
]);
|
|
48
|
+
var tongan = {
|
|
49
|
+
name: "tongan",
|
|
50
|
+
revision: "1",
|
|
51
|
+
stemmer: null,
|
|
52
|
+
stopWords,
|
|
53
|
+
tokenizer: { splitPattern: /[^\p{Script=Latin}0-9']+/giu }
|
|
54
|
+
};
|
|
55
|
+
|
|
56
|
+
export { tongan };
|