@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
|
@@ -16,7 +16,6 @@ var stopWords = /* @__PURE__ */ new Set([
|
|
|
16
16
|
"\u3044\u308B",
|
|
17
17
|
"\u3059\u308B",
|
|
18
18
|
"\u306A\u308B",
|
|
19
|
-
"\u3067\u304D\u308B",
|
|
20
19
|
"\u306A\u3044",
|
|
21
20
|
"\u3082",
|
|
22
21
|
"\u306A",
|
|
@@ -32,48 +31,17 @@ var stopWords = /* @__PURE__ */ new Set([
|
|
|
32
31
|
"\u307B\u3069",
|
|
33
32
|
"\u3060\u3051",
|
|
34
33
|
"\u3057\u304B",
|
|
35
|
-
"\u3070\u304B\u308A",
|
|
36
|
-
"\u304F\u3089\u3044",
|
|
37
|
-
"\u3050\u3089\u3044",
|
|
38
34
|
"\u307E\u307E",
|
|
39
|
-
"\u306A\u304C\u3089",
|
|
40
35
|
"\u3064\u3064",
|
|
41
36
|
"\u306E\u3067",
|
|
42
37
|
"\u306E\u306B",
|
|
43
|
-
"\u3051\u308C\u3069",
|
|
44
|
-
"\u3051\u308C\u3069\u3082",
|
|
45
38
|
"\u3051\u3069",
|
|
46
39
|
"\u3060\u304C",
|
|
47
|
-
"\u3057\u304B\u3057",
|
|
48
|
-
"\u3068\u3053\u308D\u304C",
|
|
49
40
|
"\u3067\u3082",
|
|
50
|
-
"\u305D\u308C\u3067\u3082",
|
|
51
|
-
"\u3060\u304B\u3089",
|
|
52
|
-
"\u3057\u305F\u304C\u3063\u3066",
|
|
53
|
-
"\u305D\u306E\u305F\u3081",
|
|
54
|
-
"\u3086\u3048\u306B",
|
|
55
|
-
"\u305D\u3057\u3066",
|
|
56
41
|
"\u307E\u305F",
|
|
57
|
-
"\u304A\u3088\u3073",
|
|
58
|
-
"\u306A\u3089\u3073\u306B",
|
|
59
|
-
"\u3042\u308B\u3044\u306F",
|
|
60
|
-
"\u307E\u305F\u306F",
|
|
61
|
-
"\u3082\u3057\u304F\u306F",
|
|
62
|
-
"\u305D\u308C\u3068\u3082",
|
|
63
|
-
"\u3064\u307E\u308A",
|
|
64
|
-
"\u3059\u306A\u308F\u3061",
|
|
65
|
-
"\u305F\u3068\u3048\u3070",
|
|
66
42
|
"\u306A\u304A",
|
|
67
|
-
"\u305F\u3060\u3057",
|
|
68
|
-
"\u3082\u3063\u3068\u3082",
|
|
69
43
|
"\u3055\u3066",
|
|
70
|
-
"\u3068\u3053\u308D\u3067",
|
|
71
|
-
"\u305D\u308C\u3067\u306F",
|
|
72
44
|
"\u3067\u306F",
|
|
73
|
-
"\u3055\u3089\u306B",
|
|
74
|
-
"\u3057\u304B\u3082",
|
|
75
|
-
"\u305D\u306E\u4E0A",
|
|
76
|
-
"\u305D\u308C\u306B",
|
|
77
45
|
"\u3053\u308C",
|
|
78
46
|
"\u305D\u308C",
|
|
79
47
|
"\u3042\u308C",
|
|
@@ -84,25 +52,15 @@ var stopWords = /* @__PURE__ */ new Set([
|
|
|
84
52
|
"\u3069\u306E",
|
|
85
53
|
"\u3053\u3053",
|
|
86
54
|
"\u305D\u3053",
|
|
87
|
-
"\u3042\u305D\u3053",
|
|
88
55
|
"\u3069\u3053",
|
|
89
56
|
"\u3053\u3046",
|
|
90
57
|
"\u305D\u3046",
|
|
91
58
|
"\u3042\u3042",
|
|
92
59
|
"\u3069\u3046",
|
|
93
|
-
"\u3053\u3093\u306A",
|
|
94
|
-
"\u305D\u3093\u306A",
|
|
95
|
-
"\u3042\u3093\u306A",
|
|
96
|
-
"\u3069\u3093\u306A",
|
|
97
|
-
"\u3053\u3061\u3089",
|
|
98
|
-
"\u305D\u3061\u3089",
|
|
99
|
-
"\u3042\u3061\u3089",
|
|
100
|
-
"\u3069\u3061\u3089",
|
|
101
60
|
"\u79C1",
|
|
102
61
|
"\u50D5",
|
|
103
62
|
"\u4FFA",
|
|
104
63
|
"\u81EA\u5206",
|
|
105
|
-
"\u3042\u306A\u305F",
|
|
106
64
|
"\u5F7C",
|
|
107
65
|
"\u5F7C\u5973",
|
|
108
66
|
"\u6211\u3005",
|
|
@@ -110,40 +68,15 @@ var stopWords = /* @__PURE__ */ new Set([
|
|
|
110
68
|
"\u5F7C\u3089",
|
|
111
69
|
"\u8AB0",
|
|
112
70
|
"\u4F55",
|
|
113
|
-
"\u3069\u306A\u305F",
|
|
114
71
|
"\u3044\u3064",
|
|
115
72
|
"\u306A\u305C",
|
|
116
|
-
"\u3069\u3046\u3057\u3066",
|
|
117
|
-
"\u3044\u304B\u304C",
|
|
118
73
|
"\u3067\u3059",
|
|
119
74
|
"\u307E\u3059",
|
|
120
|
-
"\u3067\u3057\u305F",
|
|
121
|
-
"\u307E\u3057\u305F",
|
|
122
|
-
"\u3067\u3057\u3087\u3046",
|
|
123
|
-
"\u3060\u308D\u3046",
|
|
124
|
-
"\u3060\u3063\u305F",
|
|
125
|
-
"\u3067\u3042\u308B",
|
|
126
|
-
"\u3067\u3042\u3063\u305F",
|
|
127
|
-
"\u3067\u3042\u308A",
|
|
128
|
-
"\u3067\u3042\u308D\u3046",
|
|
129
|
-
"\u3066\u3044\u308B",
|
|
130
|
-
"\u3066\u3044\u305F",
|
|
131
|
-
"\u3066\u304A\u304F",
|
|
132
|
-
"\u3066\u3042\u308B",
|
|
133
|
-
"\u3066\u3057\u307E\u3046",
|
|
134
|
-
"\u3066\u3044\u304F",
|
|
135
|
-
"\u3066\u304F\u308B",
|
|
136
|
-
"\u3066\u307F\u308B",
|
|
137
|
-
"\u3088\u3046\u3060",
|
|
138
|
-
"\u3089\u3057\u3044",
|
|
139
|
-
"\u305D\u3046\u3060",
|
|
140
|
-
"\u307F\u305F\u3044\u3060",
|
|
141
75
|
"\u3079\u304D",
|
|
142
76
|
"\u306F\u305A",
|
|
143
77
|
"\u308F\u3051",
|
|
144
78
|
"\u3053\u3068",
|
|
145
79
|
"\u3082\u306E",
|
|
146
|
-
"\u3068\u3053\u308D",
|
|
147
80
|
"\u3068\u304D",
|
|
148
81
|
"\u3068\u3082",
|
|
149
82
|
"\u305F\u3081",
|
|
@@ -152,58 +85,22 @@ var stopWords = /* @__PURE__ */ new Set([
|
|
|
152
85
|
"\u305F\u3061",
|
|
153
86
|
"\u3089",
|
|
154
87
|
"\u306A\u3069",
|
|
155
|
-
"\u306A\u3093\u304B",
|
|
156
|
-
"\u306A\u3093\u3066",
|
|
157
88
|
"\u3063\u3066",
|
|
158
|
-
"\u3068\u3044\u3046",
|
|
159
|
-
"\u3068\u3044\u3063\u305F",
|
|
160
|
-
"\u3068\u3057\u3066",
|
|
161
|
-
"\u306B\u3068\u3063\u3066",
|
|
162
|
-
"\u306B\u3064\u3044\u3066",
|
|
163
|
-
"\u306B\u95A2\u3057\u3066",
|
|
164
|
-
"\u306B\u5BFE\u3057\u3066",
|
|
165
|
-
"\u306B\u304A\u3044\u3066",
|
|
166
|
-
"\u306B\u304A\u3051\u308B",
|
|
167
|
-
"\u306B\u3088\u3063\u3066",
|
|
168
|
-
"\u306B\u3088\u308B",
|
|
169
|
-
"\u306B\u3088\u308A",
|
|
170
|
-
"\u3092\u3082\u3063\u3066",
|
|
171
|
-
"\u306B\u95A2\u3059\u308B",
|
|
172
|
-
"\u306B\u5BFE\u3059\u308B",
|
|
173
|
-
"\u305D\u308C\u305E\u308C",
|
|
174
|
-
"\u3059\u3079\u3066",
|
|
175
|
-
"\u307F\u3093\u306A",
|
|
176
|
-
"\u307B\u3068\u3093\u3069",
|
|
177
|
-
"\u305F\u304F\u3055\u3093",
|
|
178
|
-
"\u3044\u304F\u3064",
|
|
179
|
-
"\u3044\u304F\u3089",
|
|
180
89
|
"\u3082\u3046",
|
|
181
90
|
"\u307E\u3060",
|
|
182
|
-
"\u3082\u3063\u3068",
|
|
183
|
-
"\u305A\u3063\u3068",
|
|
184
|
-
"\u3068\u3066\u3082",
|
|
185
|
-
"\u304B\u306A\u308A",
|
|
186
|
-
"\u3084\u306F\u308A",
|
|
187
|
-
"\u3084\u3063\u3071\u308A",
|
|
188
91
|
"\u305C\u3072",
|
|
189
|
-
"\
|
|
190
|
-
"\u305F\u3076\u3093",
|
|
191
|
-
"\u304A\u305D\u3089\u304F",
|
|
192
|
-
"\u7279\u306B",
|
|
193
|
-
"\u3061\u3087\u3063\u3068",
|
|
194
|
-
"\u5C11\u3057",
|
|
195
|
-
"\u3060\u3044\u305F\u3044",
|
|
196
|
-
"\u307B\u307C",
|
|
197
|
-
"\u53CA\u3073"
|
|
92
|
+
"\u307B\u307C"
|
|
198
93
|
]);
|
|
199
94
|
var japanese = {
|
|
200
95
|
name: "japanese",
|
|
96
|
+
revision: "1",
|
|
201
97
|
stemmer: null,
|
|
202
98
|
stopWords,
|
|
203
99
|
tokenizer: {
|
|
204
|
-
splitPattern: /[^\
|
|
100
|
+
splitPattern: /[^\p{Script=Han}\u3040-\u309f\u30a0-\u30fa\u30fc-\u30ff0-9a-zA-Z]+/giu,
|
|
205
101
|
normalizeDiacritics: false,
|
|
206
|
-
minTokenLength: 1
|
|
102
|
+
minTokenLength: 1,
|
|
103
|
+
ngramSize: 2
|
|
207
104
|
}
|
|
208
105
|
};
|
|
209
106
|
|
|
@@ -0,0 +1,94 @@
|
|
|
1
|
+
// src/languages/kannada.ts
|
|
2
|
+
var stopWords = /* @__PURE__ */ new Set([
|
|
3
|
+
"\u0C85\u0C82\u0CA4",
|
|
4
|
+
"\u0C85\u0CA5\u0CB5\u0CBE",
|
|
5
|
+
"\u0C85\u0CA6\u0C95\u0CCD\u0C95\u0CC6",
|
|
6
|
+
"\u0C85\u0CA6\u0CA8\u0CCD\u0CA8\u0CC1",
|
|
7
|
+
"\u0C85\u0CA6\u0CC1",
|
|
8
|
+
"\u0C85\u0CA6\u0CC7",
|
|
9
|
+
"\u0C85\u0CB2\u0CCD\u0CB2",
|
|
10
|
+
"\u0C85\u0CB2\u0CCD\u0CB2\u0CA6\u0CC6",
|
|
11
|
+
"\u0C85\u0CB5\u0CB0",
|
|
12
|
+
"\u0C85\u0CB5\u0CB0\u0CA8\u0CCD\u0CA8\u0CC1",
|
|
13
|
+
"\u0C85\u0CB5\u0CB0\u0CBF\u0C97\u0CC6",
|
|
14
|
+
"\u0C85\u0CB5\u0CB0\u0CC1",
|
|
15
|
+
"\u0C85\u0CB7\u0CCD\u0C9F\u0CC7",
|
|
16
|
+
"\u0C86",
|
|
17
|
+
"\u0C86\u0C97",
|
|
18
|
+
"\u0C86\u0C97\u0CBF",
|
|
19
|
+
"\u0C86\u0CA6\u0CB0\u0CC6",
|
|
20
|
+
"\u0C87\u0CA4\u0CCD\u0CA4\u0CC1",
|
|
21
|
+
"\u0C87\u0CA6\u0C95\u0CCD\u0C95\u0CC6",
|
|
22
|
+
"\u0C87\u0CA6\u0CA8\u0CCD\u0CA8\u0CC1",
|
|
23
|
+
"\u0C87\u0CA6\u0CB0",
|
|
24
|
+
"\u0C87\u0CA6\u0CB0\u0CBF\u0C82\u0CA6",
|
|
25
|
+
"\u0C87\u0CA6\u0CC1",
|
|
26
|
+
"\u0C87\u0CA6\u0CC6",
|
|
27
|
+
"\u0C87\u0CA6\u0CC7",
|
|
28
|
+
"\u0C87\u0CA8\u0CCD\u0CA8\u0CC1",
|
|
29
|
+
"\u0C87\u0CA8\u0CCD\u0CA8\u0CC2",
|
|
30
|
+
"\u0C87\u0CB0\u0CC1\u0CB5",
|
|
31
|
+
"\u0C87\u0CB2\u0CCD\u0CB2",
|
|
32
|
+
"\u0C87\u0CB2\u0CCD\u0CB2\u0CBF",
|
|
33
|
+
"\u0C88",
|
|
34
|
+
"\u0C88\u0C97",
|
|
35
|
+
"\u0C88\u0C97\u0CBE\u0C97\u0CB2\u0CC7",
|
|
36
|
+
"\u0C8E\u0C82\u0CA6\u0CB0\u0CC1",
|
|
37
|
+
"\u0C8E\u0C82\u0CA6\u0CB0\u0CC6",
|
|
38
|
+
"\u0C8E\u0C82\u0CA6\u0CC1",
|
|
39
|
+
"\u0C8E\u0C82\u0CAC",
|
|
40
|
+
"\u0C8E\u0C82\u0CAC\u0CC1\u0CA6\u0CC1",
|
|
41
|
+
"\u0C8E\u0CB2\u0CCD\u0CB2",
|
|
42
|
+
"\u0C8E\u0CB2\u0CCD\u0CB2\u0CBE",
|
|
43
|
+
"\u0C8E\u0CB7\u0CCD\u0C9F\u0CC1",
|
|
44
|
+
"\u0C8F\u0CA8\u0CC1",
|
|
45
|
+
"\u0C95\u0CB3\u0CC6\u0CA6",
|
|
46
|
+
"\u0C95\u0CC2\u0CA1",
|
|
47
|
+
"\u0C95\u0CC6\u0CB2\u0CB5\u0CC1",
|
|
48
|
+
"\u0C95\u0CC7\u0CB5\u0CB2",
|
|
49
|
+
"\u0C9C\u0CCA\u0CA4\u0CC6\u0C97\u0CC6",
|
|
50
|
+
"\u0CA4\u0CA8\u0CCD\u0CA8",
|
|
51
|
+
"\u0CA4\u0CAE\u0CCD\u0CAE",
|
|
52
|
+
"\u0CA4\u0CC1\u0C82\u0CAC\u0CBE",
|
|
53
|
+
"\u0CA8\u0CA1\u0CC6\u0CA6",
|
|
54
|
+
"\u0CA8\u0CA1\u0CC6\u0CA6\u0CBF\u0CA6\u0CC6",
|
|
55
|
+
"\u0CA8\u0CA8\u0C97\u0CC6",
|
|
56
|
+
"\u0CA8\u0CA8\u0CCD\u0CA8",
|
|
57
|
+
"\u0CA8\u0CA8\u0CCD\u0CA8\u0CA8\u0CCD\u0CA8\u0CC1",
|
|
58
|
+
"\u0CA8\u0CAE\u0CCD\u0CAE",
|
|
59
|
+
"\u0CA8\u0CBE\u0CA8\u0CC1",
|
|
60
|
+
"\u0CA8\u0CBE\u0CB5\u0CC1",
|
|
61
|
+
"\u0CA8\u0CBF\u0CAE\u0CCD\u0CAE",
|
|
62
|
+
"\u0CA8\u0CC0\u0CA1\u0CBF\u0CA6\u0CB0\u0CC1",
|
|
63
|
+
"\u0CA8\u0CC0\u0CA1\u0CBF\u0CA6\u0CC6",
|
|
64
|
+
"\u0CA8\u0CC0\u0CA1\u0CBF\u0CA6\u0CCD\u0CA6\u0CBE\u0CB0\u0CC6",
|
|
65
|
+
"\u0CA8\u0CC0\u0CB5\u0CC1",
|
|
66
|
+
"\u0CAC\u0C97\u0CCD\u0C97\u0CC6",
|
|
67
|
+
"\u0CAC\u0CB3\u0CBF\u0C95",
|
|
68
|
+
"\u0CAC\u0CC7\u0CB0\u0CC6",
|
|
69
|
+
"\u0CAE\u0CA4\u0CCD\u0CA4\u0CC1",
|
|
70
|
+
"\u0CAE\u0CA4\u0CCD\u0CA4\u0CC6",
|
|
71
|
+
"\u0CAE\u0CBE\u0CA1\u0CBF",
|
|
72
|
+
"\u0CAE\u0CBE\u0CA4\u0CCD\u0CB0",
|
|
73
|
+
"\u0CAE\u0CC2\u0CB2\u0C95",
|
|
74
|
+
"\u0CAE\u0CC7\u0CB2\u0CC6",
|
|
75
|
+
"\u0CAF\u0CBE\u0CB5",
|
|
76
|
+
"\u0CAF\u0CBE\u0CB5\u0CC1\u0CA6\u0CC7",
|
|
77
|
+
"\u0CB9\u0CB2\u0CB5\u0CC1",
|
|
78
|
+
"\u0CB9\u0CBE\u0C97\u0CBE\u0C97\u0CBF",
|
|
79
|
+
"\u0CB9\u0CBE\u0C97\u0CC2",
|
|
80
|
+
"\u0CB9\u0CC0\u0C97\u0CBE\u0C97\u0CBF",
|
|
81
|
+
"\u0CB9\u0CC0\u0C97\u0CC6",
|
|
82
|
+
"\u0CB9\u0CC6\u0C9A\u0CCD\u0C9A\u0CC1",
|
|
83
|
+
"\u0CB9\u0CC7\u0CB3\u0CBF",
|
|
84
|
+
"\u0CB9\u0CC7\u0CB3\u0CBF\u0CA6\u0CCD\u0CA6\u0CBE\u0CB0\u0CC6"
|
|
85
|
+
]);
|
|
86
|
+
var kannada = {
|
|
87
|
+
name: "kannada",
|
|
88
|
+
revision: "1",
|
|
89
|
+
stemmer: null,
|
|
90
|
+
stopWords,
|
|
91
|
+
tokenizer: { splitPattern: /[^\u0c80-\u0cffa-z0-9]+/gi }
|
|
92
|
+
};
|
|
93
|
+
|
|
94
|
+
export { kannada };
|
|
@@ -0,0 +1,90 @@
|
|
|
1
|
+
// src/languages/kazakh.ts
|
|
2
|
+
var stopWords = /* @__PURE__ */ new Set([
|
|
3
|
+
"\u0430\u0437",
|
|
4
|
+
"\u0430\u043B",
|
|
5
|
+
"\u0430\u0440\u049B\u044B\u043B\u044B",
|
|
6
|
+
"\u0431\u0430\u0440",
|
|
7
|
+
"\u0431\u0430\u0440\u043B\u044B\u0493\u044B",
|
|
8
|
+
"\u0431\u0430\u0440\u043B\u044B\u049B",
|
|
9
|
+
"\u0431\u0430\u0441\u049B\u0430",
|
|
10
|
+
"\u0431\u0435\u043D",
|
|
11
|
+
"\u0431\u0435\u0440\u0456",
|
|
12
|
+
"\u0431\u043E\u0439\u044B\u043D\u0448\u0430",
|
|
13
|
+
"\u0431\u043E\u043B\u0434\u044B",
|
|
14
|
+
"\u0431\u043E\u043B\u044B\u043F",
|
|
15
|
+
"\u0431\u043E\u043B\u0493\u0430\u043D",
|
|
16
|
+
"\u0431\u0456\u0437",
|
|
17
|
+
"\u0431\u0456\u0437\u0434\u0456\u04A3",
|
|
18
|
+
"\u0431\u0456\u0440",
|
|
19
|
+
"\u0431\u0456\u0440\u0430\u049B",
|
|
20
|
+
"\u0431\u04B1\u043B",
|
|
21
|
+
"\u0434\u0430",
|
|
22
|
+
"\u0434\u0435",
|
|
23
|
+
"\u0434\u0435\u0439\u0456\u043D",
|
|
24
|
+
"\u0435\u0434\u0456",
|
|
25
|
+
"\u0435\u043A\u0435\u043D",
|
|
26
|
+
"\u0435\u043A\u0456",
|
|
27
|
+
"\u0435\u043C\u0435\u0441",
|
|
28
|
+
"\u0435\u043D\u0434\u0456",
|
|
29
|
+
"\u0435\u04A3",
|
|
30
|
+
"\u0436\u043E\u049B",
|
|
31
|
+
"\u0436\u04D9\u043D\u0435",
|
|
32
|
+
"\u043A\u0435\u0439\u0456\u043D",
|
|
33
|
+
"\u043A\u0456\u043C",
|
|
34
|
+
"\u043A\u04E9\u043F",
|
|
35
|
+
"\u043C\u0435\u043D",
|
|
36
|
+
"\u043C\u0435\u043D\u0456\u04A3",
|
|
37
|
+
"\u043D\u0435",
|
|
38
|
+
"\u043D\u0435\u0433\u0435",
|
|
39
|
+
"\u043D\u0435\u043C\u0435\u0441\u0435",
|
|
40
|
+
"\u043E\u043B",
|
|
41
|
+
"\u043E\u043B\u0430\u0440",
|
|
42
|
+
"\u043E\u043B\u0430\u0440\u0434\u044B",
|
|
43
|
+
"\u043E\u043D\u044B\u04A3",
|
|
44
|
+
"\u043E\u0441\u044B",
|
|
45
|
+
"\u043F\u0435\u043D",
|
|
46
|
+
"\u0441\u0430\u0439\u044B\u043D",
|
|
47
|
+
"\u0441\u0435\u0431\u0435\u0431\u0456",
|
|
48
|
+
"\u0441\u0435\u043D",
|
|
49
|
+
"\u0441\u0435\u043D\u0434\u0435\u0440",
|
|
50
|
+
"\u0441\u0435\u043D\u0456\u04A3",
|
|
51
|
+
"\u0441\u043E\u043B",
|
|
52
|
+
"\u0441\u043E\u043D\u0434\u0430",
|
|
53
|
+
"\u0441\u043E\u043D\u0434\u044B\u049B\u0442\u0430\u043D",
|
|
54
|
+
"\u0441\u043E\u04A3",
|
|
55
|
+
"\u0441\u0456\u0437",
|
|
56
|
+
"\u0441\u0456\u0437\u0434\u0435\u0440",
|
|
57
|
+
"\u0441\u0456\u0437\u0434\u0456\u04A3",
|
|
58
|
+
"\u0442\u0430",
|
|
59
|
+
"\u0442\u0430\u0493\u044B",
|
|
60
|
+
"\u0442\u0435",
|
|
61
|
+
"\u0442\u0435\u043A",
|
|
62
|
+
"\u0442\u0443\u0440\u0430\u043B\u044B",
|
|
63
|
+
"\u044F",
|
|
64
|
+
"\u0493\u0430\u043D\u0430",
|
|
65
|
+
"\u049B\u0430\u0437\u0456\u0440",
|
|
66
|
+
"\u049B\u0430\u0439\u0434\u0430",
|
|
67
|
+
"\u049B\u0430\u043B\u0430\u0439",
|
|
68
|
+
"\u049B\u0430\u043D\u0430",
|
|
69
|
+
"\u049B\u0430\u043D\u0434\u0430\u0439",
|
|
70
|
+
"\u049B\u0430\u043D\u0448\u0430",
|
|
71
|
+
"\u049B\u0430\u0440\u0430\u0439",
|
|
72
|
+
"\u049B\u0430\u0448\u0430\u043D",
|
|
73
|
+
"\u04AF\u0448\u0456\u043D",
|
|
74
|
+
"\u04D9\u043B\u0456",
|
|
75
|
+
"\u04D9\u0440",
|
|
76
|
+
"\u04D9\u0440\u0431\u0456\u0440",
|
|
77
|
+
"\u04D9\u0440\u0456",
|
|
78
|
+
"\u04E9\u0437",
|
|
79
|
+
"\u04E9\u0437\u0456",
|
|
80
|
+
"\u04E9\u0442\u0435"
|
|
81
|
+
]);
|
|
82
|
+
var kazakh = {
|
|
83
|
+
name: "kazakh",
|
|
84
|
+
revision: "1",
|
|
85
|
+
stemmer: null,
|
|
86
|
+
stopWords,
|
|
87
|
+
tokenizer: { splitPattern: /[^\u0400-\u04FF0-9]+/gi }
|
|
88
|
+
};
|
|
89
|
+
|
|
90
|
+
export { kazakh };
|
|
@@ -0,0 +1,181 @@
|
|
|
1
|
+
// src/languages/khmer.ts
|
|
2
|
+
var stopWords = /* @__PURE__ */ new Set([
|
|
3
|
+
"\u1780\u17CF",
|
|
4
|
+
"\u1780\u17B6\u178F\u17CB",
|
|
5
|
+
"\u1780\u17B6\u1793\u17CB",
|
|
6
|
+
"\u1780\u17B6\u179A",
|
|
7
|
+
"\u1780\u17B6\u179B",
|
|
8
|
+
"\u1780\u17BE\u178F",
|
|
9
|
+
"\u1780\u17D2\u178F\u17B8",
|
|
10
|
+
"\u1781\u17B6\u1784",
|
|
11
|
+
"\u1781\u17BB\u179F",
|
|
12
|
+
"\u1781\u17D2\u1789\u17BB\u17C6",
|
|
13
|
+
"\u1781\u17D2\u179B\u17C7",
|
|
14
|
+
"\u1782\u17B6\u178F\u17CB",
|
|
15
|
+
"\u1782\u17BA",
|
|
16
|
+
"\u1782\u17BA\u1787\u17B6",
|
|
17
|
+
"\u1782\u17C1",
|
|
18
|
+
"\u1782\u17D2\u1793\u17B6",
|
|
19
|
+
"\u1782\u17D2\u179A\u17B6",
|
|
20
|
+
"\u1783\u17BE\u1789",
|
|
21
|
+
"\u1785\u1784\u17CB",
|
|
22
|
+
"\u1785\u17C6\u1796\u17C4\u17C7",
|
|
23
|
+
"\u1785\u17B6\u17C6",
|
|
24
|
+
"\u1785\u17B6\u1794\u17CB",
|
|
25
|
+
"\u1785\u17BB\u1784",
|
|
26
|
+
"\u1785\u17BC\u179B",
|
|
27
|
+
"\u1785\u17C1\u1789",
|
|
28
|
+
"\u1786\u17B6\u1794\u17CB",
|
|
29
|
+
"\u1787\u17B6\u1784",
|
|
30
|
+
"\u1787\u17B7\u178F",
|
|
31
|
+
"\u1787\u17BC\u1793",
|
|
32
|
+
"\u1787\u17BD\u1799",
|
|
33
|
+
"\u1787\u17BE\u1784",
|
|
34
|
+
"\u1788\u1794\u17CB",
|
|
35
|
+
"\u1788\u1798",
|
|
36
|
+
"\u1789\u17B9\u1780",
|
|
37
|
+
"\u1789\u17C2\u1780",
|
|
38
|
+
"\u178A\u17CF",
|
|
39
|
+
"\u178A\u17C6",
|
|
40
|
+
"\u178A\u1780",
|
|
41
|
+
"\u178A\u1784",
|
|
42
|
+
"\u178A\u17CF\u1791\u17C3",
|
|
43
|
+
"\u178A\u179B\u17CB",
|
|
44
|
+
"\u178A\u17B6\u17C6",
|
|
45
|
+
"\u178A\u17B9\u1784",
|
|
46
|
+
"\u178A\u17BC\u1785",
|
|
47
|
+
"\u178A\u17BE\u179A",
|
|
48
|
+
"\u178A\u17C2\u179A",
|
|
49
|
+
"\u178A\u17C2\u179B",
|
|
50
|
+
"\u178A\u17C4\u1799",
|
|
51
|
+
"\u178E\u17B6",
|
|
52
|
+
"\u178E\u17B6\u1791\u17C5",
|
|
53
|
+
"\u178E\u17B6\u179F\u17CB",
|
|
54
|
+
"\u178F\u17B6\u17C6\u1784",
|
|
55
|
+
"\u178F\u17B6\u1798",
|
|
56
|
+
"\u178F\u17BE",
|
|
57
|
+
"\u178F\u17C2",
|
|
58
|
+
"\u1790\u17B6",
|
|
59
|
+
"\u1790\u17D2\u1798\u17B8",
|
|
60
|
+
"\u1791\u1791\u17C1",
|
|
61
|
+
"\u1791\u179B\u17CB",
|
|
62
|
+
"\u1791\u17B6\u17C6\u1784",
|
|
63
|
+
"\u1791\u17B6\u1793\u17CB",
|
|
64
|
+
"\u1791\u17B6\u1794",
|
|
65
|
+
"\u1791\u17B6\u179F\u17CB",
|
|
66
|
+
"\u1791\u17B8",
|
|
67
|
+
"\u1791\u17B8\u178E\u17B6",
|
|
68
|
+
"\u1791\u17B8\u1793\u17C1\u17C7",
|
|
69
|
+
"\u1791\u17B8\u1793\u17C4\u17C7",
|
|
70
|
+
"\u1791\u17BB\u1780",
|
|
71
|
+
"\u1791\u17BC\u1791\u17C5",
|
|
72
|
+
"\u1791\u17C0\u178F",
|
|
73
|
+
"\u1791\u17C1",
|
|
74
|
+
"\u1791\u17C4\u17C7",
|
|
75
|
+
"\u1791\u17C4\u17C7\u1794\u17B8",
|
|
76
|
+
"\u1791\u17C5",
|
|
77
|
+
"\u1791\u17C5\u179B\u17BE",
|
|
78
|
+
"\u1792\u17D2\u179C\u17BE",
|
|
79
|
+
"\u1793\u17B6\u1784",
|
|
80
|
+
"\u1793\u17B6\u1799",
|
|
81
|
+
"\u1793\u17B7\u1784",
|
|
82
|
+
"\u1793\u17B9\u1780",
|
|
83
|
+
"\u1793\u17B9\u1784",
|
|
84
|
+
"\u1793\u17BC\u179C",
|
|
85
|
+
"\u1793\u17C1\u17C7",
|
|
86
|
+
"\u1793\u17C1\u17C7\u1787\u17B6",
|
|
87
|
+
"\u1793\u17C3",
|
|
88
|
+
"\u1793\u17C4\u17C7",
|
|
89
|
+
"\u1793\u17C4\u17C7\u1787\u17B6",
|
|
90
|
+
"\u1793\u17C4\u17C7\u1791\u17C1",
|
|
91
|
+
"\u1793\u17C5",
|
|
92
|
+
"\u1793\u17C5\u178F\u17C2",
|
|
93
|
+
"\u1793\u17C5\u1791\u17B8",
|
|
94
|
+
"\u1793\u17C5\u179B\u17BE",
|
|
95
|
+
"\u1794\u17B8",
|
|
96
|
+
"\u1794\u17B8\u1793\u17C1\u17C7",
|
|
97
|
+
"\u1794\u17BD\u1793",
|
|
98
|
+
"\u1794\u17BE",
|
|
99
|
+
"\u1794\u17BE\u1780",
|
|
100
|
+
"\u1794\u17D2\u179A\u17B6\u17C6",
|
|
101
|
+
"\u1794\u17D2\u179A\u17BE",
|
|
102
|
+
"\u1794\u17D2\u179A\u17C2",
|
|
103
|
+
"\u1795\u1784",
|
|
104
|
+
"\u1795\u17BB\u178F",
|
|
105
|
+
"\u1796\u17B8",
|
|
106
|
+
"\u1796\u17B8\u179A",
|
|
107
|
+
"\u1796\u17B8\u179B\u17BE",
|
|
108
|
+
"\u1796\u17BB\u17C6",
|
|
109
|
+
"\u1796\u17BD\u1780",
|
|
110
|
+
"\u1796\u17C1\u1789",
|
|
111
|
+
"\u1796\u17C1\u179B",
|
|
112
|
+
"\u1796\u17C4\u179B",
|
|
113
|
+
"\u1796\u17D2\u179A\u17C4\u17C7",
|
|
114
|
+
"\u1797\u17B6\u1796",
|
|
115
|
+
"\u1797\u17D2\u179B\u17BA",
|
|
116
|
+
"\u1798\u1780",
|
|
117
|
+
"\u1798\u17B6\u1793",
|
|
118
|
+
"\u1798\u17B7\u1793",
|
|
119
|
+
"\u1798\u17BC\u179B",
|
|
120
|
+
"\u1798\u17BD\u1799",
|
|
121
|
+
"\u1798\u17C9\u17C1\u1785",
|
|
122
|
+
"\u1798\u17C2\u1793",
|
|
123
|
+
"\u1798\u17D2\u1797\u17C3",
|
|
124
|
+
"\u1798\u17D2\u179B\u17C9\u17C1\u17C7",
|
|
125
|
+
"\u1799\u1780",
|
|
126
|
+
"\u1799\u17C9\u17B6\u1784",
|
|
127
|
+
"\u1799\u17BE\u1784",
|
|
128
|
+
"\u179A\u1799\u17C8",
|
|
129
|
+
"\u179A\u17B6\u179B\u17CB",
|
|
130
|
+
"\u179A\u17BA",
|
|
131
|
+
"\u179A\u17BA\u1780\u17CF",
|
|
132
|
+
"\u179A\u17BD\u1785",
|
|
133
|
+
"\u179A\u17BD\u1798",
|
|
134
|
+
"\u179B\u17BE",
|
|
135
|
+
"\u179B\u17BE\u1780",
|
|
136
|
+
"\u179B\u17BE\u179F",
|
|
137
|
+
"\u179B\u17C2\u1784",
|
|
138
|
+
"\u179B\u17C4\u1780",
|
|
139
|
+
"\u179B\u17D2\u17A2",
|
|
140
|
+
"\u179C\u17B6",
|
|
141
|
+
"\u179C\u17B7\u1789",
|
|
142
|
+
"\u179C\u17C2\u1784",
|
|
143
|
+
"\u179F\u17C4\u17C7",
|
|
144
|
+
"\u179F\u17D2\u1793\u17BE",
|
|
145
|
+
"\u179F\u17D2\u1798\u17C4\u17C7",
|
|
146
|
+
"\u17A0\u178F\u17CB",
|
|
147
|
+
"\u17A0\u17CA\u17B6\u1793",
|
|
148
|
+
"\u17A0\u17B6\u1798",
|
|
149
|
+
"\u17A0\u17BE\u1799",
|
|
150
|
+
"\u17A0\u17C5",
|
|
151
|
+
"\u17A1\u17BE\u1784",
|
|
152
|
+
"\u17A1\u17BE\u1799",
|
|
153
|
+
"\u17A2\u17C6\u1796\u17B8",
|
|
154
|
+
"\u17A2\u179F\u17CB",
|
|
155
|
+
"\u17A2\u17B6\u1785",
|
|
156
|
+
"\u17A2\u17B8",
|
|
157
|
+
"\u17A2\u17C4\u1799",
|
|
158
|
+
"\u17A2\u17D2\u179C\u17B8",
|
|
159
|
+
"\u17A5\u178F",
|
|
160
|
+
"\u17AC",
|
|
161
|
+
"\u17AC\u1791\u17C1",
|
|
162
|
+
"\u17AF",
|
|
163
|
+
"\u17AF\u1780\u17C4",
|
|
164
|
+
"\u17AF\u1784",
|
|
165
|
+
"\u17AF\u178E\u17B6",
|
|
166
|
+
"\u17B1\u17D2\u1799"
|
|
167
|
+
]);
|
|
168
|
+
var khmer = {
|
|
169
|
+
name: "khmer",
|
|
170
|
+
revision: "1",
|
|
171
|
+
stemmer: null,
|
|
172
|
+
stopWords,
|
|
173
|
+
tokenizer: {
|
|
174
|
+
splitPattern: /[^ក-៓ៗ០-៩a-z0-9]+/giu,
|
|
175
|
+
normalizeDiacritics: false,
|
|
176
|
+
minTokenLength: 1,
|
|
177
|
+
ngramSize: 2
|
|
178
|
+
}
|
|
179
|
+
};
|
|
180
|
+
|
|
181
|
+
export { khmer };
|
|
@@ -0,0 +1,99 @@
|
|
|
1
|
+
// src/languages/kinyarwanda.ts
|
|
2
|
+
var stopWords = /* @__PURE__ */ new Set([
|
|
3
|
+
"aba",
|
|
4
|
+
"abandi",
|
|
5
|
+
"abo",
|
|
6
|
+
"aho",
|
|
7
|
+
"aka",
|
|
8
|
+
"ako",
|
|
9
|
+
"andi",
|
|
10
|
+
"ari",
|
|
11
|
+
"ariko",
|
|
12
|
+
"aya",
|
|
13
|
+
"ayo",
|
|
14
|
+
"bari",
|
|
15
|
+
"bo",
|
|
16
|
+
"bose",
|
|
17
|
+
"buri",
|
|
18
|
+
"bw",
|
|
19
|
+
"byose",
|
|
20
|
+
"cy",
|
|
21
|
+
"cyane",
|
|
22
|
+
"cyangwa",
|
|
23
|
+
"gusa",
|
|
24
|
+
"hagati",
|
|
25
|
+
"hano",
|
|
26
|
+
"hari",
|
|
27
|
+
"hariya",
|
|
28
|
+
"hasi",
|
|
29
|
+
"hejuru",
|
|
30
|
+
"hose",
|
|
31
|
+
"ibi",
|
|
32
|
+
"ibindi",
|
|
33
|
+
"ibyo",
|
|
34
|
+
"icyo",
|
|
35
|
+
"iki",
|
|
36
|
+
"ikindi",
|
|
37
|
+
"indi",
|
|
38
|
+
"iri",
|
|
39
|
+
"iryo",
|
|
40
|
+
"iyi",
|
|
41
|
+
"iyo",
|
|
42
|
+
"izindi",
|
|
43
|
+
"izo",
|
|
44
|
+
"jye",
|
|
45
|
+
"kandi",
|
|
46
|
+
"ku",
|
|
47
|
+
"kubera",
|
|
48
|
+
"kuko",
|
|
49
|
+
"kuri",
|
|
50
|
+
"mbere",
|
|
51
|
+
"mu",
|
|
52
|
+
"muri",
|
|
53
|
+
"mwe",
|
|
54
|
+
"mwese",
|
|
55
|
+
"n",
|
|
56
|
+
"na",
|
|
57
|
+
"ndi",
|
|
58
|
+
"ni",
|
|
59
|
+
"niba",
|
|
60
|
+
"njye",
|
|
61
|
+
"nka",
|
|
62
|
+
"no",
|
|
63
|
+
"noneho",
|
|
64
|
+
"nta",
|
|
65
|
+
"nyuma",
|
|
66
|
+
"rw",
|
|
67
|
+
"si",
|
|
68
|
+
"turi",
|
|
69
|
+
"twe",
|
|
70
|
+
"twese",
|
|
71
|
+
"u",
|
|
72
|
+
"ubu",
|
|
73
|
+
"ubwo",
|
|
74
|
+
"uku",
|
|
75
|
+
"ukwo",
|
|
76
|
+
"undi",
|
|
77
|
+
"uri",
|
|
78
|
+
"uru",
|
|
79
|
+
"urwo",
|
|
80
|
+
"utu",
|
|
81
|
+
"utwo",
|
|
82
|
+
"uwo",
|
|
83
|
+
"uyu",
|
|
84
|
+
"we",
|
|
85
|
+
"wowe",
|
|
86
|
+
"y",
|
|
87
|
+
"yari",
|
|
88
|
+
"yose",
|
|
89
|
+
"zose"
|
|
90
|
+
]);
|
|
91
|
+
var kinyarwanda = {
|
|
92
|
+
name: "kinyarwanda",
|
|
93
|
+
revision: "1",
|
|
94
|
+
stemmer: null,
|
|
95
|
+
stopWords,
|
|
96
|
+
tokenizer: { splitPattern: /[^a-z0-9]+/gi }
|
|
97
|
+
};
|
|
98
|
+
|
|
99
|
+
export { kinyarwanda };
|