@delali/narsil 0.1.0
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/LICENSE +201 -0
- package/dist/adapters/filesystem.d.ts +8 -0
- package/dist/adapters/filesystem.mjs +107 -0
- package/dist/adapters/indexeddb.d.ts +9 -0
- package/dist/adapters/indexeddb.mjs +153 -0
- package/dist/adapters/memory.d.ts +5 -0
- package/dist/adapters/memory.mjs +30 -0
- package/dist/adapters-XODT2F3h.d.ts +30 -0
- package/dist/chunk-5HXMVMSM.mjs +3147 -0
- package/dist/chunk-JKIBUPAY.mjs +42 -0
- package/dist/chunk-N37VMC62.mjs +18 -0
- package/dist/chunk-WYC635GI.mjs +335 -0
- package/dist/direct-executor-MQD7JRD4.mjs +3 -0
- package/dist/index.d.ts +386 -0
- package/dist/index.mjs +1099 -0
- package/dist/invalidation/broadcast-channel.d.ts +8 -0
- package/dist/invalidation/broadcast-channel.mjs +46 -0
- package/dist/invalidation/filesystem.d.ts +10 -0
- package/dist/invalidation/filesystem.mjs +159 -0
- package/dist/invalidation/noop.d.ts +5 -0
- package/dist/invalidation/noop.mjs +13 -0
- package/dist/language-ClUGLuoC.d.ts +13 -0
- package/dist/languages/arabic.d.ts +5 -0
- package/dist/languages/arabic.mjs +631 -0
- package/dist/languages/armenian.d.ts +5 -0
- package/dist/languages/armenian.mjs +306 -0
- package/dist/languages/bulgarian.d.ts +5 -0
- package/dist/languages/bulgarian.mjs +330 -0
- package/dist/languages/chinese.d.ts +5 -0
- package/dist/languages/chinese.mjs +250 -0
- package/dist/languages/dagbani.d.ts +5 -0
- package/dist/languages/dagbani.mjs +85 -0
- package/dist/languages/danish.d.ts +5 -0
- package/dist/languages/danish.mjs +224 -0
- package/dist/languages/dutch.d.ts +5 -0
- package/dist/languages/dutch.mjs +296 -0
- package/dist/languages/english.d.ts +5 -0
- package/dist/languages/english.mjs +1 -0
- package/dist/languages/ewe.d.ts +5 -0
- package/dist/languages/ewe.mjs +85 -0
- package/dist/languages/finnish.d.ts +5 -0
- package/dist/languages/finnish.mjs +536 -0
- package/dist/languages/french.d.ts +5 -0
- package/dist/languages/french.mjs +635 -0
- package/dist/languages/ga.d.ts +5 -0
- package/dist/languages/ga.mjs +69 -0
- package/dist/languages/german.d.ts +5 -0
- package/dist/languages/german.mjs +414 -0
- package/dist/languages/greek.d.ts +5 -0
- package/dist/languages/greek.mjs +888 -0
- package/dist/languages/hausa.d.ts +5 -0
- package/dist/languages/hausa.mjs +243 -0
- package/dist/languages/hindi.d.ts +5 -0
- package/dist/languages/hindi.mjs +333 -0
- package/dist/languages/hungarian.d.ts +5 -0
- package/dist/languages/hungarian.mjs +600 -0
- package/dist/languages/igbo.d.ts +5 -0
- package/dist/languages/igbo.mjs +96 -0
- package/dist/languages/indonesian.d.ts +5 -0
- package/dist/languages/indonesian.mjs +896 -0
- package/dist/languages/irish.d.ts +5 -0
- package/dist/languages/irish.mjs +268 -0
- package/dist/languages/italian.d.ts +5 -0
- package/dist/languages/italian.mjs +713 -0
- package/dist/languages/japanese.d.ts +5 -0
- package/dist/languages/japanese.mjs +210 -0
- package/dist/languages/nepali.d.ts +5 -0
- package/dist/languages/nepali.mjs +262 -0
- package/dist/languages/norwegian.d.ts +5 -0
- package/dist/languages/norwegian.mjs +274 -0
- package/dist/languages/portuguese.d.ts +5 -0
- package/dist/languages/portuguese.mjs +606 -0
- package/dist/languages/romanian.d.ts +5 -0
- package/dist/languages/romanian.mjs +634 -0
- package/dist/languages/russian.d.ts +5 -0
- package/dist/languages/russian.mjs +424 -0
- package/dist/languages/sanskrit.d.ts +5 -0
- package/dist/languages/sanskrit.mjs +155 -0
- package/dist/languages/serbian.d.ts +5 -0
- package/dist/languages/serbian.mjs +277 -0
- package/dist/languages/slovenian.d.ts +5 -0
- package/dist/languages/slovenian.mjs +522 -0
- package/dist/languages/spanish.d.ts +5 -0
- package/dist/languages/spanish.mjs +606 -0
- package/dist/languages/swahili.d.ts +5 -0
- package/dist/languages/swahili.mjs +212 -0
- package/dist/languages/swedish.d.ts +5 -0
- package/dist/languages/swedish.mjs +224 -0
- package/dist/languages/tamil.d.ts +5 -0
- package/dist/languages/tamil.mjs +186 -0
- package/dist/languages/turkish.d.ts +5 -0
- package/dist/languages/turkish.mjs +675 -0
- package/dist/languages/twi.d.ts +5 -0
- package/dist/languages/twi.mjs +84 -0
- package/dist/languages/ukrainian.d.ts +5 -0
- package/dist/languages/ukrainian.mjs +444 -0
- package/dist/languages/yoruba.d.ts +5 -0
- package/dist/languages/yoruba.mjs +103 -0
- package/dist/languages/zulu.d.ts +5 -0
- package/dist/languages/zulu.mjs +113 -0
- package/dist/workers/entry.d.ts +3 -0
- package/dist/workers/entry.mjs +106 -0
- package/package.json +87 -0
|
@@ -0,0 +1,243 @@
|
|
|
1
|
+
// src/languages/hausa.ts
|
|
2
|
+
var stopWords = /* @__PURE__ */ new Set([
|
|
3
|
+
"da",
|
|
4
|
+
"ya",
|
|
5
|
+
"a",
|
|
6
|
+
"ba",
|
|
7
|
+
"ta",
|
|
8
|
+
"na",
|
|
9
|
+
"wannan",
|
|
10
|
+
"ne",
|
|
11
|
+
"kuma",
|
|
12
|
+
"su",
|
|
13
|
+
"sai",
|
|
14
|
+
"mu",
|
|
15
|
+
"yan",
|
|
16
|
+
"ko",
|
|
17
|
+
"mai",
|
|
18
|
+
"dai",
|
|
19
|
+
"shi",
|
|
20
|
+
"ka",
|
|
21
|
+
"yayi",
|
|
22
|
+
"to",
|
|
23
|
+
"amma",
|
|
24
|
+
"ga",
|
|
25
|
+
"haka",
|
|
26
|
+
"yi",
|
|
27
|
+
"daga",
|
|
28
|
+
"ma",
|
|
29
|
+
"duk",
|
|
30
|
+
"kai",
|
|
31
|
+
"masu",
|
|
32
|
+
"wani",
|
|
33
|
+
"sun",
|
|
34
|
+
"zai",
|
|
35
|
+
"sa",
|
|
36
|
+
"ina",
|
|
37
|
+
"nan",
|
|
38
|
+
"idan",
|
|
39
|
+
"wanda",
|
|
40
|
+
"an",
|
|
41
|
+
"ce",
|
|
42
|
+
"ke",
|
|
43
|
+
"suka",
|
|
44
|
+
"cikin",
|
|
45
|
+
"wa",
|
|
46
|
+
"daya",
|
|
47
|
+
"kasa",
|
|
48
|
+
"iya",
|
|
49
|
+
"yake",
|
|
50
|
+
"domin",
|
|
51
|
+
"in",
|
|
52
|
+
"kan",
|
|
53
|
+
"yana",
|
|
54
|
+
"aka",
|
|
55
|
+
"za",
|
|
56
|
+
"me",
|
|
57
|
+
"don",
|
|
58
|
+
"har",
|
|
59
|
+
"ake",
|
|
60
|
+
"gaba",
|
|
61
|
+
"akwai",
|
|
62
|
+
"yadda",
|
|
63
|
+
"abin",
|
|
64
|
+
"allah",
|
|
65
|
+
"muke",
|
|
66
|
+
"yawa",
|
|
67
|
+
"koda",
|
|
68
|
+
"wane",
|
|
69
|
+
"kawai",
|
|
70
|
+
"kansa",
|
|
71
|
+
"yace",
|
|
72
|
+
"dole",
|
|
73
|
+
"bisani",
|
|
74
|
+
"misali",
|
|
75
|
+
"marasa",
|
|
76
|
+
"sauransu",
|
|
77
|
+
"ana",
|
|
78
|
+
"dubu",
|
|
79
|
+
"nasa",
|
|
80
|
+
"tunda",
|
|
81
|
+
"matukar",
|
|
82
|
+
"mini",
|
|
83
|
+
"kusa",
|
|
84
|
+
"zarar",
|
|
85
|
+
"ku",
|
|
86
|
+
"sam",
|
|
87
|
+
"nawa",
|
|
88
|
+
"kowane",
|
|
89
|
+
"yayin",
|
|
90
|
+
"gare",
|
|
91
|
+
"sune",
|
|
92
|
+
"lokacin",
|
|
93
|
+
"kuke",
|
|
94
|
+
"hudu",
|
|
95
|
+
"fatan",
|
|
96
|
+
"bai",
|
|
97
|
+
"dake",
|
|
98
|
+
"komai",
|
|
99
|
+
"kwarai",
|
|
100
|
+
"ki",
|
|
101
|
+
"wadancan",
|
|
102
|
+
"zata",
|
|
103
|
+
"muddin",
|
|
104
|
+
"kafin",
|
|
105
|
+
"game",
|
|
106
|
+
"kin",
|
|
107
|
+
"juna",
|
|
108
|
+
"kasance",
|
|
109
|
+
"muka",
|
|
110
|
+
"bisa",
|
|
111
|
+
"wadansu",
|
|
112
|
+
"tana",
|
|
113
|
+
"kowace",
|
|
114
|
+
"basu",
|
|
115
|
+
"saura",
|
|
116
|
+
"hatta",
|
|
117
|
+
"masa",
|
|
118
|
+
"yau",
|
|
119
|
+
"maimakon",
|
|
120
|
+
"yiwuwa",
|
|
121
|
+
"dubban",
|
|
122
|
+
"muna",
|
|
123
|
+
"yin",
|
|
124
|
+
"fa",
|
|
125
|
+
"inda",
|
|
126
|
+
"nasu",
|
|
127
|
+
"dan",
|
|
128
|
+
"basa",
|
|
129
|
+
"yakan",
|
|
130
|
+
"wadda",
|
|
131
|
+
"dinga",
|
|
132
|
+
"yawancin",
|
|
133
|
+
"kake",
|
|
134
|
+
"yiwu",
|
|
135
|
+
"dukkanin",
|
|
136
|
+
"tuni",
|
|
137
|
+
"maka",
|
|
138
|
+
"kamar",
|
|
139
|
+
"sukan",
|
|
140
|
+
"gabata",
|
|
141
|
+
"mata",
|
|
142
|
+
"suna",
|
|
143
|
+
"shima",
|
|
144
|
+
"tace",
|
|
145
|
+
"kuwa",
|
|
146
|
+
"kodayake",
|
|
147
|
+
"tafi",
|
|
148
|
+
"saboda",
|
|
149
|
+
"kun",
|
|
150
|
+
"kimanin",
|
|
151
|
+
"kanmu",
|
|
152
|
+
"data",
|
|
153
|
+
"wata",
|
|
154
|
+
"yiwuwar",
|
|
155
|
+
"ace",
|
|
156
|
+
"yanzu",
|
|
157
|
+
"misalin",
|
|
158
|
+
"bayan",
|
|
159
|
+
"namu",
|
|
160
|
+
"zamu",
|
|
161
|
+
"kana",
|
|
162
|
+
"balle",
|
|
163
|
+
"wai",
|
|
164
|
+
"musamman",
|
|
165
|
+
"ita",
|
|
166
|
+
"ashe",
|
|
167
|
+
"ni",
|
|
168
|
+
"yiwa",
|
|
169
|
+
"haifar",
|
|
170
|
+
"kasancewar",
|
|
171
|
+
"sauran",
|
|
172
|
+
"hakika",
|
|
173
|
+
"tamkar",
|
|
174
|
+
"yayinda",
|
|
175
|
+
"bin",
|
|
176
|
+
"suma",
|
|
177
|
+
"muku",
|
|
178
|
+
"zasu",
|
|
179
|
+
"zaton",
|
|
180
|
+
"abinda",
|
|
181
|
+
"bar",
|
|
182
|
+
"mafi",
|
|
183
|
+
"sosai",
|
|
184
|
+
"tun",
|
|
185
|
+
"cika",
|
|
186
|
+
"alhali",
|
|
187
|
+
"cewa",
|
|
188
|
+
"cewar",
|
|
189
|
+
"bugu",
|
|
190
|
+
"suke",
|
|
191
|
+
"baa",
|
|
192
|
+
"koina",
|
|
193
|
+
"inji",
|
|
194
|
+
"akalla",
|
|
195
|
+
"mace",
|
|
196
|
+
"kanka",
|
|
197
|
+
"wasu",
|
|
198
|
+
"ban",
|
|
199
|
+
"kansu",
|
|
200
|
+
"kanta",
|
|
201
|
+
"lalle",
|
|
202
|
+
"guda",
|
|
203
|
+
"zaka",
|
|
204
|
+
"wadanda",
|
|
205
|
+
"goma",
|
|
206
|
+
"zuwa",
|
|
207
|
+
"wadannan",
|
|
208
|
+
"dukkan",
|
|
209
|
+
"daman",
|
|
210
|
+
"lallai",
|
|
211
|
+
"mun",
|
|
212
|
+
"take",
|
|
213
|
+
"min",
|
|
214
|
+
"hakan",
|
|
215
|
+
"bane",
|
|
216
|
+
"dama",
|
|
217
|
+
"kowanne",
|
|
218
|
+
"akan",
|
|
219
|
+
"kuna",
|
|
220
|
+
"mana",
|
|
221
|
+
"duka",
|
|
222
|
+
"nake",
|
|
223
|
+
"zan",
|
|
224
|
+
"yawan",
|
|
225
|
+
"kusan",
|
|
226
|
+
"kenan",
|
|
227
|
+
"kowa",
|
|
228
|
+
"yanzun",
|
|
229
|
+
"biyu",
|
|
230
|
+
"gabatar"
|
|
231
|
+
]);
|
|
232
|
+
var hausa = {
|
|
233
|
+
name: "hausa",
|
|
234
|
+
stemmer: null,
|
|
235
|
+
stopWords,
|
|
236
|
+
tokenizer: {
|
|
237
|
+
splitPattern: /[^a-zA-Zɓɗƙƴʼ0-9]+/gi,
|
|
238
|
+
normalizeDiacritics: false,
|
|
239
|
+
minTokenLength: 1
|
|
240
|
+
}
|
|
241
|
+
};
|
|
242
|
+
|
|
243
|
+
export { hausa };
|
|
@@ -0,0 +1,333 @@
|
|
|
1
|
+
// src/languages/hindi.ts
|
|
2
|
+
var SUFFIXES_5 = ["\u0924\u093F\u092F\u094B\u0902", "\u093E\u0913\u0902\u0917\u0940", "\u093E\u0913\u0902\u0917\u0947", "\u093E\u0913\u0902\u0917\u093E", "\u093E\u0907\u090F\u0917\u0940", "\u093E\u0907\u090F\u0917\u0947", "\u093E\u0907\u090F\u0917\u093E"];
|
|
3
|
+
var SUFFIXES_4 = [
|
|
4
|
+
"\u093E\u090F\u0902\u0917\u0940",
|
|
5
|
+
"\u093E\u090F\u0902\u0917\u0947",
|
|
6
|
+
"\u093E\u090F\u0902\u0917\u093E",
|
|
7
|
+
"\u093E\u090A\u0902\u0917\u0940",
|
|
8
|
+
"\u093E\u090A\u0902\u0917\u093E",
|
|
9
|
+
"\u093F\u092F\u093E\u0901",
|
|
10
|
+
"\u093F\u092F\u094B\u0902",
|
|
11
|
+
"\u093F\u092F\u093E\u0902",
|
|
12
|
+
"\u0924\u093F\u092F\u094B",
|
|
13
|
+
"\u0942\u0901\u0917\u0940",
|
|
14
|
+
"\u0942\u0901\u0917\u093E",
|
|
15
|
+
"\u093E\u0924\u0940\u0902",
|
|
16
|
+
"\u0928\u093E\u0913\u0902",
|
|
17
|
+
"\u093E\u0913\u0917\u0940",
|
|
18
|
+
"\u093E\u0913\u0917\u0947",
|
|
19
|
+
"\u093E\u0913\u0917\u093E",
|
|
20
|
+
"\u093E\u0907\u090F\u0902",
|
|
21
|
+
"\u093E\u0907\u092F\u0947",
|
|
22
|
+
"\u093E\u0908\u092F\u0947"
|
|
23
|
+
];
|
|
24
|
+
var SUFFIXES_3 = [
|
|
25
|
+
"\u093F\u092F\u093E\u0901",
|
|
26
|
+
"\u094B\u0902\u0917\u0940",
|
|
27
|
+
"\u094B\u0902\u0917\u0947",
|
|
28
|
+
"\u094B\u0902\u0917\u093E",
|
|
29
|
+
"\u090F\u0902\u0917\u0940",
|
|
30
|
+
"\u090F\u0902\u0917\u0947",
|
|
31
|
+
"\u090F\u0902\u0917\u093E",
|
|
32
|
+
"\u0942\u0902\u0917\u0940",
|
|
33
|
+
"\u0942\u0902\u0917\u093E",
|
|
34
|
+
"\u093E\u0924\u0940\u0902",
|
|
35
|
+
"\u0935\u093E\u0932\u093E",
|
|
36
|
+
"\u0935\u093E\u0932\u0947",
|
|
37
|
+
"\u0935\u093E\u0932\u0940",
|
|
38
|
+
"\u093F\u090F\u0917\u0940",
|
|
39
|
+
"\u093F\u090F\u0917\u0947",
|
|
40
|
+
"\u093F\u090F\u0917\u093E",
|
|
41
|
+
"\u0940\u090F\u0917\u0940",
|
|
42
|
+
"\u0940\u090F\u0917\u0947",
|
|
43
|
+
"\u0940\u090F\u0917\u093E",
|
|
44
|
+
"\u093F\u092F\u094B",
|
|
45
|
+
"\u093F\u092F\u093E",
|
|
46
|
+
"\u093E\u0924\u0940",
|
|
47
|
+
"\u093E\u0924\u093E",
|
|
48
|
+
"\u093E\u0924\u0947",
|
|
49
|
+
"\u093E\u0928\u093E",
|
|
50
|
+
"\u093E\u0928\u0947",
|
|
51
|
+
"\u093E\u0928\u0940",
|
|
52
|
+
"\u093E\u0928\u094B",
|
|
53
|
+
"\u093E\u0913\u0902",
|
|
54
|
+
"\u093E\u0915\u0930",
|
|
55
|
+
"\u093E\u0907\u090F",
|
|
56
|
+
"\u093E\u0908\u0902"
|
|
57
|
+
];
|
|
58
|
+
var SUFFIXES_2 = [
|
|
59
|
+
"\u0915\u0930",
|
|
60
|
+
"\u093E\u0913",
|
|
61
|
+
"\u093F\u090F",
|
|
62
|
+
"\u093E\u0908",
|
|
63
|
+
"\u093E\u090F",
|
|
64
|
+
"\u0928\u0947",
|
|
65
|
+
"\u0928\u0940",
|
|
66
|
+
"\u0928\u093E",
|
|
67
|
+
"\u0924\u0947",
|
|
68
|
+
"\u0924\u0940",
|
|
69
|
+
"\u0924\u093E",
|
|
70
|
+
"\u093E\u0901",
|
|
71
|
+
"\u093E\u0902",
|
|
72
|
+
"\u094B\u0902",
|
|
73
|
+
"\u0947\u0902",
|
|
74
|
+
"\u0939\u093E",
|
|
75
|
+
"\u0939\u0947",
|
|
76
|
+
"\u0939\u0940",
|
|
77
|
+
"\u0939\u094B",
|
|
78
|
+
"\u0940\u0902",
|
|
79
|
+
"\u094B\u0901"
|
|
80
|
+
];
|
|
81
|
+
var SUFFIXES_1 = ["\u093E", "\u0947", "\u094B", "\u0940", "\u093F", "\u0942", "\u0941", "\u0902", "\u0903", "\u0901"];
|
|
82
|
+
var ALL_SUFFIXES = [...SUFFIXES_5, ...SUFFIXES_4, ...SUFFIXES_3, ...SUFFIXES_2, ...SUFFIXES_1];
|
|
83
|
+
function isConsonant(ch) {
|
|
84
|
+
const code = ch.charCodeAt(0);
|
|
85
|
+
return code >= 2325 && code <= 2361;
|
|
86
|
+
}
|
|
87
|
+
function stem(word) {
|
|
88
|
+
for (const suffix of ALL_SUFFIXES) {
|
|
89
|
+
if (!word.endsWith(suffix)) continue;
|
|
90
|
+
const remaining = word.slice(0, word.length - suffix.length);
|
|
91
|
+
if (remaining.length < 2) continue;
|
|
92
|
+
const lastChar = remaining[remaining.length - 1];
|
|
93
|
+
if (isConsonant(lastChar)) return remaining;
|
|
94
|
+
}
|
|
95
|
+
return word;
|
|
96
|
+
}
|
|
97
|
+
var stopWords = /* @__PURE__ */ new Set([
|
|
98
|
+
"\u0905\u0902\u0926\u0930",
|
|
99
|
+
"\u0905\u0924",
|
|
100
|
+
"\u0905\u0926\u093F",
|
|
101
|
+
"\u0905\u092A",
|
|
102
|
+
"\u0905\u092A\u0928\u093E",
|
|
103
|
+
"\u0905\u092A\u0928\u093F",
|
|
104
|
+
"\u0905\u092A\u0928\u0940",
|
|
105
|
+
"\u0905\u092A\u0928\u0947",
|
|
106
|
+
"\u0905\u092D\u093F",
|
|
107
|
+
"\u0905\u092D\u0940",
|
|
108
|
+
"\u0906\u0926\u093F",
|
|
109
|
+
"\u0906\u092A",
|
|
110
|
+
"\u0907\u0902\u0939\u093F\u0902",
|
|
111
|
+
"\u0907\u0902\u0939\u0947\u0902",
|
|
112
|
+
"\u0907\u0902\u0939\u094B\u0902",
|
|
113
|
+
"\u0907\u0924\u092F\u093E\u0926\u093F",
|
|
114
|
+
"\u0907\u0924\u094D\u092F\u093E\u0926\u093F",
|
|
115
|
+
"\u0907\u0928",
|
|
116
|
+
"\u0907\u0928\u0915\u093E",
|
|
117
|
+
"\u0907\u0928\u094D\u0939\u0940\u0902",
|
|
118
|
+
"\u0907\u0928\u094D\u0939\u0947\u0902",
|
|
119
|
+
"\u0907\u0928\u094D\u0939\u094B\u0902",
|
|
120
|
+
"\u0907\u0938",
|
|
121
|
+
"\u0907\u0938\u0915\u093E",
|
|
122
|
+
"\u0907\u0938\u0915\u093F",
|
|
123
|
+
"\u0907\u0938\u0915\u0940",
|
|
124
|
+
"\u0907\u0938\u0915\u0947",
|
|
125
|
+
"\u0907\u0938\u092E\u0947\u0902",
|
|
126
|
+
"\u0907\u0938\u093F",
|
|
127
|
+
"\u0907\u0938\u0940",
|
|
128
|
+
"\u0907\u0938\u0947",
|
|
129
|
+
"\u0909\u0902\u0939\u093F\u0902",
|
|
130
|
+
"\u0909\u0902\u0939\u0947\u0902",
|
|
131
|
+
"\u0909\u0902\u0939\u094B\u0902",
|
|
132
|
+
"\u0909\u0928",
|
|
133
|
+
"\u0909\u0928\u0915\u093E",
|
|
134
|
+
"\u0909\u0928\u0915\u093F",
|
|
135
|
+
"\u0909\u0928\u0915\u0940",
|
|
136
|
+
"\u0909\u0928\u0915\u0947",
|
|
137
|
+
"\u0909\u0928\u0915\u094B",
|
|
138
|
+
"\u0909\u0928\u094D\u0939\u0940\u0902",
|
|
139
|
+
"\u0909\u0928\u094D\u0939\u0947\u0902",
|
|
140
|
+
"\u0909\u0928\u094D\u0939\u094B\u0902",
|
|
141
|
+
"\u0909\u0938",
|
|
142
|
+
"\u0909\u0938\u0915\u0947",
|
|
143
|
+
"\u0909\u0938\u093F",
|
|
144
|
+
"\u0909\u0938\u0940",
|
|
145
|
+
"\u0909\u0938\u0947",
|
|
146
|
+
"\u090F\u0915",
|
|
147
|
+
"\u090F\u0935\u0902",
|
|
148
|
+
"\u090F\u0938",
|
|
149
|
+
"\u090F\u0938\u0947",
|
|
150
|
+
"\u0910\u0938\u0947",
|
|
151
|
+
"\u0913\u0930",
|
|
152
|
+
"\u0914\u0930",
|
|
153
|
+
"\u0915\u0907",
|
|
154
|
+
"\u0915\u0908",
|
|
155
|
+
"\u0915\u0930",
|
|
156
|
+
"\u0915\u0930\u0924\u093E",
|
|
157
|
+
"\u0915\u0930\u0924\u0947",
|
|
158
|
+
"\u0915\u0930\u0928\u093E",
|
|
159
|
+
"\u0915\u0930\u0928\u0947",
|
|
160
|
+
"\u0915\u0930\u0947\u0902",
|
|
161
|
+
"\u0915\u0939\u0924\u0947",
|
|
162
|
+
"\u0915\u0939\u093E",
|
|
163
|
+
"\u0915\u093E",
|
|
164
|
+
"\u0915\u093E\u092B\u093C\u0940",
|
|
165
|
+
"\u0915\u093E\u092B\u093F",
|
|
166
|
+
"\u0915\u093E\u095E\u0940",
|
|
167
|
+
"\u0915\u093F",
|
|
168
|
+
"\u0915\u093F\u0902\u0939\u0947\u0902",
|
|
169
|
+
"\u0915\u093F\u0902\u0939\u094B\u0902",
|
|
170
|
+
"\u0915\u093F\u0924\u0928\u093E",
|
|
171
|
+
"\u0915\u093F\u0928\u094D\u0939\u0947\u0902",
|
|
172
|
+
"\u0915\u093F\u0928\u094D\u0939\u094B\u0902",
|
|
173
|
+
"\u0915\u093F\u092F\u093E",
|
|
174
|
+
"\u0915\u093F\u0930",
|
|
175
|
+
"\u0915\u093F\u0938",
|
|
176
|
+
"\u0915\u093F\u0938\u093F",
|
|
177
|
+
"\u0915\u093F\u0938\u0940",
|
|
178
|
+
"\u0915\u093F\u0938\u0947",
|
|
179
|
+
"\u0915\u0940",
|
|
180
|
+
"\u0915\u0941\u091B",
|
|
181
|
+
"\u0915\u0941\u0932",
|
|
182
|
+
"\u0915\u0947",
|
|
183
|
+
"\u0915\u094B",
|
|
184
|
+
"\u0915\u094B\u0907",
|
|
185
|
+
"\u0915\u094B\u0908",
|
|
186
|
+
"\u0915\u094B\u0928",
|
|
187
|
+
"\u0915\u094B\u0928\u0938\u093E",
|
|
188
|
+
"\u0915\u094C\u0928",
|
|
189
|
+
"\u0915\u094C\u0928\u0938\u093E",
|
|
190
|
+
"\u0917\u092F\u093E",
|
|
191
|
+
"\u0918\u0930",
|
|
192
|
+
"\u091C\u092C",
|
|
193
|
+
"\u091C\u0939\u093E\u0901",
|
|
194
|
+
"\u091C\u0939\u093E\u0902",
|
|
195
|
+
"\u091C\u093E",
|
|
196
|
+
"\u091C\u093F\u0902\u0939\u0947\u0902",
|
|
197
|
+
"\u091C\u093F\u0902\u0939\u094B\u0902",
|
|
198
|
+
"\u091C\u093F\u0924\u0928\u093E",
|
|
199
|
+
"\u091C\u093F\u0927\u0930",
|
|
200
|
+
"\u091C\u093F\u0928",
|
|
201
|
+
"\u091C\u093F\u0928\u094D\u0939\u0947\u0902",
|
|
202
|
+
"\u091C\u093F\u0928\u094D\u0939\u094B\u0902",
|
|
203
|
+
"\u091C\u093F\u0938",
|
|
204
|
+
"\u091C\u093F\u0938\u0947",
|
|
205
|
+
"\u091C\u0940\u0927\u0930",
|
|
206
|
+
"\u091C\u0947\u0938\u093E",
|
|
207
|
+
"\u091C\u0947\u0938\u0947",
|
|
208
|
+
"\u091C\u0948\u0938\u093E",
|
|
209
|
+
"\u091C\u0948\u0938\u0947",
|
|
210
|
+
"\u091C\u094B",
|
|
211
|
+
"\u0924\u0915",
|
|
212
|
+
"\u0924\u092C",
|
|
213
|
+
"\u0924\u0930\u0939",
|
|
214
|
+
"\u0924\u093F\u0902\u0939\u0947\u0902",
|
|
215
|
+
"\u0924\u093F\u0902\u0939\u094B\u0902",
|
|
216
|
+
"\u0924\u093F\u0928",
|
|
217
|
+
"\u0924\u093F\u0928\u094D\u0939\u0947\u0902",
|
|
218
|
+
"\u0924\u093F\u0928\u094D\u0939\u094B\u0902",
|
|
219
|
+
"\u0924\u093F\u0938",
|
|
220
|
+
"\u0924\u093F\u0938\u0947",
|
|
221
|
+
"\u0924\u094B",
|
|
222
|
+
"\u0925\u093E",
|
|
223
|
+
"\u0925\u093F",
|
|
224
|
+
"\u0925\u0940",
|
|
225
|
+
"\u0925\u0947",
|
|
226
|
+
"\u0926\u092C\u093E\u0930\u093E",
|
|
227
|
+
"\u0926\u0935\u093E\u0930\u093E",
|
|
228
|
+
"\u0926\u093F\u092F\u093E",
|
|
229
|
+
"\u0926\u0941\u0938\u0930\u093E",
|
|
230
|
+
"\u0926\u0941\u0938\u0930\u0947",
|
|
231
|
+
"\u0926\u0942\u0938\u0930\u0947",
|
|
232
|
+
"\u0926\u094B",
|
|
233
|
+
"\u0926\u094D\u0935\u093E\u0930\u093E",
|
|
234
|
+
"\u0928",
|
|
235
|
+
"\u0928\u0939\u093F\u0902",
|
|
236
|
+
"\u0928\u0939\u0940\u0902",
|
|
237
|
+
"\u0928\u093E",
|
|
238
|
+
"\u0928\u093F\u091A\u0947",
|
|
239
|
+
"\u0928\u093F\u0939\u093E\u092F\u0924",
|
|
240
|
+
"\u0928\u0940\u091A\u0947",
|
|
241
|
+
"\u0928\u0947",
|
|
242
|
+
"\u092A\u0930",
|
|
243
|
+
"\u092A\u0939\u0932\u0947",
|
|
244
|
+
"\u092A\u0941\u0930\u093E",
|
|
245
|
+
"\u092A\u0942\u0930\u093E",
|
|
246
|
+
"\u092A\u0947",
|
|
247
|
+
"\u092B\u093F\u0930",
|
|
248
|
+
"\u092C\u0928\u093F",
|
|
249
|
+
"\u092C\u0928\u0940",
|
|
250
|
+
"\u092C\u0939\u093F",
|
|
251
|
+
"\u092C\u0939\u0940",
|
|
252
|
+
"\u092C\u0939\u0941\u0924",
|
|
253
|
+
"\u092C\u093E\u0926",
|
|
254
|
+
"\u092C\u093E\u0932\u093E",
|
|
255
|
+
"\u092C\u093F\u0932\u0915\u0941\u0932",
|
|
256
|
+
"\u092D\u093F",
|
|
257
|
+
"\u092D\u093F\u0924\u0930",
|
|
258
|
+
"\u092D\u0940",
|
|
259
|
+
"\u092D\u0940\u0924\u0930",
|
|
260
|
+
"\u092E\u0917\u0930",
|
|
261
|
+
"\u092E\u093E\u0928\u094B",
|
|
262
|
+
"\u092E\u0947",
|
|
263
|
+
"\u092E\u0947\u0902",
|
|
264
|
+
"\u092F\u0926\u093F",
|
|
265
|
+
"\u092F\u0939",
|
|
266
|
+
"\u092F\u0939\u093E\u0901",
|
|
267
|
+
"\u092F\u0939\u093E\u0902",
|
|
268
|
+
"\u092F\u0939\u093F",
|
|
269
|
+
"\u092F\u0939\u0940",
|
|
270
|
+
"\u092F\u093E",
|
|
271
|
+
"\u092F\u093F\u0939",
|
|
272
|
+
"\u092F\u0947",
|
|
273
|
+
"\u0930\u0916\u0947\u0902",
|
|
274
|
+
"\u0930\u0935\u093E\u0938\u093E",
|
|
275
|
+
"\u0930\u0939\u093E",
|
|
276
|
+
"\u0930\u0939\u0947",
|
|
277
|
+
"\u0931\u094D\u0935\u093E\u0938\u093E",
|
|
278
|
+
"\u0932\u093F\u090F",
|
|
279
|
+
"\u0932\u093F\u092F\u0947",
|
|
280
|
+
"\u0932\u0947\u0915\u093F\u0928",
|
|
281
|
+
"\u0935",
|
|
282
|
+
"\u0935\u0917\u093C\u0948\u0930\u0939",
|
|
283
|
+
"\u0935\u0917\u0947\u0930\u0939",
|
|
284
|
+
"\u0935\u0930\u0917",
|
|
285
|
+
"\u0935\u0930\u094D\u0917",
|
|
286
|
+
"\u0935\u0939",
|
|
287
|
+
"\u0935\u0939\u093E\u0901",
|
|
288
|
+
"\u0935\u0939\u093E\u0902",
|
|
289
|
+
"\u0935\u0939\u093F\u0902",
|
|
290
|
+
"\u0935\u0939\u0940\u0902",
|
|
291
|
+
"\u0935\u093E\u0932\u0947",
|
|
292
|
+
"\u0935\u0941\u0939",
|
|
293
|
+
"\u0935\u0947",
|
|
294
|
+
"\u0935\u095A\u0948\u0930\u0939",
|
|
295
|
+
"\u0938\u0902\u0917",
|
|
296
|
+
"\u0938\u0915\u0924\u093E",
|
|
297
|
+
"\u0938\u0915\u0924\u0947",
|
|
298
|
+
"\u0938\u092C\u0938\u0947",
|
|
299
|
+
"\u0938\u092D\u093F",
|
|
300
|
+
"\u0938\u092D\u0940",
|
|
301
|
+
"\u0938\u093E\u0925",
|
|
302
|
+
"\u0938\u093E\u092C\u0941\u0924",
|
|
303
|
+
"\u0938\u093E\u092D",
|
|
304
|
+
"\u0938\u093E\u0930\u093E",
|
|
305
|
+
"\u0938\u0947",
|
|
306
|
+
"\u0938\u094B",
|
|
307
|
+
"\u0939\u093F",
|
|
308
|
+
"\u0939\u0940",
|
|
309
|
+
"\u0939\u0941\u0905",
|
|
310
|
+
"\u0939\u0941\u0906",
|
|
311
|
+
"\u0939\u0941\u0907",
|
|
312
|
+
"\u0939\u0941\u0908",
|
|
313
|
+
"\u0939\u0941\u090F",
|
|
314
|
+
"\u0939\u0947",
|
|
315
|
+
"\u0939\u0947\u0902",
|
|
316
|
+
"\u0939\u0948",
|
|
317
|
+
"\u0939\u0948\u0902",
|
|
318
|
+
"\u0939\u094B",
|
|
319
|
+
"\u0939\u094B\u0924\u093E",
|
|
320
|
+
"\u0939\u094B\u0924\u093F",
|
|
321
|
+
"\u0939\u094B\u0924\u0940",
|
|
322
|
+
"\u0939\u094B\u0924\u0947",
|
|
323
|
+
"\u0939\u094B\u0928\u093E",
|
|
324
|
+
"\u0939\u094B\u0928\u0947"
|
|
325
|
+
]);
|
|
326
|
+
var hindi = {
|
|
327
|
+
name: "hindi",
|
|
328
|
+
stemmer: stem,
|
|
329
|
+
stopWords,
|
|
330
|
+
tokenizer: { splitPattern: /[^\u0900-\u097fa-z0-9]+/gi }
|
|
331
|
+
};
|
|
332
|
+
|
|
333
|
+
export { hindi };
|