@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 +1,306 @@
|
|
|
1
|
+
// src/languages/armenian.ts
|
|
2
|
+
var VOWELS = "\u0561\u0565\u0567\u056B\u0578\u0585";
|
|
3
|
+
function isVowel(ch) {
|
|
4
|
+
return VOWELS.includes(ch);
|
|
5
|
+
}
|
|
6
|
+
function findR2(word) {
|
|
7
|
+
let r2 = word.length;
|
|
8
|
+
let vcvcCount = 0;
|
|
9
|
+
let sawVowel = false;
|
|
10
|
+
for (let i = 0; i < word.length; i++) {
|
|
11
|
+
const v = isVowel(word[i]);
|
|
12
|
+
if (v && !sawVowel) {
|
|
13
|
+
sawVowel = true;
|
|
14
|
+
} else if (!v && sawVowel) {
|
|
15
|
+
vcvcCount++;
|
|
16
|
+
sawVowel = false;
|
|
17
|
+
if (vcvcCount === 2) {
|
|
18
|
+
r2 = i + 1;
|
|
19
|
+
break;
|
|
20
|
+
}
|
|
21
|
+
}
|
|
22
|
+
}
|
|
23
|
+
return r2;
|
|
24
|
+
}
|
|
25
|
+
function removeSuffixInRegion(word, suffixes, regionStart) {
|
|
26
|
+
for (const suffix of suffixes) {
|
|
27
|
+
if (word.endsWith(suffix) && word.length - suffix.length >= regionStart) {
|
|
28
|
+
return word.slice(0, word.length - suffix.length);
|
|
29
|
+
}
|
|
30
|
+
}
|
|
31
|
+
return null;
|
|
32
|
+
}
|
|
33
|
+
var ENDING_SUFFIXES = [
|
|
34
|
+
"\u0565\u0580\u0578\u0580\u0564",
|
|
35
|
+
"\u0580\u0578\u0580\u0564",
|
|
36
|
+
"\u0578\u0580\u0561\u056F",
|
|
37
|
+
"\u0561\u056F\u0561\u0576",
|
|
38
|
+
"\u0561\u0580\u0561\u0576",
|
|
39
|
+
"\u0565\u056F\u0565\u0576",
|
|
40
|
+
"\u0565\u0580\u0565\u0576",
|
|
41
|
+
"\u0578\u0580\u0567\u0576",
|
|
42
|
+
"\u0578\u057E\u056B\u0576",
|
|
43
|
+
"\u056C\u0561\u0575\u0576",
|
|
44
|
+
"\u057E\u0578\u0582\u0576",
|
|
45
|
+
"\u0561\u057E\u0565\u057F",
|
|
46
|
+
"\u0563\u056B\u0576",
|
|
47
|
+
"\u0561\u056C\u056B",
|
|
48
|
+
"\u0561\u056F\u056B",
|
|
49
|
+
"\u057A\u0565\u057D",
|
|
50
|
+
"\u056F\u0578\u057F",
|
|
51
|
+
"\u0562\u0561\u0580",
|
|
52
|
+
"\u0565\u0572",
|
|
53
|
+
"\u0565\u0576",
|
|
54
|
+
"\u056B\u0576",
|
|
55
|
+
"\u056B\u057E",
|
|
56
|
+
"\u0561\u057F"
|
|
57
|
+
];
|
|
58
|
+
var VERB_SUFFIXES = [
|
|
59
|
+
"\u0561\u0581\u0580\u056B\u0576\u0584",
|
|
60
|
+
"\u057E\u0565\u0581\u056B\u0576\u0584",
|
|
61
|
+
"\u0565\u0581\u056B\u0576\u0584",
|
|
62
|
+
"\u0561\u0581\u056B\u0576\u0584",
|
|
63
|
+
"\u0561\u0581\u0561\u0576\u0584",
|
|
64
|
+
"\u0561\u0581\u0580\u056B\u0584",
|
|
65
|
+
"\u057E\u0565\u0581\u056B\u0584",
|
|
66
|
+
"\u0565\u0581\u0576\u0565\u056C",
|
|
67
|
+
"\u0561\u0581\u057E\u0565\u056C",
|
|
68
|
+
"\u0565\u0581\u057E\u0565\u056C",
|
|
69
|
+
"\u0561\u0581\u0576\u0561\u056C",
|
|
70
|
+
"\u057E\u0565\u0581\u056B\u0580",
|
|
71
|
+
"\u0561\u0581\u0580\u056B\u0580",
|
|
72
|
+
"\u0561\u0581\u0580\u0565\u0581",
|
|
73
|
+
"\u0565\u056C\u0578\u0582\u0581",
|
|
74
|
+
"\u0561\u056C\u0578\u0582\u0581",
|
|
75
|
+
"\u056F\u0578\u057F\u0565\u056C",
|
|
76
|
+
"\u0561\u0581\u0561\u0584",
|
|
77
|
+
"\u0565\u0581\u056B\u0584",
|
|
78
|
+
"\u0561\u0581\u056B\u0584",
|
|
79
|
+
"\u0565\u0581\u056B\u0580",
|
|
80
|
+
"\u0561\u0581\u056B\u0580",
|
|
81
|
+
"\u057E\u0565\u0581\u056B",
|
|
82
|
+
"\u0561\u0581\u0580\u056B",
|
|
83
|
+
"\u0561\u056C\u056B\u057D",
|
|
84
|
+
"\u0565\u056C\u056B\u057D",
|
|
85
|
+
"\u0561\u0581\u0561\u0580",
|
|
86
|
+
"\u0565\u0581\u0561\u0580",
|
|
87
|
+
"\u0561\u0581\u0561\u057E",
|
|
88
|
+
"\u0565\u0581\u0561\u057E",
|
|
89
|
+
"\u0561\u0581\u0561\u0576",
|
|
90
|
+
"\u057E\u0578\u0582\u0574",
|
|
91
|
+
"\u0561\u056C\u0578\u057E",
|
|
92
|
+
"\u0565\u056C\u0578\u057E",
|
|
93
|
+
"\u0561\u056C\u0578\u0582",
|
|
94
|
+
"\u0565\u056C\u0578\u0582",
|
|
95
|
+
"\u0561\u0576\u0561\u056C",
|
|
96
|
+
"\u0565\u0576\u0561\u056C",
|
|
97
|
+
"\u0581\u0576\u0565\u056C",
|
|
98
|
+
"\u0561\u057F\u0565\u056C",
|
|
99
|
+
"\u0578\u057F\u0565\u056C",
|
|
100
|
+
"\u0581\u0561\u0576\u0584",
|
|
101
|
+
"\u0565\u0581\u056B",
|
|
102
|
+
"\u0561\u0581\u056B",
|
|
103
|
+
"\u0561\u0581\u0561",
|
|
104
|
+
"\u0565\u0581\u0561",
|
|
105
|
+
"\u0578\u0582\u0574",
|
|
106
|
+
"\u0568\u0561\u056C",
|
|
107
|
+
"\u0579\u0565\u056C",
|
|
108
|
+
"\u057E\u0565\u056C",
|
|
109
|
+
"\u057F\u0565\u056C",
|
|
110
|
+
"\u0568\u0565\u056C",
|
|
111
|
+
"\u0576\u0565\u056C",
|
|
112
|
+
"\u057E\u0561\u056E",
|
|
113
|
+
"\u0581\u0561\u0576",
|
|
114
|
+
"\u0561\u0576\u0584",
|
|
115
|
+
"\u0581\u0561\u0584",
|
|
116
|
+
"\u0561\u056C",
|
|
117
|
+
"\u0565\u056C",
|
|
118
|
+
"\u0561\u0576",
|
|
119
|
+
"\u0561\u057E",
|
|
120
|
+
"\u0561\u0580",
|
|
121
|
+
"\u0561\u0581",
|
|
122
|
+
"\u0565\u0581",
|
|
123
|
+
"\u0561\u0584",
|
|
124
|
+
"\u057E\u0565",
|
|
125
|
+
"\u0561"
|
|
126
|
+
];
|
|
127
|
+
var ADJECTIVE_SUFFIXES = [
|
|
128
|
+
"\u0578\u0582\u0569\u0575\u0578\u0582\u0576",
|
|
129
|
+
"\u0574\u0578\u0582\u0576\u0584",
|
|
130
|
+
"\u0561\u0576\u0585\u0581",
|
|
131
|
+
"\u057E\u0561\u056E\u0584",
|
|
132
|
+
"\u0578\u0582\u0575\u0584",
|
|
133
|
+
"\u057D\u057F\u0561\u0576",
|
|
134
|
+
"\u0561\u0580\u0561\u0576",
|
|
135
|
+
"\u0565\u0572\u0567\u0576",
|
|
136
|
+
"\u0575\u0578\u0582\u0576",
|
|
137
|
+
"\u0561\u0576\u0561\u056F",
|
|
138
|
+
"\u0578\u0582\u0570\u056B",
|
|
139
|
+
"\u0578\u0582\u0575\u0569",
|
|
140
|
+
"\u0578\u0582\u0576\u0584",
|
|
141
|
+
"\u0561\u056C\u056B\u0584",
|
|
142
|
+
"\u0561\u0576\u056B\u0584",
|
|
143
|
+
"\u0578\u0582\u057D\u057F",
|
|
144
|
+
"\u0561\u057E\u0578\u0580",
|
|
145
|
+
"\u0575\u0561\u056F",
|
|
146
|
+
"\u0579\u0565\u0584",
|
|
147
|
+
"\u0578\u0576\u0584",
|
|
148
|
+
"\u0565\u0576\u0584",
|
|
149
|
+
"\u0561\u0580\u0584",
|
|
150
|
+
"\u056B\u0579\u0584",
|
|
151
|
+
"\u0578\u0582\u057D",
|
|
152
|
+
"\u0578\u0582\u056F",
|
|
153
|
+
"\u0561\u056E\u0578",
|
|
154
|
+
"\u057A\u0561\u0576",
|
|
155
|
+
"\u0563\u0561\u0580",
|
|
156
|
+
"\u057E\u0578\u0580",
|
|
157
|
+
"\u0578\u0580\u0564",
|
|
158
|
+
"\u0581\u056B",
|
|
159
|
+
"\u056B\u056C",
|
|
160
|
+
"\u056B\u056F",
|
|
161
|
+
"\u0561\u056F",
|
|
162
|
+
"\u0561\u0576",
|
|
163
|
+
"\u0578\u0581",
|
|
164
|
+
"\u0578\u0582",
|
|
165
|
+
"\u056B\u0579",
|
|
166
|
+
"\u056B\u0584",
|
|
167
|
+
"\u0584"
|
|
168
|
+
];
|
|
169
|
+
var NOUN_SUFFIXES = [
|
|
170
|
+
"\u0578\u0582\u0569\u0575\u0561\u0576\u0564",
|
|
171
|
+
"\u0578\u0582\u0569\u0575\u0561\u0576\u0576",
|
|
172
|
+
"\u0578\u0582\u0569\u0575\u0561\u0576\u057D",
|
|
173
|
+
"\u0578\u0582\u0569\u0575\u0561\u0576\u0568",
|
|
174
|
+
"\u0578\u0582\u0569\u0575\u0561\u0576",
|
|
175
|
+
"\u0576\u0565\u0580\u0578\u0582\u0574",
|
|
176
|
+
"\u0561\u0576\u0578\u0582\u0574",
|
|
177
|
+
"\u0565\u0580\u0578\u0582\u0574",
|
|
178
|
+
"\u057E\u0561\u0576\u056B\u0581",
|
|
179
|
+
"\u0576\u0565\u0580\u056B\u0576",
|
|
180
|
+
"\u0576\u0565\u0580\u0578\u057E",
|
|
181
|
+
"\u0576\u0565\u0580\u056B\u0581",
|
|
182
|
+
"\u057E\u0561\u0576\u0564",
|
|
183
|
+
"\u057E\u0561\u0576\u0568",
|
|
184
|
+
"\u057E\u0561\u0576\u057D",
|
|
185
|
+
"\u0576\u0565\u0580\u0564",
|
|
186
|
+
"\u0576\u0565\u0580\u0568",
|
|
187
|
+
"\u0576\u0565\u0580\u056B",
|
|
188
|
+
"\u0576\u0565\u0580\u0576",
|
|
189
|
+
"\u0561\u0576\u0578\u057E",
|
|
190
|
+
"\u0565\u0580\u0578\u057E",
|
|
191
|
+
"\u0565\u0580\u056B\u0581",
|
|
192
|
+
"\u0578\u057B\u056B\u0581",
|
|
193
|
+
"\u057E\u0561\u0576",
|
|
194
|
+
"\u0576\u0565\u0580",
|
|
195
|
+
"\u057E\u0578\u057E",
|
|
196
|
+
"\u057E\u056B\u0581",
|
|
197
|
+
"\u0581\u056B\u0581",
|
|
198
|
+
"\u0565\u0580\u056B",
|
|
199
|
+
"\u0565\u0580\u0564",
|
|
200
|
+
"\u0565\u0580\u0568",
|
|
201
|
+
"\u0565\u0580\u0576",
|
|
202
|
+
"\u0578\u057B\u0564",
|
|
203
|
+
"\u0578\u057B\u0568",
|
|
204
|
+
"\u0578\u057B\u057D",
|
|
205
|
+
"\u0578\u0582\u0564",
|
|
206
|
+
"\u0561\u0576\u0564",
|
|
207
|
+
"\u0561\u0576\u0568",
|
|
208
|
+
"\u0578\u0582\u0576",
|
|
209
|
+
"\u0578\u0582\u0581",
|
|
210
|
+
"\u0561\u0574\u0562",
|
|
211
|
+
"\u056B\u0576",
|
|
212
|
+
"\u0578\u057B",
|
|
213
|
+
"\u0578\u057E",
|
|
214
|
+
"\u0565\u0580",
|
|
215
|
+
"\u056B\u0581",
|
|
216
|
+
"\u0578\u0581",
|
|
217
|
+
"\u057D\u0561",
|
|
218
|
+
"\u057E\u0561",
|
|
219
|
+
"\u0561\u0576",
|
|
220
|
+
"\u057E\u056B",
|
|
221
|
+
"\u0576",
|
|
222
|
+
"\u0564",
|
|
223
|
+
"\u0568",
|
|
224
|
+
"\u056B",
|
|
225
|
+
"\u057D",
|
|
226
|
+
"\u0581"
|
|
227
|
+
];
|
|
228
|
+
function stem(word) {
|
|
229
|
+
if (word.length < 3) return word;
|
|
230
|
+
const r2 = findR2(word);
|
|
231
|
+
if (word.length <= r2) return word;
|
|
232
|
+
const nounResult = removeSuffixInRegion(word, NOUN_SUFFIXES, r2);
|
|
233
|
+
if (nounResult !== null) {
|
|
234
|
+
word = nounResult;
|
|
235
|
+
} else {
|
|
236
|
+
const verbResult = removeSuffixInRegion(word, VERB_SUFFIXES, r2);
|
|
237
|
+
if (verbResult !== null) {
|
|
238
|
+
word = verbResult;
|
|
239
|
+
} else {
|
|
240
|
+
const adjResult = removeSuffixInRegion(word, ADJECTIVE_SUFFIXES, r2);
|
|
241
|
+
if (adjResult !== null) {
|
|
242
|
+
word = adjResult;
|
|
243
|
+
}
|
|
244
|
+
}
|
|
245
|
+
}
|
|
246
|
+
const endingResult = removeSuffixInRegion(word, ENDING_SUFFIXES, r2);
|
|
247
|
+
if (endingResult !== null) {
|
|
248
|
+
word = endingResult;
|
|
249
|
+
}
|
|
250
|
+
return word;
|
|
251
|
+
}
|
|
252
|
+
var stopWords = /* @__PURE__ */ new Set([
|
|
253
|
+
"\u0561\u0575\u0564",
|
|
254
|
+
"\u0561\u0575\u056C",
|
|
255
|
+
"\u0561\u0575\u0576",
|
|
256
|
+
"\u0561\u0575\u057D",
|
|
257
|
+
"\u0564\u0578\u0582",
|
|
258
|
+
"\u0564\u0578\u0582\u0584",
|
|
259
|
+
"\u0565\u0574",
|
|
260
|
+
"\u0565\u0576",
|
|
261
|
+
"\u0565\u0576\u0584",
|
|
262
|
+
"\u0565\u057D",
|
|
263
|
+
"\u0565\u0584",
|
|
264
|
+
"\u0567",
|
|
265
|
+
"\u0567\u056B",
|
|
266
|
+
"\u0567\u056B\u0576",
|
|
267
|
+
"\u0567\u056B\u0576\u0584",
|
|
268
|
+
"\u0567\u056B\u0580",
|
|
269
|
+
"\u0567\u056B\u0584",
|
|
270
|
+
"\u0567\u0580",
|
|
271
|
+
"\u0568\u057D\u057F",
|
|
272
|
+
"\u0569",
|
|
273
|
+
"\u056B",
|
|
274
|
+
"\u056B\u0576",
|
|
275
|
+
"\u056B\u057D\u056F",
|
|
276
|
+
"\u056B\u0580",
|
|
277
|
+
"\u056F\u0561\u0574",
|
|
278
|
+
"\u0570\u0561\u0574\u0561\u0580",
|
|
279
|
+
"\u0570\u0565\u057F",
|
|
280
|
+
"\u0570\u0565\u057F\u0578",
|
|
281
|
+
"\u0574\u0565\u0576\u0584",
|
|
282
|
+
"\u0574\u0565\u057B",
|
|
283
|
+
"\u0574\u056B",
|
|
284
|
+
"\u0576",
|
|
285
|
+
"\u0576\u0561",
|
|
286
|
+
"\u0576\u0561\u0587",
|
|
287
|
+
"\u0576\u0580\u0561",
|
|
288
|
+
"\u0576\u0580\u0561\u0576\u0584",
|
|
289
|
+
"\u0578\u0580",
|
|
290
|
+
"\u0578\u0580\u0568",
|
|
291
|
+
"\u0578\u0580\u0578\u0576\u0584",
|
|
292
|
+
"\u0578\u0580\u057A\u0565\u057D",
|
|
293
|
+
"\u0578\u0582",
|
|
294
|
+
"\u0578\u0582\u0574",
|
|
295
|
+
"\u057A\u056B\u057F\u056B",
|
|
296
|
+
"\u057E\u0580\u0561",
|
|
297
|
+
"\u0587"
|
|
298
|
+
]);
|
|
299
|
+
var armenian = {
|
|
300
|
+
name: "armenian",
|
|
301
|
+
stemmer: stem,
|
|
302
|
+
stopWords,
|
|
303
|
+
tokenizer: { splitPattern: /[^\u0531-\u0587a-z0-9]+/gi }
|
|
304
|
+
};
|
|
305
|
+
|
|
306
|
+
export { armenian };
|
|
@@ -0,0 +1,330 @@
|
|
|
1
|
+
// src/languages/bulgarian.ts
|
|
2
|
+
function stem(word) {
|
|
3
|
+
let len = word.length;
|
|
4
|
+
if (len < 4) return word;
|
|
5
|
+
if (len > 5 && word.endsWith("\u0438\u0449\u0430")) {
|
|
6
|
+
word = word.slice(0, -3);
|
|
7
|
+
len = word.length;
|
|
8
|
+
}
|
|
9
|
+
if (len > 6 && word.endsWith("\u0438\u044F\u0442")) {
|
|
10
|
+
word = word.slice(0, -3);
|
|
11
|
+
len = word.length;
|
|
12
|
+
} else if (len > 5 && (word.endsWith("\u044A\u0442") || word.endsWith("\u0442\u043E") || word.endsWith("\u0442\u0435") || word.endsWith("\u0442\u0430") || word.endsWith("\u0438\u044F"))) {
|
|
13
|
+
word = word.slice(0, -2);
|
|
14
|
+
len = word.length;
|
|
15
|
+
} else if (len > 4 && word.endsWith("\u044F\u0442")) {
|
|
16
|
+
word = word.slice(0, -2);
|
|
17
|
+
len = word.length;
|
|
18
|
+
}
|
|
19
|
+
if (len > 6 && word.endsWith("\u043E\u0432\u0446\u0438")) {
|
|
20
|
+
word = word.slice(0, -3);
|
|
21
|
+
len = word.length;
|
|
22
|
+
} else if (len > 6 && word.endsWith("\u043E\u0432\u0435")) {
|
|
23
|
+
word = word.slice(0, -3);
|
|
24
|
+
len = word.length;
|
|
25
|
+
} else if (len > 6 && word.endsWith("\u0435\u0432\u0435")) {
|
|
26
|
+
word = `${word.slice(0, -3)}\u0439`;
|
|
27
|
+
len = word.length;
|
|
28
|
+
} else if (len > 5 && word.endsWith("\u0438\u0449\u0430")) {
|
|
29
|
+
word = word.slice(0, -3);
|
|
30
|
+
len = word.length;
|
|
31
|
+
} else if (len > 5 && word.endsWith("\u0442\u0430")) {
|
|
32
|
+
word = word.slice(0, -2);
|
|
33
|
+
len = word.length;
|
|
34
|
+
} else if (len > 5 && word.endsWith("\u0446\u0438")) {
|
|
35
|
+
word = `${word.slice(0, -2)}\u043A`;
|
|
36
|
+
len = word.length;
|
|
37
|
+
} else if (len > 5 && word.endsWith("\u0437\u0438")) {
|
|
38
|
+
word = `${word.slice(0, -2)}\u0433`;
|
|
39
|
+
len = word.length;
|
|
40
|
+
} else if (len > 5 && word.length >= 3 && word[word.length - 1] === "\u0438" && word[word.length - 3] === "\u0435") {
|
|
41
|
+
word = `${word.slice(0, -3)}\u044F${word.slice(-2, -1)}`;
|
|
42
|
+
len = word.length;
|
|
43
|
+
} else if (len > 4 && word.endsWith("\u0441\u0438")) {
|
|
44
|
+
word = `${word.slice(0, -2)}\u0445`;
|
|
45
|
+
len = word.length;
|
|
46
|
+
} else if (len > 4 && word.endsWith("\u0438")) {
|
|
47
|
+
word = word.slice(0, -1);
|
|
48
|
+
len = word.length;
|
|
49
|
+
}
|
|
50
|
+
if (len > 3 && (word.endsWith("\u044F") || word.endsWith("\u0430") || word.endsWith("\u043E") || word.endsWith("\u0435"))) {
|
|
51
|
+
word = word.slice(0, -1);
|
|
52
|
+
len = word.length;
|
|
53
|
+
}
|
|
54
|
+
if (len > 4 && word.endsWith("\u0435\u043D")) {
|
|
55
|
+
word = `${word.slice(0, -2)}\u043D`;
|
|
56
|
+
len = word.length;
|
|
57
|
+
}
|
|
58
|
+
if (len > 5 && word[word.length - 2] === "\u044A") {
|
|
59
|
+
word = `${word.slice(0, -2)}${word[word.length - 1]}`;
|
|
60
|
+
}
|
|
61
|
+
return word;
|
|
62
|
+
}
|
|
63
|
+
var stopWords = /* @__PURE__ */ new Set([
|
|
64
|
+
"\u0430",
|
|
65
|
+
"\u0430\u0432\u0442\u0435\u043D\u0442\u0438\u0447\u0435\u043D",
|
|
66
|
+
"\u0430\u0437",
|
|
67
|
+
"\u0430\u043A\u043E",
|
|
68
|
+
"\u0430\u043B\u0430",
|
|
69
|
+
"\u0431\u0435",
|
|
70
|
+
"\u0431\u0435\u0437",
|
|
71
|
+
"\u0431\u0435\u0448\u0435",
|
|
72
|
+
"\u0431\u0438",
|
|
73
|
+
"\u0431\u0438\u0432\u0448",
|
|
74
|
+
"\u0431\u0438\u0432\u0448\u0430",
|
|
75
|
+
"\u0431\u0438\u0432\u0448\u043E",
|
|
76
|
+
"\u0431\u0438\u043B",
|
|
77
|
+
"\u0431\u0438\u043B\u0430",
|
|
78
|
+
"\u0431\u0438\u043B\u0438",
|
|
79
|
+
"\u0431\u0438\u043B\u043E",
|
|
80
|
+
"\u0431\u043B\u0430\u0433\u043E\u0434\u0430\u0440\u044F",
|
|
81
|
+
"\u0431\u043B\u0438\u0437\u043E",
|
|
82
|
+
"\u0431\u044A\u0434\u0430\u0442",
|
|
83
|
+
"\u0431\u044A\u0434\u0435",
|
|
84
|
+
"\u0431\u044F\u0445\u0430",
|
|
85
|
+
"\u0432",
|
|
86
|
+
"\u0432\u0430\u0441",
|
|
87
|
+
"\u0432\u0430\u0448",
|
|
88
|
+
"\u0432\u0430\u0448\u0430",
|
|
89
|
+
"\u0432\u0435\u0440\u043E\u044F\u0442\u043D\u043E",
|
|
90
|
+
"\u0432\u0435\u0447\u0435",
|
|
91
|
+
"\u0432\u0437\u0435\u043C\u0430",
|
|
92
|
+
"\u0432\u0438",
|
|
93
|
+
"\u0432\u0438\u0435",
|
|
94
|
+
"\u0432\u0438\u043D\u0430\u0433\u0438",
|
|
95
|
+
"\u0432\u043D\u0438\u043C\u0430\u0432\u0430",
|
|
96
|
+
"\u0432\u0440\u0435\u043C\u0435",
|
|
97
|
+
"\u0432\u0441\u0435",
|
|
98
|
+
"\u0432\u0441\u0435\u043A\u0438",
|
|
99
|
+
"\u0432\u0441\u0438\u0447\u043A\u0438",
|
|
100
|
+
"\u0432\u0441\u0438\u0447\u043A\u043E",
|
|
101
|
+
"\u0432\u0441\u044F\u043A\u0430",
|
|
102
|
+
"\u0432\u044A\u0432",
|
|
103
|
+
"\u0432\u044A\u043F\u0440\u0435\u043A\u0438",
|
|
104
|
+
"\u0432\u044A\u0440\u0445\u0443",
|
|
105
|
+
"\u0433",
|
|
106
|
+
"\u0433\u0438",
|
|
107
|
+
"\u0433\u043B\u0430\u0432\u0435\u043D",
|
|
108
|
+
"\u0433\u043B\u0430\u0432\u043D\u0430",
|
|
109
|
+
"\u0433\u043B\u0430\u0432\u043D\u043E",
|
|
110
|
+
"\u0433\u043B\u0430\u0441",
|
|
111
|
+
"\u0433\u043E",
|
|
112
|
+
"\u0433\u043E\u0434\u0438\u043D\u0430",
|
|
113
|
+
"\u0433\u043E\u0434\u0438\u043D\u0438",
|
|
114
|
+
"\u0433\u043E\u0434\u0438\u0448\u0435\u043D",
|
|
115
|
+
"\u0434",
|
|
116
|
+
"\u0434\u0430",
|
|
117
|
+
"\u0434\u0430\u043B\u0438",
|
|
118
|
+
"\u0434\u0432\u0430",
|
|
119
|
+
"\u0434\u0432\u0430\u043C\u0430",
|
|
120
|
+
"\u0434\u0432\u0430\u043C\u0430\u0442\u0430",
|
|
121
|
+
"\u0434\u0432\u0435",
|
|
122
|
+
"\u0434\u0432\u0435\u0442\u0435",
|
|
123
|
+
"\u0434\u0435\u043D",
|
|
124
|
+
"\u0434\u043D\u0435\u0441",
|
|
125
|
+
"\u0434\u043D\u0438",
|
|
126
|
+
"\u0434\u043E",
|
|
127
|
+
"\u0434\u043E\u0431\u0440\u0430",
|
|
128
|
+
"\u0434\u043E\u0431\u0440\u0435",
|
|
129
|
+
"\u0434\u043E\u0431\u0440\u043E",
|
|
130
|
+
"\u0434\u043E\u0431\u044A\u0440",
|
|
131
|
+
"\u0434\u043E\u043A\u0430\u0442\u043E",
|
|
132
|
+
"\u0434\u043E\u043A\u043E\u0433\u0430",
|
|
133
|
+
"\u0434\u043E\u0440\u0438",
|
|
134
|
+
"\u0434\u043E\u0441\u0435\u0433\u0430",
|
|
135
|
+
"\u0434\u043E\u0441\u0442\u0430",
|
|
136
|
+
"\u0434\u0440\u0443\u0433",
|
|
137
|
+
"\u0434\u0440\u0443\u0433\u0430",
|
|
138
|
+
"\u0434\u0440\u0443\u0433\u0438",
|
|
139
|
+
"\u0435",
|
|
140
|
+
"\u0435\u0432\u0442\u0438\u043D",
|
|
141
|
+
"\u0435\u0434\u0432\u0430",
|
|
142
|
+
"\u0435\u0434\u0438\u043D",
|
|
143
|
+
"\u0435\u0434\u043D\u0430",
|
|
144
|
+
"\u0435\u0434\u043D\u0430\u043A\u0432\u0430",
|
|
145
|
+
"\u0435\u0434\u043D\u0430\u043A\u0432\u0438",
|
|
146
|
+
"\u0435\u0434\u043D\u0430\u043A\u044A\u0432",
|
|
147
|
+
"\u0435\u0434\u043D\u043E",
|
|
148
|
+
"\u0435\u043A\u0438\u043F",
|
|
149
|
+
"\u0435\u0442\u043E",
|
|
150
|
+
"\u0436\u0438\u0432\u043E\u0442",
|
|
151
|
+
"\u0437\u0430",
|
|
152
|
+
"\u0437\u0430\u0431\u0430\u0432\u044F\u043C",
|
|
153
|
+
"\u0437\u0430\u0434",
|
|
154
|
+
"\u0437\u0430\u0435\u0434\u043D\u043E",
|
|
155
|
+
"\u0437\u0430\u0440\u0430\u0434\u0438",
|
|
156
|
+
"\u0437\u0430\u0441\u0435\u0433\u0430",
|
|
157
|
+
"\u0437\u0430\u0441\u043F\u0430\u043B",
|
|
158
|
+
"\u0437\u0430\u0442\u043E\u0432\u0430",
|
|
159
|
+
"\u0437\u0430\u0449\u043E",
|
|
160
|
+
"\u0437\u0430\u0449\u043E\u0442\u043E",
|
|
161
|
+
"\u0438",
|
|
162
|
+
"\u0438\u0437",
|
|
163
|
+
"\u0438\u043B\u0438",
|
|
164
|
+
"\u0438\u043C",
|
|
165
|
+
"\u0438\u043C\u0430",
|
|
166
|
+
"\u0438\u043C\u0430\u0442",
|
|
167
|
+
"\u0438\u0441\u043A\u0430",
|
|
168
|
+
"\u0439",
|
|
169
|
+
"\u043A\u0430\u0437\u0430",
|
|
170
|
+
"\u043A\u0430\u043A",
|
|
171
|
+
"\u043A\u0430\u043A\u0432\u0430",
|
|
172
|
+
"\u043A\u0430\u043A\u0432\u043E",
|
|
173
|
+
"\u043A\u0430\u043A\u0442\u043E",
|
|
174
|
+
"\u043A\u0430\u043A\u044A\u0432",
|
|
175
|
+
"\u043A\u0430\u0442\u043E",
|
|
176
|
+
"\u043A\u043E\u0433\u0430",
|
|
177
|
+
"\u043A\u043E\u0433\u0430\u0442\u043E",
|
|
178
|
+
"\u043A\u043E\u0435\u0442\u043E",
|
|
179
|
+
"\u043A\u043E\u0438\u0442\u043E",
|
|
180
|
+
"\u043A\u043E\u0439",
|
|
181
|
+
"\u043A\u043E\u0439\u0442\u043E",
|
|
182
|
+
"\u043A\u043E\u043B\u043A\u043E",
|
|
183
|
+
"\u043A\u043E\u044F\u0442\u043E",
|
|
184
|
+
"\u043A\u044A\u0434\u0435",
|
|
185
|
+
"\u043A\u044A\u0434\u0435\u0442\u043E",
|
|
186
|
+
"\u043A\u044A\u043C",
|
|
187
|
+
"\u043B\u0435\u0441\u0435\u043D",
|
|
188
|
+
"\u043B\u0435\u0441\u043D\u043E",
|
|
189
|
+
"\u043B\u0438",
|
|
190
|
+
"\u043B\u043E\u0448",
|
|
191
|
+
"\u043C",
|
|
192
|
+
"\u043C\u0430\u0439",
|
|
193
|
+
"\u043C\u0430\u043B\u043A\u043E",
|
|
194
|
+
"\u043C\u0435",
|
|
195
|
+
"\u043C\u0435\u0436\u0434\u0443",
|
|
196
|
+
"\u043C\u0435\u043A",
|
|
197
|
+
"\u043C\u0435\u043D",
|
|
198
|
+
"\u043C\u0435\u0441\u0435\u0446",
|
|
199
|
+
"\u043C\u0438",
|
|
200
|
+
"\u043C\u043D\u043E\u0433\u043E",
|
|
201
|
+
"\u043C\u043D\u043E\u0437\u0438\u043D\u0430",
|
|
202
|
+
"\u043C\u043E\u0433\u0430",
|
|
203
|
+
"\u043C\u043E\u0433\u0430\u0442",
|
|
204
|
+
"\u043C\u043E\u0436\u0435",
|
|
205
|
+
"\u043C\u043E\u043A\u044A\u0440",
|
|
206
|
+
"\u043C\u043E\u043B\u044F",
|
|
207
|
+
"\u043C\u043E\u043C\u0435\u043D\u0442\u0430",
|
|
208
|
+
"\u043C\u0443",
|
|
209
|
+
"\u043D",
|
|
210
|
+
"\u043D\u0430",
|
|
211
|
+
"\u043D\u0430\u0434",
|
|
212
|
+
"\u043D\u0430\u0437\u0430\u0434",
|
|
213
|
+
"\u043D\u0430\u0439",
|
|
214
|
+
"\u043D\u0430\u043F\u0440\u0430\u0432\u0438",
|
|
215
|
+
"\u043D\u0430\u043F\u0440\u0435\u0434",
|
|
216
|
+
"\u043D\u0430\u043F\u0440\u0438\u043C\u0435\u0440",
|
|
217
|
+
"\u043D\u0430\u0441",
|
|
218
|
+
"\u043D\u0435",
|
|
219
|
+
"\u043D\u0435\u0433\u043E",
|
|
220
|
+
"\u043D\u0435\u0449\u043E",
|
|
221
|
+
"\u043D\u0435\u044F",
|
|
222
|
+
"\u043D\u0438",
|
|
223
|
+
"\u043D\u0438\u0435",
|
|
224
|
+
"\u043D\u0438\u043A\u043E\u0439",
|
|
225
|
+
"\u043D\u0438\u0442\u043E",
|
|
226
|
+
"\u043D\u0438\u0449\u043E",
|
|
227
|
+
"\u043D\u043E",
|
|
228
|
+
"\u043D\u043E\u0432",
|
|
229
|
+
"\u043D\u043E\u0432\u0430",
|
|
230
|
+
"\u043D\u043E\u0432\u0438",
|
|
231
|
+
"\u043D\u043E\u0432\u0438\u043D\u0430",
|
|
232
|
+
"\u043D\u044F\u043A\u043E\u0438",
|
|
233
|
+
"\u043D\u044F\u043A\u043E\u0439",
|
|
234
|
+
"\u043D\u044F\u043A\u043E\u043B\u043A\u043E",
|
|
235
|
+
"\u043D\u044F\u043C\u0430",
|
|
236
|
+
"\u043E\u0431\u0430\u0447\u0435",
|
|
237
|
+
"\u043E\u043A\u043E\u043B\u043E",
|
|
238
|
+
"\u043E\u0441\u0432\u0435\u043D",
|
|
239
|
+
"\u043E\u0441\u043E\u0431\u0435\u043D\u043E",
|
|
240
|
+
"\u043E\u0442",
|
|
241
|
+
"\u043E\u0442\u0433\u043E\u0440\u0435",
|
|
242
|
+
"\u043E\u0442\u043D\u043E\u0432\u043E",
|
|
243
|
+
"\u043E\u0449\u0435",
|
|
244
|
+
"\u043F\u0430\u043A",
|
|
245
|
+
"\u043F\u043E",
|
|
246
|
+
"\u043F\u043E\u0432\u0435\u0447\u0435",
|
|
247
|
+
"\u043F\u043E\u0432\u0435\u0447\u0435\u0442\u043E",
|
|
248
|
+
"\u043F\u043E\u0434",
|
|
249
|
+
"\u043F\u043E\u043D\u0435",
|
|
250
|
+
"\u043F\u043E\u0440\u0430\u0434\u0438",
|
|
251
|
+
"\u043F\u043E\u0441\u043B\u0435",
|
|
252
|
+
"\u043F\u043E\u0447\u0442\u0438",
|
|
253
|
+
"\u043F\u0440\u0430\u0432\u0438",
|
|
254
|
+
"\u043F\u0440\u0435\u0434",
|
|
255
|
+
"\u043F\u0440\u0435\u0434\u0438",
|
|
256
|
+
"\u043F\u0440\u0435\u0437",
|
|
257
|
+
"\u043F\u0440\u0438",
|
|
258
|
+
"\u043F\u044A\u043A",
|
|
259
|
+
"\u043F\u044A\u0440\u0432\u0430\u0442\u0430",
|
|
260
|
+
"\u043F\u044A\u0440\u0432\u0438",
|
|
261
|
+
"\u043F\u044A\u0440\u0432\u043E",
|
|
262
|
+
"\u043F\u044A\u0442\u0438",
|
|
263
|
+
"\u0440\u0430\u0432\u0435\u043D",
|
|
264
|
+
"\u0440\u0430\u0432\u043D\u0430",
|
|
265
|
+
"\u0441",
|
|
266
|
+
"\u0441\u0430",
|
|
267
|
+
"\u0441\u0430\u043C",
|
|
268
|
+
"\u0441\u0430\u043C\u043E",
|
|
269
|
+
"\u0441\u0435",
|
|
270
|
+
"\u0441\u0435\u0433\u0430",
|
|
271
|
+
"\u0441\u0438",
|
|
272
|
+
"\u0441\u0438\u043D",
|
|
273
|
+
"\u0441\u043A\u043E\u0440\u043E",
|
|
274
|
+
"\u0441\u043B\u0435\u0434",
|
|
275
|
+
"\u0441\u043B\u0435\u0434\u0432\u0430\u0449",
|
|
276
|
+
"\u0441\u043C\u0435",
|
|
277
|
+
"\u0441\u043C\u044F\u0445",
|
|
278
|
+
"\u0441\u043F\u043E\u0440\u0435\u0434",
|
|
279
|
+
"\u0441\u0440\u0435\u0434",
|
|
280
|
+
"\u0441\u0440\u0435\u0449\u0443",
|
|
281
|
+
"\u0441\u0442\u0435",
|
|
282
|
+
"\u0441\u044A\u043C",
|
|
283
|
+
"\u0441\u044A\u0441",
|
|
284
|
+
"\u0441\u044A\u0449\u043E",
|
|
285
|
+
"\u0442",
|
|
286
|
+
"\u0442\u0430\u0437\u0438",
|
|
287
|
+
"\u0442\u0430\u043A\u0430",
|
|
288
|
+
"\u0442\u0430\u043A\u0438\u0432\u0430",
|
|
289
|
+
"\u0442\u0430\u043A\u044A\u0432",
|
|
290
|
+
"\u0442\u0430\u043C",
|
|
291
|
+
"\u0442\u0432\u043E\u0439",
|
|
292
|
+
"\u0442\u0435",
|
|
293
|
+
"\u0442\u0435\u0437\u0438",
|
|
294
|
+
"\u0442\u0438",
|
|
295
|
+
"\u0442\u043E",
|
|
296
|
+
"\u0442\u043E\u0432\u0430",
|
|
297
|
+
"\u0442\u043E\u0433\u0430\u0432\u0430",
|
|
298
|
+
"\u0442\u043E\u0437\u0438",
|
|
299
|
+
"\u0442\u043E\u0439",
|
|
300
|
+
"\u0442\u043E\u043B\u043A\u043E\u0432\u0430",
|
|
301
|
+
"\u0442\u043E\u0447\u043D\u043E",
|
|
302
|
+
"\u0442\u0440\u0438",
|
|
303
|
+
"\u0442\u0440\u044F\u0431\u0432\u0430",
|
|
304
|
+
"\u0442\u0443\u043A",
|
|
305
|
+
"\u0442\u044A\u0439",
|
|
306
|
+
"\u0442\u044F",
|
|
307
|
+
"\u0442\u044F\u0445",
|
|
308
|
+
"\u0443",
|
|
309
|
+
"\u0443\u0442\u0440\u0435",
|
|
310
|
+
"\u0445\u0430\u0440\u0435\u0441\u0432\u0430",
|
|
311
|
+
"\u0445\u0438\u043B\u044F\u0434\u0438",
|
|
312
|
+
"\u0447",
|
|
313
|
+
"\u0447\u0430\u0441\u0430",
|
|
314
|
+
"\u0447\u0435",
|
|
315
|
+
"\u0447\u0435\u0441\u0442\u043E",
|
|
316
|
+
"\u0447\u0440\u0435\u0437",
|
|
317
|
+
"\u0449\u0435",
|
|
318
|
+
"\u0449\u043E\u043C",
|
|
319
|
+
"\u044E\u043C\u0440\u0443\u043A",
|
|
320
|
+
"\u044F",
|
|
321
|
+
"\u044F\u043A"
|
|
322
|
+
]);
|
|
323
|
+
var bulgarian = {
|
|
324
|
+
name: "bulgarian",
|
|
325
|
+
stemmer: stem,
|
|
326
|
+
stopWords,
|
|
327
|
+
tokenizer: { splitPattern: /[^a-z0-9а-яъ]+/gi }
|
|
328
|
+
};
|
|
329
|
+
|
|
330
|
+
export { bulgarian };
|