@delali/narsil 0.1.15 → 0.2.1-next.25
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.d.ts +23 -1
- package/dist/adapters/filesystem.mjs +1 -1
- package/dist/adapters/indexeddb.d.ts +25 -1
- package/dist/adapters/indexeddb.mjs +1 -1
- package/dist/adapters/memory.d.ts +14 -1
- package/dist/adapters-CVqI5NBt.d.ts +149 -0
- package/dist/{chunk-HMO6OMZ3.mjs → chunk-ASI6AOW2.mjs} +1073 -506
- package/dist/{chunk-HRABJLPB.mjs → chunk-BKAI2WXI.mjs} +1 -0
- package/dist/{chunk-WISGW24C.mjs → chunk-FKK76OR7.mjs} +56 -55
- package/dist/chunk-GHJWC2E5.mjs +9 -0
- package/dist/chunk-KH3YGMVT.mjs +15 -0
- package/dist/chunk-TGP5M5R7.mjs +880 -0
- package/dist/chunk-U2BKYEC7.mjs +254 -0
- package/dist/{chunk-ALB35PRQ.mjs → chunk-UNJZ7ZDB.mjs} +18 -0
- package/dist/{chunk-MEW7HD75.mjs → chunk-WLR5BFL3.mjs} +2557 -1166
- package/dist/config-CAjTHzkH.d.ts +256 -0
- package/dist/direct-executor-3W4QRA4E.mjs +5 -0
- package/dist/distribution/coordinator/etcd.d.ts +44 -3
- package/dist/distribution/coordinator/etcd.mjs +2 -2
- package/dist/distribution/coordinator/in-memory.d.ts +14 -3
- package/dist/distribution.d.ts +318 -45
- package/dist/distribution.mjs +1875 -1772
- package/dist/embeddings/openai.d.ts +28 -1
- package/dist/embeddings/openai.mjs +1 -1
- package/dist/engine-CCrlfLyq.d.ts +431 -0
- package/dist/index.browser.mjs +4571 -1776
- package/dist/index.d.ts +180 -6
- package/dist/index.mjs +6 -5
- package/dist/invalidation/broadcast-channel.d.ts +20 -1
- package/dist/invalidation/filesystem.d.ts +24 -1
- package/dist/invalidation/noop.d.ts +14 -1
- package/dist/language-DB8KG1z5.d.ts +53 -0
- package/dist/languages/albanian.d.ts +15 -0
- package/dist/languages/albanian.mjs +103 -0
- package/dist/languages/amharic.d.ts +15 -0
- package/dist/languages/amharic.mjs +240 -0
- package/dist/languages/arabic.d.ts +10 -1
- package/dist/languages/arabic.mjs +939 -154
- package/dist/languages/armenian.d.ts +10 -1
- package/dist/languages/armenian.mjs +275 -243
- package/dist/languages/azerbaijani.d.ts +15 -0
- package/dist/languages/azerbaijani.mjs +152 -0
- package/dist/languages/bambara.d.ts +15 -0
- package/dist/languages/bambara.mjs +60 -0
- package/dist/languages/basque.d.ts +14 -0
- package/dist/languages/basque.mjs +726 -0
- package/dist/languages/belarusian.d.ts +15 -0
- package/dist/languages/belarusian.mjs +100 -0
- package/dist/languages/bengali.d.ts +15 -0
- package/dist/languages/bengali.mjs +128 -0
- package/dist/languages/bosnian.d.ts +15 -0
- package/dist/languages/bosnian.mjs +197 -0
- package/dist/languages/breton.d.ts +15 -0
- package/dist/languages/breton.mjs +96 -0
- package/dist/languages/bulgarian.d.ts +10 -1
- package/dist/languages/bulgarian.mjs +11 -2
- package/dist/languages/burmese.d.ts +15 -0
- package/dist/languages/burmese.mjs +57 -0
- package/dist/languages/catalan.d.ts +14 -0
- package/dist/languages/catalan.mjs +987 -0
- package/dist/languages/chinese.d.ts +11 -1
- package/dist/languages/chinese.mjs +4 -3
- package/dist/languages/croatian.d.ts +15 -0
- package/dist/languages/croatian.mjs +191 -0
- package/dist/languages/czech.d.ts +14 -0
- package/dist/languages/czech.mjs +628 -0
- package/dist/languages/dagbani.d.ts +11 -1
- package/dist/languages/dagbani.mjs +11 -2
- package/dist/languages/danish.d.ts +10 -1
- package/dist/languages/danish.mjs +215 -110
- package/dist/languages/dutch.d.ts +10 -1
- package/dist/languages/dutch.mjs +1248 -163
- package/dist/languages/english.d.ts +10 -1
- package/dist/languages/english.mjs +2 -1
- package/dist/languages/esperanto.d.ts +14 -0
- package/dist/languages/esperanto.mjs +689 -0
- package/dist/languages/estonian.d.ts +14 -0
- package/dist/languages/estonian.mjs +2160 -0
- package/dist/languages/ewe.d.ts +11 -1
- package/dist/languages/ewe.mjs +11 -2
- package/dist/languages/faroese.d.ts +15 -0
- package/dist/languages/faroese.mjs +93 -0
- package/dist/languages/fijian.d.ts +15 -0
- package/dist/languages/fijian.mjs +55 -0
- package/dist/languages/finnish.d.ts +10 -1
- package/dist/languages/finnish.mjs +528 -263
- package/dist/languages/french.d.ts +10 -1
- package/dist/languages/french.mjs +950 -437
- package/dist/languages/ga.d.ts +11 -1
- package/dist/languages/ga.mjs +2 -1
- package/dist/languages/galician.d.ts +15 -0
- package/dist/languages/galician.mjs +172 -0
- package/dist/languages/georgian.d.ts +15 -0
- package/dist/languages/georgian.mjs +91 -0
- package/dist/languages/german.d.ts +10 -1
- package/dist/languages/german.mjs +432 -162
- package/dist/languages/greek.d.ts +10 -1
- package/dist/languages/greek.mjs +1970 -281
- package/dist/languages/guarani.d.ts +15 -0
- package/dist/languages/guarani.mjs +132 -0
- package/dist/languages/gujarati.d.ts +15 -0
- package/dist/languages/gujarati.mjs +234 -0
- package/dist/languages/haitian-creole.d.ts +15 -0
- package/dist/languages/haitian-creole.mjs +106 -0
- package/dist/languages/hausa.d.ts +11 -1
- package/dist/languages/hausa.mjs +2 -1
- package/dist/languages/hawaiian.d.ts +15 -0
- package/dist/languages/hawaiian.mjs +69 -0
- package/dist/languages/hebrew.d.ts +15 -0
- package/dist/languages/hebrew.mjs +213 -0
- package/dist/languages/hindi.d.ts +10 -1
- package/dist/languages/hindi.mjs +175 -97
- package/dist/languages/hungarian.d.ts +10 -1
- package/dist/languages/hungarian.mjs +412 -326
- package/dist/languages/icelandic.d.ts +15 -0
- package/dist/languages/icelandic.mjs +161 -0
- package/dist/languages/igbo.d.ts +11 -1
- package/dist/languages/igbo.mjs +2 -1
- package/dist/languages/indonesian.d.ts +10 -1
- package/dist/languages/indonesian.mjs +329 -137
- package/dist/languages/irish.d.ts +10 -1
- package/dist/languages/irish.mjs +223 -139
- package/dist/languages/italian.d.ts +10 -1
- package/dist/languages/italian.mjs +703 -406
- package/dist/languages/japanese.d.ts +11 -1
- package/dist/languages/japanese.mjs +5 -108
- package/dist/languages/kannada.d.ts +15 -0
- package/dist/languages/kannada.mjs +94 -0
- package/dist/languages/kazakh.d.ts +15 -0
- package/dist/languages/kazakh.mjs +90 -0
- package/dist/languages/khmer.d.ts +15 -0
- package/dist/languages/khmer.mjs +181 -0
- package/dist/languages/kinyarwanda.d.ts +15 -0
- package/dist/languages/kinyarwanda.mjs +99 -0
- package/dist/languages/kirundi.d.ts +15 -0
- package/dist/languages/kirundi.mjs +92 -0
- package/dist/languages/korean.d.ts +15 -0
- package/dist/languages/korean.mjs +252 -0
- package/dist/languages/kurmanji.d.ts +15 -0
- package/dist/languages/kurmanji.mjs +50 -0
- package/dist/languages/kyrgyz.d.ts +15 -0
- package/dist/languages/kyrgyz.mjs +108 -0
- package/dist/languages/lao.d.ts +15 -0
- package/dist/languages/lao.mjs +63 -0
- package/dist/languages/latin.d.ts +15 -0
- package/dist/languages/latin.mjs +69 -0
- package/dist/languages/latvian.d.ts +15 -0
- package/dist/languages/latvian.mjs +173 -0
- package/dist/languages/lingala.d.ts +15 -0
- package/dist/languages/lingala.mjs +61 -0
- package/dist/languages/lithuanian.d.ts +14 -0
- package/dist/languages/lithuanian.mjs +571 -0
- package/dist/languages/luxembourgish.d.ts +15 -0
- package/dist/languages/luxembourgish.mjs +214 -0
- package/dist/languages/macedonian.d.ts +15 -0
- package/dist/languages/macedonian.mjs +808 -0
- package/dist/languages/malagasy.d.ts +15 -0
- package/dist/languages/malagasy.mjs +67 -0
- package/dist/languages/malay.d.ts +15 -0
- package/dist/languages/malay.mjs +482 -0
- package/dist/languages/malayalam.d.ts +15 -0
- package/dist/languages/malayalam.mjs +122 -0
- package/dist/languages/maltese.d.ts +15 -0
- package/dist/languages/maltese.mjs +139 -0
- package/dist/languages/maori.d.ts +15 -0
- package/dist/languages/maori.mjs +79 -0
- package/dist/languages/marathi.d.ts +15 -0
- package/dist/languages/marathi.mjs +111 -0
- package/dist/languages/nepali.d.ts +10 -1
- package/dist/languages/nepali.mjs +209 -99
- package/dist/languages/norwegian.d.ts +10 -1
- package/dist/languages/norwegian.mjs +218 -85
- package/dist/languages/oromo.d.ts +15 -0
- package/dist/languages/oromo.mjs +78 -0
- package/dist/languages/persian.d.ts +14 -0
- package/dist/languages/persian.mjs +710 -0
- package/dist/languages/polish.d.ts +14 -0
- package/dist/languages/polish.mjs +614 -0
- package/dist/languages/portuguese.d.ts +10 -1
- package/dist/languages/portuguese.mjs +613 -380
- package/dist/languages/punjabi.d.ts +15 -0
- package/dist/languages/punjabi.mjs +197 -0
- package/dist/languages/romanian.d.ts +10 -1
- package/dist/languages/romanian.mjs +561 -382
- package/dist/languages/russian.d.ts +10 -1
- package/dist/languages/russian.mjs +373 -244
- package/dist/languages/samoan.d.ts +15 -0
- package/dist/languages/samoan.mjs +63 -0
- package/dist/languages/sanskrit.d.ts +10 -1
- package/dist/languages/sanskrit.mjs +2 -2
- package/dist/languages/scottish-gaelic.d.ts +15 -0
- package/dist/languages/scottish-gaelic.mjs +370 -0
- package/dist/languages/serbian.d.ts +10 -1
- package/dist/languages/serbian.mjs +3635 -155
- package/dist/languages/shona.d.ts +15 -0
- package/dist/languages/shona.mjs +91 -0
- package/dist/languages/sinhala.d.ts +15 -0
- package/dist/languages/sinhala.mjs +182 -0
- package/dist/languages/slovak.d.ts +15 -0
- package/dist/languages/slovak.mjs +430 -0
- package/dist/languages/slovenian.d.ts +10 -1
- package/dist/languages/slovenian.mjs +11 -2
- package/dist/languages/sorani.d.ts +15 -0
- package/dist/languages/sorani.mjs +156 -0
- package/dist/languages/spanish.d.ts +10 -1
- package/dist/languages/spanish.mjs +676 -374
- package/dist/languages/swahili.d.ts +10 -1
- package/dist/languages/swahili.mjs +2 -1
- package/dist/languages/swedish.d.ts +10 -1
- package/dist/languages/swedish.mjs +237 -95
- package/dist/languages/tagalog.d.ts +15 -0
- package/dist/languages/tagalog.mjs +159 -0
- package/dist/languages/tamil.d.ts +10 -1
- package/dist/languages/tamil.mjs +836 -64
- package/dist/languages/tatar.d.ts +15 -0
- package/dist/languages/tatar.mjs +1017 -0
- package/dist/languages/telugu.d.ts +15 -0
- package/dist/languages/telugu.mjs +62 -0
- package/dist/languages/thai.d.ts +15 -0
- package/dist/languages/thai.mjs +65 -0
- package/dist/languages/tibetan.d.ts +15 -0
- package/dist/languages/tibetan.mjs +119 -0
- package/dist/languages/tigrinya.d.ts +15 -0
- package/dist/languages/tigrinya.mjs +194 -0
- package/dist/languages/tongan.d.ts +15 -0
- package/dist/languages/tongan.mjs +56 -0
- package/dist/languages/turkish.d.ts +10 -1
- package/dist/languages/turkish.mjs +1246 -393
- package/dist/languages/twi.d.ts +11 -1
- package/dist/languages/twi.mjs +13 -2
- package/dist/languages/ukrainian.d.ts +10 -1
- package/dist/languages/ukrainian.mjs +125 -47
- package/dist/languages/urdu.d.ts +15 -0
- package/dist/languages/urdu.mjs +528 -0
- package/dist/languages/vietnamese.d.ts +15 -0
- package/dist/languages/vietnamese.mjs +277 -0
- package/dist/languages/welsh.d.ts +15 -0
- package/dist/languages/welsh.mjs +446 -0
- package/dist/languages/wolof.d.ts +15 -0
- package/dist/languages/wolof.mjs +79 -0
- package/dist/languages/xhosa.d.ts +15 -0
- package/dist/languages/xhosa.mjs +95 -0
- package/dist/languages/yoruba.d.ts +11 -1
- package/dist/languages/yoruba.mjs +1 -0
- package/dist/languages/zulu.d.ts +11 -1
- package/dist/languages/zulu.mjs +2 -1
- package/dist/persistence/durability/checkpoint-worker.d.ts +13 -0
- package/dist/persistence/durability/checkpoint-worker.mjs +6 -5
- package/dist/schema-CE6S_aAr.d.ts +208 -0
- package/dist/search-CgQQIciB.d.ts +786 -0
- package/dist/server.d.ts +214 -12
- package/dist/server.mjs +41 -2
- package/dist/types-BV3nAKTL.d.ts +302 -0
- package/dist/vector/hnsw-build-worker.d.ts +68 -2
- package/dist/vector/hnsw-build-worker.mjs +0 -2
- package/dist/workers/entry.mjs +2 -2
- package/package.json +14 -4
- package/dist/adapters-Bo2JQg8b.d.ts +0 -37
- package/dist/chunk-TW6BXTOL.mjs +0 -338
- package/dist/config-CojEPOgS.d.ts +0 -481
- package/dist/direct-executor-E52FFGDE.mjs +0 -4
- package/dist/language-ONQ3iVDf.d.ts +0 -14
- package/dist/narsil-D-84Nvm4.d.ts +0 -72
- package/dist/schema-DCgkO1j4.d.ts +0 -65
- package/dist/types-DHX6zGeD.d.ts +0 -73
|
@@ -0,0 +1,710 @@
|
|
|
1
|
+
import { withNormalisedSpellings } from '../chunk-GHJWC2E5.mjs';
|
|
2
|
+
import { BaseStemmer } from '../chunk-U2BKYEC7.mjs';
|
|
3
|
+
|
|
4
|
+
// src/languages/snowball/persian.ts
|
|
5
|
+
var a_0 = [
|
|
6
|
+
["", 7],
|
|
7
|
+
[" ", 6, 1],
|
|
8
|
+
["\u0623", 4, 2],
|
|
9
|
+
["\u0624", 5, 3],
|
|
10
|
+
["\u0625", 4, 4],
|
|
11
|
+
["\u0626", 2, 5],
|
|
12
|
+
["\u0629", 3, 6],
|
|
13
|
+
["\u0643", 1, 7],
|
|
14
|
+
["\u064A", 2, 8],
|
|
15
|
+
["\u06C1", 3, 9],
|
|
16
|
+
["\u200D", 6, 10]
|
|
17
|
+
];
|
|
18
|
+
var a_1 = [
|
|
19
|
+
["\u0645\u06CC\u200C", 2],
|
|
20
|
+
["\u0646\u0645\u06CC\u200C", 1]
|
|
21
|
+
];
|
|
22
|
+
var a_2 = [
|
|
23
|
+
["\u0633\u062A\u0627\u0646", -1],
|
|
24
|
+
["\u0631\u0627\u0646", -1],
|
|
25
|
+
["\u0633\u0627\u0646", -1],
|
|
26
|
+
["\u0648\u0627\u0646", -1]
|
|
27
|
+
];
|
|
28
|
+
var a_3 = [
|
|
29
|
+
["\u0622\u0630\u0631\u0628\u0627\u06CC\u062C\u0627\u0646", 1],
|
|
30
|
+
["\u0647\u0645\u062F\u0627\u0646", 1],
|
|
31
|
+
["\u062E\u0627\u0646\u062F\u0627\u0646", 1],
|
|
32
|
+
["\u0632\u0646\u062F\u0627\u0646", 1],
|
|
33
|
+
["\u0645\u06CC\u0632\u0627\u0646", 1],
|
|
34
|
+
["\u062F\u0631\u062E\u0634\u0627\u0646", 1],
|
|
35
|
+
["\u0622\u062A\u0634\u0641\u0634\u0627\u0646", 1],
|
|
36
|
+
["\u0646\u0634\u0627\u0646", 1],
|
|
37
|
+
["\u06A9\u0647\u06A9\u0634\u0627\u0646", 1],
|
|
38
|
+
["\u0627\u06CC\u0634\u0627\u0646", 1],
|
|
39
|
+
["\u067E\u0631\u06CC\u0634\u0627\u0646", 1],
|
|
40
|
+
["\u0633\u0644\u0637\u0627\u0646", 1],
|
|
41
|
+
["\u06AF\u06CC\u0644\u0627\u0646", 1],
|
|
42
|
+
["\u0633\u0627\u062E\u062A\u0645\u0627\u0646", 1],
|
|
43
|
+
["\u0631\u0645\u0627\u0646", 1],
|
|
44
|
+
["\u062F\u0631\u0645\u0627\u0646", 1, 1],
|
|
45
|
+
["\u0642\u0647\u0631\u0645\u0627\u0646", 1, 2],
|
|
46
|
+
["\u06A9\u0631\u0645\u0627\u0646", 1, 3],
|
|
47
|
+
["\u0633\u0627\u0632\u0645\u0627\u0646", 1],
|
|
48
|
+
["\u0647\u0645\u0632\u0645\u0627\u0646", 1],
|
|
49
|
+
["\u0622\u0633\u0645\u0627\u0646", 1],
|
|
50
|
+
["\u0622\u0644\u0645\u0627\u0646", 1],
|
|
51
|
+
["\u0645\u0633\u0644\u0645\u0627\u0646", 1],
|
|
52
|
+
["\u0627\u06CC\u0645\u0627\u0646", 1],
|
|
53
|
+
["\u0633\u0644\u06CC\u0645\u0627\u0646", 1],
|
|
54
|
+
["\u067E\u06CC\u0645\u0627\u0646", 1],
|
|
55
|
+
["\u0644\u0628\u0646\u0627\u0646", 1],
|
|
56
|
+
["\u06CC\u0648\u0646\u0627\u0646", 1],
|
|
57
|
+
["\u0627\u0635\u0641\u0647\u0627\u0646", 1],
|
|
58
|
+
["\u0627\u0645\u06A9\u0627\u0646", 1],
|
|
59
|
+
["\u067E\u0627\u06CC\u0627\u0646", 1],
|
|
60
|
+
["\u0628\u06CC\u0627\u0646", 1],
|
|
61
|
+
["\u062C\u0631\u06CC\u0627\u0646", 1]
|
|
62
|
+
];
|
|
63
|
+
var a_4 = [
|
|
64
|
+
["\u0627\u0633\u0627\u062A\u06CC\u062F", 2],
|
|
65
|
+
["\u0627\u062E\u0628\u0627\u0631", 1]
|
|
66
|
+
];
|
|
67
|
+
var as_4 = ["\u062E\u0628\u0631", "\u0627\u0633\u062A\u0627\u062F"];
|
|
68
|
+
var a_5 = [
|
|
69
|
+
["\u0647\u0627", 1],
|
|
70
|
+
["\u0627\u062A", 1],
|
|
71
|
+
["\u06CC\u062A", 1],
|
|
72
|
+
["\u0645\u0646\u062F", 1],
|
|
73
|
+
["\u0648\u0627\u0631", 1],
|
|
74
|
+
["\u06AF\u0627\u0631", 1],
|
|
75
|
+
["\u062A\u0631", 2],
|
|
76
|
+
["\u0627\u0634", 1],
|
|
77
|
+
["\u0627\u0645", 1],
|
|
78
|
+
["\u0627\u0646", 1],
|
|
79
|
+
["\u0628\u0627\u0646", 1, 1],
|
|
80
|
+
["\u06AF\u0627\u0646", 1, 2],
|
|
81
|
+
["\u06CC\u0627\u0646", 1, 3],
|
|
82
|
+
["\u06CC\u0646", 1],
|
|
83
|
+
["\u062A\u0631\u06CC\u0646", 1, 1],
|
|
84
|
+
["\u06AF\u0627\u0647", 1],
|
|
85
|
+
["\u0627\u0646\u0647", 1],
|
|
86
|
+
["\u0646\u0627\u06A9", 1],
|
|
87
|
+
["\u0647\u0627\u06CC", 1],
|
|
88
|
+
["\u0627\u0646\u06CC", 1],
|
|
89
|
+
["\u06AF\u06CC", 1],
|
|
90
|
+
["\u06CC\u06CC", 1]
|
|
91
|
+
];
|
|
92
|
+
var a_6 = [
|
|
93
|
+
["\u0627\u0633\u062A", 1],
|
|
94
|
+
["\u0627\u0646\u062F", 1],
|
|
95
|
+
["\u06CC\u062F", 1],
|
|
96
|
+
["\u0627\u06CC\u062F", 1, 1],
|
|
97
|
+
["\u0627\u0633", 1],
|
|
98
|
+
["\u06CC\u0645", 1],
|
|
99
|
+
["\u0627\u06CC\u0645", 1, 1],
|
|
100
|
+
["\u0627\u06CC", 1]
|
|
101
|
+
];
|
|
102
|
+
var a_7 = [
|
|
103
|
+
["\u062F", 1],
|
|
104
|
+
["\u0627\u0646\u062F", 1, 1],
|
|
105
|
+
["\u0631\u0641\u062A\u0627\u0646\u062F", 2, 1],
|
|
106
|
+
["\u06CC\u062F", 1, 3],
|
|
107
|
+
["\u0631\u0641\u062A\u06CC\u062F", 2, 1],
|
|
108
|
+
["\u0645", 1],
|
|
109
|
+
["\u0627\u0645", 1, 1],
|
|
110
|
+
["\u0631\u0641\u062A\u0645", 2, 2],
|
|
111
|
+
["\u06CC\u0645", 1, 3],
|
|
112
|
+
["\u0631\u0641\u062A\u06CC\u0645", 2, 1],
|
|
113
|
+
["\u0627\u0646", 3],
|
|
114
|
+
["\u062A\u0647", 5],
|
|
115
|
+
["\u062F\u0647", 4],
|
|
116
|
+
["\u0646\u062F\u0647", 3, 1],
|
|
117
|
+
["\u0631\u0641\u062A\u06CC", 2]
|
|
118
|
+
];
|
|
119
|
+
var PersianStemmer = class extends BaseStemmer {
|
|
120
|
+
#stem() {
|
|
121
|
+
let a;
|
|
122
|
+
let I_p1;
|
|
123
|
+
let B_remove_verb_person_endings;
|
|
124
|
+
let B_saw_present_prefix;
|
|
125
|
+
B_saw_present_prefix = false;
|
|
126
|
+
const v_1 = this.c;
|
|
127
|
+
while (true) {
|
|
128
|
+
const v_2 = this.c;
|
|
129
|
+
lab1: {
|
|
130
|
+
this.bra = this.c;
|
|
131
|
+
a = this.find_among(a_0);
|
|
132
|
+
this.ket = this.c;
|
|
133
|
+
switch (a) {
|
|
134
|
+
case 1: {
|
|
135
|
+
this.slice_from("\u06A9");
|
|
136
|
+
break;
|
|
137
|
+
}
|
|
138
|
+
case 2: {
|
|
139
|
+
this.slice_from("\u06CC");
|
|
140
|
+
break;
|
|
141
|
+
}
|
|
142
|
+
case 3: {
|
|
143
|
+
this.slice_from("\u0647");
|
|
144
|
+
break;
|
|
145
|
+
}
|
|
146
|
+
case 4: {
|
|
147
|
+
this.slice_from("\u0627");
|
|
148
|
+
break;
|
|
149
|
+
}
|
|
150
|
+
case 5: {
|
|
151
|
+
this.slice_from("\u0648");
|
|
152
|
+
break;
|
|
153
|
+
}
|
|
154
|
+
case 6: {
|
|
155
|
+
this.slice_del();
|
|
156
|
+
break;
|
|
157
|
+
}
|
|
158
|
+
case 7: {
|
|
159
|
+
if (this.c >= this.limit) break lab1;
|
|
160
|
+
this.c++;
|
|
161
|
+
break;
|
|
162
|
+
}
|
|
163
|
+
}
|
|
164
|
+
continue;
|
|
165
|
+
}
|
|
166
|
+
this.c = v_2;
|
|
167
|
+
break;
|
|
168
|
+
}
|
|
169
|
+
this.c = v_1;
|
|
170
|
+
const v_3 = this.c;
|
|
171
|
+
lab2: {
|
|
172
|
+
this.bra = this.c;
|
|
173
|
+
a = this.find_among(a_1);
|
|
174
|
+
if (a === 0) break lab2;
|
|
175
|
+
this.ket = this.c;
|
|
176
|
+
switch (a) {
|
|
177
|
+
case 1: {
|
|
178
|
+
if (this.c + 2 > this.limit) break lab2;
|
|
179
|
+
this.c += 2;
|
|
180
|
+
B_saw_present_prefix = true;
|
|
181
|
+
break;
|
|
182
|
+
}
|
|
183
|
+
case 2: {
|
|
184
|
+
if (this.c + 2 > this.limit) break lab2;
|
|
185
|
+
this.c += 2;
|
|
186
|
+
this.slice_del();
|
|
187
|
+
B_saw_present_prefix = true;
|
|
188
|
+
break;
|
|
189
|
+
}
|
|
190
|
+
}
|
|
191
|
+
}
|
|
192
|
+
this.c = v_3;
|
|
193
|
+
const v_4 = this.c;
|
|
194
|
+
while (true) {
|
|
195
|
+
const v_5 = this.c;
|
|
196
|
+
lab4: {
|
|
197
|
+
while (true) {
|
|
198
|
+
const v_6 = this.c;
|
|
199
|
+
lab6: {
|
|
200
|
+
this.bra = this.c;
|
|
201
|
+
if (!this.eq_s("\u200C")) break lab6;
|
|
202
|
+
this.ket = this.c;
|
|
203
|
+
this.slice_del();
|
|
204
|
+
this.c = v_6;
|
|
205
|
+
break;
|
|
206
|
+
}
|
|
207
|
+
this.c = v_6;
|
|
208
|
+
if (this.c >= this.limit) break lab4;
|
|
209
|
+
this.c++;
|
|
210
|
+
}
|
|
211
|
+
continue;
|
|
212
|
+
}
|
|
213
|
+
this.c = v_5;
|
|
214
|
+
break;
|
|
215
|
+
}
|
|
216
|
+
this.c = v_4;
|
|
217
|
+
I_p1 = this.limit;
|
|
218
|
+
const v_7 = this.c;
|
|
219
|
+
lab7: {
|
|
220
|
+
if (this.c + 3 > this.limit) break lab7;
|
|
221
|
+
this.c += 3;
|
|
222
|
+
I_p1 = this.c;
|
|
223
|
+
}
|
|
224
|
+
this.c = v_7;
|
|
225
|
+
this.limit_backward = this.c;
|
|
226
|
+
this.c = this.limit;
|
|
227
|
+
while (true) {
|
|
228
|
+
const v_8 = this.limit - this.c;
|
|
229
|
+
lab8: {
|
|
230
|
+
const v_9 = this.limit - this.c;
|
|
231
|
+
B_remove_verb_person_endings = false;
|
|
232
|
+
lab9: {
|
|
233
|
+
if (!B_saw_present_prefix) break lab9;
|
|
234
|
+
B_remove_verb_person_endings = true;
|
|
235
|
+
}
|
|
236
|
+
{
|
|
237
|
+
const v_10 = this.limit - this.c;
|
|
238
|
+
lab10: {
|
|
239
|
+
if (this.find_among_b(a_3) === 0) break lab10;
|
|
240
|
+
if (
|
|
241
|
+
/**@type {boolean}*/
|
|
242
|
+
this.c > this.limit_backward
|
|
243
|
+
) break lab10;
|
|
244
|
+
break lab8;
|
|
245
|
+
}
|
|
246
|
+
this.c = this.limit - v_10;
|
|
247
|
+
}
|
|
248
|
+
{
|
|
249
|
+
const v_11 = this.limit - this.c;
|
|
250
|
+
lab11: {
|
|
251
|
+
if (this.find_among_b(a_2) === 0) break lab11;
|
|
252
|
+
break lab8;
|
|
253
|
+
}
|
|
254
|
+
this.c = this.limit - v_11;
|
|
255
|
+
}
|
|
256
|
+
lab12: {
|
|
257
|
+
const v_12 = this.limit - this.c;
|
|
258
|
+
lab13: {
|
|
259
|
+
lab14: {
|
|
260
|
+
const v_13 = this.limit - this.c;
|
|
261
|
+
lab15: {
|
|
262
|
+
this.ket = this.c;
|
|
263
|
+
a = this.find_among_b(a_4);
|
|
264
|
+
if (a === 0) break lab15;
|
|
265
|
+
this.bra = this.c;
|
|
266
|
+
this.slice_from(as_4[a - 1]);
|
|
267
|
+
break lab14;
|
|
268
|
+
}
|
|
269
|
+
this.c = this.limit - v_13;
|
|
270
|
+
if (this.c < I_p1) break lab13;
|
|
271
|
+
const v_14 = this.limit_backward;
|
|
272
|
+
this.limit_backward = I_p1;
|
|
273
|
+
this.ket = this.c;
|
|
274
|
+
a = this.find_among_b(a_5);
|
|
275
|
+
if (a === 0) {
|
|
276
|
+
this.limit_backward = v_14;
|
|
277
|
+
break lab13;
|
|
278
|
+
}
|
|
279
|
+
this.bra = this.c;
|
|
280
|
+
switch (a) {
|
|
281
|
+
case 1: {
|
|
282
|
+
this.slice_del();
|
|
283
|
+
break;
|
|
284
|
+
}
|
|
285
|
+
case 2: {
|
|
286
|
+
if (
|
|
287
|
+
/**@type {boolean}*/
|
|
288
|
+
this.c <= this.limit_backward
|
|
289
|
+
) {
|
|
290
|
+
this.limit_backward = v_14;
|
|
291
|
+
break lab13;
|
|
292
|
+
}
|
|
293
|
+
this.slice_del();
|
|
294
|
+
break;
|
|
295
|
+
}
|
|
296
|
+
}
|
|
297
|
+
this.limit_backward = v_14;
|
|
298
|
+
}
|
|
299
|
+
break lab12;
|
|
300
|
+
}
|
|
301
|
+
this.c = this.limit - v_12;
|
|
302
|
+
lab16: {
|
|
303
|
+
const v_15 = this.limit - this.c;
|
|
304
|
+
lab17: {
|
|
305
|
+
this.ket = this.c;
|
|
306
|
+
if (this.find_among_b(a_6) === 0) break lab17;
|
|
307
|
+
this.bra = this.c;
|
|
308
|
+
if (
|
|
309
|
+
/**@type {boolean}*/
|
|
310
|
+
I_p1 > this.c
|
|
311
|
+
) break lab17;
|
|
312
|
+
this.slice_del();
|
|
313
|
+
break lab16;
|
|
314
|
+
}
|
|
315
|
+
this.c = this.limit - v_15;
|
|
316
|
+
this.ket = this.c;
|
|
317
|
+
a = this.find_among_b(a_7);
|
|
318
|
+
if (a === 0) break lab8;
|
|
319
|
+
this.bra = this.c;
|
|
320
|
+
switch (a) {
|
|
321
|
+
case 1: {
|
|
322
|
+
if (!B_remove_verb_person_endings) break lab8;
|
|
323
|
+
if (
|
|
324
|
+
/**@type {boolean}*/
|
|
325
|
+
I_p1 > this.c
|
|
326
|
+
) break lab8;
|
|
327
|
+
this.slice_del();
|
|
328
|
+
break;
|
|
329
|
+
}
|
|
330
|
+
case 2: {
|
|
331
|
+
this.slice_from("\u0631\u0641\u062A");
|
|
332
|
+
break;
|
|
333
|
+
}
|
|
334
|
+
case 3: {
|
|
335
|
+
if (
|
|
336
|
+
/**@type {boolean}*/
|
|
337
|
+
I_p1 > this.c
|
|
338
|
+
) break lab8;
|
|
339
|
+
this.slice_del();
|
|
340
|
+
B_remove_verb_person_endings = true;
|
|
341
|
+
break;
|
|
342
|
+
}
|
|
343
|
+
case 4: {
|
|
344
|
+
if (
|
|
345
|
+
/**@type {boolean}*/
|
|
346
|
+
this.c <= this.limit_backward
|
|
347
|
+
) break lab8;
|
|
348
|
+
this.slice_from("\u062F");
|
|
349
|
+
B_remove_verb_person_endings = true;
|
|
350
|
+
break;
|
|
351
|
+
}
|
|
352
|
+
case 5: {
|
|
353
|
+
if (
|
|
354
|
+
/**@type {boolean}*/
|
|
355
|
+
this.c <= this.limit_backward
|
|
356
|
+
) break lab8;
|
|
357
|
+
this.slice_from("\u062A");
|
|
358
|
+
B_remove_verb_person_endings = true;
|
|
359
|
+
break;
|
|
360
|
+
}
|
|
361
|
+
}
|
|
362
|
+
}
|
|
363
|
+
}
|
|
364
|
+
this.c = this.limit - v_9;
|
|
365
|
+
continue;
|
|
366
|
+
}
|
|
367
|
+
this.c = this.limit - v_8;
|
|
368
|
+
break;
|
|
369
|
+
}
|
|
370
|
+
this.c = this.limit_backward;
|
|
371
|
+
return true;
|
|
372
|
+
}
|
|
373
|
+
stem(input) {
|
|
374
|
+
this.setCurrent(input);
|
|
375
|
+
this.#stem();
|
|
376
|
+
return this.getCurrent();
|
|
377
|
+
}
|
|
378
|
+
};
|
|
379
|
+
var shared = new PersianStemmer();
|
|
380
|
+
function stem(token) {
|
|
381
|
+
return shared.stem(token);
|
|
382
|
+
}
|
|
383
|
+
|
|
384
|
+
// src/languages/persian.ts
|
|
385
|
+
var stopWords = /* @__PURE__ */ new Set([
|
|
386
|
+
"\u0627\u0628\u0627\u062F",
|
|
387
|
+
"\u0627\u062B\u0631",
|
|
388
|
+
"\u0627\u0631\u0647",
|
|
389
|
+
"\u0627\u0631\u064A",
|
|
390
|
+
"\u0627\u0632",
|
|
391
|
+
"\u0627\u0633\u062A",
|
|
392
|
+
"\u0627\u0633\u062A\u0641\u0627\u062F\u0647",
|
|
393
|
+
"\u0627\u0634",
|
|
394
|
+
"\u0627\u0643\u0646\u0648\u0646",
|
|
395
|
+
"\u0627\u06AF\u0631",
|
|
396
|
+
"\u0627\u0644\u0628\u062A\u0647",
|
|
397
|
+
"\u0627\u0645",
|
|
398
|
+
"\u0627\u0645\u0627",
|
|
399
|
+
"\u0627\u0645\u062F",
|
|
400
|
+
"\u0627\u0645\u062F\u0647",
|
|
401
|
+
"\u0627\u0645\u0631\u0648\u0632",
|
|
402
|
+
"\u0627\u0645\u0633\u0627\u0644",
|
|
403
|
+
"\u0627\u0646",
|
|
404
|
+
"\u0627\u0646\u0627\u0646",
|
|
405
|
+
"\u0627\u0646\u062C\u0627",
|
|
406
|
+
"\u0627\u0646\u0686\u0647",
|
|
407
|
+
"\u0627\u0646\u062F",
|
|
408
|
+
"\u0627\u0646\u0643\u0647",
|
|
409
|
+
"\u0627\u0646\u0647\u0627",
|
|
410
|
+
"\u0627\u0648",
|
|
411
|
+
"\u0627\u0648\u0631\u062F",
|
|
412
|
+
"\u0627\u0648\u0631\u062F\u0647",
|
|
413
|
+
"\u0627\u0648\u0644",
|
|
414
|
+
"\u0627\u064A",
|
|
415
|
+
"\u0627\u064A\u0627",
|
|
416
|
+
"\u0627\u064A\u062F",
|
|
417
|
+
"\u0627\u064A\u0634\u0627\u0646",
|
|
418
|
+
"\u0627\u064A\u0645",
|
|
419
|
+
"\u0627\u064A\u0646",
|
|
420
|
+
"\u0627\u064A\u0646\u0643\u0647",
|
|
421
|
+
"\u0628\u0627",
|
|
422
|
+
"\u0628\u0627\u0631",
|
|
423
|
+
"\u0628\u0627\u0631\u0647",
|
|
424
|
+
"\u0628\u0627\u0634\u062F",
|
|
425
|
+
"\u0628\u0627\u0634\u0646\u062F",
|
|
426
|
+
"\u0628\u0627\u0634\u064A\u0645",
|
|
427
|
+
"\u0628\u0627\u0644\u0627",
|
|
428
|
+
"\u0628\u0627\u0644\u0627\u064A",
|
|
429
|
+
"\u0628\u0627\u064A\u062F",
|
|
430
|
+
"\u0628\u062F\u0648\u0646",
|
|
431
|
+
"\u0628\u0631",
|
|
432
|
+
"\u0628\u0631\u0627\u0628\u0631",
|
|
433
|
+
"\u0628\u0631\u0627\u0633\u0627\u0633",
|
|
434
|
+
"\u0628\u0631\u0627\u064A",
|
|
435
|
+
"\u0628\u0631\u062E\u0648\u0631\u062F\u0627\u0631",
|
|
436
|
+
"\u0628\u0631\u062E\u064A",
|
|
437
|
+
"\u0628\u0631\u062F\u0627\u0631\u064A",
|
|
438
|
+
"\u0628\u0631\u0648\u0632",
|
|
439
|
+
"\u0628\u0633\u064A\u0627\u0631",
|
|
440
|
+
"\u0628\u0633\u064A\u0627\u0631\u064A",
|
|
441
|
+
"\u0628\u0639\u062F",
|
|
442
|
+
"\u0628\u0639\u0631\u064A",
|
|
443
|
+
"\u0628\u0639\u0636\u064A",
|
|
444
|
+
"\u0628\u0644\u0643\u0647",
|
|
445
|
+
"\u0628\u0644\u0647",
|
|
446
|
+
"\u0628\u0644\u064A",
|
|
447
|
+
"\u0628\u0646\u0627\u0628\u0631\u0627\u064A\u0646",
|
|
448
|
+
"\u0628\u0646\u062F\u064A",
|
|
449
|
+
"\u0628\u0647",
|
|
450
|
+
"\u0628\u0647\u062A\u0631\u064A\u0646",
|
|
451
|
+
"\u0628\u0648\u062F",
|
|
452
|
+
"\u0628\u0648\u062F\u0646",
|
|
453
|
+
"\u0628\u0648\u062F\u0646\u062F",
|
|
454
|
+
"\u0628\u0648\u062F\u0647",
|
|
455
|
+
"\u0628\u064A",
|
|
456
|
+
"\u0628\u064A\u0631\u0648\u0646",
|
|
457
|
+
"\u0628\u064A\u0633\u062A",
|
|
458
|
+
"\u0628\u064A\u0634",
|
|
459
|
+
"\u0628\u064A\u0634\u062A\u0631",
|
|
460
|
+
"\u0628\u064A\u0634\u062A\u0631\u064A",
|
|
461
|
+
"\u0628\u064A\u0646",
|
|
462
|
+
"\u067E\u0627\u0639\u064A\u0646",
|
|
463
|
+
"\u067E\u0633",
|
|
464
|
+
"\u067E\u0646\u062C",
|
|
465
|
+
"\u067E\u064A\u0634",
|
|
466
|
+
"\u062A\u0627",
|
|
467
|
+
"\u062A\u0627\u0632\u0647",
|
|
468
|
+
"\u062A\u0627\u0643\u0646\u0648\u0646",
|
|
469
|
+
"\u062A\u0627\u0646",
|
|
470
|
+
"\u062A\u062D\u062A",
|
|
471
|
+
"\u062A\u0631",
|
|
472
|
+
"\u062A\u0631\u064A\u0646",
|
|
473
|
+
"\u062A\u0645\u0627\u0645",
|
|
474
|
+
"\u062A\u0645\u0627\u0645\u064A",
|
|
475
|
+
"\u062A\u0646\u0647\u0627",
|
|
476
|
+
"\u062A\u0648\u0627\u0646\u062F",
|
|
477
|
+
"\u062A\u0648\u0627\u0646\u0646\u062F",
|
|
478
|
+
"\u062A\u0648\u0633\u0637",
|
|
479
|
+
"\u062A\u0648\u0644",
|
|
480
|
+
"\u062A\u0648\u064A",
|
|
481
|
+
"\u062C\u0627",
|
|
482
|
+
"\u062C\u0627\u064A",
|
|
483
|
+
"\u062C\u0627\u064A\u064A",
|
|
484
|
+
"\u062C\u062F\u0627",
|
|
485
|
+
"\u062C\u062F\u064A\u062F",
|
|
486
|
+
"\u062C\u0631\u064A\u0627\u0646",
|
|
487
|
+
"\u062C\u0632",
|
|
488
|
+
"\u062C\u0644\u0648\u06AF\u064A\u0631\u064A",
|
|
489
|
+
"\u062C\u0644\u0648\u064A",
|
|
490
|
+
"\u0686\u0631\u0627",
|
|
491
|
+
"\u0686\u0637\u0648\u0631",
|
|
492
|
+
"\u0686\u06AF\u0648\u0646\u0647",
|
|
493
|
+
"\u0686\u0646\u062F",
|
|
494
|
+
"\u0686\u0646\u062F\u064A\u0646",
|
|
495
|
+
"\u0686\u0646\u064A\u0646",
|
|
496
|
+
"\u0686\u0647",
|
|
497
|
+
"\u0686\u0647\u0627\u0631",
|
|
498
|
+
"\u0686\u0648\u0646",
|
|
499
|
+
"\u0686\u064A\u0632",
|
|
500
|
+
"\u0686\u064A\u0632\u064A",
|
|
501
|
+
"\u0686\u064A\u0633\u062A",
|
|
502
|
+
"\u062D\u062A\u064A",
|
|
503
|
+
"\u062D\u062F\u0648\u062F",
|
|
504
|
+
"\u062D\u0642",
|
|
505
|
+
"\u062E\u0627\u0631\u062C",
|
|
506
|
+
"\u062E\u062F\u0645\u0627\u062A",
|
|
507
|
+
"\u062E\u0648\u0627\u0633\u062A",
|
|
508
|
+
"\u062E\u0648\u0627\u0647\u062F",
|
|
509
|
+
"\u062E\u0648\u0627\u0647\u0646\u062F",
|
|
510
|
+
"\u062E\u0648\u0627\u0647\u064A\u0645",
|
|
511
|
+
"\u062E\u0648\u062F",
|
|
512
|
+
"\u062E\u0648\u064A\u0634",
|
|
513
|
+
"\u062E\u064A\u0627\u0647",
|
|
514
|
+
"\u062F\u0627\u062F",
|
|
515
|
+
"\u062F\u0627\u062F\u0646",
|
|
516
|
+
"\u062F\u0627\u062F\u0646\u062F",
|
|
517
|
+
"\u062F\u0627\u062F\u0647",
|
|
518
|
+
"\u062F\u0627\u0631\u062F",
|
|
519
|
+
"\u062F\u0627\u0631\u0646\u062F",
|
|
520
|
+
"\u062F\u0627\u0631\u064A\u0645",
|
|
521
|
+
"\u062F\u0627\u0634\u062A",
|
|
522
|
+
"\u062F\u0627\u0634\u062A\u0646",
|
|
523
|
+
"\u062F\u0627\u0634\u062A\u0646\u062F",
|
|
524
|
+
"\u062F\u0627\u0634\u062A\u0647",
|
|
525
|
+
"\u062F\u0627\u0646\u0633\u062A",
|
|
526
|
+
"\u062F\u0627\u0646\u0646\u062F",
|
|
527
|
+
"\u062F\u0631",
|
|
528
|
+
"\u062F\u0631\u0628\u0627\u0631\u0647",
|
|
529
|
+
"\u062F\u0646\u0628\u0627\u0644",
|
|
530
|
+
"\u062F\u0647",
|
|
531
|
+
"\u062F\u0647\u062F",
|
|
532
|
+
"\u062F\u0647\u0646\u062F",
|
|
533
|
+
"\u062F\u0648",
|
|
534
|
+
"\u062F\u0648\u0645",
|
|
535
|
+
"\u062F\u064A\u062F\u0647",
|
|
536
|
+
"\u062F\u064A\u0631\u0648\u0632",
|
|
537
|
+
"\u062F\u064A\u06AF\u0631",
|
|
538
|
+
"\u062F\u064A\u06AF\u0631\u0627\u0646",
|
|
539
|
+
"\u062F\u064A\u06AF\u0631\u064A",
|
|
540
|
+
"\u0631\u0627",
|
|
541
|
+
"\u0631\u0627\u0647",
|
|
542
|
+
"\u0631\u0641\u062A",
|
|
543
|
+
"\u0631\u0641\u062A\u0647",
|
|
544
|
+
"\u0631\u0648\u0628",
|
|
545
|
+
"\u0631\u0648\u0632\u0647\u0627\u064A",
|
|
546
|
+
"\u0631\u0648\u064A",
|
|
547
|
+
"\u0631\u064A\u0632\u064A",
|
|
548
|
+
"\u0632\u064A\u0627\u062F",
|
|
549
|
+
"\u0632\u064A\u0631",
|
|
550
|
+
"\u0632\u064A\u0631\u0627",
|
|
551
|
+
"\u0633\u0627\u0628\u0642",
|
|
552
|
+
"\u0633\u0627\u062E\u062A\u0647",
|
|
553
|
+
"\u0633\u0627\u0632\u064A",
|
|
554
|
+
"\u0633\u067E\u0633",
|
|
555
|
+
"\u0633\u0631\u0627\u0633\u0631",
|
|
556
|
+
"\u0633\u0631\u064A",
|
|
557
|
+
"\u0633\u0639\u064A",
|
|
558
|
+
"\u0633\u0645\u062A",
|
|
559
|
+
"\u0633\u0648\u0645",
|
|
560
|
+
"\u0633\u0648\u064A",
|
|
561
|
+
"\u0634\u0627\u0646",
|
|
562
|
+
"\u0634\u0627\u064A\u062F",
|
|
563
|
+
"\u0634\u062F",
|
|
564
|
+
"\u0634\u062F\u0646",
|
|
565
|
+
"\u0634\u062F\u0646\u062F",
|
|
566
|
+
"\u0634\u062F\u0647",
|
|
567
|
+
"\u0634\u0634",
|
|
568
|
+
"\u0634\u0645\u0627",
|
|
569
|
+
"\u0634\u0646\u0627\u0633\u064A",
|
|
570
|
+
"\u0634\u0648\u062F",
|
|
571
|
+
"\u0634\u0648\u0646\u062F",
|
|
572
|
+
"\u0635\u0648\u0631\u062A",
|
|
573
|
+
"\u0636\u062F",
|
|
574
|
+
"\u0636\u0645\u0646",
|
|
575
|
+
"\u0637\u0628\u0642",
|
|
576
|
+
"\u0637\u0631\u064A\u0642",
|
|
577
|
+
"\u0637\u0648\u0631",
|
|
578
|
+
"\u0637\u064A",
|
|
579
|
+
"\u0639\u0642\u0628",
|
|
580
|
+
"\u0639\u0644\u062A",
|
|
581
|
+
"\u0639\u0646\u0648\u0627\u0646",
|
|
582
|
+
"\u063A\u064A\u0631",
|
|
583
|
+
"\u0641\u0642\u0637",
|
|
584
|
+
"\u0641\u0643\u0631",
|
|
585
|
+
"\u0641\u0648\u0642",
|
|
586
|
+
"\u0642\u0627\u0628\u0644",
|
|
587
|
+
"\u0642\u0628\u0644",
|
|
588
|
+
"\u0642\u0635\u062F",
|
|
589
|
+
"\u0643\u062C\u0627",
|
|
590
|
+
"\u0643\u062C\u0627\u0633\u062A",
|
|
591
|
+
"\u0643\u062F\u0627\u0645",
|
|
592
|
+
"\u0643\u0631\u062F",
|
|
593
|
+
"\u0643\u0631\u062F\u0645",
|
|
594
|
+
"\u0643\u0631\u062F\u0646",
|
|
595
|
+
"\u0643\u0631\u062F\u0646\u062F",
|
|
596
|
+
"\u0643\u0631\u062F\u0647",
|
|
597
|
+
"\u0643\u0633",
|
|
598
|
+
"\u0643\u0633\u064A",
|
|
599
|
+
"\u0643\u0644",
|
|
600
|
+
"\u0643\u0645\u062A\u0631",
|
|
601
|
+
"\u0643\u0646\u0627\u0631",
|
|
602
|
+
"\u0643\u0646\u062F",
|
|
603
|
+
"\u0643\u0646\u0645",
|
|
604
|
+
"\u0643\u0646\u0646\u062F",
|
|
605
|
+
"\u0643\u0646\u064A\u062F",
|
|
606
|
+
"\u0643\u0646\u064A\u0645",
|
|
607
|
+
"\u0643\u0647",
|
|
608
|
+
"\u0643\u064A",
|
|
609
|
+
"\u06AF\u0630\u0627\u0631\u064A",
|
|
610
|
+
"\u06AF\u0630\u0627\u0634\u062A\u0647",
|
|
611
|
+
"\u06AF\u0631\u062F\u062F",
|
|
612
|
+
"\u06AF\u0631\u0641\u062A",
|
|
613
|
+
"\u06AF\u0631\u0641\u062A\u0647",
|
|
614
|
+
"\u06AF\u0631\u0648\u0647\u064A",
|
|
615
|
+
"\u06AF\u0641\u062A",
|
|
616
|
+
"\u06AF\u0641\u062A\u0647",
|
|
617
|
+
"\u06AF\u0648\u064A\u062F",
|
|
618
|
+
"\u06AF\u0648\u064A\u0646\u062F",
|
|
619
|
+
"\u06AF\u064A\u0631\u062F",
|
|
620
|
+
"\u06AF\u064A\u0631\u064A",
|
|
621
|
+
"\u0644\u0637\u0641\u0627",
|
|
622
|
+
"\u0645\u0627",
|
|
623
|
+
"\u0645\u0627\u0646",
|
|
624
|
+
"\u0645\u0627\u0646\u0646\u062F",
|
|
625
|
+
"\u0645\u062B\u0644",
|
|
626
|
+
"\u0645\u062E\u062A\u0644\u0641",
|
|
627
|
+
"\u0645\u062F\u062A\u064A",
|
|
628
|
+
"\u0645\u0631\u062F\u0645",
|
|
629
|
+
"\u0645\u0631\u0633\u064A",
|
|
630
|
+
"\u0645\u0642\u0627\u0628\u0644",
|
|
631
|
+
"\u0645\u06AF\u0631",
|
|
632
|
+
"\u0645\u0646",
|
|
633
|
+
"\u0645\u0648\u0631\u062F",
|
|
634
|
+
"\u0645\u064A",
|
|
635
|
+
"\u0645\u064A\u0644\u064A\u0627\u0631\u062F",
|
|
636
|
+
"\u0645\u064A\u0644\u064A\u0648\u0646",
|
|
637
|
+
"\u0646\u0627\u0634\u064A",
|
|
638
|
+
"\u0646\u0627\u0645",
|
|
639
|
+
"\u0646\u0628\u0627\u064A\u062F",
|
|
640
|
+
"\u0646\u0628\u0648\u062F",
|
|
641
|
+
"\u0646\u062E\u0633\u062A",
|
|
642
|
+
"\u0646\u062E\u0633\u062A\u064A\u0646",
|
|
643
|
+
"\u0646\u062E\u0648\u0627\u0647\u062F",
|
|
644
|
+
"\u0646\u062F\u0627\u0631\u062F",
|
|
645
|
+
"\u0646\u062F\u0627\u0631\u0646\u062F",
|
|
646
|
+
"\u0646\u062F\u0627\u0634\u062A\u0647",
|
|
647
|
+
"\u0646\u0632\u062F",
|
|
648
|
+
"\u0646\u0632\u062F\u064A\u0643",
|
|
649
|
+
"\u0646\u0634\u0627\u0646",
|
|
650
|
+
"\u0646\u0634\u062F\u0647",
|
|
651
|
+
"\u0646\u0638\u064A\u0631",
|
|
652
|
+
"\u0646\u0643\u0631\u062F\u0647",
|
|
653
|
+
"\u0646\u0645\u0627\u064A\u062F",
|
|
654
|
+
"\u0646\u0645\u064A",
|
|
655
|
+
"\u0646\u0647",
|
|
656
|
+
"\u0646\u0648\u0639\u064A",
|
|
657
|
+
"\u0646\u064A\u0632",
|
|
658
|
+
"\u0646\u064A\u0633\u062A",
|
|
659
|
+
"\u0647",
|
|
660
|
+
"\u0647\u0627",
|
|
661
|
+
"\u0647\u0627\u064A",
|
|
662
|
+
"\u0647\u0627\u064A\u064A",
|
|
663
|
+
"\u0647\u0631",
|
|
664
|
+
"\u0647\u0631\u06AF\u0632",
|
|
665
|
+
"\u0647\u0632\u0627\u0631",
|
|
666
|
+
"\u0647\u0633\u062A",
|
|
667
|
+
"\u0647\u0633\u062A\u0646\u062F",
|
|
668
|
+
"\u0647\u0633\u062A\u064A\u0645",
|
|
669
|
+
"\u0647\u0641\u062A",
|
|
670
|
+
"\u0647\u0645",
|
|
671
|
+
"\u0647\u0645\u0627\u0646",
|
|
672
|
+
"\u0647\u0645\u0686\u0646\u0627\u0646",
|
|
673
|
+
"\u0647\u0645\u0686\u0646\u064A\u0646",
|
|
674
|
+
"\u0647\u0645\u0686\u0648\u0646",
|
|
675
|
+
"\u0647\u0645\u0647",
|
|
676
|
+
"\u0647\u0645\u0648\u0627\u0631\u0647",
|
|
677
|
+
"\u0647\u0645\u064A\u0646",
|
|
678
|
+
"\u0647\u0646\u06AF\u0627\u0645",
|
|
679
|
+
"\u0647\u0646\u06AF\u0627\u0645\u064A",
|
|
680
|
+
"\u0647\u0646\u0648\u0632",
|
|
681
|
+
"\u0647\u064A\u0686",
|
|
682
|
+
"\u0648",
|
|
683
|
+
"\u0648\u0633\u0637",
|
|
684
|
+
"\u0648\u0642\u062A\u064A",
|
|
685
|
+
"\u0648\u0642\u062A\u064A\u0643\u0647",
|
|
686
|
+
"\u0648\u06AF\u0648",
|
|
687
|
+
"\u0648\u0644\u064A",
|
|
688
|
+
"\u0648\u064A",
|
|
689
|
+
"\u064A\u0627",
|
|
690
|
+
"\u064A\u0627\u0628\u062F",
|
|
691
|
+
"\u064A\u0643",
|
|
692
|
+
"\u064A\u0643\u062F\u064A\u06AF\u0631",
|
|
693
|
+
"\u064A\u0643\u064A"
|
|
694
|
+
]);
|
|
695
|
+
var ARABIC_YEH_FORMS = /[\u064A\u0649]/g;
|
|
696
|
+
var ARABIC_KAF = /\u0643/g;
|
|
697
|
+
var TATWEEL_AND_HARAKAT = /[\u0640\u064B-\u0652]/g;
|
|
698
|
+
function normalize(token) {
|
|
699
|
+
return token.replace(ARABIC_YEH_FORMS, "\u06CC").replace(ARABIC_KAF, "\u06A9").replace(TATWEEL_AND_HARAKAT, "");
|
|
700
|
+
}
|
|
701
|
+
var persian = {
|
|
702
|
+
name: "persian",
|
|
703
|
+
revision: "99f39d265efbf595",
|
|
704
|
+
stemmer: stem,
|
|
705
|
+
stopWords: withNormalisedSpellings(stopWords, normalize),
|
|
706
|
+
normalizer: normalize,
|
|
707
|
+
tokenizer: { splitPattern: /[^\u0621-\u065f\u0660-\u0669\u066e-\u06d3\u06f0-\u06f9a-z0-9]+/gi }
|
|
708
|
+
};
|
|
709
|
+
|
|
710
|
+
export { persian };
|