@delali/narsil 0.1.15 → 0.2.0
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/NOTICE +54 -0
- package/README.md +38 -982
- package/dist/adapters/filesystem.mjs +1 -1
- package/dist/adapters/indexeddb.mjs +1 -1
- package/dist/{chunk-WISGW24C.mjs → chunk-AUMKFYTG.mjs} +1 -1
- package/dist/{chunk-HRABJLPB.mjs → chunk-BKAI2WXI.mjs} +1 -0
- package/dist/chunk-GHJWC2E5.mjs +9 -0
- package/dist/chunk-KH3YGMVT.mjs +15 -0
- package/dist/{chunk-ALB35PRQ.mjs → chunk-M2EDG55Y.mjs} +8 -0
- package/dist/{chunk-HMO6OMZ3.mjs → chunk-S6IAXOBA.mjs} +1073 -506
- package/dist/chunk-TGP5M5R7.mjs +880 -0
- package/dist/chunk-U2BKYEC7.mjs +254 -0
- package/dist/{chunk-MEW7HD75.mjs → chunk-XAKNSWIK.mjs} +2543 -1154
- package/dist/{config-CojEPOgS.d.ts → config-P27AtSn4.d.ts} +28 -7
- package/dist/direct-executor-2FV3MEDD.mjs +5 -0
- package/dist/distribution/coordinator/etcd.d.ts +2 -2
- package/dist/distribution/coordinator/etcd.mjs +2 -2
- package/dist/distribution/coordinator/in-memory.d.ts +2 -2
- package/dist/distribution.d.ts +10 -4
- package/dist/distribution.mjs +1875 -1772
- package/dist/embeddings/openai.mjs +1 -1
- package/dist/index.browser.mjs +4566 -1783
- package/dist/index.d.ts +14 -5
- package/dist/index.mjs +6 -5
- package/dist/{language-ONQ3iVDf.d.ts → language-CmJBHf1j.d.ts} +4 -1
- package/dist/languages/albanian.d.ts +5 -0
- package/dist/languages/albanian.mjs +103 -0
- package/dist/languages/amharic.d.ts +5 -0
- package/dist/languages/amharic.mjs +240 -0
- package/dist/languages/arabic.d.ts +1 -1
- package/dist/languages/arabic.mjs +939 -154
- package/dist/languages/armenian.d.ts +1 -1
- package/dist/languages/armenian.mjs +275 -243
- package/dist/languages/azerbaijani.d.ts +5 -0
- package/dist/languages/azerbaijani.mjs +152 -0
- package/dist/languages/bambara.d.ts +5 -0
- package/dist/languages/bambara.mjs +60 -0
- package/dist/languages/basque.d.ts +5 -0
- package/dist/languages/basque.mjs +726 -0
- package/dist/languages/belarusian.d.ts +5 -0
- package/dist/languages/belarusian.mjs +100 -0
- package/dist/languages/bengali.d.ts +5 -0
- package/dist/languages/bengali.mjs +128 -0
- package/dist/languages/bosnian.d.ts +5 -0
- package/dist/languages/bosnian.mjs +197 -0
- package/dist/languages/breton.d.ts +5 -0
- package/dist/languages/breton.mjs +96 -0
- package/dist/languages/bulgarian.d.ts +1 -1
- package/dist/languages/bulgarian.mjs +11 -2
- package/dist/languages/burmese.d.ts +5 -0
- package/dist/languages/burmese.mjs +57 -0
- package/dist/languages/catalan.d.ts +5 -0
- package/dist/languages/catalan.mjs +987 -0
- package/dist/languages/chinese.d.ts +1 -1
- package/dist/languages/chinese.mjs +4 -3
- package/dist/languages/croatian.d.ts +5 -0
- package/dist/languages/croatian.mjs +191 -0
- package/dist/languages/czech.d.ts +5 -0
- package/dist/languages/czech.mjs +628 -0
- package/dist/languages/dagbani.d.ts +1 -1
- package/dist/languages/dagbani.mjs +11 -2
- package/dist/languages/danish.d.ts +1 -1
- package/dist/languages/danish.mjs +215 -110
- package/dist/languages/dutch.d.ts +1 -1
- package/dist/languages/dutch.mjs +1248 -163
- package/dist/languages/english.d.ts +1 -1
- package/dist/languages/english.mjs +2 -1
- package/dist/languages/esperanto.d.ts +5 -0
- package/dist/languages/esperanto.mjs +689 -0
- package/dist/languages/estonian.d.ts +5 -0
- package/dist/languages/estonian.mjs +2160 -0
- package/dist/languages/ewe.d.ts +1 -1
- package/dist/languages/ewe.mjs +11 -2
- package/dist/languages/faroese.d.ts +5 -0
- package/dist/languages/faroese.mjs +93 -0
- package/dist/languages/fijian.d.ts +5 -0
- package/dist/languages/fijian.mjs +55 -0
- package/dist/languages/finnish.d.ts +1 -1
- package/dist/languages/finnish.mjs +528 -263
- package/dist/languages/french.d.ts +1 -1
- package/dist/languages/french.mjs +950 -437
- package/dist/languages/ga.d.ts +1 -1
- package/dist/languages/ga.mjs +2 -1
- package/dist/languages/galician.d.ts +5 -0
- package/dist/languages/galician.mjs +172 -0
- package/dist/languages/georgian.d.ts +5 -0
- package/dist/languages/georgian.mjs +91 -0
- package/dist/languages/german.d.ts +1 -1
- package/dist/languages/german.mjs +432 -162
- package/dist/languages/greek.d.ts +1 -1
- package/dist/languages/greek.mjs +1970 -281
- package/dist/languages/guarani.d.ts +5 -0
- package/dist/languages/guarani.mjs +132 -0
- package/dist/languages/gujarati.d.ts +5 -0
- package/dist/languages/gujarati.mjs +234 -0
- package/dist/languages/haitian-creole.d.ts +5 -0
- package/dist/languages/haitian-creole.mjs +106 -0
- package/dist/languages/hausa.d.ts +1 -1
- package/dist/languages/hausa.mjs +2 -1
- package/dist/languages/hawaiian.d.ts +5 -0
- package/dist/languages/hawaiian.mjs +69 -0
- package/dist/languages/hebrew.d.ts +5 -0
- package/dist/languages/hebrew.mjs +213 -0
- package/dist/languages/hindi.d.ts +1 -1
- package/dist/languages/hindi.mjs +175 -97
- package/dist/languages/hungarian.d.ts +1 -1
- package/dist/languages/hungarian.mjs +412 -326
- package/dist/languages/icelandic.d.ts +5 -0
- package/dist/languages/icelandic.mjs +161 -0
- package/dist/languages/igbo.d.ts +1 -1
- package/dist/languages/igbo.mjs +2 -1
- package/dist/languages/indonesian.d.ts +1 -1
- package/dist/languages/indonesian.mjs +329 -137
- package/dist/languages/irish.d.ts +1 -1
- package/dist/languages/irish.mjs +223 -139
- package/dist/languages/italian.d.ts +1 -1
- package/dist/languages/italian.mjs +703 -406
- package/dist/languages/japanese.d.ts +1 -1
- package/dist/languages/japanese.mjs +5 -108
- package/dist/languages/kannada.d.ts +5 -0
- package/dist/languages/kannada.mjs +94 -0
- package/dist/languages/kazakh.d.ts +5 -0
- package/dist/languages/kazakh.mjs +90 -0
- package/dist/languages/khmer.d.ts +5 -0
- package/dist/languages/khmer.mjs +181 -0
- package/dist/languages/kinyarwanda.d.ts +5 -0
- package/dist/languages/kinyarwanda.mjs +99 -0
- package/dist/languages/kirundi.d.ts +5 -0
- package/dist/languages/kirundi.mjs +92 -0
- package/dist/languages/korean.d.ts +5 -0
- package/dist/languages/korean.mjs +252 -0
- package/dist/languages/kurmanji.d.ts +5 -0
- package/dist/languages/kurmanji.mjs +50 -0
- package/dist/languages/kyrgyz.d.ts +5 -0
- package/dist/languages/kyrgyz.mjs +108 -0
- package/dist/languages/lao.d.ts +5 -0
- package/dist/languages/lao.mjs +63 -0
- package/dist/languages/latin.d.ts +5 -0
- package/dist/languages/latin.mjs +69 -0
- package/dist/languages/latvian.d.ts +5 -0
- package/dist/languages/latvian.mjs +173 -0
- package/dist/languages/lingala.d.ts +5 -0
- package/dist/languages/lingala.mjs +61 -0
- package/dist/languages/lithuanian.d.ts +5 -0
- package/dist/languages/lithuanian.mjs +571 -0
- package/dist/languages/luxembourgish.d.ts +5 -0
- package/dist/languages/luxembourgish.mjs +214 -0
- package/dist/languages/macedonian.d.ts +5 -0
- package/dist/languages/macedonian.mjs +808 -0
- package/dist/languages/malagasy.d.ts +5 -0
- package/dist/languages/malagasy.mjs +67 -0
- package/dist/languages/malay.d.ts +5 -0
- package/dist/languages/malay.mjs +482 -0
- package/dist/languages/malayalam.d.ts +5 -0
- package/dist/languages/malayalam.mjs +122 -0
- package/dist/languages/maltese.d.ts +5 -0
- package/dist/languages/maltese.mjs +139 -0
- package/dist/languages/maori.d.ts +5 -0
- package/dist/languages/maori.mjs +79 -0
- package/dist/languages/marathi.d.ts +5 -0
- package/dist/languages/marathi.mjs +111 -0
- package/dist/languages/nepali.d.ts +1 -1
- package/dist/languages/nepali.mjs +209 -99
- package/dist/languages/norwegian.d.ts +1 -1
- package/dist/languages/norwegian.mjs +218 -85
- package/dist/languages/oromo.d.ts +5 -0
- package/dist/languages/oromo.mjs +78 -0
- package/dist/languages/persian.d.ts +5 -0
- package/dist/languages/persian.mjs +710 -0
- package/dist/languages/polish.d.ts +5 -0
- package/dist/languages/polish.mjs +614 -0
- package/dist/languages/portuguese.d.ts +1 -1
- package/dist/languages/portuguese.mjs +613 -380
- package/dist/languages/punjabi.d.ts +5 -0
- package/dist/languages/punjabi.mjs +197 -0
- package/dist/languages/romanian.d.ts +1 -1
- package/dist/languages/romanian.mjs +561 -382
- package/dist/languages/russian.d.ts +1 -1
- package/dist/languages/russian.mjs +373 -244
- package/dist/languages/samoan.d.ts +5 -0
- package/dist/languages/samoan.mjs +63 -0
- package/dist/languages/sanskrit.d.ts +1 -1
- package/dist/languages/sanskrit.mjs +2 -2
- package/dist/languages/scottish-gaelic.d.ts +5 -0
- package/dist/languages/scottish-gaelic.mjs +370 -0
- package/dist/languages/serbian.d.ts +1 -1
- package/dist/languages/serbian.mjs +3635 -155
- package/dist/languages/shona.d.ts +5 -0
- package/dist/languages/shona.mjs +91 -0
- package/dist/languages/sinhala.d.ts +5 -0
- package/dist/languages/sinhala.mjs +182 -0
- package/dist/languages/slovak.d.ts +5 -0
- package/dist/languages/slovak.mjs +430 -0
- package/dist/languages/slovenian.d.ts +1 -1
- package/dist/languages/slovenian.mjs +11 -2
- package/dist/languages/sorani.d.ts +5 -0
- package/dist/languages/sorani.mjs +156 -0
- package/dist/languages/spanish.d.ts +1 -1
- package/dist/languages/spanish.mjs +676 -374
- package/dist/languages/swahili.d.ts +1 -1
- package/dist/languages/swahili.mjs +2 -1
- package/dist/languages/swedish.d.ts +1 -1
- package/dist/languages/swedish.mjs +237 -95
- package/dist/languages/tagalog.d.ts +5 -0
- package/dist/languages/tagalog.mjs +159 -0
- package/dist/languages/tamil.d.ts +1 -1
- package/dist/languages/tamil.mjs +836 -64
- package/dist/languages/tatar.d.ts +5 -0
- package/dist/languages/tatar.mjs +1017 -0
- package/dist/languages/telugu.d.ts +5 -0
- package/dist/languages/telugu.mjs +62 -0
- package/dist/languages/thai.d.ts +5 -0
- package/dist/languages/thai.mjs +65 -0
- package/dist/languages/tibetan.d.ts +5 -0
- package/dist/languages/tibetan.mjs +119 -0
- package/dist/languages/tigrinya.d.ts +5 -0
- package/dist/languages/tigrinya.mjs +194 -0
- package/dist/languages/tongan.d.ts +5 -0
- package/dist/languages/tongan.mjs +56 -0
- package/dist/languages/turkish.d.ts +1 -1
- package/dist/languages/turkish.mjs +1246 -393
- package/dist/languages/twi.d.ts +1 -1
- package/dist/languages/twi.mjs +13 -2
- package/dist/languages/ukrainian.d.ts +1 -1
- package/dist/languages/ukrainian.mjs +125 -47
- package/dist/languages/urdu.d.ts +5 -0
- package/dist/languages/urdu.mjs +528 -0
- package/dist/languages/vietnamese.d.ts +5 -0
- package/dist/languages/vietnamese.mjs +277 -0
- package/dist/languages/welsh.d.ts +5 -0
- package/dist/languages/welsh.mjs +446 -0
- package/dist/languages/wolof.d.ts +5 -0
- package/dist/languages/wolof.mjs +79 -0
- package/dist/languages/xhosa.d.ts +5 -0
- package/dist/languages/xhosa.mjs +95 -0
- package/dist/languages/yoruba.d.ts +1 -1
- package/dist/languages/yoruba.mjs +1 -0
- package/dist/languages/zulu.d.ts +1 -1
- package/dist/languages/zulu.mjs +2 -1
- package/dist/narsil-CMZRPDlQ.d.ts +159 -0
- package/dist/persistence/durability/checkpoint-worker.d.ts +13 -0
- package/dist/persistence/durability/checkpoint-worker.mjs +6 -5
- package/dist/{schema-DCgkO1j4.d.ts → schema-BnvLvjRp.d.ts} +5 -3
- package/dist/server.d.ts +4 -3
- package/dist/server.mjs +13 -2
- package/dist/{types-DHX6zGeD.d.ts → types-BRTcBgAp.d.ts} +1 -1
- package/dist/workers/entry.mjs +2 -2
- package/package.json +10 -3
- package/dist/chunk-TW6BXTOL.mjs +0 -338
- package/dist/direct-executor-E52FFGDE.mjs +0 -4
- package/dist/narsil-D-84Nvm4.d.ts +0 -72
|
@@ -0,0 +1,987 @@
|
|
|
1
|
+
import { BaseStemmer } from '../chunk-U2BKYEC7.mjs';
|
|
2
|
+
|
|
3
|
+
// src/languages/snowball/catalan.ts
|
|
4
|
+
var a_0 = [
|
|
5
|
+
["", 7],
|
|
6
|
+
["\xB7", 6, 1],
|
|
7
|
+
["\xE0", 1, 2],
|
|
8
|
+
["\xE1", 1, 3],
|
|
9
|
+
["\xE8", 2, 4],
|
|
10
|
+
["\xE9", 2, 5],
|
|
11
|
+
["\xEC", 3, 6],
|
|
12
|
+
["\xED", 3, 7],
|
|
13
|
+
["\xEF", 3, 8],
|
|
14
|
+
["\xF2", 4, 9],
|
|
15
|
+
["\xF3", 4, 10],
|
|
16
|
+
["\xFA", 5, 11],
|
|
17
|
+
["\xFC", 5, 12]
|
|
18
|
+
];
|
|
19
|
+
var a_1 = [
|
|
20
|
+
["la", 1],
|
|
21
|
+
["-la", 1, 1],
|
|
22
|
+
["sela", 1, 2],
|
|
23
|
+
["le", 1],
|
|
24
|
+
["me", 1],
|
|
25
|
+
["-me", 1, 1],
|
|
26
|
+
["se", 1],
|
|
27
|
+
["-te", 1],
|
|
28
|
+
["hi", 1],
|
|
29
|
+
["'hi", 1, 1],
|
|
30
|
+
["li", 1],
|
|
31
|
+
["-li", 1, 1],
|
|
32
|
+
["'l", 1],
|
|
33
|
+
["'m", 1],
|
|
34
|
+
["-m", 1],
|
|
35
|
+
["'n", 1],
|
|
36
|
+
["-n", 1],
|
|
37
|
+
["ho", 1],
|
|
38
|
+
["'ho", 1, 1],
|
|
39
|
+
["lo", 1],
|
|
40
|
+
["selo", 1, 1],
|
|
41
|
+
["'s", 1],
|
|
42
|
+
["las", 1],
|
|
43
|
+
["selas", 1, 1],
|
|
44
|
+
["les", 1],
|
|
45
|
+
["-les", 1, 1],
|
|
46
|
+
["'ls", 1],
|
|
47
|
+
["-ls", 1],
|
|
48
|
+
["'ns", 1],
|
|
49
|
+
["-ns", 1],
|
|
50
|
+
["ens", 1],
|
|
51
|
+
["los", 1],
|
|
52
|
+
["selos", 1, 1],
|
|
53
|
+
["nos", 1],
|
|
54
|
+
["-nos", 1, 1],
|
|
55
|
+
["vos", 1],
|
|
56
|
+
["us", 1],
|
|
57
|
+
["-us", 1, 1],
|
|
58
|
+
["'t", 1]
|
|
59
|
+
];
|
|
60
|
+
var a_2 = [
|
|
61
|
+
["ica", 4],
|
|
62
|
+
["l\xF3gica", 3, 1],
|
|
63
|
+
["enca", 1],
|
|
64
|
+
["ada", 2],
|
|
65
|
+
["ancia", 1],
|
|
66
|
+
["encia", 1],
|
|
67
|
+
["\xE8ncia", 1],
|
|
68
|
+
["\xEDcia", 1],
|
|
69
|
+
["logia", 3],
|
|
70
|
+
["inia", 1],
|
|
71
|
+
["\xEDinia", 1, 1],
|
|
72
|
+
["eria", 1],
|
|
73
|
+
["\xE0ria", 1],
|
|
74
|
+
["at\xF2ria", 1],
|
|
75
|
+
["alla", 1],
|
|
76
|
+
["ella", 1],
|
|
77
|
+
["\xEDvola", 1],
|
|
78
|
+
["ima", 1],
|
|
79
|
+
["\xEDssima", 1, 1],
|
|
80
|
+
["qu\xEDssima", 5, 1],
|
|
81
|
+
["ana", 1],
|
|
82
|
+
["ina", 1],
|
|
83
|
+
["era", 1],
|
|
84
|
+
["sfera", 1, 1],
|
|
85
|
+
["ora", 1],
|
|
86
|
+
["dora", 1, 1],
|
|
87
|
+
["adora", 1, 1],
|
|
88
|
+
["adura", 1],
|
|
89
|
+
["esa", 1],
|
|
90
|
+
["osa", 1],
|
|
91
|
+
["assa", 1],
|
|
92
|
+
["essa", 1],
|
|
93
|
+
["issa", 1],
|
|
94
|
+
["eta", 1],
|
|
95
|
+
["ita", 1],
|
|
96
|
+
["ota", 1],
|
|
97
|
+
["ista", 1],
|
|
98
|
+
["ialista", 1, 1],
|
|
99
|
+
["ionista", 1, 2],
|
|
100
|
+
["iva", 1],
|
|
101
|
+
["ativa", 1, 1],
|
|
102
|
+
["n\xE7a", 1],
|
|
103
|
+
["log\xEDa", 3],
|
|
104
|
+
["ic", 4],
|
|
105
|
+
["\xEDstic", 1, 1],
|
|
106
|
+
["enc", 1],
|
|
107
|
+
["esc", 1],
|
|
108
|
+
["ud", 1],
|
|
109
|
+
["atge", 1],
|
|
110
|
+
["ble", 1],
|
|
111
|
+
["able", 1, 1],
|
|
112
|
+
["ible", 1, 2],
|
|
113
|
+
["isme", 1],
|
|
114
|
+
["ialisme", 1, 1],
|
|
115
|
+
["ionisme", 1, 2],
|
|
116
|
+
["ivisme", 1, 3],
|
|
117
|
+
["aire", 1],
|
|
118
|
+
["icte", 1],
|
|
119
|
+
["iste", 1],
|
|
120
|
+
["ici", 1],
|
|
121
|
+
["\xEDci", 1],
|
|
122
|
+
["logi", 3],
|
|
123
|
+
["ari", 1],
|
|
124
|
+
["tori", 1],
|
|
125
|
+
["al", 1],
|
|
126
|
+
["il", 1],
|
|
127
|
+
["all", 1],
|
|
128
|
+
["ell", 1],
|
|
129
|
+
["\xEDvol", 1],
|
|
130
|
+
["isam", 1],
|
|
131
|
+
["issem", 1],
|
|
132
|
+
["\xECssem", 1],
|
|
133
|
+
["\xEDssem", 1],
|
|
134
|
+
["\xEDssim", 1],
|
|
135
|
+
["qu\xEDssim", 5, 1],
|
|
136
|
+
["amen", 1],
|
|
137
|
+
["\xECssin", 1],
|
|
138
|
+
["ar", 1],
|
|
139
|
+
["ificar", 1, 1],
|
|
140
|
+
["egar", 1, 2],
|
|
141
|
+
["ejar", 1, 3],
|
|
142
|
+
["itar", 1, 4],
|
|
143
|
+
["itzar", 1, 5],
|
|
144
|
+
["fer", 1],
|
|
145
|
+
["or", 1],
|
|
146
|
+
["dor", 1, 1],
|
|
147
|
+
["dur", 1],
|
|
148
|
+
["doras", 1],
|
|
149
|
+
["ics", 4],
|
|
150
|
+
["l\xF3gics", 3, 1],
|
|
151
|
+
["uds", 1],
|
|
152
|
+
["nces", 1],
|
|
153
|
+
["ades", 2],
|
|
154
|
+
["ancies", 1],
|
|
155
|
+
["encies", 1],
|
|
156
|
+
["\xE8ncies", 1],
|
|
157
|
+
["\xEDcies", 1],
|
|
158
|
+
["logies", 3],
|
|
159
|
+
["inies", 1],
|
|
160
|
+
["\xEDnies", 1],
|
|
161
|
+
["eries", 1],
|
|
162
|
+
["\xE0ries", 1],
|
|
163
|
+
["at\xF2ries", 1],
|
|
164
|
+
["bles", 1],
|
|
165
|
+
["ables", 1, 1],
|
|
166
|
+
["ibles", 1, 2],
|
|
167
|
+
["imes", 1],
|
|
168
|
+
["\xEDssimes", 1, 1],
|
|
169
|
+
["qu\xEDssimes", 5, 1],
|
|
170
|
+
["formes", 1],
|
|
171
|
+
["ismes", 1],
|
|
172
|
+
["ialismes", 1, 1],
|
|
173
|
+
["ines", 1],
|
|
174
|
+
["eres", 1],
|
|
175
|
+
["ores", 1],
|
|
176
|
+
["dores", 1, 1],
|
|
177
|
+
["idores", 1, 1],
|
|
178
|
+
["dures", 1],
|
|
179
|
+
["eses", 1],
|
|
180
|
+
["oses", 1],
|
|
181
|
+
["asses", 1],
|
|
182
|
+
["ictes", 1],
|
|
183
|
+
["ites", 1],
|
|
184
|
+
["otes", 1],
|
|
185
|
+
["istes", 1],
|
|
186
|
+
["ialistes", 1, 1],
|
|
187
|
+
["ionistes", 1, 2],
|
|
188
|
+
["iques", 4],
|
|
189
|
+
["l\xF3giques", 3, 1],
|
|
190
|
+
["ives", 1],
|
|
191
|
+
["atives", 1, 1],
|
|
192
|
+
["log\xEDes", 3],
|
|
193
|
+
["alleng\xFCes", 1],
|
|
194
|
+
["icis", 1],
|
|
195
|
+
["\xEDcis", 1],
|
|
196
|
+
["logis", 3],
|
|
197
|
+
["aris", 1],
|
|
198
|
+
["toris", 1],
|
|
199
|
+
["ls", 1],
|
|
200
|
+
["als", 1, 1],
|
|
201
|
+
["ells", 1, 2],
|
|
202
|
+
["ims", 1],
|
|
203
|
+
["\xEDssims", 1, 1],
|
|
204
|
+
["qu\xEDssims", 5, 1],
|
|
205
|
+
["ions", 1],
|
|
206
|
+
["cions", 1, 1],
|
|
207
|
+
["acions", 2, 1],
|
|
208
|
+
["esos", 1],
|
|
209
|
+
["osos", 1],
|
|
210
|
+
["assos", 1],
|
|
211
|
+
["issos", 1],
|
|
212
|
+
["ers", 1],
|
|
213
|
+
["ors", 1],
|
|
214
|
+
["dors", 1, 1],
|
|
215
|
+
["adors", 1, 1],
|
|
216
|
+
["idors", 1, 2],
|
|
217
|
+
["ats", 1],
|
|
218
|
+
["itats", 1, 1],
|
|
219
|
+
["bilitats", 1, 1],
|
|
220
|
+
["ivitats", 1, 2],
|
|
221
|
+
["ativitats", 1, 1],
|
|
222
|
+
["\xEFtats", 1, 5],
|
|
223
|
+
["ets", 1],
|
|
224
|
+
["ants", 1],
|
|
225
|
+
["ents", 1],
|
|
226
|
+
["ments", 1, 1],
|
|
227
|
+
["aments", 1, 1],
|
|
228
|
+
["ots", 1],
|
|
229
|
+
["uts", 1],
|
|
230
|
+
["ius", 1],
|
|
231
|
+
["trius", 1, 1],
|
|
232
|
+
["atius", 1, 2],
|
|
233
|
+
["\xE8s", 1],
|
|
234
|
+
["\xE9s", 1],
|
|
235
|
+
["\xEDs", 1],
|
|
236
|
+
["d\xEDs", 1, 1],
|
|
237
|
+
["\xF3s", 1],
|
|
238
|
+
["itat", 1],
|
|
239
|
+
["bilitat", 1, 1],
|
|
240
|
+
["ivitat", 1, 2],
|
|
241
|
+
["ativitat", 1, 1],
|
|
242
|
+
["\xEFtat", 1],
|
|
243
|
+
["et", 1],
|
|
244
|
+
["ant", 1],
|
|
245
|
+
["ent", 1],
|
|
246
|
+
["ient", 1, 1],
|
|
247
|
+
["ment", 1, 2],
|
|
248
|
+
["ament", 1, 1],
|
|
249
|
+
["isament", 1, 1],
|
|
250
|
+
["ot", 1],
|
|
251
|
+
["isseu", 1],
|
|
252
|
+
["\xECsseu", 1],
|
|
253
|
+
["\xEDsseu", 1],
|
|
254
|
+
["triu", 1],
|
|
255
|
+
["\xEDssiu", 1],
|
|
256
|
+
["atiu", 1],
|
|
257
|
+
["\xF3", 1],
|
|
258
|
+
["i\xF3", 1, 1],
|
|
259
|
+
["ci\xF3", 1, 1],
|
|
260
|
+
["aci\xF3", 1, 1]
|
|
261
|
+
];
|
|
262
|
+
var a_3 = [
|
|
263
|
+
["aba", 1],
|
|
264
|
+
["esca", 1],
|
|
265
|
+
["isca", 1],
|
|
266
|
+
["\xEFsca", 1],
|
|
267
|
+
["ada", 1],
|
|
268
|
+
["ida", 1],
|
|
269
|
+
["uda", 1],
|
|
270
|
+
["\xEFda", 1],
|
|
271
|
+
["ia", 1],
|
|
272
|
+
["aria", 1, 1],
|
|
273
|
+
["iria", 1, 2],
|
|
274
|
+
["ara", 1],
|
|
275
|
+
["iera", 1],
|
|
276
|
+
["ira", 1],
|
|
277
|
+
["adora", 1],
|
|
278
|
+
["\xEFra", 1],
|
|
279
|
+
["ava", 1],
|
|
280
|
+
["ixa", 1],
|
|
281
|
+
["itza", 1],
|
|
282
|
+
["\xEDa", 1],
|
|
283
|
+
["ar\xEDa", 1, 1],
|
|
284
|
+
["er\xEDa", 1, 2],
|
|
285
|
+
["ir\xEDa", 1, 3],
|
|
286
|
+
["\xEFa", 1],
|
|
287
|
+
["isc", 1],
|
|
288
|
+
["\xEFsc", 1],
|
|
289
|
+
["ad", 1],
|
|
290
|
+
["ed", 1],
|
|
291
|
+
["id", 1],
|
|
292
|
+
["ie", 1],
|
|
293
|
+
["re", 1],
|
|
294
|
+
["dre", 1, 1],
|
|
295
|
+
["ase", 1],
|
|
296
|
+
["iese", 1],
|
|
297
|
+
["aste", 1],
|
|
298
|
+
["iste", 1],
|
|
299
|
+
["ii", 1],
|
|
300
|
+
["ini", 1],
|
|
301
|
+
["esqui", 1],
|
|
302
|
+
["eixi", 1],
|
|
303
|
+
["itzi", 1],
|
|
304
|
+
["am", 1],
|
|
305
|
+
["em", 1],
|
|
306
|
+
["arem", 1, 1],
|
|
307
|
+
["irem", 1, 2],
|
|
308
|
+
["\xE0rem", 1, 3],
|
|
309
|
+
["\xEDrem", 1, 4],
|
|
310
|
+
["\xE0ssem", 1, 5],
|
|
311
|
+
["\xE9ssem", 1, 6],
|
|
312
|
+
["iguem", 1, 7],
|
|
313
|
+
["\xEFguem", 1, 8],
|
|
314
|
+
["avem", 1, 9],
|
|
315
|
+
["\xE0vem", 1, 10],
|
|
316
|
+
["\xE1vem", 1, 11],
|
|
317
|
+
["ir\xECem", 1, 12],
|
|
318
|
+
["\xEDem", 1, 13],
|
|
319
|
+
["ar\xEDem", 1, 1],
|
|
320
|
+
["ir\xEDem", 1, 2],
|
|
321
|
+
["assim", 1],
|
|
322
|
+
["essim", 1],
|
|
323
|
+
["issim", 1],
|
|
324
|
+
["\xE0ssim", 1],
|
|
325
|
+
["\xE8ssim", 1],
|
|
326
|
+
["\xE9ssim", 1],
|
|
327
|
+
["\xEDssim", 1],
|
|
328
|
+
["\xEFm", 1],
|
|
329
|
+
["an", 1],
|
|
330
|
+
["aban", 1, 1],
|
|
331
|
+
["arian", 1, 2],
|
|
332
|
+
["aran", 1, 3],
|
|
333
|
+
["ieran", 1, 4],
|
|
334
|
+
["iran", 1, 5],
|
|
335
|
+
["\xEDan", 1, 6],
|
|
336
|
+
["ar\xEDan", 1, 1],
|
|
337
|
+
["er\xEDan", 1, 2],
|
|
338
|
+
["ir\xEDan", 1, 3],
|
|
339
|
+
["en", 1],
|
|
340
|
+
["ien", 1, 1],
|
|
341
|
+
["arien", 1, 1],
|
|
342
|
+
["irien", 1, 2],
|
|
343
|
+
["aren", 1, 4],
|
|
344
|
+
["eren", 1, 5],
|
|
345
|
+
["iren", 1, 6],
|
|
346
|
+
["\xE0ren", 1, 7],
|
|
347
|
+
["\xEFren", 1, 8],
|
|
348
|
+
["asen", 1, 9],
|
|
349
|
+
["iesen", 1, 10],
|
|
350
|
+
["assen", 1, 11],
|
|
351
|
+
["essen", 1, 12],
|
|
352
|
+
["issen", 1, 13],
|
|
353
|
+
["\xE9ssen", 1, 14],
|
|
354
|
+
["\xEFssen", 1, 15],
|
|
355
|
+
["esquen", 1, 16],
|
|
356
|
+
["isquen", 1, 17],
|
|
357
|
+
["\xEFsquen", 1, 18],
|
|
358
|
+
["aven", 1, 19],
|
|
359
|
+
["ixen", 1, 20],
|
|
360
|
+
["eixen", 1, 1],
|
|
361
|
+
["\xEFxen", 1, 22],
|
|
362
|
+
["\xEFen", 1, 23],
|
|
363
|
+
["in", 1],
|
|
364
|
+
["inin", 1, 1],
|
|
365
|
+
["sin", 1, 2],
|
|
366
|
+
["isin", 1, 1],
|
|
367
|
+
["assin", 1, 2],
|
|
368
|
+
["essin", 1, 3],
|
|
369
|
+
["issin", 1, 4],
|
|
370
|
+
["\xEFssin", 1, 5],
|
|
371
|
+
["esquin", 1, 8],
|
|
372
|
+
["eixin", 1, 9],
|
|
373
|
+
["aron", 1],
|
|
374
|
+
["ieron", 1],
|
|
375
|
+
["ar\xE1n", 1],
|
|
376
|
+
["er\xE1n", 1],
|
|
377
|
+
["ir\xE1n", 1],
|
|
378
|
+
["i\xEFn", 1],
|
|
379
|
+
["ado", 1],
|
|
380
|
+
["ido", 1],
|
|
381
|
+
["ando", 2],
|
|
382
|
+
["iendo", 1],
|
|
383
|
+
["io", 1],
|
|
384
|
+
["ixo", 1],
|
|
385
|
+
["eixo", 1, 1],
|
|
386
|
+
["\xEFxo", 1],
|
|
387
|
+
["itzo", 1],
|
|
388
|
+
["ar", 1],
|
|
389
|
+
["tzar", 1, 1],
|
|
390
|
+
["er", 1],
|
|
391
|
+
["eixer", 1, 1],
|
|
392
|
+
["ir", 1],
|
|
393
|
+
["ador", 1],
|
|
394
|
+
["as", 1],
|
|
395
|
+
["abas", 1, 1],
|
|
396
|
+
["adas", 1, 2],
|
|
397
|
+
["idas", 1, 3],
|
|
398
|
+
["aras", 1, 4],
|
|
399
|
+
["ieras", 1, 5],
|
|
400
|
+
["\xEDas", 1, 6],
|
|
401
|
+
["ar\xEDas", 1, 1],
|
|
402
|
+
["er\xEDas", 1, 2],
|
|
403
|
+
["ir\xEDas", 1, 3],
|
|
404
|
+
["ids", 1],
|
|
405
|
+
["es", 1],
|
|
406
|
+
["ades", 1, 1],
|
|
407
|
+
["ides", 1, 2],
|
|
408
|
+
["udes", 1, 3],
|
|
409
|
+
["\xEFdes", 1, 4],
|
|
410
|
+
["atges", 1, 5],
|
|
411
|
+
["ies", 1, 6],
|
|
412
|
+
["aries", 1, 1],
|
|
413
|
+
["iries", 1, 2],
|
|
414
|
+
["ares", 1, 9],
|
|
415
|
+
["ires", 1, 10],
|
|
416
|
+
["adores", 1, 11],
|
|
417
|
+
["\xEFres", 1, 12],
|
|
418
|
+
["ases", 1, 13],
|
|
419
|
+
["ieses", 1, 14],
|
|
420
|
+
["asses", 1, 15],
|
|
421
|
+
["esses", 1, 16],
|
|
422
|
+
["isses", 1, 17],
|
|
423
|
+
["\xEFsses", 1, 18],
|
|
424
|
+
["ques", 1, 19],
|
|
425
|
+
["esques", 1, 1],
|
|
426
|
+
["\xEFsques", 1, 2],
|
|
427
|
+
["aves", 1, 22],
|
|
428
|
+
["ixes", 1, 23],
|
|
429
|
+
["eixes", 1, 1],
|
|
430
|
+
["\xEFxes", 1, 25],
|
|
431
|
+
["\xEFes", 1, 26],
|
|
432
|
+
["abais", 1],
|
|
433
|
+
["arais", 1],
|
|
434
|
+
["ierais", 1],
|
|
435
|
+
["\xEDais", 1],
|
|
436
|
+
["ar\xEDais", 1, 1],
|
|
437
|
+
["er\xEDais", 1, 2],
|
|
438
|
+
["ir\xEDais", 1, 3],
|
|
439
|
+
["aseis", 1],
|
|
440
|
+
["ieseis", 1],
|
|
441
|
+
["asteis", 1],
|
|
442
|
+
["isteis", 1],
|
|
443
|
+
["inis", 1],
|
|
444
|
+
["sis", 1],
|
|
445
|
+
["isis", 1, 1],
|
|
446
|
+
["assis", 1, 2],
|
|
447
|
+
["essis", 1, 3],
|
|
448
|
+
["issis", 1, 4],
|
|
449
|
+
["\xEFssis", 1, 5],
|
|
450
|
+
["esquis", 1],
|
|
451
|
+
["eixis", 1],
|
|
452
|
+
["itzis", 1],
|
|
453
|
+
["\xE1is", 1],
|
|
454
|
+
["ar\xE9is", 1],
|
|
455
|
+
["er\xE9is", 1],
|
|
456
|
+
["ir\xE9is", 1],
|
|
457
|
+
["ams", 1],
|
|
458
|
+
["ados", 1],
|
|
459
|
+
["idos", 1],
|
|
460
|
+
["amos", 1],
|
|
461
|
+
["\xE1bamos", 1, 1],
|
|
462
|
+
["\xE1ramos", 1, 2],
|
|
463
|
+
["i\xE9ramos", 1, 3],
|
|
464
|
+
["\xEDamos", 1, 4],
|
|
465
|
+
["ar\xEDamos", 1, 1],
|
|
466
|
+
["er\xEDamos", 1, 2],
|
|
467
|
+
["ir\xEDamos", 1, 3],
|
|
468
|
+
["aremos", 1],
|
|
469
|
+
["eremos", 1],
|
|
470
|
+
["iremos", 1],
|
|
471
|
+
["\xE1semos", 1],
|
|
472
|
+
["i\xE9semos", 1],
|
|
473
|
+
["imos", 1],
|
|
474
|
+
["adors", 1],
|
|
475
|
+
["ass", 1],
|
|
476
|
+
["erass", 1, 1],
|
|
477
|
+
["ess", 1],
|
|
478
|
+
["ats", 1],
|
|
479
|
+
["its", 1],
|
|
480
|
+
["ents", 1],
|
|
481
|
+
["\xE0s", 1],
|
|
482
|
+
["ar\xE0s", 1, 1],
|
|
483
|
+
["ir\xE0s", 1, 2],
|
|
484
|
+
["ar\xE1s", 1],
|
|
485
|
+
["er\xE1s", 1],
|
|
486
|
+
["ir\xE1s", 1],
|
|
487
|
+
["\xE9s", 1],
|
|
488
|
+
["ar\xE9s", 1, 1],
|
|
489
|
+
["\xEDs", 1],
|
|
490
|
+
["i\xEFs", 1],
|
|
491
|
+
["at", 1],
|
|
492
|
+
["it", 1],
|
|
493
|
+
["ant", 1],
|
|
494
|
+
["ent", 1],
|
|
495
|
+
["int", 1],
|
|
496
|
+
["ut", 1],
|
|
497
|
+
["\xEFt", 1],
|
|
498
|
+
["au", 1],
|
|
499
|
+
["erau", 1, 1],
|
|
500
|
+
["ieu", 1],
|
|
501
|
+
["ineu", 1],
|
|
502
|
+
["areu", 1],
|
|
503
|
+
["ireu", 1],
|
|
504
|
+
["\xE0reu", 1],
|
|
505
|
+
["\xEDreu", 1],
|
|
506
|
+
["asseu", 1],
|
|
507
|
+
["esseu", 1],
|
|
508
|
+
["eresseu", 1, 1],
|
|
509
|
+
["\xE0sseu", 1],
|
|
510
|
+
["\xE9sseu", 1],
|
|
511
|
+
["igueu", 1],
|
|
512
|
+
["\xEFgueu", 1],
|
|
513
|
+
["\xE0veu", 1],
|
|
514
|
+
["\xE1veu", 1],
|
|
515
|
+
["itzeu", 1],
|
|
516
|
+
["\xECeu", 1],
|
|
517
|
+
["ir\xECeu", 1, 1],
|
|
518
|
+
["\xEDeu", 1],
|
|
519
|
+
["ar\xEDeu", 1, 1],
|
|
520
|
+
["ir\xEDeu", 1, 2],
|
|
521
|
+
["assiu", 1],
|
|
522
|
+
["issiu", 1],
|
|
523
|
+
["\xE0ssiu", 1],
|
|
524
|
+
["\xE8ssiu", 1],
|
|
525
|
+
["\xE9ssiu", 1],
|
|
526
|
+
["\xEDssiu", 1],
|
|
527
|
+
["\xEFu", 1],
|
|
528
|
+
["ix", 1],
|
|
529
|
+
["eix", 1, 1],
|
|
530
|
+
["\xEFx", 1],
|
|
531
|
+
["itz", 1],
|
|
532
|
+
["i\xE0", 1],
|
|
533
|
+
["ar\xE0", 1],
|
|
534
|
+
["ir\xE0", 1],
|
|
535
|
+
["itz\xE0", 1],
|
|
536
|
+
["ar\xE1", 1],
|
|
537
|
+
["er\xE1", 1],
|
|
538
|
+
["ir\xE1", 1],
|
|
539
|
+
["ir\xE8", 1],
|
|
540
|
+
["ar\xE9", 1],
|
|
541
|
+
["er\xE9", 1],
|
|
542
|
+
["ir\xE9", 1],
|
|
543
|
+
["\xED", 1],
|
|
544
|
+
["i\xEF", 1],
|
|
545
|
+
["i\xF3", 1]
|
|
546
|
+
];
|
|
547
|
+
var a_4 = [
|
|
548
|
+
["a", 1],
|
|
549
|
+
["e", 1],
|
|
550
|
+
["i", 1],
|
|
551
|
+
["\xEFn", 1],
|
|
552
|
+
["o", 1],
|
|
553
|
+
["ir", 1],
|
|
554
|
+
["s", 1],
|
|
555
|
+
["is", 1, 1],
|
|
556
|
+
["os", 1, 2],
|
|
557
|
+
["\xEFs", 1, 3],
|
|
558
|
+
["it", 1],
|
|
559
|
+
["eu", 1],
|
|
560
|
+
["iu", 1],
|
|
561
|
+
["iqu", 2],
|
|
562
|
+
["itz", 1],
|
|
563
|
+
["\xE0", 1],
|
|
564
|
+
["\xE1", 1],
|
|
565
|
+
["\xE9", 1],
|
|
566
|
+
["\xEC", 1],
|
|
567
|
+
["\xED", 1],
|
|
568
|
+
["\xEF", 1],
|
|
569
|
+
["\xF3", 1]
|
|
570
|
+
];
|
|
571
|
+
var g_v = [17, 65, 16, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 128, 129, 81, 6, 10];
|
|
572
|
+
var CatalanStemmer = class extends BaseStemmer {
|
|
573
|
+
#stem() {
|
|
574
|
+
let a;
|
|
575
|
+
let I_p2;
|
|
576
|
+
let I_p1;
|
|
577
|
+
{
|
|
578
|
+
I_p1 = this.limit;
|
|
579
|
+
I_p2 = this.limit;
|
|
580
|
+
const v_1 = this.c;
|
|
581
|
+
lab1: {
|
|
582
|
+
if (!this.go_out_grouping(g_v, 97, 252)) break lab1;
|
|
583
|
+
this.c++;
|
|
584
|
+
if (!this.go_in_grouping(g_v, 97, 252)) break lab1;
|
|
585
|
+
this.c++;
|
|
586
|
+
I_p1 = this.c;
|
|
587
|
+
if (!this.go_out_grouping(g_v, 97, 252)) break lab1;
|
|
588
|
+
this.c++;
|
|
589
|
+
if (!this.go_in_grouping(g_v, 97, 252)) break lab1;
|
|
590
|
+
this.c++;
|
|
591
|
+
I_p2 = this.c;
|
|
592
|
+
}
|
|
593
|
+
this.c = v_1;
|
|
594
|
+
}
|
|
595
|
+
this.limit_backward = this.c;
|
|
596
|
+
this.c = this.limit;
|
|
597
|
+
const v_2 = this.limit - this.c;
|
|
598
|
+
lab2: {
|
|
599
|
+
this.ket = this.c;
|
|
600
|
+
if (this.find_among_b(a_1) === 0) break lab2;
|
|
601
|
+
this.bra = this.c;
|
|
602
|
+
if (
|
|
603
|
+
/**@type {boolean}*/
|
|
604
|
+
I_p1 > this.c
|
|
605
|
+
) break lab2;
|
|
606
|
+
this.slice_del();
|
|
607
|
+
}
|
|
608
|
+
this.c = this.limit - v_2;
|
|
609
|
+
const v_3 = this.limit - this.c;
|
|
610
|
+
lab3: {
|
|
611
|
+
lab4: {
|
|
612
|
+
const v_4 = this.limit - this.c;
|
|
613
|
+
lab5: {
|
|
614
|
+
this.ket = this.c;
|
|
615
|
+
a = this.find_among_b(a_2);
|
|
616
|
+
if (a === 0) break lab5;
|
|
617
|
+
this.bra = this.c;
|
|
618
|
+
switch (a) {
|
|
619
|
+
case 1: {
|
|
620
|
+
if (
|
|
621
|
+
/**@type {boolean}*/
|
|
622
|
+
I_p1 > this.c
|
|
623
|
+
) break lab5;
|
|
624
|
+
this.slice_del();
|
|
625
|
+
break;
|
|
626
|
+
}
|
|
627
|
+
case 2: {
|
|
628
|
+
if (
|
|
629
|
+
/**@type {boolean}*/
|
|
630
|
+
I_p2 > this.c
|
|
631
|
+
) break lab5;
|
|
632
|
+
this.slice_del();
|
|
633
|
+
break;
|
|
634
|
+
}
|
|
635
|
+
case 3: {
|
|
636
|
+
if (
|
|
637
|
+
/**@type {boolean}*/
|
|
638
|
+
I_p2 > this.c
|
|
639
|
+
) break lab5;
|
|
640
|
+
this.slice_from("log");
|
|
641
|
+
break;
|
|
642
|
+
}
|
|
643
|
+
case 4: {
|
|
644
|
+
if (
|
|
645
|
+
/**@type {boolean}*/
|
|
646
|
+
I_p2 > this.c
|
|
647
|
+
) break lab5;
|
|
648
|
+
this.slice_from("ic");
|
|
649
|
+
break;
|
|
650
|
+
}
|
|
651
|
+
case 5: {
|
|
652
|
+
if (
|
|
653
|
+
/**@type {boolean}*/
|
|
654
|
+
I_p1 > this.c
|
|
655
|
+
) break lab5;
|
|
656
|
+
this.slice_from("c");
|
|
657
|
+
break;
|
|
658
|
+
}
|
|
659
|
+
}
|
|
660
|
+
break lab4;
|
|
661
|
+
}
|
|
662
|
+
this.c = this.limit - v_4;
|
|
663
|
+
this.ket = this.c;
|
|
664
|
+
a = this.find_among_b(a_3);
|
|
665
|
+
if (a === 0) break lab3;
|
|
666
|
+
this.bra = this.c;
|
|
667
|
+
switch (a) {
|
|
668
|
+
case 1: {
|
|
669
|
+
if (
|
|
670
|
+
/**@type {boolean}*/
|
|
671
|
+
I_p1 > this.c
|
|
672
|
+
) break lab3;
|
|
673
|
+
this.slice_del();
|
|
674
|
+
break;
|
|
675
|
+
}
|
|
676
|
+
case 2: {
|
|
677
|
+
if (
|
|
678
|
+
/**@type {boolean}*/
|
|
679
|
+
I_p2 > this.c
|
|
680
|
+
) break lab3;
|
|
681
|
+
this.slice_del();
|
|
682
|
+
break;
|
|
683
|
+
}
|
|
684
|
+
}
|
|
685
|
+
}
|
|
686
|
+
}
|
|
687
|
+
this.c = this.limit - v_3;
|
|
688
|
+
const v_5 = this.limit - this.c;
|
|
689
|
+
lab6: {
|
|
690
|
+
this.ket = this.c;
|
|
691
|
+
a = this.find_among_b(a_4);
|
|
692
|
+
if (a === 0) break lab6;
|
|
693
|
+
this.bra = this.c;
|
|
694
|
+
switch (a) {
|
|
695
|
+
case 1: {
|
|
696
|
+
if (
|
|
697
|
+
/**@type {boolean}*/
|
|
698
|
+
I_p1 > this.c
|
|
699
|
+
) break lab6;
|
|
700
|
+
this.slice_del();
|
|
701
|
+
break;
|
|
702
|
+
}
|
|
703
|
+
case 2: {
|
|
704
|
+
if (
|
|
705
|
+
/**@type {boolean}*/
|
|
706
|
+
I_p1 > this.c
|
|
707
|
+
) break lab6;
|
|
708
|
+
this.slice_from("ic");
|
|
709
|
+
break;
|
|
710
|
+
}
|
|
711
|
+
}
|
|
712
|
+
}
|
|
713
|
+
this.c = this.limit - v_5;
|
|
714
|
+
this.c = this.limit_backward;
|
|
715
|
+
const v_6 = this.c;
|
|
716
|
+
while (true) {
|
|
717
|
+
const v_7 = this.c;
|
|
718
|
+
lab8: {
|
|
719
|
+
this.bra = this.c;
|
|
720
|
+
a = this.find_among(a_0);
|
|
721
|
+
this.ket = this.c;
|
|
722
|
+
switch (a) {
|
|
723
|
+
case 1: {
|
|
724
|
+
this.slice_from("a");
|
|
725
|
+
break;
|
|
726
|
+
}
|
|
727
|
+
case 2: {
|
|
728
|
+
this.slice_from("e");
|
|
729
|
+
break;
|
|
730
|
+
}
|
|
731
|
+
case 3: {
|
|
732
|
+
this.slice_from("i");
|
|
733
|
+
break;
|
|
734
|
+
}
|
|
735
|
+
case 4: {
|
|
736
|
+
this.slice_from("o");
|
|
737
|
+
break;
|
|
738
|
+
}
|
|
739
|
+
case 5: {
|
|
740
|
+
this.slice_from("u");
|
|
741
|
+
break;
|
|
742
|
+
}
|
|
743
|
+
case 6: {
|
|
744
|
+
this.slice_from(".");
|
|
745
|
+
break;
|
|
746
|
+
}
|
|
747
|
+
case 7: {
|
|
748
|
+
if (this.c >= this.limit) break lab8;
|
|
749
|
+
this.c++;
|
|
750
|
+
break;
|
|
751
|
+
}
|
|
752
|
+
}
|
|
753
|
+
continue;
|
|
754
|
+
}
|
|
755
|
+
this.c = v_7;
|
|
756
|
+
break;
|
|
757
|
+
}
|
|
758
|
+
this.c = v_6;
|
|
759
|
+
return true;
|
|
760
|
+
}
|
|
761
|
+
stem(input) {
|
|
762
|
+
this.setCurrent(input);
|
|
763
|
+
this.#stem();
|
|
764
|
+
return this.getCurrent();
|
|
765
|
+
}
|
|
766
|
+
};
|
|
767
|
+
var shared = new CatalanStemmer();
|
|
768
|
+
function stem(token) {
|
|
769
|
+
return shared.stem(token);
|
|
770
|
+
}
|
|
771
|
+
|
|
772
|
+
// src/languages/catalan.ts
|
|
773
|
+
var stopWords = /* @__PURE__ */ new Set([
|
|
774
|
+
"a",
|
|
775
|
+
"abans",
|
|
776
|
+
"ac\xED",
|
|
777
|
+
"ah",
|
|
778
|
+
"aix\xED",
|
|
779
|
+
"aix\xF2",
|
|
780
|
+
"al",
|
|
781
|
+
"aleshores",
|
|
782
|
+
"algun",
|
|
783
|
+
"alguna",
|
|
784
|
+
"algunes",
|
|
785
|
+
"alguns",
|
|
786
|
+
"alhora",
|
|
787
|
+
"all\xE0",
|
|
788
|
+
"all\xED",
|
|
789
|
+
"all\xF2",
|
|
790
|
+
"als",
|
|
791
|
+
"altra",
|
|
792
|
+
"altre",
|
|
793
|
+
"altres",
|
|
794
|
+
"amb",
|
|
795
|
+
"ambd\xF3s",
|
|
796
|
+
"ambdues",
|
|
797
|
+
"apa",
|
|
798
|
+
"aquell",
|
|
799
|
+
"aquella",
|
|
800
|
+
"aquelles",
|
|
801
|
+
"aquells",
|
|
802
|
+
"aquest",
|
|
803
|
+
"aquesta",
|
|
804
|
+
"aquestes",
|
|
805
|
+
"aquests",
|
|
806
|
+
"aqu\xED",
|
|
807
|
+
"baix",
|
|
808
|
+
"cada",
|
|
809
|
+
"cadasc\xFA",
|
|
810
|
+
"cadascuna",
|
|
811
|
+
"cadascunes",
|
|
812
|
+
"cadascuns",
|
|
813
|
+
"com",
|
|
814
|
+
"contra",
|
|
815
|
+
"dalt",
|
|
816
|
+
"de",
|
|
817
|
+
"del",
|
|
818
|
+
"dels",
|
|
819
|
+
"des",
|
|
820
|
+
"despr\xE9s",
|
|
821
|
+
"dins",
|
|
822
|
+
"dintre",
|
|
823
|
+
"donat",
|
|
824
|
+
"doncs",
|
|
825
|
+
"durant",
|
|
826
|
+
"e",
|
|
827
|
+
"eh",
|
|
828
|
+
"el",
|
|
829
|
+
"els",
|
|
830
|
+
"em",
|
|
831
|
+
"en",
|
|
832
|
+
"encara",
|
|
833
|
+
"ens",
|
|
834
|
+
"entre",
|
|
835
|
+
"\xE9rem",
|
|
836
|
+
"eren",
|
|
837
|
+
"\xE9reu",
|
|
838
|
+
"es",
|
|
839
|
+
"\xE9s",
|
|
840
|
+
"esta",
|
|
841
|
+
"est\xE0",
|
|
842
|
+
"est\xE0vem",
|
|
843
|
+
"estaven",
|
|
844
|
+
"est\xE0veu",
|
|
845
|
+
"esteu",
|
|
846
|
+
"et",
|
|
847
|
+
"etc",
|
|
848
|
+
"ets",
|
|
849
|
+
"fins",
|
|
850
|
+
"fora",
|
|
851
|
+
"gaireb\xE9",
|
|
852
|
+
"ha",
|
|
853
|
+
"han",
|
|
854
|
+
"has",
|
|
855
|
+
"havia",
|
|
856
|
+
"he",
|
|
857
|
+
"hem",
|
|
858
|
+
"heu",
|
|
859
|
+
"hi",
|
|
860
|
+
"ho",
|
|
861
|
+
"i",
|
|
862
|
+
"igual",
|
|
863
|
+
"iguals",
|
|
864
|
+
"ja",
|
|
865
|
+
"la",
|
|
866
|
+
"les",
|
|
867
|
+
"li",
|
|
868
|
+
"llavors",
|
|
869
|
+
"ma",
|
|
870
|
+
"mal",
|
|
871
|
+
"malgrat",
|
|
872
|
+
"mateix",
|
|
873
|
+
"mateixa",
|
|
874
|
+
"mateixes",
|
|
875
|
+
"mateixos",
|
|
876
|
+
"me",
|
|
877
|
+
"mentre",
|
|
878
|
+
"m\xE9s",
|
|
879
|
+
"meu",
|
|
880
|
+
"meus",
|
|
881
|
+
"meva",
|
|
882
|
+
"meves",
|
|
883
|
+
"molt",
|
|
884
|
+
"molta",
|
|
885
|
+
"moltes",
|
|
886
|
+
"molts",
|
|
887
|
+
"mon",
|
|
888
|
+
"mons",
|
|
889
|
+
"ne",
|
|
890
|
+
"ni",
|
|
891
|
+
"no",
|
|
892
|
+
"nogensmenys",
|
|
893
|
+
"nom\xE9s",
|
|
894
|
+
"nosaltres",
|
|
895
|
+
"nostra",
|
|
896
|
+
"nostre",
|
|
897
|
+
"nostres",
|
|
898
|
+
"o",
|
|
899
|
+
"oh",
|
|
900
|
+
"oi",
|
|
901
|
+
"on",
|
|
902
|
+
"pas",
|
|
903
|
+
"pel",
|
|
904
|
+
"pels",
|
|
905
|
+
"per",
|
|
906
|
+
"per\xF2",
|
|
907
|
+
"perqu\xE8",
|
|
908
|
+
"poc",
|
|
909
|
+
"poca",
|
|
910
|
+
"pocs",
|
|
911
|
+
"poques",
|
|
912
|
+
"potser",
|
|
913
|
+
"propi",
|
|
914
|
+
"qual",
|
|
915
|
+
"quals",
|
|
916
|
+
"quan",
|
|
917
|
+
"quant",
|
|
918
|
+
"que",
|
|
919
|
+
"qu\xE8",
|
|
920
|
+
"quelcom",
|
|
921
|
+
"qui",
|
|
922
|
+
"quin",
|
|
923
|
+
"quina",
|
|
924
|
+
"quines",
|
|
925
|
+
"quins",
|
|
926
|
+
"sa",
|
|
927
|
+
"semblant",
|
|
928
|
+
"semblants",
|
|
929
|
+
"ses",
|
|
930
|
+
"seu",
|
|
931
|
+
"seus",
|
|
932
|
+
"seva",
|
|
933
|
+
"seves",
|
|
934
|
+
"si",
|
|
935
|
+
"sobre",
|
|
936
|
+
"sobretot",
|
|
937
|
+
"s\xF3c",
|
|
938
|
+
"solament",
|
|
939
|
+
"sols",
|
|
940
|
+
"son",
|
|
941
|
+
"s\xF3n",
|
|
942
|
+
"sons",
|
|
943
|
+
"sota",
|
|
944
|
+
"sou",
|
|
945
|
+
"ta",
|
|
946
|
+
"tal",
|
|
947
|
+
"tamb\xE9",
|
|
948
|
+
"tampoc",
|
|
949
|
+
"tan",
|
|
950
|
+
"tant",
|
|
951
|
+
"tanta",
|
|
952
|
+
"tantes",
|
|
953
|
+
"teu",
|
|
954
|
+
"teus",
|
|
955
|
+
"teva",
|
|
956
|
+
"teves",
|
|
957
|
+
"ton",
|
|
958
|
+
"tons",
|
|
959
|
+
"tot",
|
|
960
|
+
"tota",
|
|
961
|
+
"totes",
|
|
962
|
+
"tots",
|
|
963
|
+
"un",
|
|
964
|
+
"una",
|
|
965
|
+
"unes",
|
|
966
|
+
"uns",
|
|
967
|
+
"us",
|
|
968
|
+
"va",
|
|
969
|
+
"vaig",
|
|
970
|
+
"vam",
|
|
971
|
+
"van",
|
|
972
|
+
"vas",
|
|
973
|
+
"veu",
|
|
974
|
+
"vosaltres",
|
|
975
|
+
"vostra",
|
|
976
|
+
"vostre",
|
|
977
|
+
"vostres"
|
|
978
|
+
]);
|
|
979
|
+
var catalan = {
|
|
980
|
+
name: "catalan",
|
|
981
|
+
revision: "1be0db9e03d63aa3",
|
|
982
|
+
stemmer: stem,
|
|
983
|
+
stopWords,
|
|
984
|
+
tokenizer: { splitPattern: /[^a-z0-9àçèéíïòóúü]+/gi }
|
|
985
|
+
};
|
|
986
|
+
|
|
987
|
+
export { catalan };
|