| blob | 0a6328ac1b2f672c7c378aa9c83739dc17f835da |
1 // Alias: hy
2 // From: http://snowballstem.org/algorithms/armenian/stemmer.html
3 // Author: Astghik Mkrtchyan
4 // Licence: 3-clause BSD, as per http://snowballstem.org/license.html
6 stringescapes {}
8 stringdef a hex '561' // 531
9 stringdef b hex '562' // 532
10 stringdef g hex '563' // 533
11 stringdef d hex '564' // 534
12 stringdef ye hex '565' // 535
13 stringdef z hex '566' // 536
14 stringdef e hex '567' // 537
15 stringdef y hex '568' // 538
16 stringdef dt hex '569' // 539
17 stringdef zh hex '56A' // 53A
18 stringdef i hex '56B' // 53B
19 stringdef l hex '56C' // 53C
20 stringdef kh hex '56D' // 53D
21 stringdef ts hex '56E' // 53E
22 stringdef k hex '56F' // 53F
23 stringdef h hex '570' // 540
24 stringdef dz hex '571' // 541
25 stringdef gh hex '572' // 542
26 stringdef djch hex '573' // 543
27 stringdef m hex '574' // 544
28 stringdef j hex '575' // 545
29 stringdef n hex '576' // 546
30 stringdef sh hex '577' // 547
31 stringdef vo hex '578' // 548
32 stringdef ch hex '579' // 549
33 stringdef p hex '57A' // 54A
34 stringdef dj hex '57B' // 54B
35 stringdef r hex '57C' // 54C
36 stringdef s hex '57D' // 54D
37 stringdef v hex '57E' // 54E
38 stringdef t hex '57F' // 54F
39 stringdef r' hex '580' // 550
40 stringdef c hex '581' // 551
41 stringdef u hex '582' // 552 //vjun
42 stringdef bp hex '583' // 553
43 stringdef q hex '584' // 554
44 stringdef ev hex '587'
45 stringdef o hex '585' // 555
46 stringdef f hex '586' // 556
48 routines ( mark_regions R2
49 adjective
50 verb
51 noun
52 ending
53 )
55 externals ( stem )
57 integers ( pV p2 )
59 groupings ( v )
61 define v '{a}{e}{i}{o}{u}{ye}{vo}{y}'
63 define mark_regions as (
65 $pV = limit
66 $p2 = limit
67 do (
68 gopast v setmark pV gopast non-v
69 gopast v gopast non-v setmark p2
70 )
71 )
73 backwardmode (
75 define R2 as $p2 <= cursor
77 define adjective as (
78 [substring] among (
79 '{b}{a}{r'}'
80 '{p}{ye}{s}'
81 '{vo}{r'}{e}{n}'
82 '{vo}{v}{i}{n}'
83 '{a}{k}{i}'
84 '{l}{a}{j}{n}'
85 '{r'}{vo}{r'}{d}'
86 '{ye}{r'}{vo}{r'}{d}'
87 '{a}{k}{a}{n}'
88 '{a}{l}{i}'
89 '{k}{vo}{t}'
90 '{ye}{k}{ye}{n}'
91 '{vo}{r'}{a}{k}'
92 '{ye}{gh}'
93 '{v}{vo}{u}{n}'
94 '{ye}{r'}{ye}{n}'
95 '{a}{r'}{a}{n}'
96 '{ye}{n}'
97 '{a}{v}{ye}{t}'
98 '{g}{i}{n}'
99 '{i}{v}'
100 '{a}{t}'
101 '{i}{n}'
103 (delete)
104 )
105 )
107 define verb as (
108 [substring] among (
109 '{vo}{u}{m}'
110 '{v}{vo}{u}{m}'
111 '{a}{l}{vo}{u}'
112 '{ye}{l}{vo}{u}'
113 '{v}{ye}{l}'
114 '{a}{n}{a}{l}'
115 '{ye}{l}{vo}{u}{c}'
116 '{a}{l}{vo}{u}{c}'
117 '{y}{a}{l}'
118 '{y}{ye}{l}'
119 '{a}{l}{vo}{v}'
120 '{ye}{l}{vo}{v}'
121 '{a}{l}{i}{s}'
122 '{ye}{l}{i}{s}'
123 '{ye}{n}{a}{l}'
124 '{a}{c}{n}{a}{l}'
125 '{ye}{c}{n}{ye}{l}'
126 '{c}{n}{ye}{l}'
127 '{n}{ye}{l}'
128 '{a}{t}{ye}{l}'
129 '{vo}{t}{ye}{l}'
130 '{k}{vo}{t}{ye}{l}'
131 '{t}{ye}{l}'
132 '{v}{a}{ts}'
133 '{ye}{c}{v}{ye}{l}'
134 '{a}{c}{v}{ye}{l}'
135 '{ye}{c}{i}{r'}'
136 '{a}{c}{i}{r'}'
137 '{ye}{c}{i}{n}{q}'
138 '{a}{c}{i}{n}{q}'
139 '{v}{ye}{c}{i}{r'}'
140 '{v}{ye}{c}{i}{n}{q}'
141 '{v}{ye}{c}{i}{q}'
142 '{v}{ye}{c}{i}{n}'
143 '{a}{c}{r'}{i}{r'}'
144 '{a}{c}{r'}{ye}{c}'
145 '{a}{c}{r'}{i}{n}{q}'
146 '{a}{c}{r'}{i}{q}'
147 '{a}{c}{r'}{i}{n}'
148 '{ye}{c}{i}{q}'
149 '{a}{c}{i}{q}'
150 '{ye}{c}{i}{n}'
151 '{a}{c}{i}{n}'
152 '{a}{c}{a}{r'}'
153 '{a}{c}{a}{v}'
154 '{a}{c}{a}{n}{q}'
155 '{a}{c}{a}{q}'
156 '{a}{c}{a}{n}'
157 '{v}{ye}{c}{i}'
158 '{a}{c}{r'}{i}'
159 '{ye}{c}{a}{r'}'
160 '{ye}{c}{a}{v}'
161 '{c}{a}{n}{q}'
162 '{c}{a}{q}'
163 '{c}{a}{n}'
164 '{a}{c}{a}'
165 '{a}{c}{i}'
166 '{ye}{c}{a}'
167 '{ch}{ye}{l}'
168 '{ye}{c}{i}'
169 '{a}{r'}'
170 '{a}{v}'
171 '{a}{n}{q}'
172 '{a}{q}'
173 '{a}{n}'
174 '{a}{l}'
175 '{ye}{l}'
176 '{ye}{c}'
177 '{a}{c}'
178 '{v}{ye}'
179 '{a}'
181 (delete)
182 )
183 )
185 define noun as (
186 [substring] among (
187 '{a}{ts}{vo}'
188 '{a}{n}{a}{k}'
189 '{a}{n}{o}{c}'
190 '{a}{r'}{a}{n}'
191 '{a}{r'}{q}'
192 '{p}{a}{n}'
193 '{s}{t}{a}{n}'
194 '{ye}{gh}{e}{n}'
195 '{ye}{n}{q}'
196 '{i}{k}'
197 '{i}{ch}'
198 '{i}{q}'
199 '{m}{vo}{u}{n}{q}'
200 '{j}{a}{k}'
201 '{j}{vo}{u}{n}'
202 '{vo}{n}{q}'
203 '{vo}{r'}{d}'
204 '{vo}{c}'
205 '{ch}{ye}{q}'
206 '{v}{a}{ts}{q}'
207 '{v}{vo}{r'}'
208 '{a}{v}{vo}{r'}'
209 '{vo}{u}{dt}{j}{vo}{u}{n}'
210 '{vo}{u}{k}'
211 '{vo}{u}{h}{i}'
212 '{vo}{u}{j}{dt}'
213 '{vo}{u}{j}{q}'
214 '{vo}{u}{s}{t}'
215 '{vo}{u}{s}'
216 '{c}{i}'
217 '{a}{l}{i}{q}'
218 '{a}{n}{i}{q}'
219 '{i}{l}'
220 '{i}{ch}{q}'
221 '{vo}{u}{n}{q}'
222 '{g}{a}{r'}'
223 '{vo}{u}'
224 '{a}{k}'
225 '{a}{n}'
226 '{q}'
228 (delete)
229 )
230 )
232 define ending as (
233 [substring] R2 among (
234 '{n}{ye}{r'}{y}'
235 '{n}{ye}{r'}{n}'
236 '{n}{ye}{r'}{i}'
237 '{n}{ye}{r'}{d}'
238 '{ye}{r'}{i}{c}'
239 '{n}{ye}{r'}{i}{c}'
240 '{ye}{r'}{i}'
241 '{ye}{r'}{d}'
242 '{ye}{r'}{n}'
243 '{ye}{r'}{y}'
244 '{n}{ye}{r'}{i}{n}'
245 '{vo}{u}{dt}{j}{a}{n}{n}'
246 '{vo}{u}{dt}{j}{a}{n}{y}'
247 '{vo}{u}{dt}{j}{a}{n}{s}'
248 '{vo}{u}{dt}{j}{a}{n}{d}'
249 '{vo}{u}{dt}{j}{a}{n}'
250 '{ye}{r'}{i}{n}'
251 '{i}{n}'
252 '{s}{a}'
253 '{vo}{dj}'
254 '{i}{c}'
255 '{ye}{r'}{vo}{v}'
256 '{n}{ye}{r'}{vo}{v}'
257 '{ye}{r'}{vo}{u}{m}'
258 '{n}{ye}{r'}{vo}{u}{m}'
259 '{vo}{u}{n}'
260 '{vo}{u}{d}'
261 '{v}{a}{n}{s}'
262 '{v}{a}{n}{y}'
263 '{v}{a}{n}{d}'
264 '{a}{n}{y}'
265 '{a}{n}{d}'
266 '{v}{a}{n}'
267 '{vo}{dj}{y}'
268 '{vo}{dj}{s}'
269 '{vo}{dj}{d}'
270 '{vo}{c}'
271 '{vo}{u}{c}'
272 '{vo}{dj}{i}{c}'
273 '{c}{i}{c}'
274 '{v}{i}{c}'
275 '{v}{i}'
276 '{v}{vo}{v}'
277 '{vo}{v}'
278 '{a}{n}{vo}{v}'
279 '{a}{n}{vo}{u}{m}'
280 '{v}{a}{n}{i}{c}'
281 '{a}{m}{b}'
282 '{a}{n}'
283 '{n}{ye}{r'}'
284 '{ye}{r'}'
285 '{v}{a}'
286 '{y}'
287 '{n}'
288 '{d}'
289 '{c}'
290 '{i}'
292 (delete)
293 )
294 )
295 )
297 define stem as (
299 do mark_regions
300 backwards setlimit tomark pV for (
301 do ending
302 do verb
303 do adjective
304 do noun
305 )
306 )