rere 0.0.1
Sign up to get free protection for your applications and to get access to all the features.
- data/.classpath +260 -0
- data/.gitignore +28 -0
- data/.project +14 -0
- data/.ruby-version +1 -0
- data/.settings/org.eclim.prefs +3 -0
- data/.settings/org.eclipse.jdt.core.prefs +5 -0
- data/.settings/org.eclipse.jdt.ui.prefs +2 -0
- data/Gemfile +7 -0
- data/LICENSE +22 -0
- data/README.md +44 -0
- data/Rakefile +65 -0
- data/bin/solr +61 -0
- data/example/config/solr.yml +23 -0
- data/example/log/.gitkeep +0 -0
- data/example/solr/README.txt +63 -0
- data/example/solr/collection1/README.txt +50 -0
- data/example/solr/collection1/conf/admin-extra.html +24 -0
- data/example/solr/collection1/conf/admin-extra.menu-bottom.html +25 -0
- data/example/solr/collection1/conf/admin-extra.menu-top.html +25 -0
- data/example/solr/collection1/conf/currency.xml +67 -0
- data/example/solr/collection1/conf/elevate.xml +38 -0
- data/example/solr/collection1/conf/lang/contractions_ca.txt +8 -0
- data/example/solr/collection1/conf/lang/contractions_fr.txt +15 -0
- data/example/solr/collection1/conf/lang/contractions_ga.txt +5 -0
- data/example/solr/collection1/conf/lang/contractions_it.txt +23 -0
- data/example/solr/collection1/conf/lang/hyphenations_ga.txt +5 -0
- data/example/solr/collection1/conf/lang/stemdict_nl.txt +6 -0
- data/example/solr/collection1/conf/lang/stoptags_ja.txt +420 -0
- data/example/solr/collection1/conf/lang/stopwords_ar.txt +125 -0
- data/example/solr/collection1/conf/lang/stopwords_bg.txt +193 -0
- data/example/solr/collection1/conf/lang/stopwords_ca.txt +220 -0
- data/example/solr/collection1/conf/lang/stopwords_cz.txt +172 -0
- data/example/solr/collection1/conf/lang/stopwords_da.txt +108 -0
- data/example/solr/collection1/conf/lang/stopwords_de.txt +292 -0
- data/example/solr/collection1/conf/lang/stopwords_el.txt +78 -0
- data/example/solr/collection1/conf/lang/stopwords_en.txt +54 -0
- data/example/solr/collection1/conf/lang/stopwords_es.txt +354 -0
- data/example/solr/collection1/conf/lang/stopwords_eu.txt +99 -0
- data/example/solr/collection1/conf/lang/stopwords_fa.txt +313 -0
- data/example/solr/collection1/conf/lang/stopwords_fi.txt +95 -0
- data/example/solr/collection1/conf/lang/stopwords_fr.txt +184 -0
- data/example/solr/collection1/conf/lang/stopwords_ga.txt +110 -0
- data/example/solr/collection1/conf/lang/stopwords_gl.txt +161 -0
- data/example/solr/collection1/conf/lang/stopwords_hi.txt +235 -0
- data/example/solr/collection1/conf/lang/stopwords_hu.txt +209 -0
- data/example/solr/collection1/conf/lang/stopwords_hy.txt +46 -0
- data/example/solr/collection1/conf/lang/stopwords_id.txt +359 -0
- data/example/solr/collection1/conf/lang/stopwords_it.txt +301 -0
- data/example/solr/collection1/conf/lang/stopwords_ja.txt +127 -0
- data/example/solr/collection1/conf/lang/stopwords_lv.txt +172 -0
- data/example/solr/collection1/conf/lang/stopwords_nl.txt +117 -0
- data/example/solr/collection1/conf/lang/stopwords_no.txt +192 -0
- data/example/solr/collection1/conf/lang/stopwords_pt.txt +251 -0
- data/example/solr/collection1/conf/lang/stopwords_ro.txt +233 -0
- data/example/solr/collection1/conf/lang/stopwords_ru.txt +241 -0
- data/example/solr/collection1/conf/lang/stopwords_sv.txt +131 -0
- data/example/solr/collection1/conf/lang/stopwords_th.txt +119 -0
- data/example/solr/collection1/conf/lang/stopwords_tr.txt +212 -0
- data/example/solr/collection1/conf/lang/userdict_ja.txt +29 -0
- data/example/solr/collection1/conf/mapping-FoldToASCII.txt +3813 -0
- data/example/solr/collection1/conf/mapping-ISOLatin1Accent.txt +246 -0
- data/example/solr/collection1/conf/protwords.txt +21 -0
- data/example/solr/collection1/conf/schema.xml +1125 -0
- data/example/solr/collection1/conf/scripts.conf +24 -0
- data/example/solr/collection1/conf/solrconfig.xml +1816 -0
- data/example/solr/collection1/conf/spellings.txt +2 -0
- data/example/solr/collection1/conf/stopwords.txt +14 -0
- data/example/solr/collection1/conf/synonyms.txt +29 -0
- data/example/solr/collection1/conf/update-script.js +53 -0
- data/example/solr/collection1/conf/velocity/VM_global_library.vm +170 -0
- data/example/solr/collection1/conf/velocity/browse.vm +50 -0
- data/example/solr/collection1/conf/velocity/cluster.vm +9 -0
- data/example/solr/collection1/conf/velocity/clusterResults.vm +12 -0
- data/example/solr/collection1/conf/velocity/debug.vm +17 -0
- data/example/solr/collection1/conf/velocity/did_you_mean.vm +4 -0
- data/example/solr/collection1/conf/velocity/facet_fields.vm +15 -0
- data/example/solr/collection1/conf/velocity/facet_pivot.vm +3 -0
- data/example/solr/collection1/conf/velocity/facet_queries.vm +3 -0
- data/example/solr/collection1/conf/velocity/facet_ranges.vm +15 -0
- data/example/solr/collection1/conf/velocity/facets.vm +5 -0
- data/example/solr/collection1/conf/velocity/footer.vm +17 -0
- data/example/solr/collection1/conf/velocity/head.vm +32 -0
- data/example/solr/collection1/conf/velocity/header.vm +3 -0
- data/example/solr/collection1/conf/velocity/hit.vm +11 -0
- data/example/solr/collection1/conf/velocity/hitGrouped.vm +24 -0
- data/example/solr/collection1/conf/velocity/join-doc.vm +4 -0
- data/example/solr/collection1/conf/velocity/jquery.autocomplete.css +48 -0
- data/example/solr/collection1/conf/velocity/jquery.autocomplete.js +763 -0
- data/example/solr/collection1/conf/velocity/layout.vm +20 -0
- data/example/solr/collection1/conf/velocity/main.css +208 -0
- data/example/solr/collection1/conf/velocity/product-doc.vm +27 -0
- data/example/solr/collection1/conf/velocity/query.vm +42 -0
- data/example/solr/collection1/conf/velocity/queryGroup.vm +19 -0
- data/example/solr/collection1/conf/velocity/querySpatial.vm +40 -0
- data/example/solr/collection1/conf/velocity/richtext-doc.vm +114 -0
- data/example/solr/collection1/conf/velocity/suggest.vm +3 -0
- data/example/solr/collection1/conf/velocity/tabs.vm +6 -0
- data/example/solr/collection1/conf/xslt/example.xsl +132 -0
- data/example/solr/collection1/conf/xslt/example_atom.xsl +67 -0
- data/example/solr/collection1/conf/xslt/example_rss.xsl +66 -0
- data/example/solr/collection1/conf/xslt/luke.xsl +337 -0
- data/example/solr/collection1/conf/xslt/updateXml.xsl +70 -0
- data/example/solr/collection1/data/index/segments.gen +0 -0
- data/example/solr/collection1/data/index/segments_1 +0 -0
- data/example/solr/data/development/index/segments.gen +0 -0
- data/example/solr/data/development/index/segments_1 +0 -0
- data/example/solr/solr.xml +53 -0
- data/example/solr/zoo.cfg +17 -0
- data/lib/rere.rb +66 -0
- data/lib/rere/railtie.rb +8 -0
- data/lib/rere/server.rb +380 -0
- data/lib/rere/tasks/solr.rake +47 -0
- data/lib/rere/version.rb +3 -0
- data/pom.xml +168 -0
- data/rere.gemspec +26 -0
- data/server/README.txt +78 -0
- data/server/cloud-scripts/zkcli.bat +12 -0
- data/server/cloud-scripts/zkcli.sh +14 -0
- data/server/contexts/solr-jetty-context.xml +8 -0
- data/server/etc/create-solrtest.keystore.sh +37 -0
- data/server/etc/jetty.xml +205 -0
- data/server/etc/logging.properties +38 -0
- data/server/etc/solrtest.keystore +0 -0
- data/server/etc/webdefault.xml +527 -0
- data/server/exampledocs/books.csv +11 -0
- data/server/exampledocs/books.json +51 -0
- data/server/exampledocs/gb18030-example.xml +32 -0
- data/server/exampledocs/hd.xml +56 -0
- data/server/exampledocs/ipod_other.xml +60 -0
- data/server/exampledocs/ipod_video.xml +40 -0
- data/server/exampledocs/manufacturers.xml +75 -0
- data/server/exampledocs/mem.xml +77 -0
- data/server/exampledocs/money.xml +65 -0
- data/server/exampledocs/monitor.xml +35 -0
- data/server/exampledocs/monitor2.xml +34 -0
- data/server/exampledocs/mp500.xml +43 -0
- data/server/exampledocs/post.jar +0 -0
- data/server/exampledocs/post.sh +30 -0
- data/server/exampledocs/sd500.xml +38 -0
- data/server/exampledocs/solr.xml +38 -0
- data/server/exampledocs/test_utf8.sh +93 -0
- data/server/exampledocs/utf8-example.xml +42 -0
- data/server/exampledocs/vidcard.xml +62 -0
- data/server/lib/ext/jcl-over-slf4j-1.6.6.jar +0 -0
- data/server/lib/ext/jul-to-slf4j-1.6.6.jar +0 -0
- data/server/lib/ext/log4j-1.2.16.jar +0 -0
- data/server/lib/ext/slf4j-api-1.6.6.jar +0 -0
- data/server/lib/ext/slf4j-log4j12-1.6.6.jar +0 -0
- data/server/lib/jetty-continuation-8.1.8.v20121106.jar +0 -0
- data/server/lib/jetty-deploy-8.1.8.v20121106.jar +0 -0
- data/server/lib/jetty-http-8.1.8.v20121106.jar +0 -0
- data/server/lib/jetty-io-8.1.8.v20121106.jar +0 -0
- data/server/lib/jetty-jmx-8.1.8.v20121106.jar +0 -0
- data/server/lib/jetty-security-8.1.8.v20121106.jar +0 -0
- data/server/lib/jetty-server-8.1.8.v20121106.jar +0 -0
- data/server/lib/jetty-servlet-8.1.8.v20121106.jar +0 -0
- data/server/lib/jetty-util-8.1.8.v20121106.jar +0 -0
- data/server/lib/jetty-webapp-8.1.8.v20121106.jar +0 -0
- data/server/lib/jetty-xml-8.1.8.v20121106.jar +0 -0
- data/server/lib/servlet-api-3.0.jar +0 -0
- data/server/resources/log4j.properties +19 -0
- data/server/solr/README.txt +63 -0
- data/server/solr/solr.xml +53 -0
- data/server/solr/zoo.cfg +17 -0
- data/server/start.jar +0 -0
- data/server/webapps/solr.war +0 -0
- data/solr/lib/solr-winds-0.1.jar +0 -0
- metadata +284 -0
@@ -0,0 +1,313 @@
|
|
1
|
+
# This file was created by Jacques Savoy and is distributed under the BSD license.
|
2
|
+
# See http://members.unine.ch/jacques.savoy/clef/index.html.
|
3
|
+
# Also see http://www.opensource.org/licenses/bsd-license.html
|
4
|
+
# Note: by default this file is used after normalization, so when adding entries
|
5
|
+
# to this file, use the arabic 'ي' instead of 'ی'
|
6
|
+
انان
|
7
|
+
نداشته
|
8
|
+
سراسر
|
9
|
+
خياه
|
10
|
+
ايشان
|
11
|
+
وي
|
12
|
+
تاكنون
|
13
|
+
بيشتري
|
14
|
+
دوم
|
15
|
+
پس
|
16
|
+
ناشي
|
17
|
+
وگو
|
18
|
+
يا
|
19
|
+
داشتند
|
20
|
+
سپس
|
21
|
+
هنگام
|
22
|
+
هرگز
|
23
|
+
پنج
|
24
|
+
نشان
|
25
|
+
امسال
|
26
|
+
ديگر
|
27
|
+
گروهي
|
28
|
+
شدند
|
29
|
+
چطور
|
30
|
+
ده
|
31
|
+
و
|
32
|
+
دو
|
33
|
+
نخستين
|
34
|
+
ولي
|
35
|
+
چرا
|
36
|
+
چه
|
37
|
+
وسط
|
38
|
+
ه
|
39
|
+
كدام
|
40
|
+
قابل
|
41
|
+
يك
|
42
|
+
رفت
|
43
|
+
هفت
|
44
|
+
همچنين
|
45
|
+
در
|
46
|
+
هزار
|
47
|
+
بله
|
48
|
+
بلي
|
49
|
+
شايد
|
50
|
+
اما
|
51
|
+
شناسي
|
52
|
+
گرفته
|
53
|
+
دهد
|
54
|
+
داشته
|
55
|
+
دانست
|
56
|
+
داشتن
|
57
|
+
خواهيم
|
58
|
+
ميليارد
|
59
|
+
وقتيكه
|
60
|
+
امد
|
61
|
+
خواهد
|
62
|
+
جز
|
63
|
+
اورده
|
64
|
+
شده
|
65
|
+
بلكه
|
66
|
+
خدمات
|
67
|
+
شدن
|
68
|
+
برخي
|
69
|
+
نبود
|
70
|
+
بسياري
|
71
|
+
جلوگيري
|
72
|
+
حق
|
73
|
+
كردند
|
74
|
+
نوعي
|
75
|
+
بعري
|
76
|
+
نكرده
|
77
|
+
نظير
|
78
|
+
نبايد
|
79
|
+
بوده
|
80
|
+
بودن
|
81
|
+
داد
|
82
|
+
اورد
|
83
|
+
هست
|
84
|
+
جايي
|
85
|
+
شود
|
86
|
+
دنبال
|
87
|
+
داده
|
88
|
+
بايد
|
89
|
+
سابق
|
90
|
+
هيچ
|
91
|
+
همان
|
92
|
+
انجا
|
93
|
+
كمتر
|
94
|
+
كجاست
|
95
|
+
گردد
|
96
|
+
كسي
|
97
|
+
تر
|
98
|
+
مردم
|
99
|
+
تان
|
100
|
+
دادن
|
101
|
+
بودند
|
102
|
+
سري
|
103
|
+
جدا
|
104
|
+
ندارند
|
105
|
+
مگر
|
106
|
+
يكديگر
|
107
|
+
دارد
|
108
|
+
دهند
|
109
|
+
بنابراين
|
110
|
+
هنگامي
|
111
|
+
سمت
|
112
|
+
جا
|
113
|
+
انچه
|
114
|
+
خود
|
115
|
+
دادند
|
116
|
+
زياد
|
117
|
+
دارند
|
118
|
+
اثر
|
119
|
+
بدون
|
120
|
+
بهترين
|
121
|
+
بيشتر
|
122
|
+
البته
|
123
|
+
به
|
124
|
+
براساس
|
125
|
+
بيرون
|
126
|
+
كرد
|
127
|
+
بعضي
|
128
|
+
گرفت
|
129
|
+
توي
|
130
|
+
اي
|
131
|
+
ميليون
|
132
|
+
او
|
133
|
+
جريان
|
134
|
+
تول
|
135
|
+
بر
|
136
|
+
مانند
|
137
|
+
برابر
|
138
|
+
باشيم
|
139
|
+
مدتي
|
140
|
+
گويند
|
141
|
+
اكنون
|
142
|
+
تا
|
143
|
+
تنها
|
144
|
+
جديد
|
145
|
+
چند
|
146
|
+
بي
|
147
|
+
نشده
|
148
|
+
كردن
|
149
|
+
كردم
|
150
|
+
گويد
|
151
|
+
كرده
|
152
|
+
كنيم
|
153
|
+
نمي
|
154
|
+
نزد
|
155
|
+
روي
|
156
|
+
قصد
|
157
|
+
فقط
|
158
|
+
بالاي
|
159
|
+
ديگران
|
160
|
+
اين
|
161
|
+
ديروز
|
162
|
+
توسط
|
163
|
+
سوم
|
164
|
+
ايم
|
165
|
+
دانند
|
166
|
+
سوي
|
167
|
+
استفاده
|
168
|
+
شما
|
169
|
+
كنار
|
170
|
+
داريم
|
171
|
+
ساخته
|
172
|
+
طور
|
173
|
+
امده
|
174
|
+
رفته
|
175
|
+
نخست
|
176
|
+
بيست
|
177
|
+
نزديك
|
178
|
+
طي
|
179
|
+
كنيد
|
180
|
+
از
|
181
|
+
انها
|
182
|
+
تمامي
|
183
|
+
داشت
|
184
|
+
يكي
|
185
|
+
طريق
|
186
|
+
اش
|
187
|
+
چيست
|
188
|
+
روب
|
189
|
+
نمايد
|
190
|
+
گفت
|
191
|
+
چندين
|
192
|
+
چيزي
|
193
|
+
تواند
|
194
|
+
ام
|
195
|
+
ايا
|
196
|
+
با
|
197
|
+
ان
|
198
|
+
ايد
|
199
|
+
ترين
|
200
|
+
اينكه
|
201
|
+
ديگري
|
202
|
+
راه
|
203
|
+
هايي
|
204
|
+
بروز
|
205
|
+
همچنان
|
206
|
+
پاعين
|
207
|
+
كس
|
208
|
+
حدود
|
209
|
+
مختلف
|
210
|
+
مقابل
|
211
|
+
چيز
|
212
|
+
گيرد
|
213
|
+
ندارد
|
214
|
+
ضد
|
215
|
+
همچون
|
216
|
+
سازي
|
217
|
+
شان
|
218
|
+
مورد
|
219
|
+
باره
|
220
|
+
مرسي
|
221
|
+
خويش
|
222
|
+
برخوردار
|
223
|
+
چون
|
224
|
+
خارج
|
225
|
+
شش
|
226
|
+
هنوز
|
227
|
+
تحت
|
228
|
+
ضمن
|
229
|
+
هستيم
|
230
|
+
گفته
|
231
|
+
فكر
|
232
|
+
بسيار
|
233
|
+
پيش
|
234
|
+
براي
|
235
|
+
روزهاي
|
236
|
+
انكه
|
237
|
+
نخواهد
|
238
|
+
بالا
|
239
|
+
كل
|
240
|
+
وقتي
|
241
|
+
كي
|
242
|
+
چنين
|
243
|
+
كه
|
244
|
+
گيري
|
245
|
+
نيست
|
246
|
+
است
|
247
|
+
كجا
|
248
|
+
كند
|
249
|
+
نيز
|
250
|
+
يابد
|
251
|
+
بندي
|
252
|
+
حتي
|
253
|
+
توانند
|
254
|
+
عقب
|
255
|
+
خواست
|
256
|
+
كنند
|
257
|
+
بين
|
258
|
+
تمام
|
259
|
+
همه
|
260
|
+
ما
|
261
|
+
باشند
|
262
|
+
مثل
|
263
|
+
شد
|
264
|
+
اري
|
265
|
+
باشد
|
266
|
+
اره
|
267
|
+
طبق
|
268
|
+
بعد
|
269
|
+
اگر
|
270
|
+
صورت
|
271
|
+
غير
|
272
|
+
جاي
|
273
|
+
بيش
|
274
|
+
ريزي
|
275
|
+
اند
|
276
|
+
زيرا
|
277
|
+
چگونه
|
278
|
+
بار
|
279
|
+
لطفا
|
280
|
+
مي
|
281
|
+
درباره
|
282
|
+
من
|
283
|
+
ديده
|
284
|
+
همين
|
285
|
+
گذاري
|
286
|
+
برداري
|
287
|
+
علت
|
288
|
+
گذاشته
|
289
|
+
هم
|
290
|
+
فوق
|
291
|
+
نه
|
292
|
+
ها
|
293
|
+
شوند
|
294
|
+
اباد
|
295
|
+
همواره
|
296
|
+
هر
|
297
|
+
اول
|
298
|
+
خواهند
|
299
|
+
چهار
|
300
|
+
نام
|
301
|
+
امروز
|
302
|
+
مان
|
303
|
+
هاي
|
304
|
+
قبل
|
305
|
+
كنم
|
306
|
+
سعي
|
307
|
+
تازه
|
308
|
+
را
|
309
|
+
هستند
|
310
|
+
زير
|
311
|
+
جلوي
|
312
|
+
عنوان
|
313
|
+
بود
|
@@ -0,0 +1,95 @@
|
|
1
|
+
| From svn.tartarus.org/snowball/trunk/website/algorithms/finnish/stop.txt
|
2
|
+
| This file is distributed under the BSD License.
|
3
|
+
| See http://snowball.tartarus.org/license.php
|
4
|
+
| Also see http://www.opensource.org/licenses/bsd-license.html
|
5
|
+
| - Encoding was converted to UTF-8.
|
6
|
+
| - This notice was added.
|
7
|
+
|
8
|
+
| forms of BE
|
9
|
+
|
10
|
+
olla
|
11
|
+
olen
|
12
|
+
olet
|
13
|
+
on
|
14
|
+
olemme
|
15
|
+
olette
|
16
|
+
ovat
|
17
|
+
ole | negative form
|
18
|
+
|
19
|
+
oli
|
20
|
+
olisi
|
21
|
+
olisit
|
22
|
+
olisin
|
23
|
+
olisimme
|
24
|
+
olisitte
|
25
|
+
olisivat
|
26
|
+
olit
|
27
|
+
olin
|
28
|
+
olimme
|
29
|
+
olitte
|
30
|
+
olivat
|
31
|
+
ollut
|
32
|
+
olleet
|
33
|
+
|
34
|
+
en | negation
|
35
|
+
et
|
36
|
+
ei
|
37
|
+
emme
|
38
|
+
ette
|
39
|
+
eivät
|
40
|
+
|
41
|
+
|Nom Gen Acc Part Iness Elat Illat Adess Ablat Allat Ess Trans
|
42
|
+
minä minun minut minua minussa minusta minuun minulla minulta minulle | I
|
43
|
+
sinä sinun sinut sinua sinussa sinusta sinuun sinulla sinulta sinulle | you
|
44
|
+
hän hänen hänet häntä hänessä hänestä häneen hänellä häneltä hänelle | he she
|
45
|
+
me meidän meidät meitä meissä meistä meihin meillä meiltä meille | we
|
46
|
+
te teidän teidät teitä teissä teistä teihin teillä teiltä teille | you
|
47
|
+
he heidän heidät heitä heissä heistä heihin heillä heiltä heille | they
|
48
|
+
|
49
|
+
tämä tämän tätä tässä tästä tähän tallä tältä tälle tänä täksi | this
|
50
|
+
tuo tuon tuotä tuossa tuosta tuohon tuolla tuolta tuolle tuona tuoksi | that
|
51
|
+
se sen sitä siinä siitä siihen sillä siltä sille sinä siksi | it
|
52
|
+
nämä näiden näitä näissä näistä näihin näillä näiltä näille näinä näiksi | these
|
53
|
+
nuo noiden noita noissa noista noihin noilla noilta noille noina noiksi | those
|
54
|
+
ne niiden niitä niissä niistä niihin niillä niiltä niille niinä niiksi | they
|
55
|
+
|
56
|
+
kuka kenen kenet ketä kenessä kenestä keneen kenellä keneltä kenelle kenenä keneksi| who
|
57
|
+
ketkä keiden ketkä keitä keissä keistä keihin keillä keiltä keille keinä keiksi | (pl)
|
58
|
+
mikä minkä minkä mitä missä mistä mihin millä miltä mille minä miksi | which what
|
59
|
+
mitkä | (pl)
|
60
|
+
|
61
|
+
joka jonka jota jossa josta johon jolla jolta jolle jona joksi | who which
|
62
|
+
jotka joiden joita joissa joista joihin joilla joilta joille joina joiksi | (pl)
|
63
|
+
|
64
|
+
| conjunctions
|
65
|
+
|
66
|
+
että | that
|
67
|
+
ja | and
|
68
|
+
jos | if
|
69
|
+
koska | because
|
70
|
+
kuin | than
|
71
|
+
mutta | but
|
72
|
+
niin | so
|
73
|
+
sekä | and
|
74
|
+
sillä | for
|
75
|
+
tai | or
|
76
|
+
vaan | but
|
77
|
+
vai | or
|
78
|
+
vaikka | although
|
79
|
+
|
80
|
+
|
81
|
+
| prepositions
|
82
|
+
|
83
|
+
kanssa | with
|
84
|
+
mukaan | according to
|
85
|
+
noin | about
|
86
|
+
poikki | across
|
87
|
+
yli | over, across
|
88
|
+
|
89
|
+
| other
|
90
|
+
|
91
|
+
kun | when
|
92
|
+
niin | so
|
93
|
+
nyt | now
|
94
|
+
itse | self
|
95
|
+
|
@@ -0,0 +1,184 @@
|
|
1
|
+
| From svn.tartarus.org/snowball/trunk/website/algorithms/french/stop.txt
|
2
|
+
| This file is distributed under the BSD License.
|
3
|
+
| See http://snowball.tartarus.org/license.php
|
4
|
+
| Also see http://www.opensource.org/licenses/bsd-license.html
|
5
|
+
| - Encoding was converted to UTF-8.
|
6
|
+
| - This notice was added.
|
7
|
+
|
8
|
+
| A French stop word list. Comments begin with vertical bar. Each stop
|
9
|
+
| word is at the start of a line.
|
10
|
+
|
11
|
+
au | a + le
|
12
|
+
aux | a + les
|
13
|
+
avec | with
|
14
|
+
ce | this
|
15
|
+
ces | these
|
16
|
+
dans | with
|
17
|
+
de | of
|
18
|
+
des | de + les
|
19
|
+
du | de + le
|
20
|
+
elle | she
|
21
|
+
en | `of them' etc
|
22
|
+
et | and
|
23
|
+
eux | them
|
24
|
+
il | he
|
25
|
+
je | I
|
26
|
+
la | the
|
27
|
+
le | the
|
28
|
+
leur | their
|
29
|
+
lui | him
|
30
|
+
ma | my (fem)
|
31
|
+
mais | but
|
32
|
+
me | me
|
33
|
+
même | same; as in moi-même (myself) etc
|
34
|
+
mes | me (pl)
|
35
|
+
moi | me
|
36
|
+
mon | my (masc)
|
37
|
+
ne | not
|
38
|
+
nos | our (pl)
|
39
|
+
notre | our
|
40
|
+
nous | we
|
41
|
+
on | one
|
42
|
+
ou | where
|
43
|
+
par | by
|
44
|
+
pas | not
|
45
|
+
pour | for
|
46
|
+
qu | que before vowel
|
47
|
+
que | that
|
48
|
+
qui | who
|
49
|
+
sa | his, her (fem)
|
50
|
+
se | oneself
|
51
|
+
ses | his (pl)
|
52
|
+
son | his, her (masc)
|
53
|
+
sur | on
|
54
|
+
ta | thy (fem)
|
55
|
+
te | thee
|
56
|
+
tes | thy (pl)
|
57
|
+
toi | thee
|
58
|
+
ton | thy (masc)
|
59
|
+
tu | thou
|
60
|
+
un | a
|
61
|
+
une | a
|
62
|
+
vos | your (pl)
|
63
|
+
votre | your
|
64
|
+
vous | you
|
65
|
+
|
66
|
+
| single letter forms
|
67
|
+
|
68
|
+
c | c'
|
69
|
+
d | d'
|
70
|
+
j | j'
|
71
|
+
l | l'
|
72
|
+
à | to, at
|
73
|
+
m | m'
|
74
|
+
n | n'
|
75
|
+
s | s'
|
76
|
+
t | t'
|
77
|
+
y | there
|
78
|
+
|
79
|
+
| forms of être (not including the infinitive):
|
80
|
+
été
|
81
|
+
étée
|
82
|
+
étées
|
83
|
+
étés
|
84
|
+
étant
|
85
|
+
suis
|
86
|
+
es
|
87
|
+
est
|
88
|
+
sommes
|
89
|
+
êtes
|
90
|
+
sont
|
91
|
+
serai
|
92
|
+
seras
|
93
|
+
sera
|
94
|
+
serons
|
95
|
+
serez
|
96
|
+
seront
|
97
|
+
serais
|
98
|
+
serait
|
99
|
+
serions
|
100
|
+
seriez
|
101
|
+
seraient
|
102
|
+
étais
|
103
|
+
était
|
104
|
+
étions
|
105
|
+
étiez
|
106
|
+
étaient
|
107
|
+
fus
|
108
|
+
fut
|
109
|
+
fûmes
|
110
|
+
fûtes
|
111
|
+
furent
|
112
|
+
sois
|
113
|
+
soit
|
114
|
+
soyons
|
115
|
+
soyez
|
116
|
+
soient
|
117
|
+
fusse
|
118
|
+
fusses
|
119
|
+
fût
|
120
|
+
fussions
|
121
|
+
fussiez
|
122
|
+
fussent
|
123
|
+
|
124
|
+
| forms of avoir (not including the infinitive):
|
125
|
+
ayant
|
126
|
+
eu
|
127
|
+
eue
|
128
|
+
eues
|
129
|
+
eus
|
130
|
+
ai
|
131
|
+
as
|
132
|
+
avons
|
133
|
+
avez
|
134
|
+
ont
|
135
|
+
aurai
|
136
|
+
auras
|
137
|
+
aura
|
138
|
+
aurons
|
139
|
+
aurez
|
140
|
+
auront
|
141
|
+
aurais
|
142
|
+
aurait
|
143
|
+
aurions
|
144
|
+
auriez
|
145
|
+
auraient
|
146
|
+
avais
|
147
|
+
avait
|
148
|
+
avions
|
149
|
+
aviez
|
150
|
+
avaient
|
151
|
+
eut
|
152
|
+
eûmes
|
153
|
+
eûtes
|
154
|
+
eurent
|
155
|
+
aie
|
156
|
+
aies
|
157
|
+
ait
|
158
|
+
ayons
|
159
|
+
ayez
|
160
|
+
aient
|
161
|
+
eusse
|
162
|
+
eusses
|
163
|
+
eût
|
164
|
+
eussions
|
165
|
+
eussiez
|
166
|
+
eussent
|
167
|
+
|
168
|
+
| Later additions (from Jean-Christophe Deschamps)
|
169
|
+
ceci | this
|
170
|
+
cela | that
|
171
|
+
celà | that
|
172
|
+
cet | this
|
173
|
+
cette | this
|
174
|
+
ici | here
|
175
|
+
ils | they
|
176
|
+
les | the (pl)
|
177
|
+
leurs | their (pl)
|
178
|
+
quel | which
|
179
|
+
quels | which
|
180
|
+
quelle | which
|
181
|
+
quelles | which
|
182
|
+
sans | without
|
183
|
+
soi | oneself
|
184
|
+
|