rbtagger 0.3.2 → 0.4.0

Sign up to get free protection for your applications and to get access to all the features.
Files changed (59) hide show
  1. data/README +44 -0
  2. data/Rakefile +78 -4
  3. data/ext/rule_tagger/registry.c +4 -4
  4. data/ext/rule_tagger/registry.h +1 -1
  5. data/ext/word_tagger/rtagger.cc +23 -1
  6. data/ext/word_tagger/tagger.cc +9 -4
  7. data/ext/word_tagger/tagger.h +2 -0
  8. data/ext/word_tagger/test.rb +2 -2
  9. data/lib/brill/brown/{LEXICON → Lexicon.rb} +0 -0
  10. data/lib/brill/tagger.rb +1 -1
  11. data/lib/rbtagger.rb +0 -3
  12. data/lib/rbtagger/version.rb +2 -2
  13. data/lib/word/tagger.rb +2 -1
  14. metadata +38 -101
  15. data/COPYING +0 -21
  16. data/History.txt +0 -4
  17. data/License.txt +0 -20
  18. data/Manifest.txt +0 -82
  19. data/PostInstall.txt +0 -1
  20. data/README.txt +0 -51
  21. data/config/hoe.rb +0 -74
  22. data/config/requirements.rb +0 -15
  23. data/ext/rule_tagger/mkmf.log +0 -46
  24. data/ext/word_tagger/mkmf.log +0 -24
  25. data/ext/word_tagger/test/Makefile +0 -22
  26. data/ext/word_tagger/test/doc.txt +0 -87
  27. data/lib/brill/brown/CONTEXTUALRULEFILE +0 -284
  28. data/lib/brill/brown/LEXICALRULEFILE +0 -148
  29. data/script/console +0 -10
  30. data/script/destroy +0 -14
  31. data/script/generate +0 -14
  32. data/script/txt2html +0 -82
  33. data/setup.rb +0 -1585
  34. data/tasks/deployment.rake +0 -34
  35. data/tasks/environment.rake +0 -7
  36. data/tasks/extconf.rake +0 -18
  37. data/tasks/extconf/rule_tagger.rake +0 -43
  38. data/tasks/extconf/word_tagger.rake +0 -43
  39. data/tasks/website.rake +0 -17
  40. data/test/docs/doc0.txt +0 -20
  41. data/test/docs/doc1.txt +0 -11
  42. data/test/docs/doc2.txt +0 -52
  43. data/test/docs/doc3.txt +0 -128
  44. data/test/docs/doc4.txt +0 -337
  45. data/test/docs/doc5.txt +0 -497
  46. data/test/docs/doc6.txt +0 -116
  47. data/test/docs/doc7.txt +0 -101
  48. data/test/docs/doc8.txt +0 -25
  49. data/test/docs/doc9.txt +0 -84
  50. data/test/fixtures/tags.txt +0 -976
  51. data/test/test_helper.rb +0 -5
  52. data/test/test_rule_tagger.rb +0 -151
  53. data/test/test_word_tagger.rb +0 -47
  54. data/tools/rakehelp.rb +0 -113
  55. data/website/index.html +0 -231
  56. data/website/index.txt +0 -70
  57. data/website/javascripts/rounded_corners_lite.inc.js +0 -285
  58. data/website/stylesheets/screen.css +0 -138
  59. data/website/template.html.erb +0 -184
@@ -1,284 +0,0 @@
1
- NN VB PREVTAG TO
2
- VBP VB PREV1OR2OR3TAG MD
3
- NN VB PREV1OR2TAG MD
4
- VB NN PREV1OR2TAG DT
5
- VBD VBN PREV1OR2OR3TAG VBZ
6
- VBN VBD PREVTAG PRP
7
- VBN VBD PREVTAG NNP
8
- VBD VBN PREVTAG VBD
9
- VBP VB PREVTAG TO
10
- POS VBZ PREVTAG PRP
11
- VB VBP PREVTAG NNS
12
- IN RB WDAND2AFT as as
13
- VBD VBN PREV1OR2WD have
14
- IN WDT NEXT1OR2TAG VB
15
- VB VBP PREVTAG PRP
16
- VBP VB PREV1OR2WD n't
17
- IN WDT NEXTTAG VBZ
18
- IN DT NEXTTAG NN
19
- JJ NNP NEXTTAG NNP
20
- IN WDT NEXTTAG VBD
21
- JJ NN NEXTWD of
22
- VBD VBN PREV1OR2WD be
23
- JJR RBR NEXTTAG JJ
24
- IN WDT NEXTTAG VBP
25
- JJS RBS WDNEXTTAG most JJ
26
- VBN VBD SURROUNDTAG NN DT
27
- NNS VBZ PREVTAG PRP
28
- POS VBZ NEXT1OR2TAG DT
29
- NNP NN SURROUNDTAG STAART NNS
30
- VBD VBN NEXTWD by
31
- VB NN PREV1OR2TAG IN
32
- VB VBP PREVTAG WDT
33
- VBG NN PREVTAG JJ
34
- NNS VBZ NEXTTAG DT
35
- VBN VBD PREVTAG WP
36
- NN VBP PREVTAG NNS
37
- VB NN PREVTAG NN
38
- NN VB PREVWD n't
39
- NN VBG NEXTTAG DT
40
- RB JJ NEXTTAG NN
41
- NN VBP PREVTAG PRP
42
- VBN VBD SURROUNDTAG NNS DT
43
- VB NN PREV1OR2TAG POS
44
- JJ NN NEXTTAG VBD
45
- RB RP WDNEXTTAG up DT
46
- JJ VB PREVTAG TO
47
- VBN VBD SURROUNDTAG , DT
48
- VBN VBD PREVWD that
49
- VB VBP PREVBIGRAM NNS RB
50
- NNP JJ SURROUNDTAG STAART NN
51
- VB VBN PREVTAG VBZ
52
- NNP JJ WDNEXTTAG American NNS
53
- JJ RB NEXTTAG JJR
54
- NNS NN CURWD yen
55
- IN WDT NEXTTAG VBD
56
- DT IN WDAND2TAGAFT that NNS
57
- POS VBZ PREVWD that
58
- JJ VB PREVTAG MD
59
- VB NN PREVTAG JJ
60
- JJR RBR NEXTTAG RB
61
- VBD VBN PREV1OR2WD are
62
- NN JJ WDNEXTTAG executive NN
63
- NNP JJ WDNEXTTAG American NN
64
- VBN VBD PREVTAG WDT
65
- VBD VBN PREVBIGRAM VBD RB
66
- JJ NN SURROUNDTAG DT .
67
- NNP JJ NEXTWD German
68
- VBN VB PREVTAG TO
69
- VBN VBD PREVBIGRAM NNP RB
70
- RB IN RBIGRAM up to
71
- VB VBP PREVTAG WP
72
- JJ NN SURROUNDTAG DT IN
73
- IN DT NEXTWD 's
74
- VBD VBN WDNEXTTAG ended NNP
75
- VBD VBN SURROUNDTAG DT NN
76
- NNS NNP NEXTTAG NNP
77
- NN NNP NEXTTAG NNP
78
- VBG NN SURROUNDTAG DT IN
79
- NNP JJ SURROUNDTAG STAART NNS
80
- RB RP WDPREVTAG VB up
81
- VBN VBD PREVBIGRAM PRP RB
82
- JJ RB NEXTTAG VBN
83
- NN VBP PREVTAG RB
84
- NNS VBZ PREVTAG RB
85
- POS VBZ PREVTAG WP
86
- VB VBN PREVWD have
87
- NN PDT WDNEXTTAG half DT
88
- IN DT PREVTAG IN
89
- IN WDT NEXTTAG MD
90
- POS VBZ PREVTAG DT
91
- NN NNP CURWD Integrated
92
- POS '' NEXT1OR2TAG ''
93
- VBD VBN PREVTAG IN
94
- JJR RBR NEXT1OR2TAG VBN
95
- JJS RBS WDNEXTTAG most RB
96
- JJ NN SURROUNDTAG JJ IN
97
- VBZ NNS PREVTAG JJ
98
- NNS VBZ WDPREVTAG JJ is
99
- JJ NN NEXTTAG VBZ
100
- VBP NN PREVTAG DT
101
- JJ NN SURROUNDTAG JJ .
102
- NNPS NNP NEXTTAG NNP
103
- WDT DT PREVTAG CC
104
- RB IN WDNEXTTAG so PRP
105
- VBP NN PREVWD earnings
106
- NN VBG PREVWD is
107
- NNS VBZ PREV1OR2WD Mr.
108
- VBZ NNS PREVWD the
109
- RB RP WDPREVTAG VBN up
110
- NNPS NNS PREVTAG STAART
111
- VBN VBD SURROUNDTAG NN JJ
112
- VBP VB PREV2TAG VB
113
- RBR JJR NEXTTAG NNS
114
- JJ NN SURROUNDTAG DT ,
115
- JJ NN SURROUNDTAG IN .
116
- NN VB PREVTAG TO
117
- VB NN PREVTAG VB
118
- NN VBP PREVWD who
119
- RB RP WDPREVTAG VBG up
120
- NN RB WDNEXTTAG right RB
121
- VBZ POS WDPREVTAG NNP 's
122
- JJ RP WDNEXTTAG up NN
123
- VBN VBD SURROUNDTAG NN NN
124
- VBN VBD SURROUNDTAG CC DT
125
- JJ NN NEXTBIGRAM MD VB
126
- JJ RB WDNEXTTAG early IN
127
- JJ VBN SURROUNDTAG STAART IN
128
- IN RB RBIGRAM though ,
129
- VBD VBN PREV1OR2WD been
130
- DT PDT WDNEXTTAG all DT
131
- VBN VBD PREVBIGRAM NN RB
132
- NN VB PREVWD help
133
- VBP VB PREV1OR2WD not
134
- VBP NN PREVTAG JJ
135
- DT WDT PREVTAG NNS
136
- NN VBP PREVTAG WDT
137
- VB RB RBIGRAM close to
138
- NNS VBZ PREVBIGRAM , WDT
139
- IN RP WDNEXTTAG out DT
140
- DT RB NEXTWD longer
141
- IN JJ SURROUNDTAG DT NN
142
- DT WDT SURROUNDTAG NN VBZ
143
- IN VB NEXT2TAG VB
144
- IN NN PREVTAG DT
145
- VBN VBD SURROUNDTAG NNS NNS
146
- IN RB RBIGRAM about $
147
- EX RB NEXT1OR2TAG IN
148
- NN VBG NEXTTAG PRP$
149
- NN VBG CURWD living
150
- VBZ NNS PREVTAG PRP$
151
- RBR JJR NEXTTAG NN
152
- RBR JJR CURWD higher
153
- VB VBP PREVBIGRAM PRP RB
154
- NN VB PREVTAG MD
155
- VB NN PREV1OR2TAG PRP$
156
- RP IN PREV1OR2TAG ,
157
- VB JJ PREVTAG DT
158
- DT IN PREVWD out
159
- POS VBZ PREVTAG EX
160
- JJ NN NEXTTAG POS
161
- NN JJ CURWD first
162
- VBD VBN PREVWD the
163
- NNS VBZ WDPREVTAG NNP plans
164
- NNP NNS SURROUNDTAG STAART IN
165
- RB JJ NEXTTAG NNS
166
- JJ RB CURWD just
167
- VBP NN PREVWD sales
168
- NNS NNPS PREVWD Orange
169
- VB VBN PREVTAG VBD
170
- WDT DT PREVTAG IN
171
- NN JJ WDNEXTTAG right NN
172
- NN VBG WDNEXTTAG operating IN
173
- JJ VBN CURWD insured
174
- JJ NNP LBIGRAM STAART U.S.
175
- IN DT NEXT1OR2OR3TAG STAART
176
- POS '' PREV1OR2OR3TAG ``
177
- NN JJ WDNEXTTAG official NN
178
- NNP JJ CURWD Irish
179
- JJ RB NEXTTAG RBR
180
- VBG NN WDPREVTAG DT selling
181
- VBP VB PREV1OR2OR3TAG MD
182
- WDT IN NEXTTAG PRP
183
- EX RB NEXTTAG .
184
- VBN VBD SURROUNDTAG NNS PRP$
185
- VBN VBD CURWD said
186
- JJ RB PREVTAG MD
187
- NN VBG NEXTBIGRAM JJ NNS
188
- JJ RB WDNEXTTAG late IN
189
- VBG NN PREVTAG PRP$
190
- VBZ NNS NEXTTAG VBP
191
- NN NNP WDPREVTAG DT CD
192
- NN VBN PREVWD be
193
- JJS RBS NEXTTAG VBN
194
- VBN VBD SURROUNDTAG NN PRP$
195
- VBN VBD SURROUNDTAG NNS JJ
196
- VBN VBD SURROUNDTAG NNS NN
197
- VBD VBN WDNEXTTAG increased NN
198
- VBZ NNS NEXTWD of
199
- IN RP WDAND2TAGAFT out NNS
200
- JJ NNP NEXTTAG POS
201
- RB RP WDNEXTTAG down DT
202
- CD NNS CURWD 1970s
203
- VBG NNP CURWD Working
204
- VBN VB PREVTAG MD
205
- JJ NN NEXTBIGRAM CC NN
206
- NN JJ SURROUNDTAG STAART NNS
207
- VBN VBD PREVBIGRAM , CC
208
- IN RB NEXTBIGRAM . STAART
209
- NN VBG PREVWD was
210
- NNP NNPS CURWD Cowboys
211
- VBZ NNS PREVWD phone
212
- NNP NNS SURROUNDTAG STAART VBP
213
- RBR JJR WDNEXTTAG lower JJ
214
- PRP$ PRP NEXTTAG IN
215
- VBD VB PREVTAG TO
216
- JJ NN WDPREVTAG NN chief
217
- JJ NN SURROUNDTAG JJ ,
218
- NN JJ WDPREVTAG DT third
219
- VBN VBD SURROUNDTAG NNS NNP
220
- NNP NN SURROUNDTAG STAART NN
221
- NNP NN CURWD HDTV
222
- VBG NN SURROUNDTAG DT ,
223
- VBG NN SURROUNDTAG DT .
224
- NNS VBZ PREVTAG WP
225
- NN VB SURROUNDTAG CC DT
226
- NNPS NNP WDAND2TAGBFR IN Securities
227
- RP IN PREVTAG NNS
228
- VBP NN LBIGRAM funds rate
229
- VBP NN WDPREVTAG NNS market
230
- DT RB RBIGRAM either .
231
- VBN NN SURROUNDTAG DT IN
232
- VBD VB PREV1OR2OR3TAG MD
233
- NN JJ NEXTWD oil
234
- VBN VBD SURROUNDTAG , $
235
- VBD VBN PREVBIGRAM DT RB
236
- VBN JJ PREVWD by
237
- NNP JJ WDNEXTTAG American JJ
238
- NN VBG PREVTAG VBP
239
- JJ RB LBIGRAM very much
240
- NN VBG RBIGRAM operating officer
241
- RB IN RBIGRAM up for
242
- NNS VBZ NEXTBIGRAM JJ NNS
243
- NNS VBZ SURROUNDTAG , IN
244
- VB VBP PREVTAG NNPS
245
- IN RP WDAND2TAGAFT out IN
246
- NNPS NNP PREVBIGRAM CC NNP
247
- NN RB RBIGRAM close to
248
- RBR RB PREVWD no
249
- JJ VBD NEXTTAG DT
250
- RB NNP PREVTAG NNP
251
- MD NN PREVWD good
252
- JJ NN WDPREVTAG NN giant
253
- NN JJ WDNEXTTAG official NNS
254
- VBN VBD SURROUNDTAG , PRP$
255
- VBN VBD SURROUNDTAG , RB
256
- VBN VBD SURROUNDTAG NN PRP
257
- NNP JJ WDNEXTTAG South JJ
258
- NN VBG PREVTAG RB
259
- NNS VBZ SURROUNDTAG , TO
260
- VBZ NNS SURROUNDTAG NN .
261
- NN VB NEXTTAG PRP$
262
- VBP VB PREV1OR2WD do
263
- VB JJ NEXTWD countries
264
- IN WDT NEXTBIGRAM RB VBZ
265
- JJ VB NEXTTAG DT
266
- WDT DT NEXTBIGRAM VBZ ,
267
- NNP RB RBIGRAM First ,
268
- DT NNP WDNEXTTAG A VBZ
269
- JJ RBR RBIGRAM further ,
270
- CD PRP WDNEXTTAG one MD
271
- POS '' PREV1OR2OR3TAG .
272
- PRP NN PREVTAG -LRB-
273
- VBN VBD SURROUNDTAG , PRP
274
- VBN VBD SURROUNDTAG NN NNS
275
- VBN VBD SURROUNDTAG NN RP
276
- NNP NN LBIGRAM STAART Business
277
- VBD VBN PREVTAG VBG
278
- IN RB RBIGRAM before ,
279
- IN RB WDAND2AFT As as
280
- NNP JJ LBIGRAM New York-based
281
- NNP JJ CURWD Mexican
282
- NNP NNPS WDNEXTTAG Motors NNP
283
- NNP NNPS WDPREVTAG NNP Enterprises
284
- JJ RB WDNEXTTAG long IN
@@ -1,148 +0,0 @@
1
- NN s fhassuf 1 NNS x
2
- NN . fchar CD x
3
- NN - fchar JJ x
4
- NN ed fhassuf 2 VBN x
5
- NN ing fhassuf 3 VBG x
6
- ly hassuf 2 RB x
7
- ly addsuf 2 JJ x
8
- NN $ fgoodright CD x
9
- NN al fhassuf 2 JJ x
10
- NN would fgoodright VB x
11
- NN 0 fchar CD x
12
- NN be fgoodright JJ x
13
- NNS us fhassuf 2 JJ x
14
- NNS it fgoodright VBZ x
15
- NN ble fhassuf 3 JJ x
16
- NN ic fhassuf 2 JJ x
17
- NN 1 fchar CD x
18
- NNS ss fhassuf 2 NN x
19
- un deletepref 2 JJ x
20
- NN ive fhassuf 3 JJ x
21
- NNP ed fhassuf 2 JJ x
22
- NN n't fgoodright VB x
23
- VB the fgoodright NN x
24
- NNS he fgoodright VBZ x
25
- VBN he fgoodright VBD x
26
- NN are fgoodright JJ x
27
- JJ was fgoodleft NN x
28
- NN est fhassuf 3 JJS x
29
- VBZ The fgoodright NNS x
30
- NNP ts fhassuf 2 NNS x
31
- NN 4 fchar CD x
32
- NN ize fhassuf 3 VB x
33
- .. hassuf 2 : x
34
- ful hassuf 3 JJ x
35
- NN ate fhassuf 3 VB x
36
- NNP ing fhassuf 3 VBG x
37
- VBG is fgoodleft NN x
38
- NN less fhassuf 4 JJ x
39
- NN ary fhassuf 3 JJ x
40
- Co. goodleft NNP x
41
- NN ant fhassuf 3 JJ x
42
- million goodleft CD x
43
- JJ their fgoodleft IN x
44
- NN he fgoodright VBD x
45
- Mr. goodright NNP x
46
- JJ of fgoodleft NN x
47
- NN so fgoodright JJ x
48
- NN y fdeletesuf 1 JJ x
49
- VBN which fgoodright VBD x
50
- VBD been fgoodright VBN x
51
- VB a fgoodright NN x
52
- NN economic fgoodleft JJ x
53
- 9 char CD x
54
- CD t fchar JJ x
55
- NN can fgoodright VB x
56
- VB the fgoodright NN x
57
- JJ S-T-A-R-T fgoodright VBN x
58
- VBN - fchar JJ x
59
- NN lar fhassuf 3 JJ x
60
- NNP ans fhassuf 3 NNPS x
61
- NN men fhassuf 3 NNS x
62
- CD d fchar JJ x
63
- JJ n fdeletesuf 1 VBN x
64
- JJ 's fgoodleft NN x
65
- NNS is fhassuf 2 NN x
66
- ES hassuf 2 NNS x
67
- JJ er fdeletesuf 2 JJR x
68
- Inc. goodleft NNP x
69
- NN 2 fchar CD x
70
- VBD be fgoodleft MD x
71
- ons hassuf 3 NNS x
72
- RB - fchar JJ x
73
- NN very fgoodright JJ x
74
- ous hassuf 3 JJ x
75
- NN a fdeletepref 1 RB x
76
- NNP people fgoodleft JJ x
77
- VB have fgoodleft RB x
78
- NNS It fgoodright VBZ x
79
- NN id fhassuf 2 JJ x
80
- JJ may fgoodleft NN x
81
- VBN but fgoodright VBD x
82
- RS hassuf 2 NNS x
83
- JJ stry fhassuf 4 NN x
84
- NNS them fgoodleft VBZ x
85
- VBZ were fgoodleft NNS x
86
- NN ing faddsuf 3 VB x
87
- JJ s faddsuf 1 NN x
88
- NN 7 fchar CD x
89
- NN d faddsuf 1 VB x
90
- VB but fgoodleft NN x
91
- NN 3 fchar CD x
92
- NN est faddsuf 3 JJ x
93
- NN en fhassuf 2 VBN x
94
- NN costs fgoodright IN x
95
- NN 8 fchar CD x
96
- VB b fhaspref 1 NN x
97
- zes hassuf 3 VBZ x
98
- VBN s faddsuf 1 NN x
99
- some hassuf 4 JJ x
100
- NN ic fhassuf 2 JJ x
101
- ly addsuf 2 JJ x
102
- ness addsuf 4 JJ x
103
- JJS s faddsuf 1 NN x
104
- NN ier fhassuf 3 JJR x
105
- NN ky fhassuf 2 JJ x
106
- tyle hassuf 4 JJ x
107
- NNS ates fhassuf 4 VBZ x
108
- fy hassuf 2 VB x
109
- body addsuf 4 DT x
110
- NN ways fgoodleft JJ x
111
- NNP ies fhassuf 3 NNPS x
112
- VB negative fgoodright NN x
113
- ders hassuf 4 NNS x
114
- ds hassuf 2 NNS x
115
- -day addsuf 4 CD x
116
- nian hassuf 4 JJ x
117
- JJR s faddsuf 1 NN x
118
- ppy hassuf 3 JJ x
119
- NN ish fhassuf 3 JJ x
120
- tors hassuf 4 NNS x
121
- oses hassuf 4 VBZ x
122
- NNS oves fhassuf 4 VBZ x
123
- VBN un fhaspref 2 JJ x
124
- lent hassuf 4 JJ x
125
- NN ward fdeletesuf 4 RB x
126
- VB k fchar NN x
127
- VB r fhassuf 1 NN x
128
- VB e fdeletesuf 1 NN x
129
- NNS Engelken fgoodright VBZ x
130
- NN ient fhassuf 4 JJ x
131
- ED hassuf 2 VBD x
132
- VBG B fchar NNP x
133
- VB le fhassuf 2 NN x
134
- ment addsuf 4 VB x
135
- ING hassuf 3 NN x
136
- JJ ery fhassuf 3 NN x
137
- JJ tus fhassuf 3 NN x
138
- JJ car fhassuf 3 NN x
139
- NN 6 fchar CD x
140
- NNS 0 fchar CD x
141
- JJ ing fdeletesuf 3 VBG x
142
- here hassuf 4 RB x
143
- VBN scr fhaspref 3 VBD x
144
- uces hassuf 4 VBZ x
145
- fies hassuf 4 VBZ x
146
- self deletesuf 4 PRP x
147
- NNP $ fchar $ x
148
- VBN wa fhaspref 2 VBD x
data/script/console DELETED
@@ -1,10 +0,0 @@
1
- #!/usr/bin/env ruby
2
- # File: script/console
3
- irb = RUBY_PLATFORM =~ /(:?mswin|mingw)/ ? 'irb.bat' : 'irb'
4
-
5
- libs = " -r irb/completion"
6
- # Perhaps use a console_lib to store any extra methods I may want available in the cosole
7
- # libs << " -r #{File.dirname(__FILE__) + '/../lib/console_lib/console_logger.rb'}"
8
- libs << " -r #{File.dirname(__FILE__) + '/../lib/rbtagger.rb'}"
9
- puts "Loading rb-brill-tagger gem"
10
- exec "#{irb} #{libs} --simple-prompt"
data/script/destroy DELETED
@@ -1,14 +0,0 @@
1
- #!/usr/bin/env ruby
2
- APP_ROOT = File.expand_path(File.join(File.dirname(__FILE__), '..'))
3
-
4
- begin
5
- require 'rubigen'
6
- rescue LoadError
7
- require 'rubygems'
8
- require 'rubigen'
9
- end
10
- require 'rubigen/scripts/destroy'
11
-
12
- ARGV.shift if ['--help', '-h'].include?(ARGV[0])
13
- RubiGen::Base.use_component_sources! [:rubygems, :newgem, :newgem_theme, :test_unit]
14
- RubiGen::Scripts::Destroy.new.run(ARGV)