grammar_cop 0.1.0 → 0.1.1
Sign up to get free protection for your applications and to get access to all the features.
- data/.DS_Store +0 -0
- data/.gitignore +1 -0
- data/README +22 -0
- data/ext/.DS_Store +0 -0
- data/ext/link_grammar/link_grammar.c +75 -0
- data/ext/link_grammar/link_grammar.h +10 -0
- data/lib/grammar_cop.rb +2 -1
- data/lib/grammar_cop/.DS_Store +0 -0
- data/lib/grammar_cop/parse_options.rb +39 -0
- data/lib/grammar_cop/sentence.rb +2 -1
- data/lib/grammar_cop/version.rb +1 -1
- data/test/grammar_cop_test.rb +10 -0
- metadata +3 -7
- data/data/.DS_Store +0 -0
- data/data/en/words/.DS_Store +0 -0
- data/ext/link_grammar/.DS_Store +0 -0
- data/lib/.DS_Store +0 -0
- data/test/.DS_Store +0 -0
data/.DS_Store
CHANGED
Binary file
|
data/.gitignore
CHANGED
data/README
ADDED
@@ -0,0 +1,22 @@
|
|
1
|
+
grammar_cop is a standalone ruby extension for the link-grammar-4.7.4 C library. You don't have to install anything extra, so this is deployable on platforms like heroku, cloudfoundry.
|
2
|
+
Example:
|
3
|
+
|
4
|
+
require 'grammar_cop'
|
5
|
+
|
6
|
+
puts dir = File.expand_path('../data', File.dirname(__FILE__))
|
7
|
+
GrammarCop::Dictionary.set_dictionary_path(dir)
|
8
|
+
|
9
|
+
dict = GrammarCop::Dictionary.create("en")
|
10
|
+
opts = GrammarCop::ParseOptions.create
|
11
|
+
text = "Grammar is useless because there's nothing to say -- Gertrude Stein."
|
12
|
+
sentence = GrammarCop::Sentence.create(text, dict)
|
13
|
+
puts num = GrammarCop::Sentence.parse(sentence, opts)
|
14
|
+
if num > 0
|
15
|
+
linkage = GrammarCop::Linkage.create(0, sentence, opts)
|
16
|
+
puts GrammarCop::Linkage.create_diagram(linkage)
|
17
|
+
puts "sublinkages=#{GrammarCop::Linkage.sublinkages_count(linkage)}"
|
18
|
+
puts "num_words=#{GrammarCop::Linkage.num_words_count(linkage)}"
|
19
|
+
puts "num_links=#{GrammarCop::Linkage.num_links_count(linkage)}"
|
20
|
+
#GrammarCop::Linkage.destroy(linkage)
|
21
|
+
end
|
22
|
+
GrammarCop::Sentence.destroy(sentence)
|
data/ext/.DS_Store
CHANGED
Binary file
|
@@ -17,6 +17,16 @@ void Init_link_grammar() {
|
|
17
17
|
rb_define_module_function(lg, "linkage_limit_get", linkage_limit_get, 1);
|
18
18
|
rb_define_module_function(lg, "short_length_set", short_length_set, 2);
|
19
19
|
rb_define_module_function(lg, "short_length_get", short_length_get, 1);
|
20
|
+
rb_define_module_function(lg, "disjunct_cost_set", disjunct_cost_set, 2);
|
21
|
+
rb_define_module_function(lg, "disjunct_cost_get", disjunct_cost_get, 1);
|
22
|
+
rb_define_module_function(lg, "min_null_count_set", min_null_count_set, 2);
|
23
|
+
rb_define_module_function(lg, "min_null_count_get", min_null_count_get, 1);
|
24
|
+
rb_define_module_function(lg, "max_null_count_set", max_null_count_set, 2);
|
25
|
+
rb_define_module_function(lg, "max_null_count_get", max_null_count_get, 1);
|
26
|
+
rb_define_module_function(lg, "options_null_block_set", options_null_block_set, 2);
|
27
|
+
rb_define_module_function(lg, "options_null_block_get", options_null_block_get, 1);
|
28
|
+
rb_define_module_function(lg, "islands_ok_set", islands_ok_set, 2);
|
29
|
+
rb_define_module_function(lg, "islands_ok_get", islands_ok_get, 1);
|
20
30
|
rb_define_module_function(lg, "create_sentence", create_sentence, 2);
|
21
31
|
rb_define_module_function(lg, "parse_sentence", parse_sentence, 2);
|
22
32
|
rb_define_module_function(lg, "delete_sentence", delete_sentence, 1);
|
@@ -136,6 +146,71 @@ VALUE short_length_get(const VALUE self, VALUE opts) {
|
|
136
146
|
return INT2FIX(l);
|
137
147
|
}
|
138
148
|
|
149
|
+
VALUE disjunct_cost_set(const VALUE self, VALUE opts, volatile VALUE cost) {
|
150
|
+
ParseOptionsPtr *ptr = retrieve_parse_options(opts);
|
151
|
+
int c = NUM2INT(cost);
|
152
|
+
parse_options_set_disjunct_cost(ptr->opts, c);
|
153
|
+
return Qnil;
|
154
|
+
}
|
155
|
+
|
156
|
+
VALUE disjunct_cost_get(const VALUE self, VALUE opts) {
|
157
|
+
ParseOptionsPtr *ptr = retrieve_parse_options(opts);
|
158
|
+
int c = parse_options_get_disjunct_cost(ptr->opts);
|
159
|
+
return INT2FIX(c);
|
160
|
+
}
|
161
|
+
|
162
|
+
VALUE min_null_count_set(const VALUE self, VALUE opts, volatile VALUE count) {
|
163
|
+
ParseOptionsPtr *ptr = retrieve_parse_options(opts);
|
164
|
+
int c = NUM2INT(count);
|
165
|
+
parse_options_set_min_null_count(ptr->opts, c);
|
166
|
+
return Qnil;
|
167
|
+
}
|
168
|
+
|
169
|
+
VALUE min_null_count_get(const VALUE self, VALUE opts) {
|
170
|
+
ParseOptionsPtr *ptr = retrieve_parse_options(opts);
|
171
|
+
int c = parse_options_get_min_null_count(ptr->opts);
|
172
|
+
return INT2FIX(c);
|
173
|
+
}
|
174
|
+
|
175
|
+
VALUE max_null_count_set(const VALUE self, VALUE opts, volatile VALUE count) {
|
176
|
+
ParseOptionsPtr *ptr = retrieve_parse_options(opts);
|
177
|
+
int c = NUM2INT(count);
|
178
|
+
parse_options_set_max_null_count(ptr->opts, c);
|
179
|
+
return Qnil;
|
180
|
+
}
|
181
|
+
|
182
|
+
VALUE max_null_count_get(const VALUE self, VALUE opts) {
|
183
|
+
ParseOptionsPtr *ptr = retrieve_parse_options(opts);
|
184
|
+
int c = parse_options_get_max_null_count(ptr->opts);
|
185
|
+
return INT2FIX(c);
|
186
|
+
}
|
187
|
+
|
188
|
+
VALUE options_null_block_set(const VALUE self, VALUE opts, volatile VALUE n_block) {
|
189
|
+
ParseOptionsPtr *ptr = retrieve_parse_options(opts);
|
190
|
+
int b = NUM2INT(n_block);
|
191
|
+
parse_options_set_null_block(ptr->opts, b);
|
192
|
+
return Qnil;
|
193
|
+
}
|
194
|
+
|
195
|
+
VALUE options_null_block_get(const VALUE self, VALUE opts) {
|
196
|
+
ParseOptionsPtr *ptr = retrieve_parse_options(opts);
|
197
|
+
int n_block = parse_options_get_null_block(ptr->opts);
|
198
|
+
return INT2FIX(n_block);
|
199
|
+
}
|
200
|
+
|
201
|
+
VALUE islands_ok_set(const VALUE self, VALUE opts, volatile VALUE islands_ok) {
|
202
|
+
ParseOptionsPtr *ptr = retrieve_parse_options(opts);
|
203
|
+
int i = NUM2INT(islands_ok);
|
204
|
+
parse_options_set_islands_ok(ptr->opts, i);
|
205
|
+
return Qnil;
|
206
|
+
}
|
207
|
+
|
208
|
+
VALUE islands_ok_get(const VALUE self, VALUE opts) {
|
209
|
+
ParseOptionsPtr *ptr = retrieve_parse_options(opts);
|
210
|
+
int i = parse_options_get_islands_ok(ptr->opts);
|
211
|
+
return INT2FIX(i);
|
212
|
+
}
|
213
|
+
|
139
214
|
VALUE create_sentence(const VALUE self, volatile VALUE str, VALUE dict) {
|
140
215
|
char *text = StringValuePtr(str);
|
141
216
|
DictionaryPtr *dict_ptr = retrieve_dictionary(dict);
|
@@ -86,6 +86,16 @@ VALUE linkage_limit_set(const VALUE self, VALUE opts, volatile VALUE limit);
|
|
86
86
|
VALUE linkage_limit_get(const VALUE self, VALUE opts);
|
87
87
|
VALUE short_length_set(const VALUE self, VALUE opts, volatile VALUE length);
|
88
88
|
VALUE short_length_get(const VALUE self, VALUE opts);
|
89
|
+
VALUE disjunct_cost_set(const VALUE self, VALUE opts, volatile VALUE cost);
|
90
|
+
VALUE disjunct_cost_get(const VALUE self, VALUE opts);
|
91
|
+
VALUE min_null_count_set(const VALUE self, VALUE opts, volatile VALUE count);
|
92
|
+
VALUE min_null_count_get(const VALUE self, VALUE opts);
|
93
|
+
VALUE max_null_count_set(const VALUE self, VALUE opts, volatile VALUE count);
|
94
|
+
VALUE max_null_count_get(const VALUE self, VALUE opts);
|
95
|
+
VALUE options_null_block_set(const VALUE self, VALUE opts, volatile VALUE n_block);
|
96
|
+
VALUE options_null_block_get(const VALUE self, VALUE opts);
|
97
|
+
VALUE islands_ok_set(const VALUE self, VALUE opts, volatile VALUE islands_ok);
|
98
|
+
VALUE islands_ok_get(const VALUE self, VALUE opts);
|
89
99
|
|
90
100
|
VALUE create_sentence(const VALUE self, volatile VALUE str, VALUE dict);
|
91
101
|
VALUE parse_sentence(const VALUE self, VALUE sentence, VALUE opts);
|
data/lib/grammar_cop.rb
CHANGED
@@ -5,5 +5,6 @@ require 'grammar_cop/sentence'
|
|
5
5
|
require 'grammar_cop/linkage'
|
6
6
|
|
7
7
|
module GrammarCop
|
8
|
-
GrammarCop::Dictionary.set_dictionary_path("#{Gem.loaded_specs['grammar_cop'].full_gem_path}/data")
|
8
|
+
#GrammarCop::Dictionary.set_dictionary_path("#{Gem.loaded_specs['grammar_cop'].full_gem_path}/data")
|
9
|
+
GrammarCop::Dictionary.set_dictionary_path(File.expand_path('../data', File.dirname(__FILE__)))
|
9
10
|
end
|
data/lib/grammar_cop/.DS_Store
CHANGED
Binary file
|
@@ -28,5 +28,44 @@ module GrammarCop
|
|
28
28
|
short_length_get(opts)
|
29
29
|
end
|
30
30
|
|
31
|
+
def self.set_disjunct_cost(opts, cost)
|
32
|
+
disjunct_cost_set(opts, cost)
|
33
|
+
end
|
34
|
+
|
35
|
+
def self.get_disjunct_cost(opts)
|
36
|
+
disjunct_cost_get(opts)
|
37
|
+
end
|
38
|
+
|
39
|
+
def self.set_min_null_count(opts, count)
|
40
|
+
min_null_count_set(opts, count)
|
41
|
+
end
|
42
|
+
|
43
|
+
def self.get_min_null_count(opts)
|
44
|
+
min_null_count_get(opts)
|
45
|
+
end
|
46
|
+
|
47
|
+
def self.set_max_null_count(opts, count)
|
48
|
+
max_null_count_set(opts, count)
|
49
|
+
end
|
50
|
+
|
51
|
+
def self.get_max_null_count(opts)
|
52
|
+
max_null_count_get(opts)
|
53
|
+
end
|
54
|
+
|
55
|
+
def self.set_null_block(opts, block)
|
56
|
+
options_null_block_set(opts, block)
|
57
|
+
end
|
58
|
+
|
59
|
+
def self.get_null_block(opts)
|
60
|
+
options_null_block_get(opts)
|
61
|
+
end
|
62
|
+
|
63
|
+
def self.set_islands_ok(opts, ok)
|
64
|
+
options_null_block_set(opts, ok)
|
65
|
+
end
|
66
|
+
|
67
|
+
def self.get_islands_ok(opts)
|
68
|
+
islands_ok_get(opts)
|
69
|
+
end
|
31
70
|
end
|
32
71
|
end
|
data/lib/grammar_cop/sentence.rb
CHANGED
@@ -25,7 +25,8 @@ module GrammarCop
|
|
25
25
|
end
|
26
26
|
|
27
27
|
#however, sometimes the split results in all incomplete clause so check one more time in those cases
|
28
|
-
|
28
|
+
#but make sure it's not a really long sentence which will cause link-grammar to choke with combinatorial explosion
|
29
|
+
if text.size < 250 and count == 0
|
29
30
|
s = create_sentence(text, dictionary)
|
30
31
|
count = parse_sentence(s, options)
|
31
32
|
delete_sentence(s)
|
data/lib/grammar_cop/version.rb
CHANGED
data/test/grammar_cop_test.rb
CHANGED
@@ -4,11 +4,21 @@ puts dir = File.expand_path('../data', File.dirname(__FILE__))
|
|
4
4
|
GrammarCop::Dictionary.set_dictionary_path(dir)
|
5
5
|
|
6
6
|
dict = GrammarCop::Dictionary.create("en")
|
7
|
+
|
8
|
+
#create and set parse options
|
7
9
|
opts = GrammarCop::ParseOptions.create
|
8
10
|
GrammarCop::ParseOptions.set_short_length(opts, 7)
|
9
11
|
puts "short_length=#{GrammarCop::ParseOptions.get_short_length(opts)}"
|
10
12
|
GrammarCop::ParseOptions.set_linkage_limit(opts, 10)
|
11
13
|
puts "linkage limit=#{GrammarCop::ParseOptions.get_linkage_limit(opts)}"
|
14
|
+
GrammarCop::ParseOptions.set_disjunct_cost(opts, 1000)
|
15
|
+
puts "disjunct cost=#{GrammarCop::ParseOptions.get_disjunct_cost(opts)}"
|
16
|
+
|
17
|
+
puts "min null count=#{GrammarCop::ParseOptions.get_min_null_count(opts)}"
|
18
|
+
puts "max null count=#{GrammarCop::ParseOptions.get_max_null_count(opts)}"
|
19
|
+
puts "null block=#{GrammarCop::ParseOptions.get_null_block(opts)}"
|
20
|
+
puts "islands ok=#{GrammarCop::ParseOptions.get_islands_ok(opts)}"
|
21
|
+
|
12
22
|
text = "Grammar is useless because there's nothing to say -- Gertrude Stein."
|
13
23
|
sentence = GrammarCop::Sentence.create(text, dict)
|
14
24
|
puts num = GrammarCop::Sentence.parse(sentence, opts)
|
metadata
CHANGED
@@ -2,7 +2,7 @@
|
|
2
2
|
name: grammar_cop
|
3
3
|
version: !ruby/object:Gem::Version
|
4
4
|
prerelease:
|
5
|
-
version: 0.1.
|
5
|
+
version: 0.1.1
|
6
6
|
platform: ruby
|
7
7
|
authors:
|
8
8
|
- Quan Nguyen
|
@@ -10,7 +10,7 @@ autorequire:
|
|
10
10
|
bindir: bin
|
11
11
|
cert_chain: []
|
12
12
|
|
13
|
-
date: 2011-05-
|
13
|
+
date: 2011-05-13 00:00:00 Z
|
14
14
|
dependencies:
|
15
15
|
- !ruby/object:Gem::Dependency
|
16
16
|
name: rake-compiler
|
@@ -36,8 +36,8 @@ files:
|
|
36
36
|
- .DS_Store
|
37
37
|
- .gitignore
|
38
38
|
- Gemfile
|
39
|
+
- README
|
39
40
|
- Rakefile
|
40
|
-
- data/.DS_Store
|
41
41
|
- data/Makefile
|
42
42
|
- data/Makefile.am
|
43
43
|
- data/Makefile.in
|
@@ -64,7 +64,6 @@ files:
|
|
64
64
|
- data/en/Makefile.in
|
65
65
|
- data/en/README
|
66
66
|
- data/en/tiny.dict
|
67
|
-
- data/en/words/.DS_Store
|
68
67
|
- data/en/words/Makefile
|
69
68
|
- data/en/words/Makefile.am
|
70
69
|
- data/en/words/Makefile.in
|
@@ -148,7 +147,6 @@ files:
|
|
148
147
|
- data/lt/Makefile.am
|
149
148
|
- data/lt/Makefile.in
|
150
149
|
- ext/.DS_Store
|
151
|
-
- ext/link_grammar/.DS_Store
|
152
150
|
- ext/link_grammar/extconf.rb
|
153
151
|
- ext/link_grammar/link-grammar/.DS_Store
|
154
152
|
- ext/link_grammar/link-grammar/.deps/analyze-linkage.Plo
|
@@ -366,7 +364,6 @@ files:
|
|
366
364
|
- ext/link_grammar/link_grammar.c
|
367
365
|
- ext/link_grammar/link_grammar.h
|
368
366
|
- grammar_cop.gemspec
|
369
|
-
- lib/.DS_Store
|
370
367
|
- lib/grammar_cop.rb
|
371
368
|
- lib/grammar_cop/.DS_Store
|
372
369
|
- lib/grammar_cop/dictionary.rb
|
@@ -374,7 +371,6 @@ files:
|
|
374
371
|
- lib/grammar_cop/parse_options.rb
|
375
372
|
- lib/grammar_cop/sentence.rb
|
376
373
|
- lib/grammar_cop/version.rb
|
377
|
-
- test/.DS_Store
|
378
374
|
- test/grammar_cop_test.rb
|
379
375
|
homepage: https://github.com/mquan
|
380
376
|
licenses: []
|
data/data/.DS_Store
DELETED
Binary file
|
data/data/en/words/.DS_Store
DELETED
Binary file
|
data/ext/link_grammar/.DS_Store
DELETED
Binary file
|
data/lib/.DS_Store
DELETED
Binary file
|
data/test/.DS_Store
DELETED
Binary file
|