mecab-light 0.2.2 → 0.2.3
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/.travis.yml +2 -0
- data/README.md +5 -5
- data/ext/mecab/light/binding.c +20 -23
- data/ext/mecab/light/extconf.rb +7 -0
- data/lib/mecab/light/result.rb +3 -2
- data/lib/mecab/light/tagger.rb +3 -3
- data/lib/mecab/light/version.rb +1 -1
- data/spec/mecab-light-morpheme_spec.rb +30 -0
- data/spec/mecab-light-result_spec.rb +62 -0
- data/spec/mecab-light-tagger_spec.rb +57 -0
- data/spec/spec_helper.rb +8 -0
- metadata +12 -5
- data/spec/mecab_spec.rb +0 -110
checksums.yaml
CHANGED
@@ -1,7 +1,7 @@
|
|
1
1
|
---
|
2
2
|
SHA1:
|
3
|
-
metadata.gz:
|
4
|
-
data.tar.gz:
|
3
|
+
metadata.gz: c610ca8725dff12d78f4c8e57aed3a669cb4dfd9
|
4
|
+
data.tar.gz: 2c7400b6c7355e26bf1cf0c11f9dbd9a98401330
|
5
5
|
SHA512:
|
6
|
-
metadata.gz:
|
7
|
-
data.tar.gz:
|
6
|
+
metadata.gz: 68e644215d1733e15152650d68923532d20e6cfeae91f0150f1e8b9c9b56503ce42cfc090c9d5cc4f6db26ce971784117d081937d2c61507b8d80c93e9ccaa6d
|
7
|
+
data.tar.gz: df72e4d97b55674761e303e218d8087bf470e9d730be9017eaef8a059bff3a53fbb5c85938f00bc00e79099b6a03e8fb4bac028fafaddc9a408cc5dd4de27cf2
|
data/.travis.yml
ADDED
data/README.md
CHANGED
@@ -18,10 +18,10 @@ Or install it yourself as:
|
|
18
18
|
|
19
19
|
## Usage
|
20
20
|
|
21
|
-
|
22
|
-
|
23
|
-
This supports only
|
24
|
-
|
21
|
+
MeCab::Light is a lightweight tool.
|
22
|
+
This gem works without the official binding.
|
23
|
+
This supports only Tagger#parse method for now.
|
24
|
+
Note that the method's feature is totally different from its original.
|
25
25
|
|
26
26
|
```ruby
|
27
27
|
require 'mecab/light'
|
@@ -31,7 +31,7 @@ string = 'この文を形態素解析してください。'
|
|
31
31
|
result = tagger.parse(string)
|
32
32
|
result[0].surface #=> "この"
|
33
33
|
result.kind_of?(Enumerable) #=> true
|
34
|
-
result.map{|morpheme| morpheme.surface }
|
34
|
+
result.map { |morpheme| morpheme.surface }
|
35
35
|
#=> ["この", "文", "を", "形態素", "解析", "し", "て", "ください", "。"]
|
36
36
|
```
|
37
37
|
|
data/ext/mecab/light/binding.c
CHANGED
@@ -2,52 +2,49 @@
|
|
2
2
|
#include <ruby/encoding.h>
|
3
3
|
#include <mecab.h>
|
4
4
|
|
5
|
-
struct
|
6
|
-
mecab_t*
|
7
|
-
};
|
5
|
+
typedef struct {
|
6
|
+
mecab_t* ptr;
|
7
|
+
} MeCab;
|
8
8
|
|
9
9
|
static VALUE
|
10
|
-
|
10
|
+
mecab_alloc(VALUE klass)
|
11
11
|
{
|
12
|
-
|
13
|
-
return Data_Wrap_Struct(klass, 0,
|
12
|
+
MeCab* mecab = ALLOC(MeCab);
|
13
|
+
return Data_Wrap_Struct(klass, 0, 0, mecab);
|
14
14
|
}
|
15
15
|
|
16
16
|
static VALUE
|
17
|
-
|
17
|
+
rb_mecab_initialize(VALUE self, VALUE arg)
|
18
18
|
{
|
19
|
-
|
19
|
+
MeCab* mecab;
|
20
20
|
|
21
|
-
Data_Get_Struct(self,
|
22
|
-
ptr
|
21
|
+
Data_Get_Struct(self, MeCab, mecab);
|
22
|
+
mecab->ptr = mecab_new2(RSTRING_PTR(arg));
|
23
23
|
return Qnil;
|
24
24
|
}
|
25
25
|
|
26
26
|
static VALUE
|
27
|
-
|
27
|
+
rb_mecab_parse_to_s(VALUE self, VALUE str)
|
28
28
|
{
|
29
|
-
|
29
|
+
MeCab* mecab;
|
30
30
|
const char* result;
|
31
31
|
rb_encoding* enc;
|
32
32
|
|
33
|
-
Data_Get_Struct(self,
|
33
|
+
Data_Get_Struct(self, MeCab, mecab);
|
34
34
|
enc = rb_enc_get(str);
|
35
|
-
result = mecab_sparse_tostr(ptr
|
35
|
+
result = mecab_sparse_tostr(mecab->ptr, RSTRING_PTR(str));
|
36
36
|
return rb_enc_associate(rb_str_new2(result), enc);
|
37
37
|
}
|
38
38
|
|
39
39
|
void
|
40
40
|
Init_binding()
|
41
41
|
{
|
42
|
-
VALUE rb_mMeCab,
|
42
|
+
VALUE rb_mMeCab, rb_mLight, rb_cBinding;
|
43
43
|
|
44
44
|
rb_mMeCab = rb_define_module("MeCab");
|
45
|
-
|
46
|
-
|
47
|
-
|
48
|
-
|
49
|
-
|
50
|
-
rb_mecab_light_binding_initialize, 1);
|
51
|
-
rb_define_method(rb_cMeCab_Light_Binding, "parse_to_s",
|
52
|
-
rb_mecab_light_binding_parse_to_s, 1);
|
45
|
+
rb_mLight = rb_define_module_under(rb_mMeCab, "Light");
|
46
|
+
rb_cBinding = rb_define_class_under(rb_mLight, "Binding", rb_cObject);
|
47
|
+
rb_define_alloc_func(rb_cBinding, mecab_alloc);
|
48
|
+
rb_define_private_method(rb_cBinding, "initialize", rb_mecab_initialize, 1);
|
49
|
+
rb_define_method(rb_cBinding, "parse_to_s", rb_mecab_parse_to_s, 1);
|
53
50
|
}
|
data/ext/mecab/light/extconf.rb
CHANGED
data/lib/mecab/light/result.rb
CHANGED
@@ -2,8 +2,9 @@ module MeCab
|
|
2
2
|
module Light
|
3
3
|
class Result
|
4
4
|
include Enumerable
|
5
|
-
|
6
|
-
|
5
|
+
|
6
|
+
def initialize(parsed)
|
7
|
+
@morphemes = parsed.each_line.map { |line| Morpheme.new(line) }
|
7
8
|
end
|
8
9
|
|
9
10
|
def each(&block)
|
data/lib/mecab/light/tagger.rb
CHANGED
@@ -6,12 +6,12 @@ module MeCab
|
|
6
6
|
end
|
7
7
|
|
8
8
|
def parse(string)
|
9
|
-
Result.new(
|
9
|
+
Result.new(parse_to_s(string))
|
10
10
|
end
|
11
11
|
|
12
12
|
private
|
13
|
-
def
|
14
|
-
@mecab.parse_to_s(string).sub(/EOS\n$/, '')
|
13
|
+
def parse_to_s(string)
|
14
|
+
@mecab.parse_to_s(string).sub(/EOS\n$/, '')
|
15
15
|
end
|
16
16
|
end
|
17
17
|
end
|
data/lib/mecab/light/version.rb
CHANGED
@@ -0,0 +1,30 @@
|
|
1
|
+
require 'spec_helper'
|
2
|
+
|
3
|
+
describe MeCab::Light::Morpheme do
|
4
|
+
let(:morpheme) { MeCab::Light::Morpheme.new("surface\tfeature\n") }
|
5
|
+
|
6
|
+
context 'the class' do
|
7
|
+
it { expect(MeCab::Light::Morpheme).to respond_to(:new).with(1).argument }
|
8
|
+
end
|
9
|
+
|
10
|
+
context 'an instance initialized with "surface\tfeature\n"' do
|
11
|
+
it { expect(morpheme).to respond_to(:surface).with(0).arguments }
|
12
|
+
it { expect(morpheme).to respond_to(:feature).with(0).arguments }
|
13
|
+
|
14
|
+
context 'surface' do
|
15
|
+
it { expect(morpheme.surface).to eq('surface') }
|
16
|
+
|
17
|
+
context 'encoding' do
|
18
|
+
it { expect(morpheme.surface.encoding).to eq(Encoding::UTF_8) }
|
19
|
+
end
|
20
|
+
end
|
21
|
+
|
22
|
+
context 'feature' do
|
23
|
+
it { expect(morpheme.feature).to eq('feature') }
|
24
|
+
|
25
|
+
context 'encoding' do
|
26
|
+
it { expect(morpheme.feature.encoding).to eq(Encoding::UTF_8) }
|
27
|
+
end
|
28
|
+
end
|
29
|
+
end
|
30
|
+
end
|
@@ -0,0 +1,62 @@
|
|
1
|
+
require 'spec_helper'
|
2
|
+
|
3
|
+
describe MeCab::Light::Result do
|
4
|
+
let(:morpheme) { double('MeCab::Light::Morpheme') }
|
5
|
+
let(:result) { MeCab::Light::Result.new("surface\tfeature\n") }
|
6
|
+
before do
|
7
|
+
MeCab::Light::Morpheme.stub(:new).and_return(morpheme)
|
8
|
+
end
|
9
|
+
|
10
|
+
context 'class' do
|
11
|
+
it { expect(MeCab::Light::Result).to respond_to(:new).with(1).argument }
|
12
|
+
|
13
|
+
context 'new with "surface\tfeature\n"' do
|
14
|
+
context 'MeCab::Light::Morpheme class' do
|
15
|
+
after do
|
16
|
+
MeCab::Light::Result.new("surface\tfeature\n")
|
17
|
+
end
|
18
|
+
|
19
|
+
it 'should receive #new with "surface\tfeature\n"' do
|
20
|
+
expect(MeCab::Light::Morpheme).to receive(:new).with("surface\tfeature\n")
|
21
|
+
end
|
22
|
+
end
|
23
|
+
end
|
24
|
+
end
|
25
|
+
|
26
|
+
context 'an instance initialized with ""' do
|
27
|
+
it { expect(result).to respond_to(:each).with(0).arguments }
|
28
|
+
it { expect(result).to be_an(Enumerable) }
|
29
|
+
|
30
|
+
context 'count' do
|
31
|
+
it { expect(result.count).to eq(1) }
|
32
|
+
end
|
33
|
+
|
34
|
+
context 'each' do
|
35
|
+
context 'with block' do
|
36
|
+
it 'should return self' do
|
37
|
+
expect(result.each{}).to eq(result)
|
38
|
+
end
|
39
|
+
it { expect { |b| result.each(&b) }.to yield_control }
|
40
|
+
it 'should yield with args(MeCab::Light::Morpheme)' do
|
41
|
+
expect { |b| result.each(&b) }.to yield_with_args(morpheme)
|
42
|
+
end
|
43
|
+
end
|
44
|
+
|
45
|
+
context 'without block' do
|
46
|
+
it { expect(result.each).to be_an_instance_of(Enumerator) }
|
47
|
+
end
|
48
|
+
end
|
49
|
+
|
50
|
+
context '[] with 0' do
|
51
|
+
it 'should be an instance of Morpheme' do
|
52
|
+
expect(result[0]).to eq(morpheme)
|
53
|
+
end
|
54
|
+
end
|
55
|
+
|
56
|
+
context 'at with 0' do
|
57
|
+
it 'should be an instance of Morpheme' do
|
58
|
+
expect(result.at(0)).to eq(morpheme)
|
59
|
+
end
|
60
|
+
end
|
61
|
+
end
|
62
|
+
end
|
@@ -0,0 +1,57 @@
|
|
1
|
+
require 'spec_helper'
|
2
|
+
|
3
|
+
describe MeCab::Light::Tagger do
|
4
|
+
let(:binding) { double( parse_to_s: "surface\tfeature\nEOS\n" ) }
|
5
|
+
let(:result) { double('MeCab::Light::Result') }
|
6
|
+
let(:tagger) { MeCab::Light::Tagger.new }
|
7
|
+
before do
|
8
|
+
MeCab::Light::Binding.stub(:new).and_return(binding)
|
9
|
+
MeCab::Light::Result.stub(:new).and_return(result)
|
10
|
+
end
|
11
|
+
|
12
|
+
context 'the class' do
|
13
|
+
it { expect(MeCab::Light::Tagger).to respond_to(:new).with(0).arguments }
|
14
|
+
|
15
|
+
context 'new' do
|
16
|
+
context 'MeCab::Light::Binding class' do
|
17
|
+
after do
|
18
|
+
MeCab::Light::Tagger.new
|
19
|
+
end
|
20
|
+
|
21
|
+
it 'should receive #new with ""' do
|
22
|
+
expect(MeCab::Light::Binding).to receive(:new).with('')
|
23
|
+
end
|
24
|
+
end
|
25
|
+
end
|
26
|
+
end
|
27
|
+
|
28
|
+
context 'an instance' do
|
29
|
+
it { expect(tagger).to respond_to(:parse).with(1).argument }
|
30
|
+
|
31
|
+
context 'parse with "surface"' do
|
32
|
+
it 'should be an instance of MeCab::Light::Result' do
|
33
|
+
expect(tagger.parse('surface')).to eq(result)
|
34
|
+
end
|
35
|
+
|
36
|
+
context 'MeCab::Light::Result class' do
|
37
|
+
after do
|
38
|
+
tagger.parse('surface')
|
39
|
+
end
|
40
|
+
|
41
|
+
it 'should receive #new with "surface\tfeature\n"' do
|
42
|
+
expect(MeCab::Light::Result).to receive(:new).with("surface\tfeature\n")
|
43
|
+
end
|
44
|
+
end
|
45
|
+
|
46
|
+
context 'a MeCab::Light::Binding object' do
|
47
|
+
after do
|
48
|
+
tagger.parse('surface')
|
49
|
+
end
|
50
|
+
|
51
|
+
it 'should receive #parse_to_s with "surface"' do
|
52
|
+
expect(binding).to receive(:parse_to_s).with('surface')
|
53
|
+
end
|
54
|
+
end
|
55
|
+
end
|
56
|
+
end
|
57
|
+
end
|
data/spec/spec_helper.rb
ADDED
metadata
CHANGED
@@ -1,14 +1,14 @@
|
|
1
1
|
--- !ruby/object:Gem::Specification
|
2
2
|
name: mecab-light
|
3
3
|
version: !ruby/object:Gem::Version
|
4
|
-
version: 0.2.
|
4
|
+
version: 0.2.3
|
5
5
|
platform: ruby
|
6
6
|
authors:
|
7
7
|
- Hajime WAKAHARA
|
8
8
|
autorequire:
|
9
9
|
bindir: bin
|
10
10
|
cert_chain: []
|
11
|
-
date: 2013-
|
11
|
+
date: 2013-10-22 00:00:00.000000000 Z
|
12
12
|
dependencies:
|
13
13
|
- !ruby/object:Gem::Dependency
|
14
14
|
name: rake
|
@@ -47,6 +47,7 @@ extensions:
|
|
47
47
|
extra_rdoc_files: []
|
48
48
|
files:
|
49
49
|
- .gitignore
|
50
|
+
- .travis.yml
|
50
51
|
- Gemfile
|
51
52
|
- LICENSE.txt
|
52
53
|
- README.md
|
@@ -59,7 +60,10 @@ files:
|
|
59
60
|
- lib/mecab/light/tagger.rb
|
60
61
|
- lib/mecab/light/version.rb
|
61
62
|
- mecab-light.gemspec
|
62
|
-
- spec/
|
63
|
+
- spec/mecab-light-morpheme_spec.rb
|
64
|
+
- spec/mecab-light-result_spec.rb
|
65
|
+
- spec/mecab-light-tagger_spec.rb
|
66
|
+
- spec/spec_helper.rb
|
63
67
|
homepage: https://github.com/hadzimme/mecab-light
|
64
68
|
licenses: []
|
65
69
|
metadata: {}
|
@@ -79,9 +83,12 @@ required_rubygems_version: !ruby/object:Gem::Requirement
|
|
79
83
|
version: '0'
|
80
84
|
requirements: []
|
81
85
|
rubyforge_project:
|
82
|
-
rubygems_version: 2.
|
86
|
+
rubygems_version: 2.1.9
|
83
87
|
signing_key:
|
84
88
|
specification_version: 4
|
85
89
|
summary: An simple interface for MeCab (UNOFFICIAL)
|
86
90
|
test_files:
|
87
|
-
- spec/
|
91
|
+
- spec/mecab-light-morpheme_spec.rb
|
92
|
+
- spec/mecab-light-result_spec.rb
|
93
|
+
- spec/mecab-light-tagger_spec.rb
|
94
|
+
- spec/spec_helper.rb
|
data/spec/mecab_spec.rb
DELETED
@@ -1,110 +0,0 @@
|
|
1
|
-
# coding: utf-8
|
2
|
-
|
3
|
-
require 'rspec'
|
4
|
-
begin
|
5
|
-
require 'mecab/light'
|
6
|
-
rescue LoadError
|
7
|
-
puts 'loaded mecab-light but MeCab does not exist.'
|
8
|
-
puts 'dummy binding will be loaded.'
|
9
|
-
end
|
10
|
-
|
11
|
-
module MeCab
|
12
|
-
module Light
|
13
|
-
class Binding
|
14
|
-
def initialize(option)
|
15
|
-
end
|
16
|
-
|
17
|
-
def parse_to_s(string)
|
18
|
-
"見る\t動詞,自立,*,*,一段,基本形,見る,ミル,ミル\nEOS\n"
|
19
|
-
end
|
20
|
-
end
|
21
|
-
end
|
22
|
-
end
|
23
|
-
|
24
|
-
describe MeCab::Light::Morpheme do
|
25
|
-
context 'when initialized with the result line of the word "見る"' do
|
26
|
-
before do
|
27
|
-
@morpheme = MeCab::Light::Morpheme.new(
|
28
|
-
"見る\t動詞,自立,*,*,一段,基本形,見る,ミル,ミル\n")
|
29
|
-
end
|
30
|
-
subject { @morpheme }
|
31
|
-
its(:surface){ should eq '見る' }
|
32
|
-
its(:feature){ should eq '動詞,自立,*,*,一段,基本形,見る,ミル,ミル' }
|
33
|
-
describe 'surface' do
|
34
|
-
subject { @morpheme.surface }
|
35
|
-
its(:encoding){ should be Encoding::UTF_8 }
|
36
|
-
end
|
37
|
-
describe 'feature' do
|
38
|
-
subject { @morpheme.feature }
|
39
|
-
its(:encoding){ should be Encoding::UTF_8 }
|
40
|
-
end
|
41
|
-
end
|
42
|
-
end
|
43
|
-
describe MeCab::Light::Result do
|
44
|
-
context 'when initialized with the result Enumerator of the word "見る"' do
|
45
|
-
before do
|
46
|
-
@result = MeCab::Light::Result.new(
|
47
|
-
["見る\t動詞,自立,*,*,一段,基本形,見る,ミル,ミル\n"].to_enum)
|
48
|
-
end
|
49
|
-
subject { @result }
|
50
|
-
it { should respond_to :each }
|
51
|
-
it { should be_an Enumerable }
|
52
|
-
its(:count){ should be 1 }
|
53
|
-
describe 'each' do
|
54
|
-
context 'with block' do
|
55
|
-
subject { @result.each{} }
|
56
|
-
it 'should return self' do
|
57
|
-
should be @result
|
58
|
-
end
|
59
|
-
end
|
60
|
-
context 'without block' do
|
61
|
-
subject { @result.each }
|
62
|
-
it 'should return Enumerator' do
|
63
|
-
should be_an_instance_of Enumerator
|
64
|
-
end
|
65
|
-
end
|
66
|
-
end
|
67
|
-
describe '[]' do
|
68
|
-
context 'when argument 0' do
|
69
|
-
subject { @result[0] }
|
70
|
-
it { should be_an_instance_of MeCab::Light::Morpheme }
|
71
|
-
end
|
72
|
-
end
|
73
|
-
describe 'at' do
|
74
|
-
context 'when argument 0' do
|
75
|
-
subject { @result.at(0) }
|
76
|
-
it { should be_an_instance_of MeCab::Light::Morpheme }
|
77
|
-
end
|
78
|
-
end
|
79
|
-
end
|
80
|
-
end
|
81
|
-
describe MeCab::Light::Tagger do
|
82
|
-
before { @tagger = MeCab::Light::Tagger.new }
|
83
|
-
subject { @tagger }
|
84
|
-
it { should respond_to :parse }
|
85
|
-
describe 'parse' do
|
86
|
-
context 'when argument "見る"' do
|
87
|
-
before { @result = @tagger.parse('見る') }
|
88
|
-
subject { @result }
|
89
|
-
it 'should return MeCab::Light::Result object' do
|
90
|
-
should be_an_instance_of MeCab::Light::Result
|
91
|
-
end
|
92
|
-
its(:count){ should be 1 }
|
93
|
-
describe 'MeCab::Light::Result (returned)' do
|
94
|
-
context '[0]' do
|
95
|
-
before { @morpheme = @result[0] }
|
96
|
-
subject { @morpheme }
|
97
|
-
it 'should return MeCab::Light::Morpheme object' do
|
98
|
-
should be_an_instance_of MeCab::Light::Morpheme
|
99
|
-
end
|
100
|
-
describe 'MeCab::Light::Morpheme (returned)' do
|
101
|
-
context 'surface' do
|
102
|
-
subject { @morpheme.surface }
|
103
|
-
it { should eq '見る' }
|
104
|
-
end
|
105
|
-
end
|
106
|
-
end
|
107
|
-
end
|
108
|
-
end
|
109
|
-
end
|
110
|
-
end
|