opener-kaf-naf-parser 1.0.0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (97) hide show
  1. checksums.yaml +7 -0
  2. data/LICENSE +674 -0
  3. data/README.md +46 -0
  4. data/bin/kaf-naf-parser +8 -0
  5. data/bin/kaf-naf-parser-server +10 -0
  6. data/bin/kaf-to-naf +7 -0
  7. data/bin/naf-to-kaf +7 -0
  8. data/config.ru +4 -0
  9. data/core/kaf-naf-parser.py +42 -0
  10. data/core/packages/KafNafParser-1.2.tar.gz +0 -0
  11. data/core/packages/VUA_pylib-1.3.tar.gz +0 -0
  12. data/core/site-packages/pre_build/KafNafParser-1.2-py2.7.egg-info/PKG-INFO +10 -0
  13. data/core/site-packages/pre_build/KafNafParser-1.2-py2.7.egg-info/SOURCES.txt +22 -0
  14. data/core/site-packages/pre_build/KafNafParser-1.2-py2.7.egg-info/dependency_links.txt +1 -0
  15. data/core/site-packages/pre_build/KafNafParser-1.2-py2.7.egg-info/installed-files.txt +47 -0
  16. data/core/site-packages/pre_build/KafNafParser-1.2-py2.7.egg-info/top_level.txt +1 -0
  17. data/core/site-packages/pre_build/KafNafParser/KafNafParserMod.py +338 -0
  18. data/core/site-packages/pre_build/KafNafParser/KafNafParserMod.pyc +0 -0
  19. data/core/site-packages/pre_build/KafNafParser/__init__.py +14 -0
  20. data/core/site-packages/pre_build/KafNafParser/__init__.pyc +0 -0
  21. data/core/site-packages/pre_build/KafNafParser/constituency_data.py +125 -0
  22. data/core/site-packages/pre_build/KafNafParser/constituency_data.pyc +0 -0
  23. data/core/site-packages/pre_build/KafNafParser/coreference_data.py +52 -0
  24. data/core/site-packages/pre_build/KafNafParser/coreference_data.pyc +0 -0
  25. data/core/site-packages/pre_build/KafNafParser/dependency_data.py +80 -0
  26. data/core/site-packages/pre_build/KafNafParser/dependency_data.pyc +0 -0
  27. data/core/site-packages/pre_build/KafNafParser/entity_data.py +59 -0
  28. data/core/site-packages/pre_build/KafNafParser/entity_data.pyc +0 -0
  29. data/core/site-packages/pre_build/KafNafParser/external_references_data.py +41 -0
  30. data/core/site-packages/pre_build/KafNafParser/external_references_data.pyc +0 -0
  31. data/core/site-packages/pre_build/KafNafParser/feature_extractor/__init__.py +2 -0
  32. data/core/site-packages/pre_build/KafNafParser/feature_extractor/__init__.pyc +0 -0
  33. data/core/site-packages/pre_build/KafNafParser/feature_extractor/constituency.py +205 -0
  34. data/core/site-packages/pre_build/KafNafParser/feature_extractor/constituency.pyc +0 -0
  35. data/core/site-packages/pre_build/KafNafParser/feature_extractor/dependency.py +300 -0
  36. data/core/site-packages/pre_build/KafNafParser/feature_extractor/dependency.pyc +0 -0
  37. data/core/site-packages/pre_build/KafNafParser/features_data.py +71 -0
  38. data/core/site-packages/pre_build/KafNafParser/features_data.pyc +0 -0
  39. data/core/site-packages/pre_build/KafNafParser/header_data.py +127 -0
  40. data/core/site-packages/pre_build/KafNafParser/header_data.pyc +0 -0
  41. data/core/site-packages/pre_build/KafNafParser/opinion_data.py +200 -0
  42. data/core/site-packages/pre_build/KafNafParser/opinion_data.pyc +0 -0
  43. data/core/site-packages/pre_build/KafNafParser/references_data.py +15 -0
  44. data/core/site-packages/pre_build/KafNafParser/references_data.pyc +0 -0
  45. data/core/site-packages/pre_build/KafNafParser/span_data.py +63 -0
  46. data/core/site-packages/pre_build/KafNafParser/span_data.pyc +0 -0
  47. data/core/site-packages/pre_build/KafNafParser/term_data.py +111 -0
  48. data/core/site-packages/pre_build/KafNafParser/term_data.pyc +0 -0
  49. data/core/site-packages/pre_build/KafNafParser/term_sentiment_data.py +42 -0
  50. data/core/site-packages/pre_build/KafNafParser/term_sentiment_data.pyc +0 -0
  51. data/core/site-packages/pre_build/KafNafParser/text_data.py +90 -0
  52. data/core/site-packages/pre_build/KafNafParser/text_data.pyc +0 -0
  53. data/core/site-packages/pre_build/VUA_pylib-1.3-py2.7.egg-info/PKG-INFO +10 -0
  54. data/core/site-packages/pre_build/VUA_pylib-1.3-py2.7.egg-info/SOURCES.txt +14 -0
  55. data/core/site-packages/pre_build/VUA_pylib-1.3-py2.7.egg-info/dependency_links.txt +1 -0
  56. data/core/site-packages/pre_build/VUA_pylib-1.3-py2.7.egg-info/installed-files.txt +23 -0
  57. data/core/site-packages/pre_build/VUA_pylib-1.3-py2.7.egg-info/top_level.txt +1 -0
  58. data/core/site-packages/pre_build/VUA_pylib/__init__.py +1 -0
  59. data/core/site-packages/pre_build/VUA_pylib/__init__.pyc +0 -0
  60. data/core/site-packages/pre_build/VUA_pylib/common/__init__.py +1 -0
  61. data/core/site-packages/pre_build/VUA_pylib/common/__init__.pyc +0 -0
  62. data/core/site-packages/pre_build/VUA_pylib/common/common.py +28 -0
  63. data/core/site-packages/pre_build/VUA_pylib/common/common.pyc +0 -0
  64. data/core/site-packages/pre_build/VUA_pylib/corpus_reader/__init__.py +1 -0
  65. data/core/site-packages/pre_build/VUA_pylib/corpus_reader/__init__.pyc +0 -0
  66. data/core/site-packages/pre_build/VUA_pylib/corpus_reader/google_web_nl.py +156 -0
  67. data/core/site-packages/pre_build/VUA_pylib/corpus_reader/google_web_nl.pyc +0 -0
  68. data/core/site-packages/pre_build/VUA_pylib/io_utils/__init__.py +1 -0
  69. data/core/site-packages/pre_build/VUA_pylib/io_utils/__init__.pyc +0 -0
  70. data/core/site-packages/pre_build/VUA_pylib/io_utils/feature_file.py +121 -0
  71. data/core/site-packages/pre_build/VUA_pylib/io_utils/feature_file.pyc +0 -0
  72. data/core/site-packages/pre_build/VUA_pylib/lexicon/__init__.py +1 -0
  73. data/core/site-packages/pre_build/VUA_pylib/lexicon/__init__.pyc +0 -0
  74. data/core/site-packages/pre_build/VUA_pylib/lexicon/lexicon.py +72 -0
  75. data/core/site-packages/pre_build/VUA_pylib/lexicon/lexicon.pyc +0 -0
  76. data/core/site-packages/pre_build/VUKafParserPy-1.0-py2.7.egg-info/PKG-INFO +10 -0
  77. data/core/site-packages/pre_build/VUKafParserPy-1.0-py2.7.egg-info/SOURCES.txt +7 -0
  78. data/core/site-packages/pre_build/VUKafParserPy-1.0-py2.7.egg-info/dependency_links.txt +1 -0
  79. data/core/site-packages/pre_build/VUKafParserPy-1.0-py2.7.egg-info/installed-files.txt +11 -0
  80. data/core/site-packages/pre_build/VUKafParserPy-1.0-py2.7.egg-info/top_level.txt +1 -0
  81. data/core/site-packages/pre_build/VUKafParserPy/KafDataObjectsMod.py +165 -0
  82. data/core/site-packages/pre_build/VUKafParserPy/KafDataObjectsMod.pyc +0 -0
  83. data/core/site-packages/pre_build/VUKafParserPy/KafParserMod.py +439 -0
  84. data/core/site-packages/pre_build/VUKafParserPy/KafParserMod.pyc +0 -0
  85. data/core/site-packages/pre_build/VUKafParserPy/__init__.py +7 -0
  86. data/core/site-packages/pre_build/VUKafParserPy/__init__.pyc +0 -0
  87. data/ext/hack/support.rb +38 -0
  88. data/lib/opener/kaf_naf_parser.rb +77 -0
  89. data/lib/opener/kaf_naf_parser/cli.rb +92 -0
  90. data/lib/opener/kaf_naf_parser/public/markdown.css +284 -0
  91. data/lib/opener/kaf_naf_parser/server.rb +16 -0
  92. data/lib/opener/kaf_naf_parser/version.rb +5 -0
  93. data/lib/opener/kaf_naf_parser/views/index.erb +103 -0
  94. data/lib/opener/kaf_naf_parser/views/result.erb +15 -0
  95. data/opener-kaf-naf-parser.gemspec +38 -0
  96. data/pre_build_requirements.txt +3 -0
  97. metadata +283 -0
@@ -0,0 +1,7 @@
1
+ ## version = 0.2
2
+ ## Added timestamp to function addLinguisitcProcessor
3
+ ## 24-april-2013 --> getSingleEntieies and getSingleProperties reads both entities/props in format
4
+ ## entities -> entity -> span -> target and entities -> entity -> references -> span
5
+ ####
6
+
7
+ from KafParserMod import KafParser
@@ -0,0 +1,38 @@
1
+ require 'opener/build-tools'
2
+
3
+ include Opener::BuildTools::Requirements
4
+ include Opener::BuildTools::Python
5
+ include Opener::BuildTools::Files
6
+
7
+ # Directory where packages will be installed to.
8
+ PYTHON_SITE_PACKAGES = File.expand_path(
9
+ '../../../core/site-packages',
10
+ __FILE__
11
+ )
12
+
13
+ # Directory containing the temporary files.
14
+ TMP_DIRECTORY = File.expand_path('../../../tmp', __FILE__)
15
+
16
+ # Path to the pip requirements file used to install requirements before
17
+ # packaging the Gem.
18
+ PRE_BUILD_REQUIREMENTS = File.expand_path(
19
+ '../../../pre_build_requirements.txt',
20
+ __FILE__
21
+ )
22
+
23
+ # Path to the pip requirements file used to install requirements upon Gem
24
+ # installation.
25
+ PRE_INSTALL_REQUIREMENTS = File.expand_path(
26
+ '../../../pre_install_requirements.txt',
27
+ __FILE__
28
+ )
29
+
30
+ ##
31
+ # Verifies the requirements to install thi Gem.
32
+ #
33
+ def verify_requirements
34
+ require_executable('python')
35
+ require_version('python', python_version, '2.7.0')
36
+ require_executable('pip')
37
+ require_version('pip', pip_version, '1.3.1')
38
+ end
@@ -0,0 +1,77 @@
1
+ require 'open3'
2
+ require 'optparse'
3
+
4
+ require_relative 'kaf_naf_parser/version'
5
+ require_relative 'kaf_naf_parser/cli'
6
+
7
+ module Opener
8
+ ##
9
+ # Ruby wrapper around the Python based KafNafParser.
10
+ #
11
+ # @!attribute [r] options
12
+ # @return [Hash]
13
+ #
14
+ class KafNafParser
15
+ attr_reader :options
16
+
17
+ ##
18
+ # Hash containing the default options to use.
19
+ #
20
+ # @return [Hash]
21
+ #
22
+ DEFAULT_OPTIONS = {
23
+ :args => [],
24
+ :conversion => "to-kaf"
25
+ }.freeze
26
+
27
+ ##
28
+ # @param [Hash] options
29
+ #
30
+ # @option options [Array] :args Collection of arbitrary arguments to pass
31
+ # to the underlying kernel.
32
+ #
33
+ def initialize(options = {})
34
+ @options = DEFAULT_OPTIONS.merge(options)
35
+ end
36
+
37
+ ##
38
+ # Returns a String containing the command to use for executing the kernel.
39
+ #
40
+ # @return [String]
41
+ #
42
+ def command
43
+ return "python -E -OO #{kernel} #{options[:args].join(' ')} #{conversion}"
44
+ end
45
+
46
+ def conversion
47
+ "--#{options[:conversion].gsub(/-/,'')}"
48
+ end
49
+
50
+ ##
51
+ # Processes the input and returns an Array containing the output of STDOUT,
52
+ # STDERR and an object containing process information.
53
+ #
54
+ # @param [String] input The text of which to detect the language.
55
+ # @return [Array]
56
+ #
57
+ def run(input)
58
+ return Open3.capture3(command, :stdin_data => input)
59
+ end
60
+
61
+ protected
62
+
63
+ ##
64
+ # @return [String]
65
+ #
66
+ def core_dir
67
+ return File.expand_path('../../../core', __FILE__)
68
+ end
69
+
70
+ ##
71
+ # @return [String]
72
+ #
73
+ def kernel
74
+ return File.join(core_dir, 'kaf-naf-parser.py')
75
+ end
76
+ end # KafNafParser
77
+ end # Opener
@@ -0,0 +1,92 @@
1
+ module Opener
2
+ class KafNafParser
3
+ ##
4
+ # CLI wrapper around {Opener::LanguageIdentifier} using OptionParser.
5
+ #
6
+ # @!attribute [r] options
7
+ # @return [Hash]
8
+ # @!attribute [r] option_parser
9
+ # @return [OptionParser]
10
+ #
11
+ class CLI
12
+ attr_reader :options, :option_parser
13
+
14
+ ##
15
+ # @param [Hash] options
16
+ #
17
+ def initialize(options = {})
18
+ @options = DEFAULT_OPTIONS.merge(options)
19
+
20
+ @option_parser = OptionParser.new do |opts|
21
+ opts.program_name = 'kaf-naf-parser'
22
+ opts.summary_indent = ' '
23
+
24
+ opts.on('-h', '--help', 'Shows this help message') do
25
+ show_help
26
+ end
27
+
28
+ opts.on('-v', '--version', 'Shows the current version') do
29
+ show_version
30
+ end
31
+
32
+ opts.on('-k', '--tokaf', 'Parses input to KAF') do
33
+ @options[:conversion] = "to-kaf"
34
+ end
35
+
36
+ opts.on('-n', '--tonaf', 'Parses input to NAF') do
37
+ @options[:conversion] = "to-naf"
38
+ end
39
+
40
+ opts.on('-l', '--log', 'Enable logging to STDERR') do
41
+ @options[:logging] = true
42
+ end
43
+
44
+ opts.separator <<-EOF
45
+
46
+ Examples:
47
+
48
+ cat example.kaf | #{opts.program_name} --tonaf # Basic usage
49
+ cat example.kaf | #{opts.program_name} --tonaf -l # Logs information to STDERR
50
+ EOF
51
+ end
52
+ end
53
+
54
+ ##
55
+ # @param [String] input
56
+ #
57
+ def run(input)
58
+ option_parser.parse!(options[:args])
59
+
60
+ tagger = KafNafParser.new(options)
61
+
62
+ stdout, stderr, process = tagger.run(input)
63
+
64
+ if process.success?
65
+ puts stdout
66
+
67
+ if options[:logging] and !stderr.empty?
68
+ STDERR.puts(stderr)
69
+ end
70
+ else
71
+ abort stderr
72
+ end
73
+ end
74
+
75
+ private
76
+
77
+ ##
78
+ # Shows the help message and exits the program.
79
+ #
80
+ def show_help
81
+ abort option_parser.to_s
82
+ end
83
+
84
+ ##
85
+ # Shows the version and exits the program.
86
+ #
87
+ def show_version
88
+ abort "#{option_parser.program_name} v#{VERSION} on #{RUBY_DESCRIPTION}"
89
+ end
90
+ end # CLI
91
+ end # KafNafParser
92
+ end # Opener
@@ -0,0 +1,284 @@
1
+
2
+ input[type="text"], textarea
3
+ {
4
+ width: 500px;
5
+ }
6
+
7
+ body {
8
+ font-family: Helvetica, arial, sans-serif;
9
+ font-size: 14px;
10
+ line-height: 1.6;
11
+ padding-top: 10px;
12
+ padding-bottom: 10px;
13
+ background-color: white;
14
+ padding: 30px; }
15
+
16
+ body > *:first-child {
17
+ margin-top: 0 !important; }
18
+ body > *:last-child {
19
+ margin-bottom: 0 !important; }
20
+
21
+ a {
22
+ color: #4183C4; }
23
+ a.absent {
24
+ color: #cc0000; }
25
+ a.anchor {
26
+ display: block;
27
+ padding-left: 30px;
28
+ margin-left: -30px;
29
+ cursor: pointer;
30
+ position: absolute;
31
+ top: 0;
32
+ left: 0;
33
+ bottom: 0; }
34
+
35
+ h1, h2, h3, h4, h5, h6 {
36
+ margin: 20px 0 10px;
37
+ padding: 0;
38
+ font-weight: bold;
39
+ -webkit-font-smoothing: antialiased;
40
+ cursor: text;
41
+ position: relative; }
42
+
43
+ h1:hover a.anchor, h2:hover a.anchor, h3:hover a.anchor, h4:hover a.anchor, h5:hover a.anchor, h6:hover a.anchor {
44
+ background: url("../../images/modules/styleguide/para.png") no-repeat 10px center;
45
+ text-decoration: none; }
46
+
47
+ h1 tt, h1 code {
48
+ font-size: inherit; }
49
+
50
+ h2 tt, h2 code {
51
+ font-size: inherit; }
52
+
53
+ h3 tt, h3 code {
54
+ font-size: inherit; }
55
+
56
+ h4 tt, h4 code {
57
+ font-size: inherit; }
58
+
59
+ h5 tt, h5 code {
60
+ font-size: inherit; }
61
+
62
+ h6 tt, h6 code {
63
+ font-size: inherit; }
64
+
65
+ h1 {
66
+ font-size: 28px;
67
+ color: black; }
68
+
69
+ h2 {
70
+ font-size: 24px;
71
+ border-bottom: 1px solid #cccccc;
72
+ color: black; }
73
+
74
+ h3 {
75
+ font-size: 18px; }
76
+
77
+ h4 {
78
+ font-size: 16px; }
79
+
80
+ h5 {
81
+ font-size: 14px; }
82
+
83
+ h6 {
84
+ color: #777777;
85
+ font-size: 14px; }
86
+
87
+ p, blockquote, ul, ol, dl, li, table, pre {
88
+ margin: 15px 0; }
89
+
90
+ hr {
91
+ background: transparent url("../../images/modules/pulls/dirty-shade.png") repeat-x 0 0;
92
+ border: 0 none;
93
+ color: #cccccc;
94
+ height: 4px;
95
+ padding: 0; }
96
+
97
+ body > h2:first-child {
98
+ margin-top: 0;
99
+ padding-top: 0; }
100
+ body > h1:first-child {
101
+ margin-top: 0;
102
+ padding-top: 0; }
103
+ body > h1:first-child + h2 {
104
+ margin-top: 0;
105
+ padding-top: 0; }
106
+ body > h3:first-child, body > h4:first-child, body > h5:first-child, body > h6:first-child {
107
+ margin-top: 0;
108
+ padding-top: 0; }
109
+
110
+ a:first-child h1, a:first-child h2, a:first-child h3, a:first-child h4, a:first-child h5, a:first-child h6 {
111
+ margin-top: 0;
112
+ padding-top: 0; }
113
+
114
+ h1 p, h2 p, h3 p, h4 p, h5 p, h6 p {
115
+ margin-top: 0; }
116
+
117
+ li p.first {
118
+ display: inline-block; }
119
+
120
+ ul, ol {
121
+ padding-left: 30px; }
122
+
123
+ ul :first-child, ol :first-child {
124
+ margin-top: 0; }
125
+
126
+ ul :last-child, ol :last-child {
127
+ margin-bottom: 0; }
128
+
129
+ dl {
130
+ padding: 0; }
131
+ dl dt {
132
+ font-size: 14px;
133
+ font-weight: bold;
134
+ font-style: italic;
135
+ padding: 0;
136
+ margin: 15px 0 5px; }
137
+ dl dt:first-child {
138
+ padding: 0; }
139
+ dl dt > :first-child {
140
+ margin-top: 0; }
141
+ dl dt > :last-child {
142
+ margin-bottom: 0; }
143
+ dl dd {
144
+ margin: 0 0 15px;
145
+ padding: 0 15px; }
146
+ dl dd > :first-child {
147
+ margin-top: 0; }
148
+ dl dd > :last-child {
149
+ margin-bottom: 0; }
150
+
151
+ blockquote {
152
+ border-left: 4px solid #dddddd;
153
+ padding: 0 15px;
154
+ color: #777777; }
155
+ blockquote > :first-child {
156
+ margin-top: 0; }
157
+ blockquote > :last-child {
158
+ margin-bottom: 0; }
159
+
160
+ table {
161
+ padding: 0; }
162
+ table tr {
163
+ border-top: 1px solid #cccccc;
164
+ background-color: white;
165
+ margin: 0;
166
+ padding: 0; }
167
+ table tr:nth-child(2n) {
168
+ background-color: #f8f8f8; }
169
+ table tr th {
170
+ font-weight: bold;
171
+ border: 1px solid #cccccc;
172
+ text-align: left;
173
+ margin: 0;
174
+ padding: 6px 13px; }
175
+ table tr td {
176
+ border: 1px solid #cccccc;
177
+ text-align: left;
178
+ margin: 0;
179
+ padding: 6px 13px; }
180
+ table tr th :first-child, table tr td :first-child {
181
+ margin-top: 0; }
182
+ table tr th :last-child, table tr td :last-child {
183
+ margin-bottom: 0; }
184
+
185
+ img {
186
+ max-width: 100%; }
187
+
188
+ span.frame {
189
+ display: block;
190
+ overflow: hidden; }
191
+ span.frame > span {
192
+ border: 1px solid #dddddd;
193
+ display: block;
194
+ float: left;
195
+ overflow: hidden;
196
+ margin: 13px 0 0;
197
+ padding: 7px;
198
+ width: auto; }
199
+ span.frame span img {
200
+ display: block;
201
+ float: left; }
202
+ span.frame span span {
203
+ clear: both;
204
+ color: #333333;
205
+ display: block;
206
+ padding: 5px 0 0; }
207
+ span.align-center {
208
+ display: block;
209
+ overflow: hidden;
210
+ clear: both; }
211
+ span.align-center > span {
212
+ display: block;
213
+ overflow: hidden;
214
+ margin: 13px auto 0;
215
+ text-align: center; }
216
+ span.align-center span img {
217
+ margin: 0 auto;
218
+ text-align: center; }
219
+ span.align-right {
220
+ display: block;
221
+ overflow: hidden;
222
+ clear: both; }
223
+ span.align-right > span {
224
+ display: block;
225
+ overflow: hidden;
226
+ margin: 13px 0 0;
227
+ text-align: right; }
228
+ span.align-right span img {
229
+ margin: 0;
230
+ text-align: right; }
231
+ span.float-left {
232
+ display: block;
233
+ margin-right: 13px;
234
+ overflow: hidden;
235
+ float: left; }
236
+ span.float-left span {
237
+ margin: 13px 0 0; }
238
+ span.float-right {
239
+ display: block;
240
+ margin-left: 13px;
241
+ overflow: hidden;
242
+ float: right; }
243
+ span.float-right > span {
244
+ display: block;
245
+ overflow: hidden;
246
+ margin: 13px auto 0;
247
+ text-align: right; }
248
+
249
+ code, tt {
250
+ margin: 0 2px;
251
+ padding: 0 5px;
252
+ white-space: nowrap;
253
+ border: 1px solid #eaeaea;
254
+ background-color: #f8f8f8;
255
+ border-radius: 3px; }
256
+
257
+ pre code {
258
+ margin: 0;
259
+ padding: 0;
260
+ white-space: pre;
261
+ border: none;
262
+ background: transparent; }
263
+
264
+ .highlight pre {
265
+ background-color: #f8f8f8;
266
+ border: 1px solid #cccccc;
267
+ font-size: 13px;
268
+ line-height: 19px;
269
+ overflow: auto;
270
+ padding: 6px 10px;
271
+ border-radius: 3px; }
272
+
273
+ pre {
274
+ background-color: #f8f8f8;
275
+ border: 1px solid #cccccc;
276
+ font-size: 13px;
277
+ line-height: 19px;
278
+ overflow: auto;
279
+ padding: 6px 10px;
280
+ border-radius: 3px; }
281
+ pre code, pre tt {
282
+ background-color: transparent;
283
+ border: none; }
284
+