ocrsdk 0.1.0 → 0.1.1
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.
- data/.gitignore +3 -1
- data/README.md +80 -1
- data/VERSION +1 -1
- data/lib/ocrsdk/abstract_entity.rb +2 -5
- data/lib/ocrsdk/image.rb +4 -4
- data/lib/ocrsdk/promise.rb +4 -4
- data/lib/ocrsdk/verifiers/format.rb +2 -2
- data/lib/ocrsdk/verifiers/language.rb +1 -1
- data/lib/ocrsdk/verifiers/profile.rb +1 -1
- data/lib/ocrsdk/verifiers/status.rb +1 -1
- data/lib/ocrsdk.rb +7 -0
- data/spec/abstract_entity_spec.rb +19 -0
- data/spec/ocrsdk/image_spec.rb +8 -1
- data/spec/ocrsdk/pdf_spec.rb +7 -0
- data/spec/ocrsdk/promise_spec.rb +7 -1
- data/spec/ocrsdk_spec.rb +10 -0
- metadata +4 -19
- data/coverage/index.css +0 -56
- data/coverage/index.html +0 -244
- data/coverage/jquery.js +0 -154
- data/coverage/jquery.tablesorter.js +0 -2
- data/coverage/lib/ocrsdk/abstract_entity.rb.html +0 -207
- data/coverage/lib/ocrsdk/document.rb.html +0 -159
- data/coverage/lib/ocrsdk/errors.rb.html +0 -199
- data/coverage/lib/ocrsdk/image.rb.html +0 -543
- data/coverage/lib/ocrsdk/pdf.rb.html +0 -311
- data/coverage/lib/ocrsdk/promise.rb.html +0 -791
- data/coverage/lib/ocrsdk/verifiers/format.rb.html +0 -303
- data/coverage/lib/ocrsdk/verifiers/language.rb.html +0 -439
- data/coverage/lib/ocrsdk/verifiers/profile.rb.html +0 -215
- data/coverage/lib/ocrsdk/verifiers/status.rb.html +0 -247
- data/coverage/lib/ocrsdk/verifiers.rb.html +0 -167
- data/coverage/lib/ocrsdk.rb.html +0 -263
- data/coverage/report.css +0 -97
|
@@ -1,311 +0,0 @@
|
|
|
1
|
-
<html>
|
|
2
|
-
<head>
|
|
3
|
-
<title>lib/ocrsdk/pdf.rb</title>
|
|
4
|
-
|
|
5
|
-
<link href="../../report.css?0.4158445434855237" media="screen" rel="stylesheet" type="text/css" />
|
|
6
|
-
<script src='../../jquery.js?0.4247975075509465'></script>
|
|
7
|
-
<script>
|
|
8
|
-
$(function() {
|
|
9
|
-
|
|
10
|
-
$('#toggle_test_file').click(function(e) {
|
|
11
|
-
e.preventDefault();
|
|
12
|
-
if ($('#main_file').hasClass('side_by_side')) {
|
|
13
|
-
$('#main_file').removeClass('side_by_side');
|
|
14
|
-
} else {
|
|
15
|
-
$('#main_file').addClass('side_by_side');
|
|
16
|
-
}
|
|
17
|
-
|
|
18
|
-
$('#test_file').toggle();
|
|
19
|
-
});
|
|
20
|
-
|
|
21
|
-
$('#test_file').hide();
|
|
22
|
-
|
|
23
|
-
function cleanString(str) {
|
|
24
|
-
return $.trim(str.replace(/<[^(<|>)]+>/gi, ''));
|
|
25
|
-
}
|
|
26
|
-
|
|
27
|
-
$('.line_number').hover(function() {
|
|
28
|
-
$(this).attr('title', 'Line: ' + cleanString($(this).html()));
|
|
29
|
-
});
|
|
30
|
-
|
|
31
|
-
$('.hit_number').hover(function() {
|
|
32
|
-
$(this).attr('title', 'Execution Count: ' + cleanString($(this).html()));
|
|
33
|
-
});
|
|
34
|
-
|
|
35
|
-
$('.code_line').hover(function() {
|
|
36
|
-
$(this).attr('title', cleanString($(this).html()));
|
|
37
|
-
});
|
|
38
|
-
|
|
39
|
-
});
|
|
40
|
-
</script>
|
|
41
|
-
</head>
|
|
42
|
-
<body>
|
|
43
|
-
<h1><a href="../../index.html">Coverage Report</a></h1>
|
|
44
|
-
|
|
45
|
-
<table cellpadding='0' cellspacing='1'>
|
|
46
|
-
<tr class='header'>
|
|
47
|
-
<th>File</th>
|
|
48
|
-
<th>Lines</th>
|
|
49
|
-
<th>Lines Of Code</th>
|
|
50
|
-
<th>Untested Lines of Code</th>
|
|
51
|
-
<th>Tested %</th>
|
|
52
|
-
</tr>
|
|
53
|
-
<tr valign='top'>
|
|
54
|
-
<td>
|
|
55
|
-
lib/ocrsdk/pdf.rb
|
|
56
|
-
|
|
57
|
-
</td>
|
|
58
|
-
<td>27</td>
|
|
59
|
-
<td>12</td>
|
|
60
|
-
<td>0</td>
|
|
61
|
-
<td>100.0%</td>
|
|
62
|
-
</tr>
|
|
63
|
-
</table>
|
|
64
|
-
|
|
65
|
-
<div> </div>
|
|
66
|
-
|
|
67
|
-
<table cellpadding='0' cellspacing='1'>
|
|
68
|
-
<tr>
|
|
69
|
-
<th>Legend</th>
|
|
70
|
-
</tr>
|
|
71
|
-
<tr>
|
|
72
|
-
<td class='hit'>This line was executed.</td>
|
|
73
|
-
</tr>
|
|
74
|
-
<tr>
|
|
75
|
-
<td class='miss'>This line was not executed!</td>
|
|
76
|
-
</tr>
|
|
77
|
-
<tr>
|
|
78
|
-
<td class='never'>This line doesn't matter.</td>
|
|
79
|
-
</tr>
|
|
80
|
-
</table>
|
|
81
|
-
|
|
82
|
-
<div> </div>
|
|
83
|
-
|
|
84
|
-
<div id='content'>
|
|
85
|
-
<div id='main_file'>
|
|
86
|
-
<table id='main'>
|
|
87
|
-
|
|
88
|
-
<tr class="hit" data-hits='#{cov}'"">
|
|
89
|
-
<td class='line_number'>1</td>
|
|
90
|
-
<td class='code_line'>
|
|
91
|
-
<pre>class OCRSDK::PDF < OCRSDK::Image</pre>
|
|
92
|
-
</td>
|
|
93
|
-
<td class='hit_number'>23</td>
|
|
94
|
-
</tr>
|
|
95
|
-
|
|
96
|
-
<tr class="never" "">
|
|
97
|
-
<td class='line_number'>2</td>
|
|
98
|
-
<td class='code_line'>
|
|
99
|
-
<pre> # We're on a shaky ground regarding what kind of pdfs</pre>
|
|
100
|
-
</td>
|
|
101
|
-
<td class='hit_number'></td>
|
|
102
|
-
</tr>
|
|
103
|
-
|
|
104
|
-
<tr class="never" "">
|
|
105
|
-
<td class='line_number'>3</td>
|
|
106
|
-
<td class='code_line'>
|
|
107
|
-
<pre> # should be recognized and what shouldn't.</pre>
|
|
108
|
-
</td>
|
|
109
|
-
<td class='hit_number'></td>
|
|
110
|
-
</tr>
|
|
111
|
-
|
|
112
|
-
<tr class="never" "">
|
|
113
|
-
<td class='line_number'>4</td>
|
|
114
|
-
<td class='code_line'>
|
|
115
|
-
<pre> # Currently we count that if there are</pre>
|
|
116
|
-
</td>
|
|
117
|
-
<td class='hit_number'></td>
|
|
118
|
-
</tr>
|
|
119
|
-
|
|
120
|
-
<tr class="never" "">
|
|
121
|
-
<td class='line_number'>5</td>
|
|
122
|
-
<td class='code_line'>
|
|
123
|
-
<pre> # images * 20 > length of text</pre>
|
|
124
|
-
</td>
|
|
125
|
-
<td class='hit_number'></td>
|
|
126
|
-
</tr>
|
|
127
|
-
|
|
128
|
-
<tr class="never" "">
|
|
129
|
-
<td class='line_number'>6</td>
|
|
130
|
-
<td class='code_line'>
|
|
131
|
-
<pre> # then this document might need recognition.</pre>
|
|
132
|
-
</td>
|
|
133
|
-
<td class='hit_number'></td>
|
|
134
|
-
</tr>
|
|
135
|
-
|
|
136
|
-
<tr class="never" "">
|
|
137
|
-
<td class='line_number'>7</td>
|
|
138
|
-
<td class='code_line'>
|
|
139
|
-
<pre> # Assumption is that there might be a title,</pre>
|
|
140
|
-
</td>
|
|
141
|
-
<td class='hit_number'></td>
|
|
142
|
-
</tr>
|
|
143
|
-
|
|
144
|
-
<tr class="never" "">
|
|
145
|
-
<td class='line_number'>8</td>
|
|
146
|
-
<td class='code_line'>
|
|
147
|
-
<pre> # page numbers or credits along with images.</pre>
|
|
148
|
-
</td>
|
|
149
|
-
<td class='hit_number'></td>
|
|
150
|
-
</tr>
|
|
151
|
-
|
|
152
|
-
<tr class="hit" data-hits='#{cov}'"">
|
|
153
|
-
<td class='line_number'>9</td>
|
|
154
|
-
<td class='code_line'>
|
|
155
|
-
<pre> def recognizeable?</pre>
|
|
156
|
-
</td>
|
|
157
|
-
<td class='hit_number'>23</td>
|
|
158
|
-
</tr>
|
|
159
|
-
|
|
160
|
-
<tr class="hit" data-hits='#{cov}'"">
|
|
161
|
-
<td class='line_number'>10</td>
|
|
162
|
-
<td class='code_line'>
|
|
163
|
-
<pre> reader = PDF::Reader.new @image_path</pre>
|
|
164
|
-
</td>
|
|
165
|
-
<td class='hit_number'>20</td>
|
|
166
|
-
</tr>
|
|
167
|
-
|
|
168
|
-
<tr class="never" "">
|
|
169
|
-
<td class='line_number'>11</td>
|
|
170
|
-
<td class='code_line'>
|
|
171
|
-
<pre></pre>
|
|
172
|
-
</td>
|
|
173
|
-
<td class='hit_number'></td>
|
|
174
|
-
</tr>
|
|
175
|
-
|
|
176
|
-
<tr class="hit" data-hits='#{cov}'"">
|
|
177
|
-
<td class='line_number'>12</td>
|
|
178
|
-
<td class='code_line'>
|
|
179
|
-
<pre> images = 0</pre>
|
|
180
|
-
</td>
|
|
181
|
-
<td class='hit_number'>12</td>
|
|
182
|
-
</tr>
|
|
183
|
-
|
|
184
|
-
<tr class="hit" data-hits='#{cov}'"">
|
|
185
|
-
<td class='line_number'>13</td>
|
|
186
|
-
<td class='code_line'>
|
|
187
|
-
<pre> text = 0</pre>
|
|
188
|
-
</td>
|
|
189
|
-
<td class='hit_number'>12</td>
|
|
190
|
-
</tr>
|
|
191
|
-
|
|
192
|
-
<tr class="hit" data-hits='#{cov}'"">
|
|
193
|
-
<td class='line_number'>14</td>
|
|
194
|
-
<td class='code_line'>
|
|
195
|
-
<pre> chars = Set.new</pre>
|
|
196
|
-
</td>
|
|
197
|
-
<td class='hit_number'>12</td>
|
|
198
|
-
</tr>
|
|
199
|
-
|
|
200
|
-
<tr class="hit" data-hits='#{cov}'"">
|
|
201
|
-
<td class='line_number'>15</td>
|
|
202
|
-
<td class='code_line'>
|
|
203
|
-
<pre> reader.pages.each do |page|</pre>
|
|
204
|
-
</td>
|
|
205
|
-
<td class='hit_number'>12</td>
|
|
206
|
-
</tr>
|
|
207
|
-
|
|
208
|
-
<tr class="hit" data-hits='#{cov}'"">
|
|
209
|
-
<td class='line_number'>16</td>
|
|
210
|
-
<td class='code_line'>
|
|
211
|
-
<pre> text += page.text.length</pre>
|
|
212
|
-
</td>
|
|
213
|
-
<td class='hit_number'>51</td>
|
|
214
|
-
</tr>
|
|
215
|
-
|
|
216
|
-
<tr class="hit" data-hits='#{cov}'"">
|
|
217
|
-
<td class='line_number'>17</td>
|
|
218
|
-
<td class='code_line'>
|
|
219
|
-
<pre> chars += page.text.split('').map(&:ord).uniq</pre>
|
|
220
|
-
</td>
|
|
221
|
-
<td class='hit_number'>51</td>
|
|
222
|
-
</tr>
|
|
223
|
-
|
|
224
|
-
<tr class="hit" data-hits='#{cov}'"">
|
|
225
|
-
<td class='line_number'>18</td>
|
|
226
|
-
<td class='code_line'>
|
|
227
|
-
<pre> images += page.xobjects.map {|k, v| v.hash[:Subtype]}.count(:Image)</pre>
|
|
228
|
-
</td>
|
|
229
|
-
<td class='hit_number'>99</td>
|
|
230
|
-
</tr>
|
|
231
|
-
|
|
232
|
-
<tr class="never" "">
|
|
233
|
-
<td class='line_number'>19</td>
|
|
234
|
-
<td class='code_line'>
|
|
235
|
-
<pre> end</pre>
|
|
236
|
-
</td>
|
|
237
|
-
<td class='hit_number'></td>
|
|
238
|
-
</tr>
|
|
239
|
-
|
|
240
|
-
<tr class="never" "">
|
|
241
|
-
<td class='line_number'>20</td>
|
|
242
|
-
<td class='code_line'>
|
|
243
|
-
<pre></pre>
|
|
244
|
-
</td>
|
|
245
|
-
<td class='hit_number'></td>
|
|
246
|
-
</tr>
|
|
247
|
-
|
|
248
|
-
<tr class="never" "">
|
|
249
|
-
<td class='line_number'>21</td>
|
|
250
|
-
<td class='code_line'>
|
|
251
|
-
<pre> # count number of distinct characters</pre>
|
|
252
|
-
</td>
|
|
253
|
-
<td class='hit_number'></td>
|
|
254
|
-
</tr>
|
|
255
|
-
|
|
256
|
-
<tr class="never" "">
|
|
257
|
-
<td class='line_number'>22</td>
|
|
258
|
-
<td class='code_line'>
|
|
259
|
-
<pre> # in case of "searchable", but incorrectly recognized document</pre>
|
|
260
|
-
</td>
|
|
261
|
-
<td class='hit_number'></td>
|
|
262
|
-
</tr>
|
|
263
|
-
|
|
264
|
-
<tr class="hit" data-hits='#{cov}'"">
|
|
265
|
-
<td class='line_number'>23</td>
|
|
266
|
-
<td class='code_line'>
|
|
267
|
-
<pre> images * 20 > text || chars.length < 10</pre>
|
|
268
|
-
</td>
|
|
269
|
-
<td class='hit_number'>12</td>
|
|
270
|
-
</tr>
|
|
271
|
-
|
|
272
|
-
<tr class="never" "">
|
|
273
|
-
<td class='line_number'>24</td>
|
|
274
|
-
<td class='code_line'>
|
|
275
|
-
<pre> rescue PDF::Reader::MalformedPDFError, PDF::Reader::UnsupportedFeatureError</pre>
|
|
276
|
-
</td>
|
|
277
|
-
<td class='hit_number'></td>
|
|
278
|
-
</tr>
|
|
279
|
-
|
|
280
|
-
<tr class="hit" data-hits='#{cov}'"">
|
|
281
|
-
<td class='line_number'>25</td>
|
|
282
|
-
<td class='code_line'>
|
|
283
|
-
<pre> false</pre>
|
|
284
|
-
</td>
|
|
285
|
-
<td class='hit_number'>3</td>
|
|
286
|
-
</tr>
|
|
287
|
-
|
|
288
|
-
<tr class="never" "">
|
|
289
|
-
<td class='line_number'>26</td>
|
|
290
|
-
<td class='code_line'>
|
|
291
|
-
<pre> end</pre>
|
|
292
|
-
</td>
|
|
293
|
-
<td class='hit_number'></td>
|
|
294
|
-
</tr>
|
|
295
|
-
|
|
296
|
-
<tr class="never" "">
|
|
297
|
-
<td class='line_number'>27</td>
|
|
298
|
-
<td class='code_line'>
|
|
299
|
-
<pre>end</pre>
|
|
300
|
-
</td>
|
|
301
|
-
<td class='hit_number'></td>
|
|
302
|
-
</tr>
|
|
303
|
-
|
|
304
|
-
</table>
|
|
305
|
-
</div>
|
|
306
|
-
|
|
307
|
-
</div>
|
|
308
|
-
|
|
309
|
-
<p>Generated on: 2012-12-01 01:40:30 +0400</p>
|
|
310
|
-
</body>
|
|
311
|
-
</html>
|