|
1
|
|
|
/* |
|
2
|
|
|
* Copyright 2014, Armenak Grigoryan, and individual contributors as indicated |
|
3
|
|
|
* by the @authors tag. See the copyright.txt in the distribution for a |
|
4
|
|
|
* full listing of individual contributors. |
|
5
|
|
|
* |
|
6
|
|
|
* This is free software; you can redistribute it and/or modify it |
|
7
|
|
|
* under the terms of the GNU Lesser General Public License as |
|
8
|
|
|
* published by the Free Software Foundation; either version 2.1 of |
|
9
|
|
|
* the License, or (at your option) any later version. |
|
10
|
|
|
* |
|
11
|
|
|
* This software is distributed in the hope that it will be useful, |
|
12
|
|
|
* but WITHOUT ANY WARRANTY; without even the implied warranty of |
|
13
|
|
|
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
|
14
|
|
|
* Lesser General Public License for more details. |
|
15
|
|
|
*/ |
|
16
|
|
|
package com.strider.datadefender.anonymizer.functions; |
|
17
|
|
|
|
|
18
|
|
|
import com.strider.datadefender.functions.NamedParameter; |
|
19
|
|
|
import com.strider.datadefender.requirement.registry.RequirementFunction; |
|
20
|
|
|
|
|
21
|
|
|
import java.io.InputStreamReader; |
|
22
|
|
|
import java.io.BufferedReader; |
|
23
|
|
|
import java.io.IOException; |
|
24
|
|
|
import java.util.ArrayList; |
|
25
|
|
|
import java.util.List; |
|
26
|
|
|
import java.util.Random; |
|
27
|
|
|
|
|
28
|
|
|
import org.apache.commons.lang3.StringUtils; |
|
29
|
|
|
|
|
30
|
|
|
import lombok.extern.log4j.Log4j2; |
|
31
|
|
|
|
|
32
|
|
|
/** |
|
33
|
|
|
* Helpers for generating text using Lorem Ipsum... |
|
34
|
|
|
* |
|
35
|
|
|
* @author Armenak Grigoryan |
|
36
|
|
|
*/ |
|
37
|
|
|
@Log4j2 |
|
38
|
|
|
public class Lipsum extends RequirementFunction { |
|
39
|
|
|
|
|
40
|
|
|
private Random rand = new Random(); |
|
41
|
|
|
private static final List<String> lipsumParagraphs = new ArrayList<>(); |
|
42
|
|
|
|
|
43
|
1 |
|
static { |
|
44
|
1 |
|
final BufferedReader br = new BufferedReader(new InputStreamReader(Lipsum.class.getResourceAsStream("lipsum.txt"))); |
|
45
|
1 |
|
final StringBuilder sb = new StringBuilder(); |
|
46
|
1 |
|
try { |
|
47
|
1 |
|
for (String line; (line = br.readLine()) != null; ) { |
|
48
|
598 |
|
if (line.trim().length() == 0) { |
|
49
|
149 |
|
lipsumParagraphs.add(sb.toString()); |
|
50
|
149 |
|
sb.setLength(0); |
|
51
|
149 |
|
continue; |
|
52
|
|
|
} |
|
53
|
150 |
|
sb.append(line); |
|
54
|
|
|
} |
|
55
|
|
|
} catch (IOException ex) { |
|
56
|
|
|
log.fatal("Error initializing lipsum.txt", ex); |
|
57
|
|
|
System.exit(1); |
|
58
|
|
|
} |
|
59
|
|
|
} |
|
60
|
|
|
|
|
61
|
|
|
/** |
|
62
|
|
|
* Generates between min and max (inclusive) lorem ipsum sentences. |
|
63
|
|
|
* |
|
64
|
|
|
* The sentences are generated from the beginning of paragraphs, although |
|
65
|
|
|
* the first paragraph chosen is random. Paragraphs are joined together to |
|
66
|
|
|
* form sentences without line breaks. |
|
67
|
|
|
* |
|
68
|
|
|
* @param min Minimum number of sentences to generate |
|
69
|
|
|
* @param max Maximum number of sentences to generate |
|
70
|
|
|
* @return the generated sentences. |
|
71
|
|
|
* @throws IOException if an error occurs reading from the lipsum text file |
|
72
|
|
|
*/ |
|
73
|
5 |
|
public String sentences( |
|
74
|
|
|
@NamedParameter("pattern") int min, |
|
75
|
|
|
@NamedParameter("pattern") int max |
|
76
|
|
|
) throws IOException { |
|
77
|
|
|
|
|
78
|
5 |
|
final List<String> lp = lipsumParagraphs; |
|
79
|
5 |
|
final StringBuilder sb = new StringBuilder(); |
|
80
|
|
|
|
|
81
|
5 |
|
final int nSentences = max - rand.nextInt((max + 1) - min); |
|
82
|
5 |
|
String separator = ""; |
|
83
|
5 |
|
mainLoop: |
|
84
|
5 |
|
for (int i = 0, start = rand.nextInt(lp.size()); i < nSentences; ++start) { |
|
85
|
5 |
|
final String para = lp.get(start % lp.size()); |
|
86
|
5 |
|
final String[] sentences = para.split("\\."); |
|
87
|
5 |
|
for (String s : sentences) { |
|
88
|
11 |
|
s = s.trim().replaceAll("\\s+", " "); |
|
89
|
22 |
|
if (s.isEmpty()) { |
|
90
|
|
|
sb.append('.'); |
|
91
|
|
|
continue; |
|
92
|
|
|
} |
|
93
|
|
|
|
|
94
|
11 |
|
sb.append(separator).append(s).append('.'); |
|
95
|
11 |
|
separator = " "; |
|
96
|
11 |
|
++i; |
|
97
|
|
|
|
|
98
|
22 |
|
if (i == nSentences) { |
|
99
|
5 |
|
break mainLoop; |
|
100
|
|
|
} |
|
101
|
|
|
} |
|
102
|
|
|
} |
|
103
|
|
|
|
|
104
|
5 |
|
return sb.toString(); |
|
105
|
|
|
} |
|
106
|
|
|
|
|
107
|
|
|
/** |
|
108
|
|
|
* Generates the specified number of paragraphs. |
|
109
|
|
|
* |
|
110
|
|
|
* The paragraphs are generated from the loaded lorem ipsum text. The start |
|
111
|
|
|
* position of the text is randomized, however the following paragraphs |
|
112
|
|
|
* appear in sequence - restarting at the beginning if all paragraphs have |
|
113
|
|
|
* been used. |
|
114
|
|
|
* |
|
115
|
|
|
* @param paragraphs the number of paragraphs to generate |
|
116
|
|
|
* @return the paragraphs |
|
117
|
|
|
* @throws IOException if an error occurs reading from the lipsum text file |
|
118
|
|
|
*/ |
|
119
|
1 |
|
public String paragraphs(@NamedParameter("paragraphs") int paragraphs) throws IOException { |
|
120
|
1 |
|
final List<String> lp = lipsumParagraphs; |
|
121
|
1 |
|
final StringBuilder sb = new StringBuilder(); |
|
122
|
5 |
|
for (int i = 0, start = rand.nextInt(lp.size()); i < paragraphs; ++i, ++start) { |
|
123
|
3 |
|
sb.append(lp.get(start % lp.size())).append("\r\n\r\n"); |
|
124
|
|
|
} |
|
125
|
1 |
|
return sb.toString().trim(); |
|
126
|
|
|
} |
|
127
|
|
|
|
|
128
|
|
|
/** |
|
129
|
|
|
* Generates an amount of text similar to the text passed as a parameter. |
|
130
|
|
|
* |
|
131
|
|
|
* The method counts the number of paragraphs in the text, generating the |
|
132
|
|
|
* same number of "lorem ipsum" paragraphs. If the text doesn't contain |
|
133
|
|
|
* paragraphs, the method counts the number of sentences and generates a |
|
134
|
|
|
* similar amount of sentences (+/- 1 sentence). |
|
135
|
|
|
* |
|
136
|
|
|
* @param text the text to use as a basis for generation |
|
137
|
|
|
* @return the generated lorem ipsum text |
|
138
|
|
|
* @throws IOException if an error occurs reading from the lipsum text file |
|
139
|
|
|
*/ |
|
140
|
5 |
|
public String similar(@NamedParameter("text") String text) throws IOException { |
|
141
|
5 |
|
final String sParas = text.replaceAll("\r\n", "\n"); |
|
142
|
5 |
|
final int nParas = StringUtils.countMatches(sParas, "\n"); |
|
143
|
10 |
|
if (nParas > 0) { |
|
144
|
1 |
|
final StringBuilder sb = new StringBuilder(); |
|
145
|
1 |
|
for (final String para : sParas.split("\n")) { |
|
146
|
10 |
|
if (para.trim().isEmpty()) { |
|
147
|
2 |
|
sb.append("\r\n"); |
|
148
|
2 |
|
continue; |
|
149
|
|
|
} |
|
150
|
3 |
|
sb.append(similar(para)).append("\r\n"); |
|
151
|
|
|
} |
|
152
|
1 |
|
return sb.toString().trim(); |
|
153
|
|
|
} |
|
154
|
4 |
|
final int nSent = StringUtils.countMatches(text.replaceAll("\\.{2,}", "."), "."); |
|
155
|
4 |
|
return sentences(Math.max(1, nSent - 1), Math.max(1, nSent + 1)); |
|
156
|
|
|
} |
|
157
|
|
|
} |
|
158
|
|
|
|
armenak /
DataDefender
