1 /**
2 * This file Copyright (c) 2003-2018 Magnolia International
3 * Ltd. (http://www.magnolia-cms.com). All rights reserved.
4 *
5 *
6 * This file is dual-licensed under both the Magnolia
7 * Network Agreement and the GNU General Public License.
8 * You may elect to use one or the other of these licenses.
9 *
10 * This file is distributed in the hope that it will be
11 * useful, but AS-IS and WITHOUT ANY WARRANTY; without even the
12 * implied warranty of MERCHANTABILITY or FITNESS FOR A
13 * PARTICULAR PURPOSE, TITLE, or NONINFRINGEMENT.
14 * Redistribution, except as permitted by whichever of the GPL
15 * or MNA you select, is prohibited.
16 *
17 * 1. For the GPL license (GPL), you can redistribute and/or
18 * modify this file under the terms of the GNU General
19 * Public License, Version 3, as published by the Free Software
20 * Foundation. You should have received a copy of the GNU
21 * General Public License, Version 3 along with this program;
22 * if not, write to the Free Software Foundation, Inc., 51
23 * Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
24 *
25 * 2. For the Magnolia Network Agreement (MNA), this file
26 * and the accompanying materials are made available under the
27 * terms of the MNA which accompanies this distribution, and
28 * is available at http://www.magnolia-cms.com/mna.html
29 *
30 * Any modifications to this file must keep this entire header
31 * intact.
32 *
33 */
34 package info.magnolia.cms.util;
35
36 import java.util.regex.Pattern;
37
38 import org.apache.commons.lang3.StringUtils;
39
40
41 /**
42 * An implementation of URLPattern which matches strings using simple <code>*</code> or <code>?</code> wildcards.
43 */
44 public final class SimpleUrlPattern implements UrlPattern {
45
46 /**
47 * @deprecated URL_CHAR_PATTERN used to be a common prefix for both single/multiple-characters wildcard patterns—
48 * before 4.3.1, it used to be a complex white-listing regex. Standalone, it is however of no use.
49 */
50 @Deprecated
51 public static final String URL_CHAR_PATTERN = ".";
52
53 /**
54 * Regexp pattern used for the simple keyword <code>*</code>. Matches 0 or more characters.
55 */
56 public static final String MULTIPLE_CHAR_PATTERN = ".*";
57
58 /**
59 * Regexp pattern used for the simple keyword <code>?</code>. Matches 0 or 1 character.
60 */
61 public static final String SINGLE_CHAR_PATTERN = ".?";
62
63 /**
64 * Regexp pattern used in match().
65 */
66 private Pattern pattern;
67
68 /**
69 * Pattern length. Longer patterns have higher priority.
70 */
71 private int length;
72
73 /**
74 * internal pattern string.
75 */
76 private String patternString;
77
78 /**
79 * Default constructor used by ContentToBean.
80 */
81 public SimpleUrlPattern() {
82 }
83
84 /**
85 * Compile a regexp pattern handling <code>*</code> and <code>?</code> chars.
86 *
87 * @param string input string
88 */
89 public SimpleUrlPattern(String string) {
90 setPatternString(string);
91 }
92
93 /**
94 * Replace all "*" with <code>RegexWildcardPattern.MULTIPLE_CHAR_PATTERN</code>.
95 *
96 * @param str input string
97 * @return string where all the occurrences of <code>*</code> and <code>?</code> are replaced with a regexp
98 * pattern.
99 */
100 public static String getEncodedString(String str) {
101 final StringBuilder builder = new StringBuilder();
102 char[] chars = str.toCharArray();
103 int i = 0, last = 0;
104 while (i < chars.length) {
105 char c = chars[i];
106 if (c == '*') {
107 builder.append('(').append(chars, last, i - last).append(')').append(MULTIPLE_CHAR_PATTERN);
108 last = i + 1;
109 } else if (c == '?') {
110 builder.append('(').append(chars, last, i - last).append(')').append(SINGLE_CHAR_PATTERN);
111 last = i + 1;
112 } else if (c == '.') {
113 builder.append('(').append(chars, last, i - last).append(')').append("\\.");
114 last = i + 1;
115 }
116 i++;
117 }
118 builder.append(chars, last, i - last);
119 return builder.toString();
120 }
121
122 @Override
123 public boolean match(String str) {
124 return this.pattern.matcher(str).matches();
125 }
126
127 @Override
128 public int getLength() {
129 return this.length;
130 }
131
132 @Override
133 public String getPatternString() {
134 return patternString;
135 }
136
137 /**
138 * Mainly used by ContentToBean.
139 */
140 public void setPatternString(String patternString) {
141 this.length = StringUtils.removeEnd(patternString, "*").length();
142 this.pattern = Pattern.compile(getEncodedString(patternString), Pattern.DOTALL);
143 this.patternString = patternString;
144 }
145
146 @Override
147 public String toString() {
148 // don't use pattern.pattern(), but keep the original string.
149 // The "compiled" pattern will display the ugly patterns like MULTIPLE_CHAR_PATTERN instead of simple *
150 return "SimpleUrlPattern{" + patternString + '}';
151 }
152 }