1 /**
2 * This file Copyright (c) 2003-2011 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.lang.StringUtils;
39
40
41 /**
42 * An implementation of URLPattern which matches strings using simple <code>*</code> or <code>?</code> wildcards.
43 * @author Sameer Charles
44 * @author Fabrizio Giustina
45 * TODO rewrite this class using ant-style path comparison and avoiding regexp. See
46 * org.springframework.util.AntPathMatcher in spring 1.2 for a nice implementation.
47 * @version $Revision $ ($Author $)
48 */
49 public final class SimpleUrlPattern implements UrlPattern {
50
51 /**
52 * Stable serialVersionUID.
53 */
54 private static final long serialVersionUID = 223L;
55
56 /**
57 * Any char, newline included.
58 */
59 public static final String URL_CHAR_PATTERN = "."; //$NON-NLS-1$
60
61 /**
62 * Regexp pattern used for the simple keyword <code>*</code>. Matches 0 or more characters.
63 */
64 public static final String MULTIPLE_CHAR_PATTERN = URL_CHAR_PATTERN + "*"; //$NON-NLS-1$
65
66 /**
67 * Regexp pattern used for the simple keyword <code>?</code>. Matches 0 or 1 character.
68 */
69 public static final String SINGLE_CHAR_PATTERN = URL_CHAR_PATTERN + "?"; //$NON-NLS-1$
70
71 /**
72 * Regexp pattern used in match().
73 */
74 private Pattern pattern;
75
76 /**
77 * Pattern length. Longer patterns have higher priority.
78 */
79 private int length;
80
81 /**
82 * internal pattern string.
83 */
84 private String patternString;
85
86 /**
87 * Default constructor used by ContentToBean.
88 */
89 public SimpleUrlPattern() {
90 }
91
92 /**
93 * Compile a regexp pattern handling <code>*</code> and <code>?</code> chars.
94 * @param string input string
95 * @return a RegExp pattern
96 */
97 public SimpleUrlPattern(String string) {
98 this.length = StringUtils.removeEnd(string, "*").length();
99 this.pattern = Pattern.compile(getEncodedString(string), Pattern.DOTALL);
100 this.patternString = string;
101 }
102
103 /**
104 * Replace all "*" with <code>RegexWildcardPattern.MULTIPLE_CHAR_PATTERN</code>.
105 * @param str input string
106 * @return string where all the occurrences of <code>*</code> and <code>?</code> are replaced with a regexp
107 * pattern.
108 */
109 public static String getEncodedString(String str) {
110 StringBuffer stringBuffer = new StringBuffer();
111 char[] chars = str.toCharArray();
112 int i = 0, last = 0;
113 while (i < chars.length) {
114 char c = chars[i];
115 if (c == '*') {
116 stringBuffer.append('(');
117 stringBuffer.append(chars, last, i - last);
118 stringBuffer.append(')');
119 stringBuffer.append(MULTIPLE_CHAR_PATTERN);
120 last = i + 1;
121 }
122 else if (c == '?') {
123 stringBuffer.append('(');
124 stringBuffer.append(chars, last, i - last);
125 stringBuffer.append(')');
126 stringBuffer.append(SINGLE_CHAR_PATTERN);
127 last = i + 1;
128 }
129 i++;
130 }
131 stringBuffer.append(chars, last, i - last);
132 return stringBuffer.toString();
133 }
134
135 /**
136 * @see info.magnolia.cms.util.UrlPattern#match(java.lang.String)
137 */
138 @Override
139 public boolean match(String str) {
140 return this.pattern.matcher(str).matches();
141 }
142
143 /**
144 * @see info.magnolia.cms.util.UrlPattern#getLength()
145 */
146 @Override
147 public int getLength() {
148 return this.length;
149 }
150
151 /**
152 * @see info.magnolia.cms.util.UrlPattern#getString()
153 */
154 @Override
155 public String getPatternString() {
156 return patternString;
157 }
158
159 /**
160 * Mainly used by ContentToBean.
161 */
162 public void setPatternString(String patternString) {
163 this.length = StringUtils.removeEnd(patternString, "*").length();
164 this.pattern = Pattern.compile(getEncodedString(patternString), Pattern.DOTALL);
165 this.patternString = patternString;
166 }
167
168 @Override
169 public String toString() {
170 // don't use pattern.pattern(), but keep the original string.
171 // The "compiled" pattern will display the ugly patterns like MULTIPLE_CHAR_PATTERN instead of simple *
172 return "SimpleUrlPattern{" + patternString + '}';
173 }
174 }