1 /**
2 * This file Copyright (c) 2009-2010 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 info.magnolia.cms.core.SystemProperty;
37 import info.magnolia.objectfactory.Components;
38
39 import java.lang.reflect.InvocationTargetException;
40 import java.lang.reflect.Method;
41
42 /**
43 * A wrapper around java.text.Normalizer and com.ibm.icu.text.Normalizer; uses the former if present, or none
44 * if none is present.
45 *
46 * <strong>note:</strong> if not running under Java >=6, and without ICU, this does nothing.
47 * If needed, one can use their own implementation, by setting the info.magnolia.cms.util.UnicodeNormalizer$Normalizer
48 * system property.
49 *
50 * @see java.text.Normalizer
51 * @see com.ibm.icu.text.Normalizer
52 * @see <a href="http://www.icu-project.org/">http://www.icu-project.org/</a> to get the ICU4J library.
53 * @see <a href="http://en.wikipedia.org/wiki/Unicode_equivalence#Normal_forms">http://en.wikipedia.org/wiki/Unicode_equivalence#Normal_forms</a> for more information.
54 *
55 * @author gjoseph
56 * @version $Revision: $ ($Author: $)
57 */
58 public class UnicodeNormalizer {
59 private static final org.slf4j.Logger log = org.slf4j.LoggerFactory.getLogger(UnicodeNormalizer.class);
60
61 private static final String JAVA6_NORMALIZER_CLASS = "java.text.Normalizer";
62 private static final String JAVA6_FORMPARAM_CLASS = "java.text.Normalizer$Form";
63 private static final String ICU_NORMALIZER_CLASS = "com.ibm.icu.text.Normalizer";
64
65 private static final Normalizer normalizer = Components.getSingleton(Normalizer.class);
66
67 /**
68 * Normalizes the given String to the NFC form.
69 */
70 public static String normalizeNFC(String in) {
71 return normalizer.normalizeNFC(in);
72 /* if you're in dire need to debug:
73 try {
74 log.debug("not normalized: " + Arrays.toString(in.getBytes("UTF-8")) + " (" + in + ")");
75 String out = normalizer.normalizeNFC(in);
76 log.debug(" normalized: " + Arrays.toString(out.getBytes("UTF-8")) + " (" + out + ")");
77 return out;
78 } catch (UnsupportedEncodingException e) {
79 // do nothing
80 }
81 return in;
82 */
83 }
84
85 public interface Normalizer {
86 String normalizeNFC(String in);
87 }
88
89 /**
90 * This uses reflection, since we're still compiling with Java 5.
91 * This implementation could be externalized to a "java 6 only" module if needed.
92 */
93 public static final class Java6ReflectionNormalizer implements Normalizer {
94 private final Method normalize;
95 private final Object nfc;
96
97 public Java6ReflectionNormalizer() {
98 try {
99 final Class<?> normalizer = Class.forName(JAVA6_NORMALIZER_CLASS);
100 final Class<?> form = Class.forName(JAVA6_FORMPARAM_CLASS);
101 normalize = normalizer.getMethod("normalize", CharSequence.class, form);
102 nfc = form.getField("NFC").get(null);
103 } catch (ClassNotFoundException e) {
104 throw new RuntimeException(e);
105 } catch (IllegalAccessException e) {
106 throw new RuntimeException(e);
107 } catch (NoSuchFieldException e) {
108 throw new RuntimeException(e);
109 } catch (NoSuchMethodException e) {
110 throw new RuntimeException(e);
111 }
112
113 }
114
115 public String normalizeNFC(String in) {
116 try {
117 return (String) normalize.invoke(null, in, nfc);
118 } catch (IllegalAccessException e) {
119 throw new RuntimeException(e);
120 } catch (InvocationTargetException e) {
121 throw new RuntimeException(e);
122 }
123 }
124 }
125
126 public static final class ICUNormalizer implements UnicodeNormalizer.Normalizer {
127 public String normalizeNFC(String in) {
128 return com.ibm.icu.text.Normalizer.normalize(in, com.ibm.icu.text.Normalizer.NFC);
129 }
130 }
131
132 public static final class NonNormalizer implements UnicodeNormalizer.Normalizer {
133 public String normalizeNFC(String in) {
134 return in;
135 }
136 }
137
138 public static final class AutoDetectNormalizer implements Normalizer {
139 private final Normalizer delegate;
140
141 public AutoDetectNormalizer() {
142 Normalizer candidate;
143 if (!SystemProperty.getBooleanProperty(SystemProperty.MAGNOLIA_UTF8_ENABLED)) {
144 candidate = new NonNormalizer();
145 } else {
146 try {
147 Class.forName(JAVA6_NORMALIZER_CLASS);
148 candidate = new Java6ReflectionNormalizer();
149 log.info("Running on Java 6, using {} for unicode form normalization.", candidate.getClass());
150 } catch (ClassNotFoundException e) {
151 log.warn("Not running on Java 6 ({} not found). Attempting to locate the ICU4J library.", JAVA6_NORMALIZER_CLASS);
152 try {
153 Class.forName(ICU_NORMALIZER_CLASS);
154 candidate = new ICUNormalizer();
155 log.info("ICU4J found, using {} for Unicode form normalization.", candidate.getClass());
156 } catch (ClassNotFoundException e2) {
157 log.warn("ICU4J not found ({} not found), Unicode will not be 100% supported; no Unicode form normalization available. If Java 6 is not an option, you can get the ICU4J library from http://www.icu-project.org/.", ICU_NORMALIZER_CLASS);
158 candidate = new NonNormalizer();
159 }
160 }
161 }
162 this.delegate = candidate;
163 }
164
165 public String normalizeNFC(String in) {
166 return delegate.normalizeNFC(in);
167 }
168 }
169
170 }