View Javadoc
1   /*
2    * Licensed to the Apache Software Foundation (ASF) under one or more
3    * contributor license agreements.  See the NOTICE file distributed with
4    * this work for additional information regarding copyright ownership.
5    * The ASF licenses this file to You under the Apache License, Version 2.0
6    * (the "License"); you may not use this file except in compliance with
7    * the License.  You may obtain a copy of the License at
8    *
9    *      http://www.apache.org/licenses/LICENSE-2.0
10   *
11   * Unless required by applicable law or agreed to in writing, software
12   * distributed under the License is distributed on an "AS IS" BASIS,
13   * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
14   * See the License for the specific language governing permissions and
15   * limitations under the License.
16   */
17  
18  package org.apache.commons.codec.language.bm;
19  
20  import org.apache.commons.codec.EncoderException;
21  import org.apache.commons.codec.StringEncoder;
22  
23  /**
24   * Encodes strings into their Beider-Morse phonetic encoding.
25   * <p>
26   * Beider-Morse phonetic encodings are optimised for family names. However, they may be useful for a wide range of
27   * words.
28   * <p>
29   * This encoder is intentionally mutable to allow dynamic configuration through bean properties. As such, it is mutable,
30   * and may not be thread-safe. If you require a guaranteed thread-safe encoding then use {@link PhoneticEngine}
31   * directly.
32   * <p>
33   * <b>Encoding overview</b>
34   * <p>
35   * Beider-Morse phonetic encodings is a multi-step process. Firstly, a table of rules is consulted to guess what
36   * language the word comes from. For example, if it ends in "<code>ault</code>" then it infers that the word is French.
37   * Next, the word is translated into a phonetic representation using a language-specific phonetics table. Some runs of
38   * letters can be pronounced in multiple ways, and a single run of letters may be potentially broken up into phonemes at
39   * different places, so this stage results in a set of possible language-specific phonetic representations. Lastly, this
40   * language-specific phonetic representation is processed by a table of rules that re-writes it phonetically taking into
41   * account systematic pronunciation differences between languages, to move it towards a pan-indo-european phonetic
42   * representation. Again, sometimes there are multiple ways this could be done and sometimes things that can be
43   * pronounced in several ways in the source language have only one way to represent them in this average phonetic
44   * language, so the result is again a set of phonetic spellings.
45   * <p>
46   * Some names are treated as having multiple parts. This can be due to two things. Firstly, they may be hyphenated. In
47   * this case, each individual hyphenated word is encoded, and then these are combined end-to-end for the final encoding.
48   * Secondly, some names have standard prefixes, for example, "<code>Mac/Mc</code>" in Scottish (English) names. As
49   * sometimes it is ambiguous whether the prefix is intended or is an accident of the spelling, the word is encoded once
50   * with the prefix and once without it. The resulting encoding contains one and then the other result.
51   * <p>
52   * <b>Encoding format</b>
53   * <p>
54   * Individual phonetic spellings of an input word are represented in upper- and lower-case roman characters. Where there
55   * are multiple possible phonetic representations, these are joined with a pipe (<code>|</code>) character. If multiple
56   * hyphenated words where found, or if the word may contain a name prefix, each encoded word is placed in elipses and
57   * these blocks are then joined with hyphens. For example, "<code>d'ortley</code>" has a possible prefix. The form
58   * without prefix encodes to "<code>ortlaj|ortlej</code>", while the form with prefix encodes to "
59   * <code>dortlaj|dortlej</code>". Thus, the full, combined encoding is "<code>(ortlaj|ortlej)-(dortlaj|dortlej)</code>".
60   * <p>
61   * The encoded forms are often quite a bit longer than the input strings. This is because a single input may have many
62   * potential phonetic interpretations. For example, "<code>Renault</code>" encodes to "
63   * <code>rYnDlt|rYnalt|rYnult|rinDlt|rinalt|rinult</code>". The <code>APPROX</code> rules will tend to produce larger
64   * encodings as they consider a wider range of possible, approximate phonetic interpretations of the original word.
65   * Down-stream applications may wish to further process the encoding for indexing or lookup purposes, for example, by
66   * splitting on pipe (<code>|</code>) and indexing under each of these alternatives.
67   * <p>
68   * <b>Note</b>: this version of the Beider-Morse encoding is equivalent with v3.4 of the reference implementation.
69   *
70   * @see <a href="http://stevemorse.org/phonetics/bmpm.htm">Beider-Morse Phonetic Matching</a>
71   * @see <a href="http://stevemorse.org/phoneticinfo.htm">Reference implementation</a>
72   *
73   * @since 1.6
74   * @version $Id: BeiderMorseEncoder.java 1636703 2014-11-04 19:49:40Z tn $
75   */
76  public class BeiderMorseEncoder implements StringEncoder {
77      // Implementation note: This class is a spring-friendly facade to PhoneticEngine. It allows read/write configuration
78      // of an immutable PhoneticEngine instance that will be delegated to for the actual encoding.
79  
80      // a cached object
81      private PhoneticEngine engine = new PhoneticEngine(NameType.GENERIC, RuleType.APPROX, true);
82  
83      @Override
84      public Object encode(final Object source) throws EncoderException {
85          if (!(source instanceof String)) {
86              throw new EncoderException("BeiderMorseEncoder encode parameter is not of type String");
87          }
88          return encode((String) source);
89      }
90  
91      @Override
92      public String encode(final String source) throws EncoderException {
93          if (source == null) {
94              return null;
95          }
96          return this.engine.encode(source);
97      }
98  
99      /**
100      * Gets the name type currently in operation.
101      *
102      * @return the NameType currently being used
103      */
104     public NameType getNameType() {
105         return this.engine.getNameType();
106     }
107 
108     /**
109      * Gets the rule type currently in operation.
110      *
111      * @return the RuleType currently being used
112      */
113     public RuleType getRuleType() {
114         return this.engine.getRuleType();
115     }
116 
117     /**
118      * Discovers if multiple possible encodings are concatenated.
119      *
120      * @return true if multiple encodings are concatenated, false if just the first one is returned
121      */
122     public boolean isConcat() {
123         return this.engine.isConcat();
124     }
125 
126     /**
127      * Sets how multiple possible phonetic encodings are combined.
128      *
129      * @param concat
130      *            true if multiple encodings are to be combined with a '|', false if just the first one is
131      *            to be considered
132      */
133     public void setConcat(final boolean concat) {
134         this.engine = new PhoneticEngine(this.engine.getNameType(),
135                                          this.engine.getRuleType(),
136                                          concat,
137                                          this.engine.getMaxPhonemes());
138     }
139 
140     /**
141      * Sets the type of name. Use {@link NameType#GENERIC} unless you specifically want phonetic encodings
142      * optimized for Ashkenazi or Sephardic Jewish family names.
143      *
144      * @param nameType
145      *            the NameType in use
146      */
147     public void setNameType(final NameType nameType) {
148         this.engine = new PhoneticEngine(nameType,
149                                          this.engine.getRuleType(),
150                                          this.engine.isConcat(),
151                                          this.engine.getMaxPhonemes());
152     }
153 
154     /**
155      * Sets the rule type to apply. This will widen or narrow the range of phonetic encodings considered.
156      *
157      * @param ruleType
158      *            {@link RuleType#APPROX} or {@link RuleType#EXACT} for approximate or exact phonetic matches
159      */
160     public void setRuleType(final RuleType ruleType) {
161         this.engine = new PhoneticEngine(this.engine.getNameType(),
162                                          ruleType,
163                                          this.engine.isConcat(),
164                                          this.engine.getMaxPhonemes());
165     }
166 
167     /**
168      * Sets the number of maximum of phonemes that shall be considered by the engine.
169      *
170      * @param maxPhonemes
171      *            the maximum number of phonemes returned by the engine
172      * @since 1.7
173      */
174     public void setMaxPhonemes(final int maxPhonemes) {
175         this.engine = new PhoneticEngine(this.engine.getNameType(),
176                                          this.engine.getRuleType(),
177                                          this.engine.isConcat(),
178                                          maxPhonemes);
179     }
180 
181 }