1 | package com.swabunga.spell.engine;
|
---|
2 |
|
---|
3 | import java.io.*;
|
---|
4 | import java.util.*;
|
---|
5 |
|
---|
6 | /**
|
---|
7 | * Another implementation of <code>SpellDictionary</code> that doesn't cache any words in memory. Avoids the huge
|
---|
8 | * footprint of <code>SpellDictionaryHashMap</code> at the cost of relatively minor latency. A future version
|
---|
9 | * of this class that implements some caching strategies might be a good idea in the future, if there's any
|
---|
10 | * demand for it.
|
---|
11 | *
|
---|
12 | * This implementation requires a special dictionary file, with "code*word" lines sorted by code.
|
---|
13 | * It's using a dichotomy algorithm to search for words in the dictionary
|
---|
14 | *
|
---|
15 | * @author Damien Guillaume
|
---|
16 | * @version 0.1
|
---|
17 | */
|
---|
18 | public class SpellDictionaryDichoDisk extends SpellDictionaryASpell {
|
---|
19 |
|
---|
20 | /** Holds the dictionary file for reading*/
|
---|
21 | private RandomAccessFile dictFile = null;
|
---|
22 |
|
---|
23 | /** dictionary and phonetic file encoding */
|
---|
24 | private String encoding = null;
|
---|
25 |
|
---|
26 | /**
|
---|
27 | * Dictionary Convienence Constructor.
|
---|
28 | */
|
---|
29 | public SpellDictionaryDichoDisk(File wordList)
|
---|
30 | throws FileNotFoundException, IOException {
|
---|
31 | super((File) null);
|
---|
32 | dictFile = new RandomAccessFile(wordList, "r");
|
---|
33 | }
|
---|
34 |
|
---|
35 | /**
|
---|
36 | * Dictionary Convienence Constructor.
|
---|
37 | */
|
---|
38 | public SpellDictionaryDichoDisk(File wordList, String encoding)
|
---|
39 | throws FileNotFoundException, IOException {
|
---|
40 | super((File) null);
|
---|
41 | this.encoding = encoding;
|
---|
42 | dictFile = new RandomAccessFile(wordList, "r");
|
---|
43 | }
|
---|
44 |
|
---|
45 | /**
|
---|
46 | * Dictionary constructor that uses an aspell phonetic file to
|
---|
47 | * build the transformation table.
|
---|
48 | */
|
---|
49 | public SpellDictionaryDichoDisk(File wordList, File phonetic)
|
---|
50 | throws FileNotFoundException, IOException {
|
---|
51 | super(phonetic);
|
---|
52 | dictFile = new RandomAccessFile(wordList, "r");
|
---|
53 | }
|
---|
54 |
|
---|
55 | /**
|
---|
56 | * Dictionary constructor that uses an aspell phonetic file to
|
---|
57 | * build the transformation table.
|
---|
58 | */
|
---|
59 | public SpellDictionaryDichoDisk(File wordList, File phonetic, String encoding)
|
---|
60 | throws FileNotFoundException, IOException {
|
---|
61 | super(phonetic, encoding);
|
---|
62 | this.encoding = encoding;
|
---|
63 | dictFile = new RandomAccessFile(wordList, "r");
|
---|
64 | }
|
---|
65 |
|
---|
66 | /**
|
---|
67 | * Add a word permanantly to the dictionary (and the dictionary file).
|
---|
68 | * <i>not implemented !</i>
|
---|
69 | */
|
---|
70 | public void addWord(String word) {
|
---|
71 | System.err.println("error: addWord is not implemented for SpellDictionaryDichoDisk");
|
---|
72 | }
|
---|
73 |
|
---|
74 | /**
|
---|
75 | * Search the dictionary file for the words corresponding to the code
|
---|
76 | * within positions p1 - p2
|
---|
77 | */
|
---|
78 | private LinkedList dichoFind(String code, long p1, long p2) throws IOException {
|
---|
79 | //System.out.println("dichoFind("+code+","+p1+","+p2+")");
|
---|
80 | long pm = (p1 + p2) / 2;
|
---|
81 | dictFile.seek(pm);
|
---|
82 | String l;
|
---|
83 | if (encoding == null)
|
---|
84 | l = dictFile.readLine();
|
---|
85 | else
|
---|
86 | l = dictReadLine();
|
---|
87 | pm = dictFile.getFilePointer();
|
---|
88 | if (encoding == null)
|
---|
89 | l = dictFile.readLine();
|
---|
90 | else
|
---|
91 | l = dictReadLine();
|
---|
92 | long pm2 = dictFile.getFilePointer();
|
---|
93 | if (pm2 >= p2)
|
---|
94 | return(seqFind(code, p1, p2));
|
---|
95 | int istar = l.indexOf('*');
|
---|
96 | if (istar == -1)
|
---|
97 | throw new IOException("bad format: no * !");
|
---|
98 | String testcode = l.substring(0, istar);
|
---|
99 | int comp = code.compareTo(testcode);
|
---|
100 | if (comp < 0)
|
---|
101 | return(dichoFind(code, p1, pm-1));
|
---|
102 | else if (comp > 0)
|
---|
103 | return(dichoFind(code, pm2, p2));
|
---|
104 | else {
|
---|
105 | LinkedList l1 = dichoFind(code, p1, pm-1);
|
---|
106 | LinkedList l2 = dichoFind(code, pm2, p2);
|
---|
107 | String word = l.substring(istar+1);
|
---|
108 | l1.add(word);
|
---|
109 | l1.addAll(l2);
|
---|
110 | return(l1);
|
---|
111 | }
|
---|
112 | }
|
---|
113 |
|
---|
114 | private LinkedList seqFind(String code, long p1, long p2) throws IOException {
|
---|
115 | //System.out.println("seqFind("+code+","+p1+","+p2+")");
|
---|
116 | LinkedList list = new LinkedList();
|
---|
117 | dictFile.seek(p1);
|
---|
118 | while (dictFile.getFilePointer() < p2) {
|
---|
119 | String l;
|
---|
120 | if (encoding == null)
|
---|
121 | l = dictFile.readLine();
|
---|
122 | else
|
---|
123 | l = dictReadLine();
|
---|
124 | int istar = l.indexOf('*');
|
---|
125 | if (istar == -1)
|
---|
126 | throw new IOException("bad format: no * !");
|
---|
127 | String testcode = l.substring(0, istar);
|
---|
128 | if (testcode.equals(code)) {
|
---|
129 | String word = l.substring(istar+1);
|
---|
130 | list.add(word);
|
---|
131 | }
|
---|
132 | }
|
---|
133 | return(list);
|
---|
134 | }
|
---|
135 |
|
---|
136 | /**
|
---|
137 | * Read a line of dictFile with a specific encoding
|
---|
138 | */
|
---|
139 | private String dictReadLine() throws IOException {
|
---|
140 | int max = 255;
|
---|
141 | byte b=0;
|
---|
142 | byte[] buf = new byte[max];
|
---|
143 | int i=0;
|
---|
144 | try {
|
---|
145 | for (; b != '\n' && b != '\r' && i<max-1; i++) {
|
---|
146 | b = dictFile.readByte();
|
---|
147 | buf[i] = b;
|
---|
148 | }
|
---|
149 | } catch (EOFException ex) {
|
---|
150 | }
|
---|
151 | if (i == 0)
|
---|
152 | return("");
|
---|
153 | String s = new String(buf, 0, i-1, encoding);
|
---|
154 | return(s);
|
---|
155 | }
|
---|
156 |
|
---|
157 | /**
|
---|
158 | * Returns a list of strings (words) for the code.
|
---|
159 | */
|
---|
160 | public List getWords(String code) {
|
---|
161 | //System.out.println("getWords("+code+")");
|
---|
162 | LinkedList list;
|
---|
163 | try {
|
---|
164 | list = dichoFind(code, 0, dictFile.length()-1);
|
---|
165 | //System.out.println(list);
|
---|
166 | } catch (IOException ex) {
|
---|
167 | System.err.println("IOException: " + ex.getMessage());
|
---|
168 | list = new LinkedList();
|
---|
169 | }
|
---|
170 | return list;
|
---|
171 | }
|
---|
172 |
|
---|
173 | }
|
---|