1 package com.hughes.android.dictionary;
\r
3 import java.io.IOException;
\r
4 import java.io.RandomAccessFile;
\r
5 import java.util.ArrayList;
\r
6 import java.util.List;
\r
7 import java.util.concurrent.atomic.AtomicBoolean;
\r
9 import com.hughes.util.CachingList;
\r
10 import com.hughes.util.raf.FileList;
\r
11 import com.hughes.util.raf.RAFFactory;
\r
12 import com.hughes.util.raf.RAFSerializable;
\r
13 import com.hughes.util.raf.RAFSerializableSerializer;
\r
14 import com.hughes.util.raf.RAFSerializer;
\r
15 import com.hughes.util.raf.UniformFileList;
\r
17 public final class Dictionary implements RAFSerializable<Dictionary> {
\r
19 static final RAFSerializer<Entry> ENTRY_SERIALIZER = new RAFSerializableSerializer<Entry>(
\r
21 static final RAFSerializer<Row> ROW_SERIALIZER = new RAFSerializableSerializer<Row>(
\r
23 static final RAFSerializer<IndexEntry> INDEX_ENTRY_SERIALIZER = new RAFSerializableSerializer<IndexEntry>(
\r
24 IndexEntry.RAF_FACTORY);
\r
26 final List<Entry> entries;
\r
27 final LanguageData[] languageDatas = new LanguageData[2];
\r
29 public Dictionary(final Language language0, final Language language1) {
\r
30 languageDatas[0] = new LanguageData(this, language0, Entry.LANG1);
\r
31 languageDatas[1] = new LanguageData(this, language1, Entry.LANG2);
\r
32 entries = new ArrayList<Entry>();
\r
35 public Dictionary(final RandomAccessFile raf) throws IOException {
\r
36 entries = CachingList.create(FileList.create(raf, ENTRY_SERIALIZER, raf
\r
37 .getFilePointer()), 10000);
\r
38 languageDatas[0] = new LanguageData(this, raf, Entry.LANG1);
\r
39 languageDatas[1] = new LanguageData(this, raf, Entry.LANG2);
\r
42 public void write(RandomAccessFile raf) throws IOException {
\r
43 FileList.write(raf, entries, ENTRY_SERIALIZER);
\r
44 languageDatas[0].write(raf);
\r
45 languageDatas[1].write(raf);
\r
48 final class LanguageData implements RAFSerializable<LanguageData> {
\r
49 final Dictionary dictionary;
\r
50 final Language language;
\r
52 final List<Row> rows;
\r
53 final List<IndexEntry> sortedIndex;
\r
55 LanguageData(final Dictionary dictionary, final Language language, final byte lang) {
\r
56 this.dictionary = dictionary;
\r
57 this.language = language;
\r
59 rows = new ArrayList<Row>();
\r
60 sortedIndex = new ArrayList<IndexEntry>();
\r
63 LanguageData(final Dictionary dictionary, final RandomAccessFile raf, final byte lang) throws IOException {
\r
64 this.dictionary = dictionary;
\r
65 language = Language.lookup(raf.readUTF());
\r
66 if (language == null) {
\r
67 throw new RuntimeException("Unknown language.");
\r
70 rows = CachingList.create(UniformFileList.create(raf, ROW_SERIALIZER, raf
\r
71 .getFilePointer()), 10000);
\r
72 sortedIndex = CachingList.create(FileList.create(raf,
\r
73 INDEX_ENTRY_SERIALIZER, raf.getFilePointer()), 10000);
\r
76 public void write(final RandomAccessFile raf) throws IOException {
\r
77 raf.writeUTF(language.symbol);
\r
78 UniformFileList.write(raf, rows, ROW_SERIALIZER, 4);
\r
79 FileList.write(raf, sortedIndex, INDEX_ENTRY_SERIALIZER);
\r
82 String rowToString(final Row row) {
\r
83 return row.isToken() ? sortedIndex.get(row.getIndex()).word : entries
\r
84 .get(row.getIndex()).toString();
\r
87 int lookup(String word, final AtomicBoolean interrupted) {
\r
88 word = word.toLowerCase();
\r
91 int end = sortedIndex.size();
\r
92 while (start < end) {
\r
93 final int mid = (start + end) / 2;
\r
94 if (interrupted.get()) {
\r
97 final IndexEntry midEntry = sortedIndex.get(mid);
\r
99 final int comp = language.tokenComparator.compare(word, midEntry.word.toLowerCase());
\r
102 while (result > 0 && language.tokenComparator.compare(word, sortedIndex.get(result - 1).word.toLowerCase()) == 0) {
\r
104 if (interrupted.get()) {
\r
109 } else if (comp < 0) {
\r
115 return Math.min(sortedIndex.size() - 1, start);
\r
118 public IndexEntry getIndexEntryForRow(final int rowIndex) {
\r
119 // TODO: this kinda blows.
\r
123 row = rows.get(r);
\r
124 if (row.isToken() || row.indexEntry != null) {
\r
129 final IndexEntry indexEntry = row.isToken() ? sortedIndex.get(row.getIndex()) : row.indexEntry;
\r
130 for (; r <= rowIndex; ++r) {
\r
131 rows.get(r).indexEntry = indexEntry;
\r
133 assert false && rows.get(indexEntry.startRow).isToken();
\r
138 public static final class Row implements RAFSerializable<Row> {
\r
141 IndexEntry indexEntry = null;
\r
143 public Row(final int index) {
\r
144 this.index = index;
\r
147 static final RAFFactory<Row> RAF_FACTORY = new RAFFactory<Row>() {
\r
148 public Row create(RandomAccessFile raf) throws IOException {
\r
149 return new Row(raf.readInt());
\r
153 public void write(RandomAccessFile raf) throws IOException {
\r
154 raf.writeInt(index);
\r
157 boolean isToken() {
\r
161 public int getIndex() {
\r
169 public static final class IndexEntry implements RAFSerializable<IndexEntry> {
\r
171 final int startRow;
\r
173 public IndexEntry(final String word, final int startRow) {
\r
175 this.startRow = startRow;
\r
178 static final RAFFactory<IndexEntry> RAF_FACTORY = new RAFFactory<IndexEntry>() {
\r
179 public IndexEntry create(RandomAccessFile raf) throws IOException {
\r
180 final String word = raf.readUTF();
\r
181 final int startRow = raf.readInt();
\r
182 return new IndexEntry(word, startRow);
\r
186 public void write(final RandomAccessFile raf) throws IOException {
\r
187 raf.writeUTF(word);
\r
188 raf.writeInt(startRow);
\r
192 public String toString() {
\r
193 return word + "@" + startRow;
\r