src/com/hughes/android/dictionary/engine/Dictionary.java

   1 // Copyright 2011 Google Inc. All Rights Reserved.
   2 //
   3 // Licensed under the Apache License, Version 2.0 (the "License");
   4 // you may not use this file except in compliance with the License.
   5 // You may obtain a copy of the License at
   6 //
   7 //     http://www.apache.org/licenses/LICENSE-2.0
   8 //
   9 // Unless required by applicable law or agreed to in writing, software
  10 // distributed under the License is distributed on an "AS IS" BASIS,
  11 // WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
  12 // See the License for the specific language governing permissions and
  13 // limitations under the License.
  14
  15 package com.hughes.android.dictionary.engine;
  16
  17 import com.hughes.android.dictionary.DictionaryInfo;
  18 import com.hughes.util.CachingList;
  19 import com.hughes.util.raf.RAFList;
  20 import com.hughes.util.raf.RAFListSerializer;
  21 import com.hughes.util.raf.RAFSerializable;
  22
  23 import java.io.DataInput;
  24 import java.io.DataOutput;
  25 import java.io.File;
  26 import java.io.IOException;
  27 import java.io.PrintStream;
  28 import java.io.RandomAccessFile;
  29 import java.util.ArrayList;
  30 import java.util.Collections;
  31 import java.util.List;
  32
  33 public class Dictionary implements RAFSerializable<Dictionary> {
  34
  35     static final int CACHE_SIZE = 5000;
  36
  37     static final int CURRENT_DICT_VERSION = 6;
  38     static final String END_OF_DICTIONARY = "END OF DICTIONARY";
  39
  40     // persisted
  41     final int dictFileVersion;
  42     final long creationMillis;
  43     public final String dictInfo;
  44     public final List<PairEntry> pairEntries;
  45     public final List<TextEntry> textEntries;
  46     public final List<HtmlEntry> htmlEntries;
  47     public final List<EntrySource> sources;
  48     public final List<Index> indices;
  49
  50     /**
  51      * dictFileVersion 1 adds: <li>links to sources? dictFileVersion 2 adds: <li>
  52      * counts of tokens in indices.
  53      */
  54
  55     public Dictionary(final String dictInfo) {
  56         this.dictFileVersion = CURRENT_DICT_VERSION;
  57         this.creationMillis = System.currentTimeMillis();
  58         this.dictInfo = dictInfo;
  59         pairEntries = new ArrayList<PairEntry>();
  60         textEntries = new ArrayList<TextEntry>();
  61         htmlEntries = new ArrayList<HtmlEntry>();
  62         sources = new ArrayList<EntrySource>();
  63         indices = new ArrayList<Index>();
  64     }
  65
  66     public Dictionary(final RandomAccessFile raf) throws IOException {
  67         dictFileVersion = raf.readInt();
  68         if (dictFileVersion < 0 || dictFileVersion > CURRENT_DICT_VERSION) {
  69             throw new IOException("Invalid dictionary version: " + dictFileVersion);
  70         }
  71         creationMillis = raf.readLong();
  72         dictInfo = raf.readUTF();
  73
  74         // Load the sources, then seek past them, because reading them later
  75         // disrupts the offset.
  76         try {
  77             final RAFList<EntrySource> rafSources = RAFList.create(raf, new EntrySource.Serializer(
  78                     this), raf.getFilePointer());
  79             sources = new ArrayList<EntrySource>(rafSources);
  80             raf.seek(rafSources.getEndOffset());
  81
  82             pairEntries = CachingList.create(
  83                     RAFList.create(raf, new PairEntry.Serializer(this), raf.getFilePointer()),
  84                     CACHE_SIZE);
  85             textEntries = CachingList.create(
  86                     RAFList.create(raf, new TextEntry.Serializer(this), raf.getFilePointer()),
  87                     CACHE_SIZE);
  88             if (dictFileVersion >= 5) {
  89                 htmlEntries = CachingList.create(
  90                         RAFList.create(raf, new HtmlEntry.Serializer(this), raf.getFilePointer()),
  91                         CACHE_SIZE);
  92             } else {
  93                 htmlEntries = Collections.emptyList();
  94             }
  95             indices = CachingList.createFullyCached(RAFList.create(raf, indexSerializer,
  96                     raf.getFilePointer()));
  97         } catch (RuntimeException e) {
  98             final IOException ioe = new IOException("RuntimeException loading dictionary");
  99             ioe.initCause(e);
 100             throw ioe;
 101         }
 102         final String end = raf.readUTF();
 103         if (!end.equals(END_OF_DICTIONARY)) {
 104             throw new IOException("Dictionary seems corrupt: " + end);
 105         }
 106     }
 107
 108     @Override
 109     public void write(DataOutput out) throws IOException {
 110         RandomAccessFile raf = (RandomAccessFile)out;
 111         raf.writeInt(dictFileVersion);
 112         raf.writeLong(creationMillis);
 113         raf.writeUTF(dictInfo);
 114         RAFList.write(raf, sources, new EntrySource.Serializer(this));
 115         RAFList.write(raf, pairEntries, new PairEntry.Serializer(this));
 116         RAFList.write(raf, textEntries, new TextEntry.Serializer(this));
 117         RAFList.write(raf, htmlEntries, new HtmlEntry.Serializer(this));
 118         RAFList.write(raf, indices, indexSerializer);
 119         raf.writeUTF(END_OF_DICTIONARY);
 120     }
 121
 122     private final RAFListSerializer<Index> indexSerializer = new RAFListSerializer<Index>() {
 123         @Override
 124         public Index read(DataInput raf, final int readIndex) throws IOException {
 125             return new Index(Dictionary.this, raf);
 126         }
 127
 128         @Override
 129         public void write(DataOutput raf, Index t) throws IOException {
 130             t.write(raf);
 131         }
 132     };
 133
 134     final RAFListSerializer<HtmlEntry> htmlEntryIndexSerializer = new RAFListSerializer<HtmlEntry>() {
 135         @Override
 136         public void write(DataOutput raf, HtmlEntry t) throws IOException {
 137             if (t.index() == -1)
 138                 throw new IndexOutOfBoundsException();
 139             raf.writeInt(t.index());
 140         }
 141
 142         @Override
 143         public HtmlEntry read(DataInput raf, int readIndex) throws IOException {
 144             return htmlEntries.get(raf.readInt());
 145         }
 146     };
 147
 148     public void print(final PrintStream out) {
 149         out.println("dictInfo=" + dictInfo);
 150         for (final EntrySource entrySource : sources) {
 151             out.printf("EntrySource: %s %d\n", entrySource.name, entrySource.numEntries);
 152         }
 153         out.println();
 154         for (final Index index : indices) {
 155             out.printf("Index: %s %s\n", index.shortName, index.longName);
 156             index.print(out);
 157             out.println();
 158         }
 159     }
 160
 161     public DictionaryInfo getDictionaryInfo() {
 162         final DictionaryInfo result = new DictionaryInfo();
 163         result.creationMillis = this.creationMillis;
 164         result.dictInfo = this.dictInfo;
 165         for (final Index index : indices) {
 166             result.indexInfos.add(index.getIndexInfo());
 167         }
 168         return result;
 169     }
 170
 171     public static DictionaryInfo getDictionaryInfo(final File file) {
 172         RandomAccessFile raf = null;
 173         try {
 174             raf = new RandomAccessFile(file, "r");
 175             final Dictionary dict = new Dictionary(raf);
 176             final DictionaryInfo dictionaryInfo = dict.getDictionaryInfo();
 177             dictionaryInfo.uncompressedFilename = file.getName();
 178             dictionaryInfo.uncompressedBytes = file.length();
 179             raf.close();
 180             return dictionaryInfo;
 181         } catch (IOException e) {
 182             return null;
 183         } finally {
 184             if (raf != null) {
 185                 try {
 186                     raf.close();
 187                 } catch (IOException e) {
 188                     e.printStackTrace();
 189                 }
 190             }
 191         }
 192     }
 193
 194 }