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.DataInputStream;
  25 import java.io.DataOutput;
  26 import java.io.File;
  27 import java.io.IOException;
  28 import java.io.PrintStream;
  29 import java.io.RandomAccessFile;
  30 import java.nio.channels.Channels;
  31 import java.nio.channels.FileChannel;
  32 import java.util.ArrayList;
  33 import java.util.Collections;
  34 import java.util.List;
  35
  36 public class Dictionary implements RAFSerializable<Dictionary> {
  37
  38     private static final int CACHE_SIZE = 5000;
  39
  40     private static final int CURRENT_DICT_VERSION = 7;
  41     private static final String END_OF_DICTIONARY = "END OF DICTIONARY";
  42
  43     // persisted
  44     final int dictFileVersion;
  45     private final long creationMillis;
  46     public final String dictInfo;
  47     public final List<PairEntry> pairEntries;
  48     public final List<TextEntry> textEntries;
  49     public final List<HtmlEntry> htmlEntries;
  50     public final List<byte[]> htmlData;
  51     public final List<EntrySource> sources;
  52     public final List<Index> indices;
  53
  54     /**
  55      * dictFileVersion 1 adds: <li>links to sources? dictFileVersion 2 adds: <li>
  56      * counts of tokens in indices.
  57      */
  58
  59     public Dictionary(final String dictInfo) {
  60         this.dictFileVersion = CURRENT_DICT_VERSION;
  61         this.creationMillis = System.currentTimeMillis();
  62         this.dictInfo = dictInfo;
  63         pairEntries = new ArrayList<>();
  64         textEntries = new ArrayList<>();
  65         htmlEntries = new ArrayList<>();
  66         htmlData = null;
  67         sources = new ArrayList<>();
  68         indices = new ArrayList<>();
  69     }
  70
  71     public Dictionary(final FileChannel ch) throws IOException {
  72         DataInput raf = new DataInputStream(Channels.newInputStream(ch));
  73         dictFileVersion = raf.readInt();
  74         if (dictFileVersion < 0 || dictFileVersion > CURRENT_DICT_VERSION) {
  75             throw new IOException("Invalid dictionary version: " + dictFileVersion);
  76         }
  77         creationMillis = raf.readLong();
  78         dictInfo = raf.readUTF();
  79
  80         // Load the sources, then seek past them, because reading them later
  81         // disrupts the offset.
  82         try {
  83             final RAFList<EntrySource> rafSources = RAFList.create(ch, new EntrySource.Serializer(
  84                     this), ch.position(), dictFileVersion, dictInfo + " sources: ");
  85             sources = new ArrayList<>(rafSources);
  86             ch.position(rafSources.getEndOffset());
  87
  88             pairEntries = CachingList.create(
  89                               RAFList.create(ch, new PairEntry.Serializer(this), ch.position(), dictFileVersion, dictInfo + " pairs: "),
  90                               CACHE_SIZE, false);
  91             textEntries = CachingList.create(
  92                               RAFList.create(ch, new TextEntry.Serializer(this), ch.position(), dictFileVersion, dictInfo + " text: "),
  93                               CACHE_SIZE, true);
  94             if (dictFileVersion >= 5) {
  95                 htmlEntries = CachingList.create(
  96                                   RAFList.create(ch, new HtmlEntry.Serializer(this, ch), ch.position(), dictFileVersion, dictInfo + " html: "),
  97                                   CACHE_SIZE, false);
  98             } else {
  99                 htmlEntries = Collections.emptyList();
 100             }
 101             if (dictFileVersion >= 7) {
 102                 htmlData = RAFList.create(ch, new HtmlEntry.DataDeserializer(), ch.position(), dictFileVersion, dictInfo + " html: ");
 103             } else {
 104                 htmlData = null;
 105             }
 106             indices = CachingList.createFullyCached(RAFList.create(ch, new IndexSerializer(ch),
 107                                                     ch.position(), dictFileVersion, dictInfo + " index: "));
 108         } catch (RuntimeException e) {
 109             throw new IOException("RuntimeException loading dictionary", e);
 110         }
 111         final String end = raf.readUTF();
 112         if (!end.equals(END_OF_DICTIONARY)) {
 113             throw new IOException("Dictionary seems corrupt: " + end);
 114         }
 115     }
 116
 117     @Override
 118     public void write(DataOutput out) throws IOException {
 119         RandomAccessFile raf = (RandomAccessFile)out;
 120         raf.writeInt(dictFileVersion);
 121         raf.writeLong(creationMillis);
 122         raf.writeUTF(dictInfo);
 123         System.out.println("sources start: " + raf.getFilePointer());
 124         RAFList.write(raf, sources, new EntrySource.Serializer(this));
 125         System.out.println("pair start: " + raf.getFilePointer());
 126         RAFList.write(raf, pairEntries, new PairEntry.Serializer(this), 64, true);
 127         System.out.println("text start: " + raf.getFilePointer());
 128         RAFList.write(raf, textEntries, new TextEntry.Serializer(this));
 129         System.out.println("html index start: " + raf.getFilePointer());
 130         RAFList.write(raf, htmlEntries, new HtmlEntry.Serializer(this, null), 64, true);
 131         System.out.println("html data start: " + raf.getFilePointer());
 132         assert htmlData == null;
 133         RAFList.write(raf, htmlEntries, new HtmlEntry.DataSerializer(), 128, true);
 134         System.out.println("indices start: " + raf.getFilePointer());
 135         RAFList.write(raf, indices, new IndexSerializer(null));
 136         System.out.println("end: " + raf.getFilePointer());
 137         raf.writeUTF(END_OF_DICTIONARY);
 138     }
 139
 140     private final class IndexSerializer implements RAFListSerializer<Index> {
 141         private final FileChannel ch;
 142
 143         IndexSerializer(FileChannel ch) {
 144             this.ch = ch;
 145         }
 146
 147         @Override
 148         public Index read(DataInput raf, final int readIndex) throws IOException {
 149             return new Index(Dictionary.this, ch, raf);
 150         }
 151
 152         @Override
 153         public void write(DataOutput raf, Index t) throws IOException {
 154             t.write(raf);
 155         }
 156     }
 157
 158     final RAFListSerializer<HtmlEntry> htmlEntryIndexSerializer = new RAFListSerializer<HtmlEntry>() {
 159         @Override
 160         public void write(DataOutput raf, HtmlEntry t) {
 161             assert false;
 162         }
 163
 164         @Override
 165         public HtmlEntry read(DataInput raf, int readIndex) throws IOException {
 166             return htmlEntries.get(raf.readInt());
 167         }
 168     };
 169
 170     public void print(final PrintStream out) {
 171         out.println("dictInfo=" + dictInfo);
 172         for (final EntrySource entrySource : sources) {
 173             out.printf("EntrySource: %s %d\n", entrySource.name, entrySource.numEntries);
 174         }
 175         out.println();
 176         for (final Index index : indices) {
 177             out.printf("Index: %s %s\n", index.shortName, index.longName);
 178             index.print(out);
 179             out.println();
 180         }
 181     }
 182
 183     public DictionaryInfo getDictionaryInfo() {
 184         final DictionaryInfo result = new DictionaryInfo();
 185         result.creationMillis = this.creationMillis;
 186         result.dictInfo = this.dictInfo;
 187         for (final Index index : indices) {
 188             result.indexInfos.add(index.getIndexInfo());
 189         }
 190         return result;
 191     }
 192
 193     public static DictionaryInfo getDictionaryInfo(final File file) {
 194         RandomAccessFile raf = null;
 195         try {
 196             raf = new RandomAccessFile(file, "r");
 197             final Dictionary dict = new Dictionary(raf.getChannel());
 198             final DictionaryInfo dictionaryInfo = dict.getDictionaryInfo();
 199             dictionaryInfo.uncompressedFilename = file.getName();
 200             dictionaryInfo.uncompressedBytes = file.length();
 201             raf.close();
 202             return dictionaryInfo;
 203         } catch (IOException e) {
 204             final DictionaryInfo dictionaryInfo = new DictionaryInfo();
 205             dictionaryInfo.uncompressedFilename = file.getName();
 206             dictionaryInfo.uncompressedBytes = file.length();
 207             return dictionaryInfo;
 208         } finally {
 209             if (raf != null) {
 210                 try {
 211                     raf.close();
 212                 } catch (IOException e) {
 213                     e.printStackTrace();
 214                 }
 215             }
 216         }
 217     }
 218
 219 }