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 java.io.DataInput;
  18 import java.io.DataInputStream;
  19 import java.io.DataOutput;
  20 import java.io.File;
  21 import java.io.IOException;
  22 import java.io.ObjectOutputStream;
  23 import java.io.PrintStream;
  24 import java.io.RandomAccessFile;
  25 import java.nio.channels.Channels;
  26 import java.nio.channels.FileChannel;
  27 import java.nio.charset.StandardCharsets;
  28 import java.util.ArrayList;
  29 import java.util.Collections;
  30 import java.util.List;
  31
  32 import com.hughes.android.dictionary.DictionaryInfo;
  33 import com.hughes.util.CachingList;
  34 import com.hughes.util.raf.RAFList;
  35 import com.hughes.util.raf.RAFListSerializer;
  36 import com.hughes.util.raf.RAFSerializable;
  37
  38 public class Dictionary implements RAFSerializable<Dictionary> {
  39
  40     private static final int CACHE_SIZE = 5000;
  41
  42     private static final int CURRENT_DICT_VERSION = 7;
  43     private static final String END_OF_DICTIONARY = "END OF DICTIONARY";
  44
  45     // persisted
  46     final int dictFileVersion;
  47     public final long creationMillis;
  48     public final String dictInfo;
  49     public final List<PairEntry> pairEntries;
  50     public final List<TextEntry> textEntries;
  51     public final List<HtmlEntry> htmlEntries;
  52     public final List<byte[]> htmlData;
  53     public final List<EntrySource> sources;
  54     public final List<Index> indices;
  55
  56     /**
  57      * dictFileVersion 1 adds: <li>links to sources? dictFileVersion 2 adds: <li>
  58      * counts of tokens in indices.
  59      */
  60
  61     public Dictionary(final String dictInfo) {
  62         this.dictFileVersion = CURRENT_DICT_VERSION;
  63         this.creationMillis = System.currentTimeMillis();
  64         this.dictInfo = dictInfo;
  65         pairEntries = new ArrayList<>();
  66         textEntries = new ArrayList<>();
  67         htmlEntries = new ArrayList<>();
  68         htmlData = null;
  69         sources = new ArrayList<>();
  70         indices = new ArrayList<>();
  71     }
  72
  73     public Dictionary(final FileChannel ch) throws IOException {
  74         DataInput raf = new DataInputStream(Channels.newInputStream(ch));
  75         dictFileVersion = raf.readInt();
  76         if (dictFileVersion < 0 || dictFileVersion > CURRENT_DICT_VERSION) {
  77             throw new IOException("Invalid dictionary version: " + dictFileVersion);
  78         }
  79         creationMillis = raf.readLong();
  80         dictInfo = raf.readUTF();
  81
  82         // Load the sources, then seek past them, because reading them later
  83         // disrupts the offset.
  84         try {
  85             final RAFList<EntrySource> rafSources = RAFList.create(ch, new EntrySource.Serializer(
  86                     this), ch.position(), dictFileVersion, dictInfo + " sources: ");
  87             sources = new ArrayList<>(rafSources);
  88             ch.position(rafSources.getEndOffset());
  89
  90             pairEntries = CachingList.create(
  91                               RAFList.create(ch, new PairEntry.Serializer(this), ch.position(), dictFileVersion, dictInfo + " pairs: "),
  92                               CACHE_SIZE, false);
  93             textEntries = CachingList.create(
  94                               RAFList.create(ch, new TextEntry.Serializer(this), ch.position(), dictFileVersion, dictInfo + " text: "),
  95                               CACHE_SIZE, true);
  96             if (dictFileVersion >= 5) {
  97                 htmlEntries = CachingList.create(
  98                                   RAFList.create(ch, new HtmlEntry.Serializer(this, ch), ch.position(), dictFileVersion, dictInfo + " html: "),
  99                                   CACHE_SIZE, false);
 100             } else {
 101                 htmlEntries = Collections.emptyList();
 102             }
 103             if (dictFileVersion >= 7) {
 104                 htmlData = RAFList.create(ch, new HtmlEntry.DataDeserializer(), ch.position(), dictFileVersion, dictInfo + " html: ");
 105             } else {
 106                 htmlData = null;
 107             }
 108             indices = CachingList.createFullyCached(RAFList.create(ch, new IndexSerializer(ch),
 109                                                     ch.position(), dictFileVersion, dictInfo + " index: "));
 110         } catch (RuntimeException e) {
 111             throw new IOException("RuntimeException loading dictionary", e);
 112         }
 113         final String end = raf.readUTF();
 114         if (!end.equals(END_OF_DICTIONARY)) {
 115             throw new IOException("Dictionary seems corrupt: " + end);
 116         }
 117     }
 118
 119     @Override
 120     public void write(DataOutput out) throws IOException {
 121         RandomAccessFile raf = (RandomAccessFile)out;
 122         if (dictFileVersion < 7) throw new RuntimeException("write function cannot write formats older than v7!");
 123         raf.writeInt(dictFileVersion);
 124         raf.writeLong(creationMillis);
 125         raf.writeUTF(dictInfo);
 126         System.out.println("sources start: " + raf.getFilePointer());
 127         RAFList.write(raf, sources, new EntrySource.Serializer(this));
 128         System.out.println("pair start: " + raf.getFilePointer());
 129         RAFList.write(raf, pairEntries, new PairEntry.Serializer(this), 64, true);
 130         System.out.println("text start: " + raf.getFilePointer());
 131         RAFList.write(raf, textEntries, new TextEntry.Serializer(this));
 132         System.out.println("html index start: " + raf.getFilePointer());
 133         RAFList.write(raf, htmlEntries, new HtmlEntry.Serializer(this, null), 64, true);
 134         System.out.println("html data start: " + raf.getFilePointer());
 135         assert htmlData == null;
 136         RAFList.write(raf, htmlEntries, new HtmlEntry.DataSerializer(), 128, true);
 137         System.out.println("indices start: " + raf.getFilePointer());
 138         RAFList.write(raf, indices, new IndexSerializer(null));
 139         System.out.println("end: " + raf.getFilePointer());
 140         raf.writeUTF(END_OF_DICTIONARY);
 141     }
 142
 143     private final class IndexSerializer implements RAFListSerializer<Index> {
 144         private final FileChannel ch;
 145
 146         IndexSerializer(FileChannel ch) {
 147             this.ch = ch;
 148         }
 149
 150         @Override
 151         public Index read(DataInput raf, final int readIndex) throws IOException {
 152             return new Index(Dictionary.this, ch, raf);
 153         }
 154
 155         @Override
 156         public void write(DataOutput raf, Index t) throws IOException {
 157             t.write(raf);
 158         }
 159     }
 160
 161     final RAFListSerializer<HtmlEntry> htmlEntryIndexSerializer = new RAFListSerializer<HtmlEntry>() {
 162         @Override
 163         public void write(DataOutput raf, HtmlEntry t) {
 164             assert false;
 165         }
 166
 167         @Override
 168         public HtmlEntry read(DataInput raf, int readIndex) throws IOException {
 169             return htmlEntries.get(raf.readInt());
 170         }
 171     };
 172
 173     public void print(final PrintStream out) {
 174         out.println("dictInfo=" + dictInfo);
 175         for (final EntrySource entrySource : sources) {
 176             out.printf("EntrySource: %s %d\n", entrySource.name, entrySource.numEntries);
 177         }
 178         out.println();
 179         for (final Index index : indices) {
 180             out.printf("Index: %s %s\n", index.shortName, index.longName);
 181             index.print(out);
 182             out.println();
 183         }
 184     }
 185
 186     public DictionaryInfo getDictionaryInfo() {
 187         final DictionaryInfo result = new DictionaryInfo();
 188         result.creationMillis = this.creationMillis;
 189         result.dictInfo = this.dictInfo;
 190         for (final Index index : indices) {
 191             result.indexInfos.add(index.getIndexInfo());
 192         }
 193         return result;
 194     }
 195
 196     public static DictionaryInfo getDictionaryInfo(final File file) {
 197         RandomAccessFile raf = null;
 198         try {
 199             raf = new RandomAccessFile(file, "r");
 200             final Dictionary dict = new Dictionary(raf.getChannel());
 201             final DictionaryInfo dictionaryInfo = dict.getDictionaryInfo();
 202             dictionaryInfo.uncompressedFilename = file.getName();
 203             dictionaryInfo.uncompressedBytes = file.length();
 204             raf.close();
 205             return dictionaryInfo;
 206         } catch (IOException e) {
 207             final DictionaryInfo dictionaryInfo = new DictionaryInfo();
 208             dictionaryInfo.uncompressedFilename = file.getName();
 209             dictionaryInfo.uncompressedBytes = file.length();
 210             return dictionaryInfo;
 211         } finally {
 212             if (raf != null) {
 213                 try {
 214                     raf.close();
 215                 } catch (IOException e) {
 216                     e.printStackTrace();
 217                 }
 218             }
 219         }
 220     }
 221
 222 }