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