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     // Could be a local variable in constructor, but
  54     // this way avoids a native-image VM bug.
  55     private final MappedByteBuffer wholefile;
  56
  57     /**
  58      * dictFileVersion 1 adds: <li>links to sources? dictFileVersion 2 adds: <li>
  59      * counts of tokens in indices.
  60      */
  61
  62     public Dictionary(final String dictInfo) {
  63         this.dictFileVersion = CURRENT_DICT_VERSION;
  64         this.creationMillis = System.currentTimeMillis();
  65         this.dictInfo = dictInfo;
  66         pairEntries = new ArrayList<>();
  67         textEntries = new ArrayList<>();
  68         htmlEntries = new ArrayList<>();
  69         htmlData = null;
  70         sources = new ArrayList<>();
  71         indices = new ArrayList<>();
  72         wholefile = null;
  73     }
  74
  75     public Dictionary(final FileChannel ch) throws IOException {
  76         wholefile = ch.map(FileChannel.MapMode.READ_ONLY, 0, ch.size());
  77         DataInputBuffer in = new DataInputBuffer(wholefile, 0);
  78         dictFileVersion = in.readInt();
  79         if (dictFileVersion < 0 || dictFileVersion > CURRENT_DICT_VERSION) {
  80             throw new IOException("Invalid dictionary version: " + dictFileVersion);
  81         }
  82         creationMillis = in.readLong();
  83         dictInfo = in.readUTF();
  84
  85         // Load the sources, then seek past them, because reading them later
  86         // disrupts the offset.
  87         try {
  88             final RAFList<EntrySource> rafSources = RAFList.create(in, new EntrySource.Serializer(
  89                     this), dictFileVersion, dictInfo + " sources: ");
  90             sources = new ArrayList<>(rafSources);
  91             ch.position(rafSources.getEndOffset());
  92
  93             pairEntries = CachingList.create(
  94                               RAFList.create(in, new PairEntry.Serializer(this), dictFileVersion, dictInfo + " pairs: "),
  95                               CACHE_SIZE, false);
  96             textEntries = CachingList.create(
  97                               RAFList.create(in, new TextEntry.Serializer(this), dictFileVersion, dictInfo + " text: "),
  98                               CACHE_SIZE, true);
  99             if (dictFileVersion >= 5) {
 100                 htmlEntries = CachingList.create(
 101                                   RAFList.create(in, new HtmlEntry.Serializer(this), dictFileVersion, dictInfo + " html: "),
 102                                   CACHE_SIZE, false);
 103             } else {
 104                 htmlEntries = Collections.emptyList();
 105             }
 106             if (dictFileVersion >= 7) {
 107                 htmlData = RAFList.create(in, new HtmlEntry.DataDeserializer(), dictFileVersion, dictInfo + " html: ");
 108             } else {
 109                 htmlData = null;
 110             }
 111             indices = CachingList.createFullyCached(RAFList.create(in, new IndexSerializer(),
 112                                                     dictFileVersion, dictInfo + " index: "));
 113         } catch (RuntimeException e) {
 114             throw new IOException("RuntimeException loading dictionary", e);
 115         }
 116         final String end = in.readUTF();
 117         if (!end.equals(END_OF_DICTIONARY)) {
 118             throw new IOException("Dictionary seems corrupt: " + end);
 119         }
 120     }
 121
 122     @Override
 123     public void write(DataOutput out) throws IOException {
 124         RandomAccessFile raf = (RandomAccessFile)out;
 125         if (dictFileVersion < 7) throw new RuntimeException("write function cannot write formats older than v7!");
 126         raf.writeInt(dictFileVersion);
 127         raf.writeLong(creationMillis);
 128         raf.writeUTF(dictInfo);
 129         System.out.println("sources start: " + raf.getFilePointer());
 130         RAFList.write(raf, sources, new EntrySource.Serializer(this));
 131         System.out.println("pair start: " + raf.getFilePointer());
 132         RAFList.write(raf, pairEntries, new PairEntry.Serializer(this), 64, true);
 133         System.out.println("text start: " + raf.getFilePointer());
 134         RAFList.write(raf, textEntries, new TextEntry.Serializer(this));
 135         System.out.println("html index start: " + raf.getFilePointer());
 136         RAFList.write(raf, htmlEntries, new HtmlEntry.Serializer(this), 64, true);
 137         System.out.println("html data start: " + raf.getFilePointer());
 138         assert htmlData == null;
 139         RAFList.write(raf, htmlEntries, new HtmlEntry.DataSerializer(), 128, true);
 140         System.out.println("indices start: " + raf.getFilePointer());
 141         RAFList.write(raf, indices, new IndexSerializer());
 142         System.out.println("end: " + raf.getFilePointer());
 143         raf.writeUTF(END_OF_DICTIONARY);
 144     }
 145
 146     private final class IndexSerializer implements RAFListSerializer<Index> {
 147         @Override
 148         public Index read(DataInput raf, final int readIndex) throws IOException {
 149             return new Index(Dictionary.this, (DataInputBuffer)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 }