import java.io.RandomAccessFile;\r
import java.util.ArrayList;\r
import java.util.List;\r
+import java.util.concurrent.atomic.AtomicBoolean;\r
\r
import com.hughes.util.CachingList;\r
import com.hughes.util.raf.FileList;\r
import com.hughes.util.raf.RAFSerializable;\r
import com.hughes.util.raf.RAFSerializableSerializer;\r
import com.hughes.util.raf.RAFSerializer;\r
+import com.hughes.util.raf.UniformFileList;\r
\r
public final class Dictionary implements RAFSerializable<Dictionary> {\r
\r
- static final RAFSerializer<Entry> ENTRY_SERIALIZER = new RAFSerializableSerializer<Entry>(Entry.RAF_FACTORY);\r
- static final RAFSerializer<Row> ROW_SERIALIZER = new RAFSerializableSerializer<Row>(Row.RAF_FACTORY);\r
- static final RAFSerializer<IndexEntry> INDEX_ENTRY_SERIALIZER = new RAFSerializableSerializer<IndexEntry>(IndexEntry.RAF_FACTORY);\r
+ private static final String VERSION_CODE = "DictionaryVersion=2.0";\r
\r
+ static final RAFSerializer<SimpleEntry> ENTRY_SERIALIZER = new RAFSerializableSerializer<SimpleEntry>(\r
+ SimpleEntry.RAF_FACTORY);\r
+ static final RAFSerializer<Row> ROW_SERIALIZER = new RAFSerializableSerializer<Row>(\r
+ Row.RAF_FACTORY);\r
+ static final RAFSerializer<IndexEntry> INDEX_ENTRY_SERIALIZER = new RAFSerializableSerializer<IndexEntry>(\r
+ IndexEntry.RAF_FACTORY);\r
+\r
+ final String dictionaryInfo;\r
+ final List<String> sources;\r
final List<Entry> entries;\r
- final Language[] languages = new Language[2];\r
- \r
- Language activeLanguage = null;\r
+ final LanguageData[] languageDatas = new LanguageData[2];\r
\r
- public Dictionary(final String lang0, final String lang1) {\r
- languages[0] = new Language(lang0);\r
- languages[1] = new Language(lang1);\r
+ public Dictionary(final String dictionaryInfo, final Language language0, final Language language1) {\r
+ this.dictionaryInfo = dictionaryInfo;\r
+ sources = new ArrayList<String>();\r
+ languageDatas[0] = new LanguageData(this, language0, SimpleEntry.LANG1);\r
+ languageDatas[1] = new LanguageData(this, language1, SimpleEntry.LANG2);\r
entries = new ArrayList<Entry>();\r
}\r
- \r
+\r
public Dictionary(final RandomAccessFile raf) throws IOException {\r
- entries = CachingList.create(FileList.create(raf, ENTRY_SERIALIZER, raf.getFilePointer()), 10000);\r
- languages[0] = new Language(raf);\r
- languages[1] = new Language(raf);\r
+ dictionaryInfo = raf.readUTF();\r
+ sources = new ArrayList<String>(FileList.create(raf, RAFSerializer.STRING, raf.getFilePointer()));\r
+ entries = CachingList.create(FileList.create(raf, ENTRY_SERIALIZER, raf\r
+ .getFilePointer()), 10000);\r
+ languageDatas[0] = new LanguageData(this, raf, SimpleEntry.LANG1);\r
+ languageDatas[1] = new LanguageData(this, raf, SimpleEntry.LANG2);\r
+ final String version = raf.readUTF();\r
+ if (!VERSION_CODE.equals(version)) {\r
+ throw new IOException("Invalid dictionary version, found " + version + ", expected: " + VERSION_CODE);\r
+ }\r
}\r
+\r
public void write(RandomAccessFile raf) throws IOException {\r
+ raf.writeUTF(dictionaryInfo);\r
+ FileList.write(raf, sources, RAFSerializer.STRING);\r
FileList.write(raf, entries, ENTRY_SERIALIZER);\r
- languages[0].write(raf);\r
- languages[1].write(raf);\r
+ languageDatas[0].write(raf);\r
+ languageDatas[1].write(raf);\r
+ raf.writeUTF(VERSION_CODE);\r
}\r
- \r
- static final class Language implements RAFSerializable<Language> {\r
- final String symbol;\r
+\r
+ final class LanguageData implements RAFSerializable<LanguageData> {\r
+ final Dictionary dictionary;\r
+ final Language language;\r
+ final byte lang;\r
final List<Row> rows;\r
final List<IndexEntry> sortedIndex;\r
- \r
- public Language(final String symbol) {\r
- this.symbol = symbol;\r
+\r
+ LanguageData(final Dictionary dictionary, final Language language, final byte lang) {\r
+ this.dictionary = dictionary;\r
+ this.language = language;\r
+ this.lang = lang;\r
rows = new ArrayList<Row>();\r
sortedIndex = new ArrayList<IndexEntry>();\r
}\r
\r
- public Language(final RandomAccessFile raf) throws IOException {\r
- symbol = raf.readUTF();\r
- rows = CachingList.create(FileList.create(raf, ROW_SERIALIZER, raf.getFilePointer()), 10000);\r
- sortedIndex = CachingList.create(FileList.create(raf, INDEX_ENTRY_SERIALIZER, raf.getFilePointer()), 10000);\r
+ LanguageData(final Dictionary dictionary, final RandomAccessFile raf, final byte lang) throws IOException {\r
+ this.dictionary = dictionary;\r
+ language = Language.lookup(raf.readUTF());\r
+ if (language == null) {\r
+ throw new RuntimeException("Unknown language.");\r
+ }\r
+ this.lang = lang;\r
+ rows = CachingList.create(UniformFileList.create(raf, ROW_SERIALIZER, raf\r
+ .getFilePointer()), 10000);\r
+ sortedIndex = CachingList.create(FileList.create(raf,\r
+ INDEX_ENTRY_SERIALIZER, raf.getFilePointer()), 10000);\r
}\r
+\r
public void write(final RandomAccessFile raf) throws IOException {\r
- raf.writeUTF(symbol);\r
- FileList.write(raf, rows, ROW_SERIALIZER);\r
+ raf.writeUTF(language.symbol);\r
+ UniformFileList.write(raf, rows, ROW_SERIALIZER, 4);\r
FileList.write(raf, sortedIndex, INDEX_ENTRY_SERIALIZER);\r
}\r
+\r
+ String rowToString(final Row row, final boolean onlyFirstSubentry) {\r
+ return row.isToken() ? sortedIndex.get(row.getIndex()).word : entries\r
+ .get(row.getIndex()).getRawText(onlyFirstSubentry);\r
+ }\r
+\r
+ int lookup(String word, final AtomicBoolean interrupted) {\r
+ word = word.toLowerCase();\r
+\r
+ int start = 0;\r
+ int end = sortedIndex.size();\r
+ while (start < end) {\r
+ final int mid = (start + end) / 2;\r
+ if (interrupted.get()) {\r
+ return mid;\r
+ }\r
+ final IndexEntry midEntry = sortedIndex.get(mid);\r
+ if (midEntry.word.equals("pre-print")) {\r
+ System.out.println();\r
+ }\r
+\r
+ final int comp = language.sortComparator.compare(word, midEntry.word.toLowerCase());\r
+ if (comp == 0) {\r
+ int result = mid;\r
+ while (result > 0 && language.findComparator.compare(word, sortedIndex.get(result - 1).word.toLowerCase()) == 0) {\r
+ --result;\r
+ if (interrupted.get()) {\r
+ return result;\r
+ }\r
+ }\r
+ return result;\r
+ } else if (comp < 0) {\r
+// Log.d("THAD", "Upper bound: " + midEntry);\r
+ end = mid;\r
+ } else {\r
+// Log.d("THAD", "Lower bound: " + midEntry);\r
+ start = mid + 1;\r
+ }\r
+ }\r
+ return Math.min(sortedIndex.size() - 1, start);\r
+ }\r
+ \r
+ public int getPrevTokenRow(final int rowIndex) {\r
+ final IndexEntry indexEntry = getIndexEntryForRow(rowIndex);\r
+ final Row tokenRow = rows.get(indexEntry.startRow);\r
+ assert tokenRow.isToken();\r
+ final int prevTokenIndex = tokenRow.getIndex() - 1;\r
+ if (indexEntry.startRow == rowIndex && prevTokenIndex >= 0) {\r
+ return sortedIndex.get(prevTokenIndex).startRow;\r
+ }\r
+ return indexEntry.startRow;\r
+ }\r
+\r
+ public int getNextTokenRow(final int rowIndex) {\r
+ final IndexEntry indexEntry = getIndexEntryForRow(rowIndex);\r
+ final Row tokenRow = rows.get(indexEntry.startRow);\r
+ assert tokenRow.isToken();\r
+ final int nextTokenIndex = tokenRow.getIndex() + 1;\r
+ if (nextTokenIndex < sortedIndex.size()) {\r
+ return sortedIndex.get(nextTokenIndex).startRow;\r
+ }\r
+ return rows.size() - 1;\r
+ }\r
+\r
+ public IndexEntry getIndexEntryForRow(final int rowIndex) {\r
+ // TODO: this kinda blows.\r
+ int r = rowIndex;\r
+ Row row;\r
+ while (true) {\r
+ row = rows.get(r); \r
+ if (row.isToken() || row.indexEntry != null) {\r
+ break;\r
+ }\r
+ --r;\r
+ }\r
+ final IndexEntry indexEntry = row.isToken() ? sortedIndex.get(row.getIndex()) : row.indexEntry;\r
+ for (; r <= rowIndex; ++r) {\r
+ rows.get(r).indexEntry = indexEntry;\r
+ }\r
+ assert rows.get(indexEntry.startRow).isToken();\r
+ return indexEntry;\r
+ }\r
}\r
- \r
+\r
public static final class Row implements RAFSerializable<Row> {\r
final int index;\r
\r
- public Row(int index) {\r
+ IndexEntry indexEntry = null;\r
+\r
+ public Row(final int index) {\r
this.index = index;\r
}\r
\r
static final RAFFactory<Row> RAF_FACTORY = new RAFFactory<Row>() {\r
public Row create(RandomAccessFile raf) throws IOException {\r
return new Row(raf.readInt());\r
- }};\r
+ }\r
+ };\r
+\r
public void write(RandomAccessFile raf) throws IOException {\r
raf.writeInt(index);\r
}\r
+\r
+ boolean isToken() {\r
+ return index < 0;\r
+ }\r
+\r
+ public int getIndex() {\r
+ if (index >= 0) {\r
+ return index;\r
+ }\r
+ return -index - 1;\r
+ }\r
}\r
\r
public static final class IndexEntry implements RAFSerializable<IndexEntry> {\r
final String word;\r
final int startRow;\r
- \r
+\r
public IndexEntry(final String word, final int startRow) {\r
this.word = word;\r
this.startRow = startRow;\r
final String word = raf.readUTF();\r
final int startRow = raf.readInt();\r
return new IndexEntry(word, startRow);\r
- }};\r
+ }\r
+ };\r
+\r
public void write(final RandomAccessFile raf) throws IOException {\r
raf.writeUTF(word);\r
raf.writeInt(startRow);\r
public String toString() {\r
return word + "@" + startRow;\r
}\r
- \r
- \r
+\r
}\r
\r
}\r