package com.hughes.android.dictionary;\r
\r
+import java.io.FileNotFoundException;\r
import java.io.IOException;\r
import java.io.RandomAccessFile;\r
import java.io.Serializable;\r
import java.util.ArrayList;\r
import java.util.Collections;\r
-import java.util.LinkedHashSet;\r
+import java.util.LinkedHashMap;\r
import java.util.List;\r
import java.util.Map;\r
-import java.util.Set;\r
import java.util.TreeMap;\r
import java.util.regex.Pattern;\r
\r
System.err.println("No input file.");\r
return;\r
}\r
+ final String dictionaryFileName = args[0];\r
+ createIndex(dictionaryFileName, Entry.LANG1);\r
+ createIndex(dictionaryFileName, Entry.LANG2);\r
+ }\r
\r
- final String file = args[0];\r
- final byte lang = Entry.LANG1;\r
+ private static void createIndex(final String dictionaryFileName,\r
+ final byte lang) throws IOException, FileNotFoundException,\r
+ ClassNotFoundException {\r
Node rootBuilder;\r
- rootBuilder = createIndex(file, lang);\r
- FileUtil.write(rootBuilder, String.format("%s_builder_%d.serialized", file, lang));\r
- rootBuilder = (Node) FileUtil.read(String.format("%s_builder_%d.serialized", file, lang));\r
+ rootBuilder = processDictionaryLines(dictionaryFileName, lang);\r
+ FileUtil.write(rootBuilder, String.format("%s_builder_%d.serialized", dictionaryFileName, lang));\r
+ rootBuilder = (Node) FileUtil.read(String.format("%s_builder_%d.serialized", dictionaryFileName, lang));\r
\r
rootBuilder.forEachNode(new Function<Node>() {\r
@Override\r
\r
// Dump twice to get accurate file locations.\r
for (int i = 0; i < 2; ++i) {\r
- final RandomAccessFile raf = new RandomAccessFile(String.format(Dictionary.INDEX_FORMAT, file, lang), "rw"); \r
+ final RandomAccessFile raf = new RandomAccessFile(String.format(Dictionary.INDEX_FORMAT, dictionaryFileName, lang), "rw"); \r
rootBuilder.dump(raf);\r
raf.close();\r
}\r
- \r
}\r
\r
// ----------------------------------------------------------------\r
}\r
}\r
\r
- static final class Node implements Serializable {\r
- final String normalizedWord;\r
- \r
- final TreeMap<String, Node> children = new TreeMap<String, Node>();\r
- final TreeMap<String,List<EntryDescriptor>> entryDescriptorsMap = new TreeMap<String, List<EntryDescriptor>>();\r
- \r
-// final List<EntryDescriptor> offsets = new ArrayList<EntryDescriptor>();\r
- \r
- int descendantOffsetCount = 0;\r
- \r
- int indexFileLocation = -1;\r
-\r
- public Node(final String normalizedWord) {\r
- if (normalizedWord.length() == 0) {\r
- System.out.println("Created root.");\r
- }\r
- this.normalizedWord = normalizedWord.intern();\r
- }\r
-\r
- public Node getNode(final String nWord, final int pos,\r
- final boolean create) {\r
- assert this.normalizedWord.equals(nWord.substring(0, pos));\r
-\r
- if (pos == nWord.length()) {\r
- assert normalizedWord.equals(nWord);\r
- return this;\r
- }\r
-\r
- final String rest = nWord.substring(pos);\r
- assert rest.length() > 0;\r
-\r
- final Map.Entry<String, Node> lcsEntry;\r
- final String lcs;\r
- {\r
- final Map.Entry<String, Node> floorEntry = children.floorEntry(rest);\r
- final Map.Entry<String, Node> ceilingEntry = children\r
- .ceilingEntry(rest);\r
- final String floorLcs = floorEntry == null ? "" : StringUtil\r
- .longestCommonSubstring(rest, floorEntry.getKey());\r
- final String ceilingLcs = ceilingEntry == null ? "" : StringUtil\r
- .longestCommonSubstring(rest, ceilingEntry.getKey());\r
- if (floorLcs.length() > ceilingLcs.length()) {\r
- lcsEntry = floorEntry;\r
- lcs = floorLcs;\r
- } else {\r
- lcsEntry = ceilingEntry;\r
- lcs = ceilingLcs;\r
- }\r
- }\r
-\r
- // No LCS, have to add everything.\r
- if (lcs.length() == 0) {\r
- if (!create) {\r
- return null;\r
- }\r
- final Node result = new Node(nWord);\r
- final Object old = children.put(rest.intern(), result);\r
- assert old == null;\r
- // System.out.println(" Adding final chunk: " + rest);\r
- return result;\r
- }\r
-\r
- assert lcsEntry != null;\r
-\r
- // The map already contained the LCS.\r
- if (lcs.length() == lcsEntry.getKey().length()) {\r
- assert lcs.equals(lcsEntry.getKey());\r
- final Node result = lcsEntry.getValue().getNode(nWord,\r
- pos + lcs.length(), create);\r
- assert result.normalizedWord.equals(nWord);\r
- return result;\r
- }\r
-\r
- if (!create) {\r
- return null;\r
- }\r
-\r
- // Have to split, inserting the LCS.\r
- // System.out.println(" Splitting " + lcsEntry + "/" + word + " @ " +\r
- // lcs);\r
- final Node newChild = new Node(nWord.substring(0, pos + lcs.length()));\r
- final Object old = children.put(lcs.intern(), newChild);\r
- assert old == null;\r
- children.remove(lcsEntry.getKey());\r
- newChild.children.put(lcsEntry.getKey().substring(lcs.length())\r
- .intern(), lcsEntry.getValue());\r
-\r
- if (lcs.equals(rest)) {\r
- return newChild;\r
- }\r
- final Node result = new Node(nWord);\r
- final Object old2 = newChild.children.put(rest.substring(lcs.length())\r
- .intern(), result);\r
- assert old2 == null;\r
- // System.out.println(" newchildren=" + newChild.children);\r
-\r
- return result;\r
- }\r
-\r
- void forEachNode(final Function<Node> f) {\r
- f.invoke(this);\r
- for (final Node child : children.values()) {\r
- child.forEachNode(f);\r
- }\r
- }\r
-\r
- int descendantCount() {\r
- int count = 1;\r
- for (final Node child : children.values()) {\r
- count += child.descendantCount();\r
- }\r
- return count;\r
- }\r
-\r
- void recursiveSetDescendantOffsetCount() {\r
- descendantOffsetCount = offsets.size();\r
- for (final Node child : children.values()) {\r
- child.recursiveSetDescendantOffsetCount();\r
- descendantOffsetCount += child.descendantOffsetCount;\r
- }\r
- }\r
-\r
- @Override\r
- public String toString() {\r
- return normalizedWord;\r
- }\r
- \r
- void dump(final RandomAccessFile file) throws IOException {\r
- if (indexFileLocation == -1) {\r
- indexFileLocation = (int) file.getFilePointer();\r
- } else {\r
- assert indexFileLocation == file.getFilePointer();\r
- }\r
- \r
- // Children.\r
- file.writeInt(children.size());\r
- for (final Map.Entry<String, Node> child : children.entrySet()) {\r
- file.writeUTF(child.getKey());\r
- file.writeInt(child.getValue().indexFileLocation);\r
- }\r
- \r
- // Entries.\r
- file.writeInt(entryDescriptorsMap.size());\r
- for (final Map.Entry<String, List<EntryDescriptor>> entry : entryDescriptorsMap.entrySet()) {\r
- file.writeUTF(entry.getKey());\r
- file.writeInt(entry.getValue().size());\r
- \r
- file.writeInt(offsets.get(i).offset);\r
- }\r
- \r
- // Dump children.\r
- for (final Map.Entry<String, Node> child : children.entrySet()) {\r
- child.getValue().dump(file);\r
- }\r
- }\r
- }\r
\r
// ----------------------------------------------------------------\r
\r
- static Node createIndex(final String file, final byte lang) throws IOException {\r
+ static Node processDictionaryLines(final String dictionaryFileName, final byte lang) throws IOException {\r
final Node root = new Node("");\r
- final RandomAccessFile raf = new RandomAccessFile(file, "r");\r
+ final RandomAccessFile dictionaryFile = new RandomAccessFile(dictionaryFileName, "r");\r
String line;\r
final Entry entry = new Entry();\r
int lineCount = 0;\r
long fileLocation = 0;\r
- while ((line = raf.readLine()) != null) {\r
+ while ((line = dictionaryFile.readLine()) != null) {\r
assert ((int) fileLocation) == fileLocation;\r
\r
line = line.trim();\r
}\r
final String text = entry.getIndexableText(Entry.LANG1);\r
final String[] tokens = WHITESPACE.split(text);\r
- final Set<String> tokenSet = new LinkedHashSet<String>();\r
+ final Map<String,String> tokenToNormalizedMap = new LinkedHashMap<String,String>();\r
for (String token : tokens) {\r
if (token.length() <= 1 || !Character.isLetter(token.charAt(0))) {\r
continue;\r
}\r
- tokenSet.add(EntryFactory.entryFactory.normalizeToken(token, lang));\r
+ tokenToNormalizedMap.put(token, EntryFactory.entryFactory.normalizeToken(token));\r
}\r
- for (final String normalized : tokenSet) {\r
- // System.out.println("Inserting: " + normalized);\r
- if ("die".equals(normalized) || "eine".equals(normalized)) {\r
- // System.out.println("hello");\r
- }\r
- final Node node = root.getNode(normalized, 0, true);\r
- node.offsets.add(new EntryDescriptor((int) fileLocation, tokens.length));\r
- assert node == root.getNode(normalized, 0, false);\r
- assert normalized\r
- .equals(root.getNode(normalized, 0, false).normalizedWord);\r
+ for (final Map.Entry<String, String> tokenToNormalized : tokenToNormalizedMap.entrySet()) {\r
+ final String normalizedToken = tokenToNormalized.getValue();\r
+ final Node node = root.getNode(normalizedToken, 0, true);\r
+ node.addToken(tokenToNormalized.getKey(), new EntryDescriptor((int) fileLocation, tokens.length));\r
+ assert node == root.getNode(normalizedToken, 0, false);\r
+ assert normalizedToken\r
+ .equals(root.getNode(normalizedToken, 0, false).normalizedToken);\r
}\r
\r
if (lineCount % 10000 == 0) {\r
}\r
\r
lineCount++;\r
- fileLocation = raf.getFilePointer();\r
+ fileLocation = dictionaryFile.getFilePointer();\r
}\r
- raf.close();\r
+ dictionaryFile.close();\r
+ \r
+ root.recursiveSetDescendantCounts();\r
+ \r
return root;\r
}\r
\r