1 package com.hughes.android.dictionary.engine;
3 import java.util.ArrayList;
4 import java.util.Collections;
5 import java.util.Comparator;
6 import java.util.EnumMap;
7 import java.util.HashSet;
11 import java.util.SortedMap;
12 import java.util.TreeMap;
14 import com.hughes.android.dictionary.engine.Index.IndexEntry;
17 public class IndexBuilder {
19 final DictionaryBuilder dictionaryBuilder;
20 public final Index index;
22 final SortedMap<String, TokenData> tokenToData;
24 IndexBuilder(final DictionaryBuilder dictionaryBuilder, final String shortName, final String longName, final Language language, final String normalizerRules, final boolean swapPairEntries) {
25 this.dictionaryBuilder = dictionaryBuilder;
26 index = new Index(dictionaryBuilder.dictionary, shortName, longName, language, normalizerRules, swapPairEntries);
27 tokenToData = new TreeMap<String, TokenData>(new NormalizeComparator(index.normalizer, language.collator));
31 final Set<EntryData> tokenEntryDatas = new HashSet<EntryData>();
32 final List<RowBase> rows = index.rows;
33 for (final TokenData tokenData : tokenToData.values()) {
34 tokenEntryDatas.clear();
35 final int indexIndex = index.sortedIndexEntries.size();
36 final int startRow = rows.size();
37 rows.add(new TokenRow(indexIndex, rows.size(), index));
38 // System.out.println("Added TokenRow: " + rows.get(rows.size() - 1));
40 // System.out.println("TOKEN: " + tokenData.token);
41 for (final Map.Entry<EntryTypeName, List<EntryData>> typeToEntry : tokenData.typeToEntries.entrySet()) {
42 for (final EntryData entryData : typeToEntry.getValue()) {
43 if (tokenEntryDatas.add(entryData)) {
44 rows.add(new PairEntry.Row(entryData.index(), rows.size(), index));
47 // System.out.print(" " + typeToEntry.getKey() + ": ");
48 // rows.get(rows.size() - 1).print(System.out);
49 // System.out.println();
53 index.sortedIndexEntries.add(new Index.IndexEntry(tokenData.token, startRow, numRows));
56 final List<IndexEntry> sortedEntries = new ArrayList<IndexEntry>(index.sortedIndexEntries);
57 Collections.sort(sortedEntries, new Comparator<IndexEntry>() {
59 public int compare(IndexEntry object1, IndexEntry object2) {
60 return object2.numRows - object1.numRows;
62 System.out.println("Most common tokens:");
63 for (int i = 0; i < 50 && i < sortedEntries.size(); ++i) {
64 System.out.println(" " + sortedEntries.get(i));
68 static class TokenData {
71 final Map<EntryTypeName, List<EntryData>> typeToEntries = new EnumMap<EntryTypeName, List<EntryData>>(EntryTypeName.class);
73 TokenData(final String token) {
74 assert token.equals(token.trim());
75 assert token.length() > 0;
80 public TokenData getOrCreateTokenData(final String token) {
81 TokenData tokenData = tokenToData.get(token);
82 if (tokenData == null) {
83 tokenData = new TokenData(token);
84 tokenToData.put(token, tokenData);
89 public List<EntryData> getOrCreateEntries(final String token, final EntryTypeName entryTypeName) {
90 final TokenData tokenData = getOrCreateTokenData(token);
91 List<EntryData> entries = tokenData.typeToEntries.get(entryTypeName);
92 if (entries == null) {
93 entries = new ArrayList<EntryData>();
94 tokenData.typeToEntries.put(entryTypeName, entries);