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.getCollator()));
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, index
54 .normalizer().transliterate(tokenData.token), startRow, numRows));
57 final List<IndexEntry> sortedEntries = new ArrayList<IndexEntry>(index.sortedIndexEntries);
58 Collections.sort(sortedEntries, new Comparator<IndexEntry>() {
60 public int compare(IndexEntry object1, IndexEntry object2) {
61 return object2.numRows - object1.numRows;
63 System.out.println("Most common tokens:");
64 for (int i = 0; i < 50 && i < sortedEntries.size(); ++i) {
65 System.out.println(" " + sortedEntries.get(i));
69 static class TokenData {
72 final Map<EntryTypeName, List<EntryData>> typeToEntries = new EnumMap<EntryTypeName, List<EntryData>>(EntryTypeName.class);
74 TokenData(final String token) {
75 assert token.equals(token.trim());
76 assert token.length() > 0;
81 public TokenData getOrCreateTokenData(final String token) {
82 TokenData tokenData = tokenToData.get(token);
83 if (tokenData == null) {
84 tokenData = new TokenData(token);
85 tokenToData.put(token, tokenData);
90 public List<EntryData> getOrCreateEntries(final String token, final EntryTypeName entryTypeName) {
91 final TokenData tokenData = getOrCreateTokenData(token);
92 List<EntryData> entries = tokenData.typeToEntries.get(entryTypeName);
93 if (entries == null) {
94 entries = new ArrayList<EntryData>();
95 tokenData.typeToEntries.put(entryTypeName, entries);
100 public void addEntryWithTokens(final EntryData entryData, final Set<String> tokens,
101 final EntryTypeName entryTypeName) {
102 for (final String token : tokens) {
103 getOrCreateEntries(token, entryTypeName).add(entryData);