X-Git-Url: http://gitweb.fperrin.net/?p=Dictionary.git;a=blobdiff_plain;f=src%2Fcom%2Fhughes%2Fandroid%2Fdictionary%2Fengine%2FIndex.java;h=b58384d1d4981849cb7966c9ca923e507c8d6bae;hp=c254d303dcdbe452829d3e9b8e7494538a4f09fc;hb=e79165503392ed6a7cb7a8eadc15eaae0cda9443;hpb=49583e178d3673fd1cade2f306d97303ffbd4a0a diff --git a/src/com/hughes/android/dictionary/engine/Index.java b/src/com/hughes/android/dictionary/engine/Index.java index c254d30..b58384d 100644 --- a/src/com/hughes/android/dictionary/engine/Index.java +++ b/src/com/hughes/android/dictionary/engine/Index.java @@ -12,10 +12,6 @@ // See the License for the specific language governing permissions and // limitations under the License. -/** - * - */ - package com.hughes.android.dictionary.engine; import com.hughes.android.dictionary.DictionaryInfo; @@ -32,12 +28,10 @@ import com.hughes.util.raf.UniformRAFList; import com.ibm.icu.text.Transliterator; import java.io.DataInput; -import java.io.DataInputStream; import java.io.DataOutput; import java.io.IOException; import java.io.PrintStream; import java.io.RandomAccessFile; -import java.nio.channels.Channels; import java.nio.channels.FileChannel; import java.util.AbstractList; import java.util.ArrayList; @@ -55,7 +49,7 @@ import java.util.regex.Pattern; public final class Index implements RAFSerializable { - static final int CACHE_SIZE = 5000; + private static final int CACHE_SIZE = 5000; public final Dictionary dict; @@ -64,7 +58,7 @@ public final class Index implements RAFSerializable { // persisted: tells how the entries are sorted. public final Language sortLanguage; - final String normalizerRules; + public final String normalizerRules; // Built from the two above. private Transliterator normalizer; @@ -82,7 +76,8 @@ public final class Index implements RAFSerializable { public final boolean swapPairEntries; // Version 2: - int mainTokenCount = -1; + @SuppressWarnings("WeakerAccess") + public int mainTokenCount = -1; // -------------------------------------------------------------------------- @@ -95,9 +90,9 @@ public final class Index implements RAFSerializable { this.sortLanguage = sortLanguage; this.normalizerRules = normalizerRules; this.swapPairEntries = swapPairEntries; - sortedIndexEntries = new ArrayList(); + sortedIndexEntries = new ArrayList<>(); this.stoplist = stoplist; - rows = new ArrayList(); + rows = new ArrayList<>(); normalizer = null; } @@ -105,6 +100,7 @@ public final class Index implements RAFSerializable { /** * Deferred initialization because it can be slow. */ + @SuppressWarnings("WeakerAccess") public synchronized Transliterator normalizer() { if (normalizer == null) { normalizer = TransliteratorManager.get(normalizerRules); @@ -116,6 +112,7 @@ public final class Index implements RAFSerializable { * Note that using this comparator probably involves doing too many text * normalizations. */ + @SuppressWarnings("WeakerAccess") public NormalizeComparator getSortComparator() { return new NormalizeComparator(normalizer(), sortLanguage.getCollator(), dict.dictFileVersion); } @@ -128,18 +125,15 @@ public final class Index implements RAFSerializable { sortLanguage = Language.lookup(languageCode); normalizerRules = raf.readUTF(); swapPairEntries = raf.readBoolean(); - if (sortLanguage == null) { - throw new IOException("Unsupported language: " + languageCode); - } if (dict.dictFileVersion >= 2) { mainTokenCount = raf.readInt(); } sortedIndexEntries = CachingList.create( RAFList.create(inp, new IndexEntrySerializer(dict.dictFileVersion == 6 ? inp : null), inp.position(), - dict.dictFileVersion, dict.dictInfo + " idx " + languageCode + ": "), CACHE_SIZE); + dict.dictFileVersion, dict.dictInfo + " idx " + languageCode + ": "), CACHE_SIZE, true); if (dict.dictFileVersion >= 7) { int count = StringUtil.readVarInt(raf); - stoplist = new HashSet(count); + stoplist = new HashSet<>(count); for (int i = 0; i < count; ++i) { stoplist.add(raf.readUTF()); } @@ -150,7 +144,7 @@ public final class Index implements RAFSerializable { } rows = CachingList.create( UniformRAFList.create(inp, new RowBase.Serializer(this), inp.position()), - CACHE_SIZE); + CACHE_SIZE, true); } @Override @@ -161,9 +155,7 @@ public final class Index implements RAFSerializable { raf.writeUTF(sortLanguage.getIsoCode()); raf.writeUTF(normalizerRules); raf.writeBoolean(swapPairEntries); - if (dict.dictFileVersion >= 2) { - raf.writeInt(mainTokenCount); - } + raf.writeInt(mainTokenCount); RAFList.write(raf, sortedIndexEntries, new IndexEntrySerializer(null), 32, true); StringUtil.writeVarInt(raf, stoplist.size()); for (String i : stoplist) { @@ -181,7 +173,7 @@ public final class Index implements RAFSerializable { private final class IndexEntrySerializer implements RAFSerializer { private final FileChannel ch; - public IndexEntrySerializer(FileChannel ch) { + IndexEntrySerializer(FileChannel ch) { this.ch = ch; } @@ -194,13 +186,13 @@ public final class Index implements RAFSerializable { public void write(DataOutput raf, IndexEntry t) throws IOException { t.write(raf); } - }; + } public static final class IndexEntry implements RAFSerializable { public final String token; private final String normalizedToken; public final int startRow; - public final int numRows; // doesn't count the token row! + final int numRows; // doesn't count the token row! public List htmlEntries; public IndexEntry(final Index index, final String token, final String normalizedToken, @@ -211,10 +203,10 @@ public final class Index implements RAFSerializable { this.normalizedToken = normalizedToken; this.startRow = startRow; this.numRows = numRows; - this.htmlEntries = new ArrayList(); + this.htmlEntries = new ArrayList<>(); } - public IndexEntry(final Index index, final FileChannel ch, final DataInput raf) throws IOException { + IndexEntry(final Index index, final FileChannel ch, final DataInput raf) throws IOException { token = raf.readUTF(); if (index.dict.dictFileVersion >= 7) { startRow = StringUtil.readVarInt(raf); @@ -249,7 +241,7 @@ public final class Index implements RAFSerializable { this.htmlEntries = CachingList.create( RAFList.create(ch, index.dict.htmlEntryIndexSerializer, ch.position(), index.dict.dictFileVersion, - index.dict.dictInfo + " htmlEntries: "), 1); + index.dict.dictInfo + " htmlEntries: "), 1, false); } else { this.htmlEntries = Collections.emptyList(); } @@ -273,12 +265,12 @@ public final class Index implements RAFSerializable { return String.format("%s@%d(%d)", token, startRow, numRows); } - public String normalizedToken() { + String normalizedToken() { return normalizedToken; } } - static final TransformingList.Transformer INDEX_ENTRY_TO_TOKEN = new TransformingList.Transformer() { + private static final TransformingList.Transformer INDEX_ENTRY_TO_TOKEN = new TransformingList.Transformer() { @Override public String transform(IndexEntry t1) { return t1.token; @@ -300,18 +292,32 @@ public final class Index implements RAFSerializable { return index != -1 ? sortedIndexEntries.get(index) : null; } - private int compareIdx(String token, final Comparator sortCollator, int idx) { + private int compareIdx(String token, final Comparator sortCollator, int idx) { final IndexEntry entry = sortedIndexEntries.get(idx); return NormalizeComparator.compareWithoutDash(token, entry.normalizedToken(), sortCollator, dict.dictFileVersion); } - public int findInsertionPointIndex(String token, final AtomicBoolean interrupted) { + private int findMatchLen(final Comparator sortCollator, String a, String b) { + int start = 0; + int end = Math.min(a.length(), b.length()); + while (start < end) + { + int mid = (start + end + 1) / 2; + if (sortCollator.compare(a.substring(0, mid), b.substring(0, mid)) == 0) + start = mid; + else + end = mid - 1; + } + return start; + } + + private int findInsertionPointIndex(String token, final AtomicBoolean interrupted) { token = normalizeToken(token); int start = 0; int end = sortedIndexEntries.size(); - final Comparator sortCollator = sortLanguage.getCollator(); + final Comparator sortCollator = sortLanguage.getCollator(); while (start < end) { final int mid = (start + end) / 2; if (interrupted.get()) { @@ -323,8 +329,7 @@ public final class Index implements RAFSerializable { if (comp == 0) comp = sortCollator.compare(token, midEntry.normalizedToken()); if (comp == 0) { - final int result = windBackCase(token, mid, interrupted); - return result; + return windBackCase(token, mid, interrupted); } else if (comp < 0) { // System.out.println("Upper bound: " + midEntry + ", norm=" + // midEntry.normalizedToken() + ", mid=" + mid); @@ -352,6 +357,15 @@ public final class Index implements RAFSerializable { } } + // if the word before is the better match, move + // our result to it + if (start > 0 && start < sortedIndexEntries.size()) { + String prev = sortedIndexEntries.get(start - 1).normalizedToken(); + String next = sortedIndexEntries.get(start).normalizedToken(); + if (findMatchLen(sortCollator, token, prev) >= findMatchLen(sortCollator, token, next)) + start--; + } + // If we search for a substring of a string that's in there, return // that. int result = Math.min(start, sortedIndexEntries.size() - 1); @@ -375,7 +389,7 @@ public final class Index implements RAFSerializable { private static final int MAX_SEARCH_ROWS = 1000; - private final Map prefixToNumRows = new HashMap(); + private final Map prefixToNumRows = new HashMap<>(); private synchronized final int getUpperBoundOnRowsStartingWith(final String normalizedPrefix, final int maxRows, final AtomicBoolean interrupted) { @@ -391,7 +405,8 @@ public final class Index implements RAFSerializable { return -1; } final IndexEntry indexEntry = sortedIndexEntries.get(index); - if (!indexEntry.normalizedToken.startsWith(normalizedPrefix)) { + if (!indexEntry.normalizedToken.startsWith(normalizedPrefix) && + !NormalizeComparator.withoutDash(indexEntry.normalizedToken).startsWith(normalizedPrefix)) { break; } rowCount += indexEntry.numRows + indexEntry.htmlEntries.size(); @@ -400,7 +415,7 @@ public final class Index implements RAFSerializable { break; } } - prefixToNumRows.put(normalizedPrefix, numRows); + prefixToNumRows.put(normalizedPrefix, rowCount); return rowCount; } @@ -408,9 +423,9 @@ public final class Index implements RAFSerializable { final String searchText, final List searchTokens, final AtomicBoolean interrupted) { final long startMills = System.currentTimeMillis(); - final List result = new ArrayList(); + final List result = new ArrayList<>(); - final Set normalizedNonStoplist = new HashSet(); + final Set normalizedNonStoplist = new HashSet<>(); String bestPrefix = null; int leastRows = Integer.MAX_VALUE; @@ -454,8 +469,8 @@ public final class Index implements RAFSerializable { + ", searchTokens=" + searchTokens); // Place to store the things that match. - final Map> matches = new EnumMap>( - RowMatchType.class); + final Map> matches = new EnumMap<>( + RowMatchType.class); for (final RowMatchType rowMatchType : RowMatchType.values()) { if (rowMatchType != RowMatchType.NO_MATCH) { matches.put(rowMatchType, new ArrayList()); @@ -474,14 +489,15 @@ public final class Index implements RAFSerializable { final String searchToken = bestPrefix; final int insertionPointIndex = findInsertionPointIndex(searchToken, interrupted); - final Set rowsAlreadySeen = new HashSet(); + final Set rowsAlreadySeen = new HashSet<>(); for (int index = insertionPointIndex; index < sortedIndexEntries.size() && matchCount < MAX_SEARCH_ROWS; ++index) { if (interrupted.get()) { return null; } final IndexEntry indexEntry = sortedIndexEntries.get(index); - if (!indexEntry.normalizedToken.startsWith(searchToken)) { + if (!indexEntry.normalizedToken.startsWith(searchToken) && + !NormalizeComparator.withoutDash(indexEntry.normalizedToken).startsWith(searchToken)) { break; } @@ -514,7 +530,7 @@ public final class Index implements RAFSerializable { final RowBase.LengthComparator lengthComparator = new RowBase.LengthComparator( swapPairEntries); for (final Collection rows : matches.values()) { - final List ordered = new ArrayList(rows); + final List ordered = new ArrayList<>(rows); Collections.sort(ordered, lengthComparator); result.addAll(ordered); }