import com.hughes.util.StringUtil;
import com.hughes.util.raf.RAFListSerializer;
-import com.hughes.util.raf.RAFSerializable;
+import com.hughes.util.raf.RAFListSerializerSkippable;
import com.ibm.icu.text.Transliterator;
import java.io.DataInput;
import java.io.DataOutput;
import java.io.IOException;
import java.io.PrintStream;
-import java.io.RandomAccessFile;
+import java.io.UnsupportedEncodingException;
import java.lang.ref.SoftReference;
+import java.nio.channels.FileChannel;
+import java.nio.charset.StandardCharsets;
import java.util.List;
import java.util.regex.Pattern;
-public class HtmlEntry extends AbstractEntry implements RAFSerializable<HtmlEntry>,
- Comparable<HtmlEntry> {
+public class HtmlEntry extends AbstractEntry implements Comparable<HtmlEntry> {
// Title is not HTML escaped.
public final String title;
- public final LazyHtmlLoader lazyHtmlLoader;
+ private final LazyHtmlLoader lazyHtmlLoader;
+ @SuppressWarnings("WeakerAccess")
public String html;
public HtmlEntry(final EntrySource entrySource, String title) {
lazyHtmlLoader = null;
}
- public HtmlEntry(Dictionary dictionary, DataInput raf, final int index)
- throws IOException {
+ public HtmlEntry(Dictionary dictionary, FileChannel ch, DataInput raf, final int index)
+ throws IOException {
super(dictionary, raf, index);
title = raf.readUTF();
- lazyHtmlLoader = new LazyHtmlLoader(raf, dictionary.dictFileVersion);
+ lazyHtmlLoader = new LazyHtmlLoader(ch, raf, dictionary.htmlData, index);
html = null;
}
- @Override
- public void write(DataOutput raf) throws IOException {
+ private void writeBase(DataOutput raf) throws IOException {
super.write(raf);
raf.writeUTF(title);
+ }
+
+ private void writeData(DataOutput raf) throws IOException {
+ final byte[] bytes = getHtml().getBytes(StandardCharsets.UTF_8);
+ StringUtil.writeVarInt(raf, bytes.length);
+ raf.write(bytes);
+ }
- final byte[] bytes = getHtml().getBytes("UTF-8");
- final byte[] zipBytes = StringUtil.zipBytes(bytes);
- StringUtil.writeVarInt(raf, zipBytes.length);
- raf.write(zipBytes);
+ private static byte[] readData(DataInput raf) throws IOException {
+ int len = StringUtil.readVarInt(raf);
+ final byte[] bytes = new byte[Math.min(len, 20 * 1024 * 1024)];
+ raf.readFully(bytes);
+ return bytes;
}
String getHtml() {
return new Row(this.index, rowIndex, dictionaryIndex);
}
- static final class Serializer implements RAFListSerializer<HtmlEntry> {
+ static final class Serializer implements RAFListSerializerSkippable<HtmlEntry> {
final Dictionary dictionary;
+ final FileChannel ch;
- Serializer(Dictionary dictionary) {
+ Serializer(Dictionary dictionary, FileChannel ch) {
this.dictionary = dictionary;
+ this.ch = ch;
}
@Override
public HtmlEntry read(DataInput raf, final int index) throws IOException {
- return new HtmlEntry(dictionary, raf, index);
+ return new HtmlEntry(dictionary, ch, raf, index);
+ }
+
+ @Override
+ public void skip(DataInput raf, final int index) throws IOException {
+ if (dictionary.dictFileVersion >= 7)
+ {
+ StringUtil.readVarInt(raf);
+ }
+ else
+ {
+ raf.skipBytes(2);
+ }
+ int l = raf.readUnsignedShort();
+ raf.skipBytes(l);
+ }
+
+ @Override
+ public void write(DataOutput raf, HtmlEntry t) throws IOException {
+ t.writeBase(raf);
+ }
+ }
+
+ static final class DataSerializer implements RAFListSerializer<HtmlEntry> {
+ @Override
+ public HtmlEntry read(DataInput raf, final int index) {
+ assert false;
+ return null;
}
@Override
public void write(DataOutput raf, HtmlEntry t) throws IOException {
- t.write(raf);
+ t.writeData(raf);
}
}
- public String getRawText(final boolean compact) {
+ static final class DataDeserializer implements RAFListSerializer<byte[]> {
+ @Override
+ public byte[] read(DataInput raf, final int index) throws IOException {
+ return HtmlEntry.readData(raf);
+ }
+
+ @Override
+ public void write(DataOutput raf, byte[] t) {
+ assert false;
+ }
+ }
+
+ private String getRawText(final boolean compact) {
return title + ":\n" + getHtml();
}
@Override
- public int compareTo(HtmlEntry another) {
+ public int compareTo(/*@NonNull*/ HtmlEntry another) {
if (title.compareTo(another.title) != 0) {
return title.compareTo(another.title);
}
public static class Row extends RowBase {
- boolean isExpanded = false;
-
Row(final DataInput raf, final int thisRowIndex,
- final Index index, int extra) throws IOException {
+ final Index index, int extra) throws IOException {
super(raf, thisRowIndex, index, extra);
}
Row(final int referenceIndex, final int thisRowIndex,
- final Index index) {
+ final Index index) {
super(referenceIndex, thisRowIndex, index);
}
@Override
public RowMatchType matches(final List<String> searchTokens,
- final Pattern orderedMatchPattern, final Transliterator normalizer,
- final boolean swapPairEntries) {
+ final Pattern orderedMatchPattern, final Transliterator normalizer,
+ final boolean swapPairEntries) {
final String text = normalizer.transform(getRawText(false));
if (orderedMatchPattern.matcher(text).find()) {
return RowMatchType.ORDERED_MATCH;
for (final HtmlEntry htmlEntry : htmlEntries) {
final String titleEscaped = StringUtil.escapeUnicodeToPureHtml(htmlEntry.title);
result.append(String.format("<h1><a href=\"%s\">%s</a></h1>\n<p>%s\n",
- formatQuickdicUrl(indexShortName, htmlEntry.title), titleEscaped,
- htmlEntry.getHtml()));
+ formatQuickdicUrl(indexShortName, htmlEntry.title), titleEscaped,
+ htmlEntry.getHtml()));
}
return result.toString();
}
+ @SuppressWarnings("WeakerAccess")
public static String formatQuickdicUrl(final String indexShortName, final String text) {
assert !indexShortName.contains(":");
assert text.length() > 0;
// --------------------------------------------------------------------
+ @SuppressWarnings("WeakerAccess")
public static final class LazyHtmlLoader {
- final RandomAccessFile raf;
+ final DataInput raf;
+ final FileChannel ch;
final long offset;
final int numBytes;
final int numZipBytes;
+ final List<byte[]> data;
+ final int index;
// Not sure this volatile is right, but oh well.
- volatile SoftReference<String> htmlRef = new SoftReference<String>(null);
-
- private LazyHtmlLoader(final DataInput inp, int version) throws IOException {
- raf = (RandomAccessFile)inp;
- if (version >= 7) {
- numBytes = -1;
- numZipBytes = StringUtil.readVarInt(raf);
- } else {
- numBytes = raf.readInt();
- numZipBytes = raf.readInt();
+ volatile SoftReference<String> htmlRef = new SoftReference<>(null);
+
+ private LazyHtmlLoader(FileChannel ch, final DataInput inp, List<byte[]> data, int index) throws IOException {
+ this.data = data;
+ this.index = index;
+ if (data != null) {
+ this.raf = null;
+ this.ch = null;
+ this.offset = 0;
+ this.numBytes = -1;
+ this.numZipBytes = -1;
+ return;
}
- offset = raf.getFilePointer();
+ raf = inp;
+ this.ch = ch;
+ numBytes = Math.min(raf.readInt(), 20 * 1024 * 1024);
+ numZipBytes = Math.min(raf.readInt(), 20 * 1024 * 1024);
+ offset = ch.position();
raf.skipBytes(numZipBytes);
}
- public String getHtml() {
+ String getHtml() {
String html = htmlRef.get();
if (html != null) {
return html;
}
+ if (data != null) {
+ html = new String(data.get(index), StandardCharsets.UTF_8);
+ htmlRef = new SoftReference<>(html);
+ return html;
+ }
System.out.println("Loading Html: numBytes=" + numBytes + ", numZipBytes="
- + numZipBytes);
+ + numZipBytes);
final byte[] zipBytes = new byte[numZipBytes];
- synchronized (raf) {
+ synchronized (ch) {
try {
- raf.seek(offset);
- raf.read(zipBytes);
+ ch.position(offset);
+ raf.readFully(zipBytes);
} catch (IOException e) {
- throw new RuntimeException(e);
+ throw new RuntimeException("Failed to read HTML data from dictionary", e);
}
}
try {
final byte[] bytes = StringUtil.unzipFully(zipBytes, numBytes);
- html = new String(bytes, "UTF-8");
+ html = new String(bytes, StandardCharsets.UTF_8);
} catch (IOException e) {
- throw new RuntimeException(e);
+ throw new RuntimeException("Dictionary HTML data corrupted", e);
}
- htmlRef = new SoftReference<String>(html);
+ htmlRef = new SoftReference<>(html);
return html;
}
}
projects / Dictionary.git / blobdiff