src/com/hughes/android/dictionary/engine/HtmlEntry.java

   1
   2 package com.hughes.android.dictionary.engine;
   3
   4 import java.io.DataInput;
   5 import java.io.DataOutput;
   6 import java.io.IOException;
   7 import java.io.PrintStream;
   8 import java.lang.ref.SoftReference;
   9 import java.nio.charset.StandardCharsets;
  10 import java.util.List;
  11 import java.util.regex.Pattern;
  12
  13 import com.hughes.util.DataInputBuffer;
  14 import com.hughes.util.StringUtil;
  15 import com.hughes.util.raf.RAFListSerializer;
  16 import com.hughes.util.raf.RAFListSerializerSkippable;
  17 import com.ibm.icu.text.Transliterator;
  18
  19 public class HtmlEntry extends AbstractEntry implements Comparable<HtmlEntry> {
  20
  21     // Title is not HTML escaped.
  22     public final String title;
  23     private final LazyHtmlLoader lazyHtmlLoader;
  24     @SuppressWarnings("WeakerAccess")
  25     public String html;
  26
  27     public HtmlEntry(final EntrySource entrySource, String title) {
  28         super(entrySource);
  29         this.title = title;
  30         lazyHtmlLoader = null;
  31     }
  32
  33     public HtmlEntry(Dictionary dictionary, DataInput raf, final int index)
  34     throws IOException {
  35         super(dictionary, raf, index);
  36         title = raf.readUTF();
  37         lazyHtmlLoader = new LazyHtmlLoader(raf, dictionary.htmlData, index);
  38         html = null;
  39     }
  40
  41     private void writeBase(DataOutput raf) throws IOException {
  42         super.write(raf);
  43         raf.writeUTF(title);
  44     }
  45
  46     private void writeData(DataOutput raf) throws IOException {
  47         final byte[] bytes = getHtml().getBytes(StandardCharsets.UTF_8);
  48         StringUtil.writeVarInt(raf, bytes.length);
  49         raf.write(bytes);
  50     }
  51
  52     private static DataInputBuffer readData(DataInput raf) throws IOException {
  53         int len = StringUtil.readVarInt(raf);
  54         return ((DataInputBuffer)raf).slice(len);
  55     }
  56
  57     String getHtml() {
  58         return html != null ? html : lazyHtmlLoader.getHtml();
  59     }
  60
  61     @Override
  62     public void addToDictionary(Dictionary dictionary) {
  63         assert index == -1;
  64         dictionary.htmlEntries.add(this);
  65         index = dictionary.htmlEntries.size() - 1;
  66     }
  67
  68     @Override
  69     public RowBase CreateRow(int rowIndex, Index dictionaryIndex) {
  70         return new Row(this.index, rowIndex, dictionaryIndex);
  71     }
  72
  73     static final class Serializer implements RAFListSerializerSkippable<HtmlEntry> {
  74
  75         final Dictionary dictionary;
  76
  77         Serializer(Dictionary dictionary) {
  78             this.dictionary = dictionary;
  79         }
  80
  81         @Override
  82         public HtmlEntry read(DataInput raf, final int index) throws IOException {
  83             return new HtmlEntry(dictionary, raf, index);
  84         }
  85
  86         @Override
  87         public void skip(DataInput raf, final int index) throws IOException {
  88             if (dictionary.dictFileVersion >= 7)
  89             {
  90                 StringUtil.readVarInt(raf);
  91             }
  92             else
  93             {
  94                 raf.skipBytes(2);
  95             }
  96             int l = raf.readUnsignedShort();
  97             raf.skipBytes(l);
  98         }
  99
 100         @Override
 101         public void write(DataOutput raf, HtmlEntry t) throws IOException {
 102             t.writeBase(raf);
 103         }
 104     }
 105
 106     static final class DataSerializer implements RAFListSerializer<HtmlEntry> {
 107         @Override
 108         public HtmlEntry read(DataInput raf, final int index) {
 109             assert false;
 110             return null;
 111         }
 112
 113         @Override
 114         public void write(DataOutput raf, HtmlEntry t) throws IOException {
 115             t.writeData(raf);
 116         }
 117     }
 118
 119     static final class DataDeserializer implements RAFListSerializer<DataInputBuffer> {
 120         @Override
 121         public DataInputBuffer read(DataInput raf, final int index) throws IOException {
 122             return readData(raf);
 123         }
 124
 125         @Override
 126         public void write(DataOutput raf, DataInputBuffer t) {
 127             assert false;
 128         }
 129     }
 130
 131     private String getRawText(final boolean compact) {
 132         return title + ":\n" + getHtml();
 133     }
 134
 135     @Override
 136     public int compareTo(/*@NonNull*/ HtmlEntry another) {
 137         if (title.compareTo(another.title) != 0) {
 138             return title.compareTo(another.title);
 139         }
 140         return getHtml().compareTo(another.getHtml());
 141     }
 142
 143     @Override
 144     public String toString() {
 145         return getRawText(false);
 146     }
 147
 148     // --------------------------------------------------------------------
 149
 150     public static class Row extends RowBase {
 151
 152         Row(final DataInput raf, final int thisRowIndex,
 153             final Index index, int extra) throws IOException {
 154             super(raf, thisRowIndex, index, extra);
 155         }
 156
 157         Row(final int referenceIndex, final int thisRowIndex,
 158             final Index index) {
 159             super(referenceIndex, thisRowIndex, index);
 160         }
 161
 162         @Override
 163         public String toString() {
 164             return getRawText(false);
 165         }
 166
 167         public HtmlEntry getEntry() {
 168             return index.dict.htmlEntries.get(referenceIndex);
 169         }
 170
 171         @Override
 172         public void print(PrintStream out) {
 173             final HtmlEntry entry = getEntry();
 174             out.println("See also HtmlEntry:" + entry.title);
 175         }
 176
 177         @Override
 178         public String getRawText(boolean compact) {
 179             final HtmlEntry entry = getEntry();
 180             return entry.getRawText(compact);
 181         }
 182
 183         @Override
 184         public RowMatchType matches(final List<String> searchTokens,
 185                                     final Pattern orderedMatchPattern, final Transliterator normalizer,
 186                                     final boolean swapPairEntries) {
 187             final String text = normalizer.transform(getRawText(false));
 188             if (orderedMatchPattern.matcher(text).find()) {
 189                 return RowMatchType.ORDERED_MATCH;
 190             }
 191             for (int i = searchTokens.size() - 1; i >= 0; --i) {
 192                 final String searchToken = searchTokens.get(i);
 193                 if (!text.contains(searchToken)) {
 194                     return RowMatchType.NO_MATCH;
 195                 }
 196             }
 197             return RowMatchType.BAG_OF_WORDS_MATCH;
 198         }
 199     }
 200
 201     public static String htmlBody(final List<HtmlEntry> htmlEntries, final String indexShortName) {
 202         final StringBuilder result = new StringBuilder();
 203         for (final HtmlEntry htmlEntry : htmlEntries) {
 204             final String titleEscaped = StringUtil.escapeUnicodeToPureHtml(htmlEntry.title);
 205             result.append(String.format("<h1><a href=\"%s\">%s</a></h1>\n<p>%s\n",
 206                                         formatQuickdicUrl(indexShortName, htmlEntry.title), titleEscaped,
 207                                         htmlEntry.getHtml()));
 208         }
 209         return result.toString();
 210     }
 211
 212     @SuppressWarnings("WeakerAccess")
 213     public static String formatQuickdicUrl(final String indexShortName, final String text) {
 214         assert !indexShortName.contains(":");
 215         assert text.length() > 0;
 216         StringBuilder s = new StringBuilder("q://d?");
 217         s.append(indexShortName);
 218         s.append("&");
 219         s.append(StringUtil.encodeForUrl(text));
 220         return s.toString();
 221     }
 222
 223     public static boolean isQuickdicUrl(String url) {
 224         return url.startsWith("q://d?");
 225     }
 226
 227     // --------------------------------------------------------------------
 228
 229     @SuppressWarnings("WeakerAccess")
 230     public static final class LazyHtmlLoader {
 231         final DataInputBuffer buf;
 232         final int numBytes;
 233         final List<DataInputBuffer> data;
 234         final int index;
 235
 236         // Not sure this volatile is right, but oh well.
 237         volatile SoftReference<String> htmlRef = new SoftReference<>(null);
 238
 239         private LazyHtmlLoader(final DataInput inp, List<DataInputBuffer> data, int index) throws IOException {
 240             this.data = data;
 241             this.index = index;
 242             if (data != null) {
 243                 buf = null;
 244                 this.numBytes = -1;
 245                 return;
 246             }
 247             numBytes = Math.min(inp.readInt(), 20 * 1024 * 1024);
 248             int numZipBytes = Math.min(inp.readInt(), 20 * 1024 * 1024);
 249             DataInputBuffer b = (DataInputBuffer)inp;
 250             buf = b.slice(numZipBytes);
 251         }
 252
 253         String getHtml() {
 254             String html = htmlRef.get();
 255             if (html != null) {
 256                 return html;
 257             }
 258             if (data != null) {
 259                 html = data.get(index).asString();
 260                 htmlRef = new SoftReference<>(html);
 261                 return html;
 262             }
 263             System.out.println("Loading Html: numBytes=" + numBytes + ", numZipBytes="
 264                                + buf.limit());
 265             final byte[] zipBytes = new byte[buf.limit()];
 266             buf.rewind();
 267             buf.readFully(zipBytes);
 268             try {
 269                 final byte[] bytes = StringUtil.unzipFully(zipBytes, numBytes);
 270                 html = new String(bytes, StandardCharsets.UTF_8);
 271             } catch (IOException e) {
 272                 throw new RuntimeException("Dictionary HTML data corrupted", e);
 273             }
 274             htmlRef = new SoftReference<>(html);
 275             return html;
 276         }
 277     }
 278
 279 }