+ public String toString() {
+ return getRawText(false);
+ }
+
+ // --------------------------------------------------------------------
+
+ public static class Row extends RowBase {
+
+ Row(final DataInput raf, final int thisRowIndex,
+ final Index index, int extra) throws IOException {
+ super(raf, thisRowIndex, index, extra);
+ }
+
+ Row(final int referenceIndex, final int thisRowIndex,
+ final Index index) {
+ super(referenceIndex, thisRowIndex, index);
+ }
+
+ @Override
+ public String toString() {
+ return getRawText(false);
+ }
+
+ public HtmlEntry getEntry() {
+ return index.dict.htmlEntries.get(referenceIndex);
+ }
+
+ @Override
+ public void print(PrintStream out) {
+ final HtmlEntry entry = getEntry();
+ out.println("See also HtmlEntry:" + entry.title);
+ }
+
+ @Override
+ public String getRawText(boolean compact) {
+ final HtmlEntry entry = getEntry();
+ return entry.getRawText(compact);
+ }
+
+ @Override
+ public RowMatchType matches(final List<String> searchTokens,
+ 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 (int i = searchTokens.size() - 1; i >= 0; --i) {
+ final String searchToken = searchTokens.get(i);
+ if (!text.contains(searchToken)) {
+ return RowMatchType.NO_MATCH;
+ }
+ }
+ return RowMatchType.BAG_OF_WORDS_MATCH;
+ }
+ }
+
+ public static String htmlBody(final List<HtmlEntry> htmlEntries, final String indexShortName) {
+ final StringBuilder result = new StringBuilder();
+ 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()));
+ }
+ return result.toString();
+ }
+
+ @SuppressWarnings("WeakerAccess")
+ public static String formatQuickdicUrl(final String indexShortName, final String text) {
+ assert !indexShortName.contains(":");
+ assert text.length() > 0;
+ StringBuilder s = new StringBuilder("q://d?");
+ s.append(indexShortName);
+ s.append("&");
+ s.append(StringUtil.encodeForUrl(text));
+ return s.toString();
+ }
+
+ public static boolean isQuickdicUrl(String url) {
+ return url.startsWith("q://d?");
+ }
+
+ // --------------------------------------------------------------------
+
+ @SuppressWarnings("WeakerAccess")
+ public static final class LazyHtmlLoader {
+ final DataInputBuffer buf;
+ final int numBytes;
+ final List<DataInputBuffer> data;
+ final int index;
+
+ // Not sure this volatile is right, but oh well.
+ volatile SoftReference<String> htmlRef = new SoftReference<>(null);
+
+ private LazyHtmlLoader(final DataInput inp, List<DataInputBuffer> data, int index) throws IOException {
+ this.data = data;
+ this.index = index;
+ if (data != null) {
+ buf = null;
+ this.numBytes = -1;
+ return;
+ }
+ numBytes = Math.min(inp.readInt(), 20 * 1024 * 1024);
+ int numZipBytes = Math.min(inp.readInt(), 20 * 1024 * 1024);
+ DataInputBuffer b = (DataInputBuffer)inp;
+ buf = b.slice(numZipBytes);
+ }
+
+ String getHtml() {
+ String html = htmlRef.get();
+ if (html != null) {
+ return html;
+ }
+ if (data != null) {
+ html = data.get(index).asString();
+ htmlRef = new SoftReference<>(html);
+ return html;
+ }
+ System.out.println("Loading Html: numBytes=" + numBytes + ", numZipBytes="
+ + buf.limit());
+ final byte[] zipBytes = new byte[buf.limit()];
+ buf.rewind();
+ buf.readFully(zipBytes);
+ try {
+ final byte[] bytes = StringUtil.unzipFully(zipBytes, numBytes);
+ html = new String(bytes, StandardCharsets.UTF_8);
+ } catch (IOException e) {
+ throw new RuntimeException("Dictionary HTML data corrupted", e);
+ }
+ htmlRef = new SoftReference<>(html);
+ return html;
+ }
+ }