--- /dev/null
+/*
+ * Reactor - KickAssTorrentsFilter.java - Copyright © 2013 David Roden
+ *
+ * This program is free software: you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation, either version 3 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program. If not, see <http://www.gnu.org/licenses/>.
+ */
+
+package net.pterodactylus.reactor.filters;
+
+import static com.google.common.base.Preconditions.checkState;
+
+import java.io.UnsupportedEncodingException;
+import java.net.URI;
+import java.net.URISyntaxException;
+import java.net.URLEncoder;
+
+import net.pterodactylus.reactor.Filter;
+import net.pterodactylus.reactor.State;
+import net.pterodactylus.reactor.queries.HttpQuery;
+import net.pterodactylus.reactor.states.FailedState;
+import net.pterodactylus.reactor.states.HtmlState;
+import net.pterodactylus.reactor.states.TorrentState;
+import net.pterodactylus.reactor.states.TorrentState.TorrentFile;
+
+import org.jsoup.nodes.Document;
+import org.jsoup.nodes.Element;
+import org.jsoup.select.Elements;
+
+/**
+ * {@link Filter} implementation that parses a {@link TorrentState} from an
+ * {@link HtmlState} which was generated by a {@link HttpQuery} to
+ * {@code kickasstorrents.ph}.
+ *
+ * @author <a href="mailto:bombe@pterodactylus.net">David ‘Bombe’ Roden</a>
+ */
+public abstract class TorrentSiteFilter implements Filter {
+
+ /**
+ * {@inheritDoc}
+ */
+ @Override
+ public State filter(State state) {
+ if (!state.success()) {
+ return FailedState.from(state);
+ }
+ checkState(state instanceof HtmlState, "state is not an HtmlState but a %s", state.getClass().getName());
+
+ /* get result table. */
+ Document document = ((HtmlState) state).document();
+
+ /* iterate over all rows. */
+ Elements dataRows = getDataRows(document);
+ TorrentState torrentState = new TorrentState();
+ for (Element dataRow : dataRows) {
+ String name = extractName(dataRow);
+ String size = extractSize(dataRow);
+ String magnetUri = extractMagnetUri(dataRow);
+ String downloadUri;
+ int fileCount = extractFileCount(dataRow);
+ int seedCount = extractSeedCount(dataRow);
+ int leechCount = extractLeechCount(dataRow);
+ try {
+ downloadUri = new URI(((HtmlState) state).uri()).resolve(URLEncoder.encode(extractDownloadUri(dataRow), "UTF-8").replace("%2F", "/")).toString();
+ TorrentFile torrentFile = new TorrentFile(name, size, magnetUri, downloadUri, fileCount, seedCount, leechCount);
+ torrentState.addTorrentFile(torrentFile);
+ } catch (URISyntaxException use1) {
+ /* ignore; if uri was wrong, we wouldn’t be here. */
+ } catch (UnsupportedEncodingException uee1) {
+ /* ignore, all JVMs can do UTF-8. */
+ }
+ }
+
+ return torrentState;
+ }
+
+ //
+ // ABSTRACT METHODS
+ //
+
+ /**
+ * Returns the data rows from the given document.
+ *
+ * @param document
+ * The document to get the data rows from
+ * @return The data rows
+ */
+ protected abstract Elements getDataRows(Document document);
+
+ /**
+ * Extracts the name from the given row.
+ *
+ * @param dataRow
+ * The row to extract the name from
+ * @return The extracted name
+ */
+ protected abstract String extractName(Element dataRow);
+
+ /**
+ * Extracts the size from the given row.
+ *
+ * @param dataRow
+ * The row to extract the size from
+ * @return The extracted size
+ */
+ protected abstract String extractSize(Element dataRow);
+
+ /**
+ * Extracts the magnet URI from the given row.
+ *
+ * @param dataRow
+ * The row to extract the magnet URI from
+ * @return The extracted magnet URI
+ */
+ protected abstract String extractMagnetUri(Element dataRow);
+
+ /**
+ * Extracts the download URI from the given row.
+ *
+ * @param dataRow
+ * The row to extract the download URI from
+ * @return The extracted download URI
+ */
+ protected abstract String extractDownloadUri(Element dataRow);
+
+ /**
+ * Extracts the file count from the given row.
+ *
+ * @param dataRow
+ * The row to extract the file count from
+ * @return The extracted file count, or {@code 0} if the file count can not
+ * be extracted
+ */
+ protected abstract int extractFileCount(Element dataRow);
+
+ /**
+ * Extracts the seed count from the given row.
+ *
+ * @param dataRow
+ * The row to extract the seed count from
+ * @return The extracted seed count
+ */
+ protected abstract int extractSeedCount(Element dataRow);
+
+ /**
+ * Extracts the leech count from the given row.
+ *
+ * @param dataRow
+ * The row to extract the leech count from
+ * @return The extracted leech count
+ */
+ protected abstract int extractLeechCount(Element dataRow);
+
+}