src/main/java/net/pterodactylus/sone/text/SoneTextParser.java

   1 /*
   2  * Sone - SoneTextParser.java - Copyright © 2010–2013 David Roden
   3  *
   4  * This program is free software: you can redistribute it and/or modify
   5  * it under the terms of the GNU General Public License as published by
   6  * the Free Software Foundation, either version 3 of the License, or
   7  * (at your option) any later version.
   8  *
   9  * This program is distributed in the hope that it will be useful,
  10  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  11  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  12  * GNU General Public License for more details.
  13  *
  14  * You should have received a copy of the GNU General Public License
  15  * along with this program.  If not, see <http://www.gnu.org/licenses/>.
  16  */
  17
  18 package net.pterodactylus.sone.text;
  19
  20 import static java.util.logging.Logger.getLogger;
  21
  22 import java.io.BufferedReader;
  23 import java.io.IOException;
  24 import java.io.Reader;
  25 import java.net.MalformedURLException;
  26 import java.util.logging.Level;
  27 import java.util.logging.Logger;
  28 import java.util.regex.Matcher;
  29 import java.util.regex.Pattern;
  30
  31 import net.pterodactylus.sone.data.Post;
  32 import net.pterodactylus.sone.data.Sone;
  33 import net.pterodactylus.sone.data.impl.IdOnlySone;
  34 import net.pterodactylus.sone.database.PostProvider;
  35 import net.pterodactylus.sone.database.SoneProvider;
  36 import net.pterodactylus.util.io.Closer;
  37
  38 import com.google.common.base.Optional;
  39
  40 import freenet.keys.FreenetURI;
  41
  42 /**
  43  * {@link Parser} implementation that can recognize Freenet URIs.
  44  *
  45  * @author <a href="mailto:bombe@pterodactylus.net">David ‘Bombe’ Roden</a>
  46  */
  47 public class SoneTextParser implements Parser<SoneTextParserContext> {
  48
  49         /** The logger. */
  50         private static final Logger logger = getLogger(SoneTextParser.class.getName());
  51
  52         /** Pattern to detect whitespace. */
  53         private static final Pattern whitespacePattern = Pattern.compile("[\\u000a\u0020\u00a0\u1680\u180e\u2000\u2001\u2002\u2003\u2004\u2005\u2006\u2007\u2008\u2009\u200a\u200b\u200c\u200d\u202f\u205f\u2060\u2800\u3000]");
  54
  55         /**
  56          * Enumeration for all recognized link types.
  57          *
  58          * @author <a href="mailto:bombe@pterodactylus.net">David ‘Bombe’ Roden</a>
  59          */
  60         private enum LinkType {
  61
  62                 KSK("KSK@", true),
  63                 CHK("CHK@", true),
  64                 SSK("SSK@", true),
  65                 USK("USK@", true),
  66                 HTTP("http://", false),
  67                 HTTPS("https://", false),
  68                 SONE("sone://", false),
  69                 POST("post://", false);
  70
  71                 private final String scheme;
  72                 private final boolean freenetLink;
  73
  74                 LinkType(String scheme, boolean freenetLink) {
  75                         this.scheme = scheme;
  76                         this.freenetLink = freenetLink;
  77                 }
  78
  79                 /**
  80                  * Returns the scheme of this link type.
  81                  *
  82                  * @return The scheme of this link type
  83                  */
  84                 public String getScheme() {
  85                         return scheme;
  86                 }
  87
  88                 public boolean isFreenetLink() {
  89                         return freenetLink;
  90                 }
  91
  92         }
  93
  94         /** The Sone provider. */
  95         private final SoneProvider soneProvider;
  96
  97         /** The post provider. */
  98         private final PostProvider postProvider;
  99
 100         /**
 101          * Creates a new freenet link parser.
 102          *
 103          * @param soneProvider
 104          *            The Sone provider
 105          * @param postProvider
 106          *            The post provider
 107          */
 108         public SoneTextParser(SoneProvider soneProvider, PostProvider postProvider) {
 109                 this.soneProvider = soneProvider;
 110                 this.postProvider = postProvider;
 111         }
 112
 113         //
 114         // PART METHODS
 115         //
 116
 117         /**
 118          * {@inheritDoc}
 119          */
 120         @Override
 121         public Iterable<Part> parse(SoneTextParserContext context, Reader source) throws IOException {
 122                 PartContainer parts = new PartContainer();
 123                 BufferedReader bufferedReader = (source instanceof BufferedReader) ? (BufferedReader) source : new BufferedReader(source);
 124                 try {
 125                         String line;
 126                         boolean lastLineEmpty = true;
 127                         int emptyLines = 0;
 128                         while ((line = bufferedReader.readLine()) != null) {
 129                                 if (line.trim().length() == 0) {
 130                                         if (lastLineEmpty) {
 131                                                 continue;
 132                                         }
 133                                         parts.add(new PlainTextPart("\n"));
 134                                         ++emptyLines;
 135                                         lastLineEmpty = emptyLines == 2;
 136                                         continue;
 137                                 }
 138                                 emptyLines = 0;
 139                                 /*
 140                                  * lineComplete tracks whether the block you are parsing is the
 141                                  * first block of the line. this is important because sometimes
 142                                  * you have to add an additional line break.
 143                                  */
 144                                 boolean lineComplete = true;
 145                                 while (line.length() > 0) {
 146                                         Optional<NextLink> nextLink = NextLink.findNextLink(line);
 147                                         if (!nextLink.isPresent()) {
 148                                                 if (lineComplete && !lastLineEmpty) {
 149                                                         parts.add(new PlainTextPart("\n" + line));
 150                                                 } else {
 151                                                         parts.add(new PlainTextPart(line));
 152                                                 }
 153                                                 break;
 154                                         }
 155                                         LinkType linkType = nextLink.get().getLinkType();
 156                                         int next = nextLink.get().getPosition();
 157
 158                                         /* cut off “freenet:” from before keys. */
 159                                         if (linkType.isFreenetLink() && (next >= 8) && (line.substring(next - 8, next).equals("freenet:"))) {
 160                                                 next -= 8;
 161                                                 line = line.substring(0, next) + line.substring(next + 8);
 162                                         }
 163
 164                                         /* if there is text before the next item, write it out. */
 165                                         if (lineComplete && !lastLineEmpty) {
 166                                                 parts.add(new PlainTextPart("\n"));
 167                                         }
 168                                         if (next > 0) {
 169                                                 parts.add(new PlainTextPart(line.substring(0, next)));
 170                                                 line = line.substring(next);
 171                                         }
 172                                         lineComplete = false;
 173
 174                                         Matcher matcher = whitespacePattern.matcher(line);
 175                                         int nextSpace = matcher.find(0) ? matcher.start() : line.length();
 176                                         String link = line.substring(0, nextSpace);
 177                                         String name = link;
 178                                         logger.log(Level.FINER, String.format("Found link: %s", link));
 179
 180                                         /* if there is no text after the scheme, it’s not a link! */
 181                                         if (link.equals(linkType.getScheme())) {
 182                                                 parts.add(new PlainTextPart(linkType.getScheme()));
 183                                                 line = line.substring(linkType.getScheme().length());
 184                                                 continue;
 185                                         }
 186
 187                                         if (linkType == LinkType.SONE) {
 188                                                 if (line.length() >= (7 + 43)) {
 189                                                         String soneId = line.substring(7, 50);
 190                                                         Optional<Sone> sone = soneProvider.getSone(soneId);
 191                                                         if (!sone.isPresent()) {
 192                                                                 /*
 193                                                                  * don’t use create=true above, we don’t want
 194                                                                  * the empty shell.
 195                                                                  */
 196                                                                 sone = Optional.<Sone>of(new IdOnlySone(soneId));
 197                                                         }
 198                                                         parts.add(new SonePart(sone.get()));
 199                                                         line = line.substring(50);
 200                                                 } else {
 201                                                         parts.add(new PlainTextPart(line));
 202                                                         line = "";
 203                                                 }
 204                                                 continue;
 205                                         }
 206                                         if (linkType == LinkType.POST) {
 207                                                 Optional<Post> post = postProvider.getPost(link.substring(7));
 208                                                 if (post.isPresent()) {
 209                                                         parts.add(new PostPart(post.get(), link.substring(7).equals(post.get().getInternalId())));
 210                                                 } else {
 211                                                         parts.add(new PlainTextPart(link));
 212                                                 }
 213                                                 line = line.substring(link.length());
 214                                                 continue;
 215                                         }
 216
 217                                         if (linkType.isFreenetLink()) {
 218                                                 FreenetURI uri;
 219                                                 if (name.indexOf('?') > -1) {
 220                                                         name = name.substring(0, name.indexOf('?'));
 221                                                 }
 222                                                 if (name.endsWith("/")) {
 223                                                         name = name.substring(0, name.length() - 1);
 224                                                 }
 225                                                 try {
 226                                                         uri = new FreenetURI(name);
 227                                                         name = uri.lastMetaString();
 228                                                         if (name == null) {
 229                                                                 name = uri.getDocName();
 230                                                         }
 231                                                         if (name == null) {
 232                                                                 name = link.substring(0, Math.min(9, link.length()));
 233                                                         }
 234                                                         boolean fromPostingSone = ((linkType == LinkType.SSK) || (linkType == LinkType.USK)) && (context != null) && (context.getPostingSone() != null) && link.substring(4, Math.min(link.length(), 47)).equals(context.getPostingSone().getId());
 235                                                         parts.add(new FreenetLinkPart(link, name, fromPostingSone));
 236                                                 } catch (MalformedURLException mue1) {
 237                                                         /* not a valid link, insert as plain text. */
 238                                                         parts.add(new PlainTextPart(link));
 239                                                 } catch (NullPointerException npe1) {
 240                                                         /* FreenetURI sometimes throws these, too. */
 241                                                         parts.add(new PlainTextPart(link));
 242                                                 } catch (ArrayIndexOutOfBoundsException aioobe1) {
 243                                                         /* oh, and these, too. */
 244                                                         parts.add(new PlainTextPart(link));
 245                                                 }
 246                                         } else if ((linkType == LinkType.HTTP) || (linkType == LinkType.HTTPS)) {
 247                                                 name = link.substring(linkType == LinkType.HTTP ? 7 : 8);
 248                                                 int firstSlash = name.indexOf('/');
 249                                                 int lastSlash = name.lastIndexOf('/');
 250                                                 if ((lastSlash - firstSlash) > 3) {
 251                                                         name = name.substring(0, firstSlash + 1) + "…" + name.substring(lastSlash);
 252                                                 }
 253                                                 if (name.endsWith("/")) {
 254                                                         name = name.substring(0, name.length() - 1);
 255                                                 }
 256                                                 if (((name.indexOf('/') > -1) && (name.indexOf('.') < name.lastIndexOf('.', name.indexOf('/'))) || ((name.indexOf('/') == -1) && (name.indexOf('.') < name.lastIndexOf('.')))) && name.startsWith("www.")) {
 257                                                         name = name.substring(4);
 258                                                 }
 259                                                 if (name.indexOf('?') > -1) {
 260                                                         name = name.substring(0, name.indexOf('?'));
 261                                                 }
 262                                                 parts.add(new LinkPart(link, name));
 263                                         }
 264                                         line = line.substring(nextSpace);
 265                                 }
 266                                 lastLineEmpty = false;
 267                         }
 268                 } finally {
 269                         if (bufferedReader != source) {
 270                                 Closer.close(bufferedReader);
 271                         }
 272                 }
 273                 for (int partIndex = parts.size() - 1; partIndex >= 0; --partIndex) {
 274                         Part part = parts.getPart(partIndex);
 275                         if (!(part instanceof PlainTextPart) || !"\n".equals(part.getText())) {
 276                                 break;
 277                         }
 278                         parts.removePart(partIndex);
 279                 }
 280                 return parts;
 281         }
 282
 283         private static class NextLink {
 284
 285                 private final int position;
 286                 private final LinkType linkType;
 287
 288                 private NextLink(int position, LinkType linkType) {
 289                         this.position = position;
 290                         this.linkType = linkType;
 291                 }
 292
 293                 public int getPosition() {
 294                         return position;
 295                 }
 296
 297                 public LinkType getLinkType() {
 298                         return linkType;
 299                 }
 300
 301                 public static Optional<NextLink> findNextLink(String line) {
 302                         int earliestLinkPosition = Integer.MAX_VALUE;
 303                         LinkType linkType = null;
 304                         for (LinkType possibleLinkType : LinkType.values()) {
 305                                 int nextLinkPosition = line.indexOf(possibleLinkType.getScheme());
 306                                 if (nextLinkPosition > -1) {
 307                                         if (nextLinkPosition < earliestLinkPosition) {
 308                                                 earliestLinkPosition = nextLinkPosition;
 309                                                 linkType = possibleLinkType;
 310                                         }
 311                                 }
 312                         }
 313                         return earliestLinkPosition < Integer.MAX_VALUE ?
 314                                         Optional.of(new NextLink(earliestLinkPosition, linkType)) : Optional.<NextLink>absent();
 315                 }
 316
 317         }
 318
 319 }