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

   1 /*
   2  * Sone - SoneTextParser.java - Copyright © 2010–2016 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.io.StringReader;
  26 import java.net.MalformedURLException;
  27 import java.util.logging.Level;
  28 import java.util.logging.Logger;
  29 import java.util.regex.Matcher;
  30 import java.util.regex.Pattern;
  31
  32 import javax.annotation.Nonnull;
  33 import javax.annotation.Nullable;
  34
  35 import net.pterodactylus.sone.data.Post;
  36 import net.pterodactylus.sone.data.Sone;
  37 import net.pterodactylus.sone.data.impl.IdOnlySone;
  38 import net.pterodactylus.sone.database.PostProvider;
  39 import net.pterodactylus.sone.database.SoneProvider;
  40
  41 import com.google.common.base.Optional;
  42
  43 import freenet.keys.FreenetURI;
  44
  45 /**
  46  * {@link Parser} implementation that can recognize Freenet URIs.
  47  *
  48  * @author <a href="mailto:bombe@pterodactylus.net">David ‘Bombe’ Roden</a>
  49  */
  50 public class SoneTextParser implements Parser<SoneTextParserContext> {
  51
  52         /** The logger. */
  53         private static final Logger logger = getLogger(SoneTextParser.class.getName());
  54
  55         /** Pattern to detect whitespace. */
  56         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]");
  57
  58         /**
  59          * Enumeration for all recognized link types.
  60          *
  61          * @author <a href="mailto:bombe@pterodactylus.net">David ‘Bombe’ Roden</a>
  62          */
  63         private enum LinkType {
  64
  65                 KSK("KSK@", true),
  66                 CHK("CHK@", true),
  67                 SSK("SSK@", true),
  68                 USK("USK@", true),
  69                 HTTP("http://", false),
  70                 HTTPS("https://", false),
  71                 SONE("sone://", false),
  72                 POST("post://", false);
  73
  74                 private final String scheme;
  75                 private final boolean freenetLink;
  76
  77                 LinkType(String scheme, boolean freenetLink) {
  78                         this.scheme = scheme;
  79                         this.freenetLink = freenetLink;
  80                 }
  81
  82                 /**
  83                  * Returns the scheme of this link type.
  84                  *
  85                  * @return The scheme of this link type
  86                  */
  87                 public String getScheme() {
  88                         return scheme;
  89                 }
  90
  91                 public boolean isFreenetLink() {
  92                         return freenetLink;
  93                 }
  94
  95         }
  96
  97         /** The Sone provider. */
  98         private final SoneProvider soneProvider;
  99
 100         /** The post provider. */
 101         private final PostProvider postProvider;
 102
 103         /**
 104          * Creates a new freenet link parser.
 105          *
 106          * @param soneProvider
 107          *            The Sone provider
 108          * @param postProvider
 109          *            The post provider
 110          */
 111         public SoneTextParser(SoneProvider soneProvider, PostProvider postProvider) {
 112                 this.soneProvider = soneProvider;
 113                 this.postProvider = postProvider;
 114         }
 115
 116         //
 117         // PART METHODS
 118         //
 119
 120         /**
 121          * {@inheritDoc}
 122          */
 123         @Nonnull
 124         @Override
 125         public Iterable<Part> parse(@Nonnull String source, @Nullable SoneTextParserContext context) {
 126                 PartContainer parts = new PartContainer();
 127                 try (Reader sourceReader = new StringReader(source);
 128                                 BufferedReader bufferedReader = new BufferedReader(sourceReader)) {
 129                         String line;
 130                         boolean lastLineEmpty = true;
 131                         int emptyLines = 0;
 132                         while ((line = bufferedReader.readLine()) != null) {
 133                                 if (line.trim().length() == 0) {
 134                                         if (lastLineEmpty) {
 135                                                 continue;
 136                                         }
 137                                         parts.add(new PlainTextPart("\n"));
 138                                         ++emptyLines;
 139                                         lastLineEmpty = emptyLines == 2;
 140                                         continue;
 141                                 }
 142                                 emptyLines = 0;
 143                                 /*
 144                                  * lineComplete tracks whether the block you are parsing is the
 145                                  * first block of the line. this is important because sometimes
 146                                  * you have to add an additional line break.
 147                                  */
 148                                 boolean lineComplete = true;
 149                                 while (line.length() > 0) {
 150                                         Optional<NextLink> nextLink = NextLink.findNextLink(line);
 151                                         if (!nextLink.isPresent()) {
 152                                                 if (lineComplete && !lastLineEmpty) {
 153                                                         parts.add(new PlainTextPart("\n" + line));
 154                                                 } else {
 155                                                         parts.add(new PlainTextPart(line));
 156                                                 }
 157                                                 break;
 158                                         }
 159                                         LinkType linkType = nextLink.get().getLinkType();
 160                                         int next = nextLink.get().getPosition();
 161
 162                                         /* cut off “freenet:” from before keys. */
 163                                         if (linkType.isFreenetLink() && (next >= 8) && (line.substring(next - 8, next).equals("freenet:"))) {
 164                                                 next -= 8;
 165                                                 line = line.substring(0, next) + line.substring(next + 8);
 166                                         }
 167
 168                                         /* if there is text before the next item, write it out. */
 169                                         if (lineComplete && !lastLineEmpty) {
 170                                                 parts.add(new PlainTextPart("\n"));
 171                                         }
 172                                         if (next > 0) {
 173                                                 parts.add(new PlainTextPart(line.substring(0, next)));
 174                                                 line = line.substring(next);
 175                                         }
 176                                         lineComplete = false;
 177
 178                                         int endOfLink = findEndOfLink(line);
 179                                         String link = line.substring(0, endOfLink);
 180                                         logger.log(Level.FINER, String.format("Found link: %s", link));
 181
 182                                         /* if there is no text after the scheme, it’s not a link! */
 183                                         if (link.equals(linkType.getScheme())) {
 184                                                 parts.add(new PlainTextPart(linkType.getScheme()));
 185                                                 line = line.substring(linkType.getScheme().length());
 186                                                 continue;
 187                                         }
 188
 189                                         switch (linkType) {
 190                                                 case SONE:
 191                                                         renderSoneLink(parts, link);
 192                                                         break;
 193                                                 case POST:
 194                                                         renderPostLink(parts, link);
 195                                                         break;
 196                                                 case KSK:
 197                                                 case CHK:
 198                                                 case SSK:
 199                                                 case USK:
 200                                                         renderFreenetLink(parts, link, linkType, context);
 201                                                         break;
 202                                                 case HTTP:
 203                                                 case HTTPS:
 204                                                         renderHttpLink(parts, link, linkType);
 205                                                         break;
 206                                         }
 207
 208                                         line = line.substring(endOfLink);
 209                                 }
 210                                 lastLineEmpty = false;
 211                         }
 212                 } catch (IOException ioe1) {
 213                         // a buffered reader around a string reader should never throw.
 214                         throw new RuntimeException(ioe1);
 215                 }
 216                 for (int partIndex = parts.size() - 1; partIndex >= 0; --partIndex) {
 217                         Part part = parts.getPart(partIndex);
 218                         if (!(part instanceof PlainTextPart) || !"\n".equals(part.getText())) {
 219                                 break;
 220                         }
 221                         parts.removePart(partIndex);
 222                 }
 223                 return parts;
 224         }
 225
 226         private void renderSoneLink(PartContainer parts, String line) {
 227                 if (line.length() >= (7 + 43)) {
 228                         String soneId = line.substring(7, 50);
 229                         Optional<Sone> sone = soneProvider.getSone(soneId);
 230                         parts.add(new SonePart(sone.or(new IdOnlySone(soneId))));
 231                 } else {
 232                         parts.add(new PlainTextPart(line));
 233                 }
 234         }
 235
 236         private void renderPostLink(PartContainer parts, String line) {
 237                 if (line.length() >= (7 + 36)) {
 238                         String postId = line.substring(7, 43);
 239                         Optional<Post> post = postProvider.getPost(postId);
 240                         if (post.isPresent()) {
 241                                 parts.add(new PostPart(post.get()));
 242                         } else {
 243                                 parts.add(new PlainTextPart(line.substring(0, 43)));
 244                         }
 245                 } else {
 246                         parts.add(new PlainTextPart(line));
 247                 }
 248         }
 249
 250         private void renderFreenetLink(PartContainer parts, String link, LinkType linkType, @Nullable SoneTextParserContext context) {
 251                 String name = link;
 252                 if (name.indexOf('?') > -1) {
 253                         name = name.substring(0, name.indexOf('?'));
 254                 }
 255                 if (name.endsWith("/")) {
 256                         name = name.substring(0, name.length() - 1);
 257                 }
 258                 try {
 259                         FreenetURI uri = new FreenetURI(name);
 260                         name = uri.lastMetaString();
 261                         if (name == null) {
 262                                 name = uri.getDocName();
 263                         }
 264                         if (name == null) {
 265                                 name = link.substring(0, Math.min(9, link.length()));
 266                         }
 267                         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());
 268                         parts.add(new FreenetLinkPart(link, name, fromPostingSone));
 269                 } catch (MalformedURLException mue1) {
 270                         /* not a valid link, insert as plain text. */
 271                         parts.add(new PlainTextPart(link));
 272                 } catch (NullPointerException npe1) {
 273                         /* FreenetURI sometimes throws these, too. */
 274                         parts.add(new PlainTextPart(link));
 275                 } catch (ArrayIndexOutOfBoundsException aioobe1) {
 276                         /* oh, and these, too. */
 277                         parts.add(new PlainTextPart(link));
 278                 }
 279         }
 280
 281         private void renderHttpLink(PartContainer parts, String link, LinkType linkType) {
 282                 String name;
 283                 name = link.substring(linkType == LinkType.HTTP ? 7 : 8);
 284                 int firstSlash = name.indexOf('/');
 285                 int lastSlash = name.lastIndexOf('/');
 286                 if ((lastSlash - firstSlash) > 3) {
 287                         name = name.substring(0, firstSlash + 1) + "…" + name.substring(lastSlash);
 288                 }
 289                 if (name.endsWith("/")) {
 290                         name = name.substring(0, name.length() - 1);
 291                 }
 292                 if (((name.indexOf('/') > -1) && (name.indexOf('.') < name.lastIndexOf('.', name.indexOf('/'))) || ((name.indexOf('/') == -1) && (name.indexOf('.') < name.lastIndexOf('.')))) && name.startsWith("www.")) {
 293                         name = name.substring(4);
 294                 }
 295                 if (name.indexOf('?') > -1) {
 296                         name = name.substring(0, name.indexOf('?'));
 297                 }
 298                 parts.add(new LinkPart(link, name));
 299         }
 300
 301         private int findEndOfLink(String line) {
 302                 Matcher matcher = whitespacePattern.matcher(line);
 303                 int endOfLink = matcher.find() ? matcher.start() : line.length();
 304                 while ((endOfLink > 0) && isPunctuation(line.charAt(endOfLink - 1))) {
 305                         endOfLink--;
 306                 }
 307                 int openParens = 0;
 308                 for (int i = 0; i < endOfLink; i++) {
 309                         switch (line.charAt(i)) {
 310                                 case '(':
 311                                         openParens++;
 312                                         break;
 313                                 case ')':
 314                                         openParens--;
 315                                         if (openParens < 0) {
 316                                                 return i;
 317                                         }
 318                                 default:
 319                         }
 320                 }
 321                 return endOfLink;
 322         }
 323
 324         private static boolean isPunctuation(char character) {
 325                 return (character == '.') || (character == ',');
 326         }
 327
 328         private static class NextLink {
 329
 330                 private final int position;
 331                 private final LinkType linkType;
 332
 333                 private NextLink(int position, LinkType linkType) {
 334                         this.position = position;
 335                         this.linkType = linkType;
 336                 }
 337
 338                 public int getPosition() {
 339                         return position;
 340                 }
 341
 342                 public LinkType getLinkType() {
 343                         return linkType;
 344                 }
 345
 346                 public static Optional<NextLink> findNextLink(String line) {
 347                         int earliestLinkPosition = Integer.MAX_VALUE;
 348                         LinkType linkType = null;
 349                         for (LinkType possibleLinkType : LinkType.values()) {
 350                                 int nextLinkPosition = line.indexOf(possibleLinkType.getScheme());
 351                                 if (nextLinkPosition > -1) {
 352                                         if (nextLinkPosition < earliestLinkPosition) {
 353                                                 earliestLinkPosition = nextLinkPosition;
 354                                                 linkType = possibleLinkType;
 355                                         }
 356                                 }
 357                         }
 358                         return earliestLinkPosition < Integer.MAX_VALUE ?
 359                                         Optional.of(new NextLink(earliestLinkPosition, linkType)) : Optional.<NextLink>absent();
 360                 }
 361
 362         }
 363
 364 }