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.StringReader;
  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 javax.annotation.Nonnull;
  32 import javax.annotation.Nullable;
  33
  34 import net.pterodactylus.sone.data.Post;
  35 import net.pterodactylus.sone.data.Sone;
  36 import net.pterodactylus.sone.data.impl.IdOnlySone;
  37 import net.pterodactylus.sone.database.PostProvider;
  38 import net.pterodactylus.sone.database.SoneProvider;
  39 import net.pterodactylus.util.io.Closer;
  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                 BufferedReader bufferedReader = new BufferedReader(new StringReader(source));
 128                 try {
 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                                         String name = link;
 181                                         logger.log(Level.FINER, String.format("Found link: %s", link));
 182
 183                                         /* if there is no text after the scheme, it’s not a link! */
 184                                         if (link.equals(linkType.getScheme())) {
 185                                                 parts.add(new PlainTextPart(linkType.getScheme()));
 186                                                 line = line.substring(linkType.getScheme().length());
 187                                                 continue;
 188                                         }
 189
 190                                         if (linkType == LinkType.SONE) {
 191                                                 if (line.length() >= (7 + 43)) {
 192                                                         String soneId = line.substring(7, 50);
 193                                                         Optional<Sone> sone = soneProvider.getSone(soneId);
 194                                                         if (!sone.isPresent()) {
 195                                                                 /*
 196                                                                  * don’t use create=true above, we don’t want
 197                                                                  * the empty shell.
 198                                                                  */
 199                                                                 sone = Optional.<Sone>of(new IdOnlySone(soneId));
 200                                                         }
 201                                                         parts.add(new SonePart(sone.get()));
 202                                                         line = line.substring(50);
 203                                                 } else {
 204                                                         parts.add(new PlainTextPart(line));
 205                                                         line = "";
 206                                                 }
 207                                                 continue;
 208                                         }
 209                                         if (linkType == LinkType.POST) {
 210                                                 if (line.length() >= (7 + 36)) {
 211                                                         String postId = line.substring(7, 43);
 212                                                         Optional<Post> post = postProvider.getPost(postId);
 213                                                         if (post.isPresent()) {
 214                                                                 parts.add(new PostPart(post.get()));
 215                                                         } else {
 216                                                                 parts.add(new PlainTextPart(line.substring(0, 43)));
 217                                                         }
 218                                                         line = line.substring(43);
 219                                                 } else {
 220                                                         parts.add(new PlainTextPart(line));
 221                                                         line = "";
 222                                                 }
 223                                                 continue;
 224                                         }
 225
 226                                         if (linkType.isFreenetLink()) {
 227                                                 FreenetURI uri;
 228                                                 if (name.indexOf('?') > -1) {
 229                                                         name = name.substring(0, name.indexOf('?'));
 230                                                 }
 231                                                 if (name.endsWith("/")) {
 232                                                         name = name.substring(0, name.length() - 1);
 233                                                 }
 234                                                 try {
 235                                                         uri = new FreenetURI(name);
 236                                                         name = uri.lastMetaString();
 237                                                         if (name == null) {
 238                                                                 name = uri.getDocName();
 239                                                         }
 240                                                         if (name == null) {
 241                                                                 name = link.substring(0, Math.min(9, link.length()));
 242                                                         }
 243                                                         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());
 244                                                         parts.add(new FreenetLinkPart(link, name, fromPostingSone));
 245                                                 } catch (MalformedURLException mue1) {
 246                                                         /* not a valid link, insert as plain text. */
 247                                                         parts.add(new PlainTextPart(link));
 248                                                 } catch (NullPointerException npe1) {
 249                                                         /* FreenetURI sometimes throws these, too. */
 250                                                         parts.add(new PlainTextPart(link));
 251                                                 } catch (ArrayIndexOutOfBoundsException aioobe1) {
 252                                                         /* oh, and these, too. */
 253                                                         parts.add(new PlainTextPart(link));
 254                                                 }
 255                                         } else if ((linkType == LinkType.HTTP) || (linkType == LinkType.HTTPS)) {
 256                                                 name = link.substring(linkType == LinkType.HTTP ? 7 : 8);
 257                                                 int firstSlash = name.indexOf('/');
 258                                                 int lastSlash = name.lastIndexOf('/');
 259                                                 if ((lastSlash - firstSlash) > 3) {
 260                                                         name = name.substring(0, firstSlash + 1) + "…" + name.substring(lastSlash);
 261                                                 }
 262                                                 if (name.endsWith("/")) {
 263                                                         name = name.substring(0, name.length() - 1);
 264                                                 }
 265                                                 if (((name.indexOf('/') > -1) && (name.indexOf('.') < name.lastIndexOf('.', name.indexOf('/'))) || ((name.indexOf('/') == -1) && (name.indexOf('.') < name.lastIndexOf('.')))) && name.startsWith("www.")) {
 266                                                         name = name.substring(4);
 267                                                 }
 268                                                 if (name.indexOf('?') > -1) {
 269                                                         name = name.substring(0, name.indexOf('?'));
 270                                                 }
 271                                                 parts.add(new LinkPart(link, name));
 272                                         }
 273                                         line = line.substring(endOfLink);
 274                                 }
 275                                 lastLineEmpty = false;
 276                         }
 277                 } catch (IOException ioe1) {
 278                         // a buffered reader around a string reader should never throw.
 279                         throw new RuntimeException(ioe1);
 280                 } finally {
 281                         Closer.close(bufferedReader);
 282                 }
 283                 for (int partIndex = parts.size() - 1; partIndex >= 0; --partIndex) {
 284                         Part part = parts.getPart(partIndex);
 285                         if (!(part instanceof PlainTextPart) || !"\n".equals(part.getText())) {
 286                                 break;
 287                         }
 288                         parts.removePart(partIndex);
 289                 }
 290                 return parts;
 291         }
 292
 293         private int findEndOfLink(String line) {
 294                 Matcher matcher = whitespacePattern.matcher(line);
 295                 if (!matcher.find(0)) {
 296                         return line.length();
 297                 }
 298                 int nextWhitespace = matcher.start();
 299                 int lastPunctuation = nextWhitespace;
 300                 while (isPunctuation(line.charAt(lastPunctuation - 1))) {
 301                         lastPunctuation -= 1;
 302                 }
 303                 if (lastPunctuation < nextWhitespace) {
 304                         return lastPunctuation;
 305                 }
 306                 int openParens = 0;
 307                 for (int i = 0; i < nextWhitespace; i++) {
 308                         switch (line.charAt(i)) {
 309                                 case '(':
 310                                         openParens++;
 311                                         break;
 312                                 case ')':
 313                                         openParens--;
 314                                         if (openParens < 0) {
 315                                                 return i;
 316                                         }
 317                                 default:
 318                         }
 319                 }
 320                 return nextWhitespace;
 321         }
 322
 323         private boolean isPunctuation(char character) {
 324                 return (character == '.') || (character == ',');
 325         }
 326
 327         private static class NextLink {
 328
 329                 private final int position;
 330                 private final LinkType linkType;
 331
 332                 private NextLink(int position, LinkType linkType) {
 333                         this.position = position;
 334                         this.linkType = linkType;
 335                 }
 336
 337                 public int getPosition() {
 338                         return position;
 339                 }
 340
 341                 public LinkType getLinkType() {
 342                         return linkType;
 343                 }
 344
 345                 public static Optional<NextLink> findNextLink(String line) {
 346                         int earliestLinkPosition = Integer.MAX_VALUE;
 347                         LinkType linkType = null;
 348                         for (LinkType possibleLinkType : LinkType.values()) {
 349                                 int nextLinkPosition = line.indexOf(possibleLinkType.getScheme());
 350                                 if (nextLinkPosition > -1) {
 351                                         if (nextLinkPosition < earliestLinkPosition) {
 352                                                 earliestLinkPosition = nextLinkPosition;
 353                                                 linkType = possibleLinkType;
 354                                         }
 355                                 }
 356                         }
 357                         return earliestLinkPosition < Integer.MAX_VALUE ?
 358                                         Optional.of(new NextLink(earliestLinkPosition, linkType)) : Optional.<NextLink>absent();
 359                 }
 360
 361         }
 362
 363 }