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 com.google.common.base.Optional.absent;
  21 import static com.google.common.base.Optional.of;
  22 import static java.util.logging.Logger.getLogger;
  23
  24 import java.io.BufferedReader;
  25 import java.io.IOException;
  26 import java.io.Reader;
  27 import java.io.StringReader;
  28 import java.net.MalformedURLException;
  29 import java.util.ArrayList;
  30 import java.util.List;
  31 import java.util.logging.Level;
  32 import java.util.logging.Logger;
  33 import java.util.regex.Matcher;
  34 import java.util.regex.Pattern;
  35
  36 import javax.annotation.Nonnull;
  37 import javax.annotation.Nullable;
  38
  39 import net.pterodactylus.sone.data.Post;
  40 import net.pterodactylus.sone.data.Sone;
  41 import net.pterodactylus.sone.data.impl.IdOnlySone;
  42 import net.pterodactylus.sone.database.PostProvider;
  43 import net.pterodactylus.sone.database.SoneProvider;
  44
  45 import com.google.common.base.Optional;
  46 import org.bitpedia.util.Base32;
  47
  48 import freenet.keys.FreenetURI;
  49 import freenet.support.Base64;
  50
  51 /**
  52  * {@link Parser} implementation that can recognize Freenet URIs.
  53  *
  54  * @author <a href="mailto:bombe@pterodactylus.net">David ‘Bombe’ Roden</a>
  55  */
  56 public class SoneTextParser implements Parser<SoneTextParserContext> {
  57
  58         /** The logger. */
  59         private static final Logger logger = getLogger(SoneTextParser.class.getName());
  60
  61         /** Pattern to detect whitespace. */
  62         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]");
  63
  64         private static class NextLink {
  65
  66                 private final int position;
  67                 private final String link;
  68                 private final String remainder;
  69                 private final LinkType linkType;
  70
  71                 private NextLink(int position, String link, String remainder, LinkType linkType) {
  72                         this.position = position;
  73                         this.link = link;
  74                         this.remainder = remainder;
  75                         this.linkType = linkType;
  76                 }
  77
  78                 public int getPosition() {
  79                         return position;
  80                 }
  81
  82                 public String getLink() {
  83                         return link;
  84                 }
  85
  86                 public String getRemainder() {
  87                         return remainder;
  88                 }
  89
  90                 public LinkType getLinkType() {
  91                         return linkType;
  92                 }
  93
  94         }
  95
  96         /**
  97          * Enumeration for all recognized link types.
  98          *
  99          * @author <a href="mailto:bombe@pterodactylus.net">David ‘Bombe’ Roden</a>
 100          */
 101         private enum LinkType {
 102
 103                 KSK("KSK@", true),
 104                 CHK("CHK@", true),
 105                 SSK("SSK@", true),
 106                 USK("USK@", true),
 107                 HTTP("http://", false),
 108                 HTTPS("https://", false),
 109                 SONE("sone://", false),
 110                 POST("post://", false),
 111
 112                 FREEMAIL("", true) {
 113                         @Override
 114                         public Optional<NextLink> findNext(String line) {
 115                                 int nextFreemailSuffix = line.indexOf(".freemail");
 116                                 if (nextFreemailSuffix < 54) {
 117                                         /* 52 chars for the id, 1 on @, at least 1 for the local part. */
 118                                         return absent();
 119                                 }
 120                                 if (line.charAt(nextFreemailSuffix - 53) != '@') {
 121                                         return absent();
 122                                 }
 123                                 if (!line.substring(nextFreemailSuffix - 52, nextFreemailSuffix).matches("^[a-z2-7]*$")) {
 124                                         return absent();
 125                                 }
 126                                 int startOfLocalPart = nextFreemailSuffix - 54;
 127                                 if (!isAllowedInLocalPart(line.charAt(startOfLocalPart))) {
 128                                         return absent();
 129                                 }
 130                                 while ((startOfLocalPart > 0) && isAllowedInLocalPart(line.charAt(startOfLocalPart - 1))) {
 131                                         startOfLocalPart--;
 132                                 }
 133                                 return of(new NextLink(startOfLocalPart, line.substring(startOfLocalPart, nextFreemailSuffix + 9), line.substring(nextFreemailSuffix + 9), this));
 134                         }
 135
 136                         private boolean isAllowedInLocalPart(char character) {
 137                                 return ((character >= 'A') && (character <= 'Z'))
 138                                                 || ((character >= 'a') && (character <= 'z'))
 139                                                 || ((character >= '0') && (character <= '9'))
 140                                                 || (character == '.') || (character == '-') || (character == '_');
 141                         }
 142                 };
 143
 144                 private final String scheme;
 145                 private final boolean freenetLink;
 146
 147                 LinkType(String scheme, boolean freenetLink) {
 148                         this.scheme = scheme;
 149                         this.freenetLink = freenetLink;
 150                 }
 151
 152                 /**
 153                  * Returns the scheme of this link type.
 154                  *
 155                  * @return The scheme of this link type
 156                  */
 157                 public String getScheme() {
 158                         return scheme;
 159                 }
 160
 161                 public boolean isFreenetLink() {
 162                         return freenetLink;
 163                 }
 164
 165                 public Optional<NextLink> findNext(String line) {
 166                         int nextLinkPosition = line.indexOf(getScheme());
 167                         if (nextLinkPosition == -1) {
 168                                 return absent();
 169                         }
 170                         int endOfLink = findEndOfLink(line.substring(nextLinkPosition));
 171                         return of(new NextLink(nextLinkPosition, line.substring(nextLinkPosition, nextLinkPosition + endOfLink), line.substring(nextLinkPosition + endOfLink), this));
 172                 }
 173
 174                 private static int findEndOfLink(String line) {
 175                         Matcher matcher = whitespacePattern.matcher(line);
 176                         int endOfLink = matcher.find() ? matcher.start() : line.length();
 177                         while (isPunctuation(line.charAt(endOfLink - 1))) {
 178                                 endOfLink--;
 179                         }
 180                         int openParens = 0;
 181                         for (int i = 0; i < endOfLink; i++) {
 182                                 switch (line.charAt(i)) {
 183                                         case '(':
 184                                                 openParens++;
 185                                                 break;
 186                                         case ')':
 187                                                 openParens--;
 188                                                 if (openParens < 0) {
 189                                                         return i;
 190                                                 }
 191                                         default:
 192                                 }
 193                         }
 194                         return endOfLink;
 195                 }
 196
 197         }
 198
 199         /** The Sone provider. */
 200         private final SoneProvider soneProvider;
 201
 202         /** The post provider. */
 203         private final PostProvider postProvider;
 204
 205         /**
 206          * Creates a new freenet link parser.
 207          *
 208          * @param soneProvider
 209          *            The Sone provider
 210          * @param postProvider
 211          *            The post provider
 212          */
 213         public SoneTextParser(SoneProvider soneProvider, PostProvider postProvider) {
 214                 this.soneProvider = soneProvider;
 215                 this.postProvider = postProvider;
 216         }
 217
 218         //
 219         // PART METHODS
 220         //
 221
 222         /**
 223          * {@inheritDoc}
 224          */
 225         @Nonnull
 226         @Override
 227         public Iterable<Part> parse(@Nonnull String source, @Nullable SoneTextParserContext context) {
 228                 List<Part> parts = new ArrayList<>();
 229                 try (Reader sourceReader = new StringReader(source);
 230                                 BufferedReader bufferedReader = new BufferedReader(sourceReader)) {
 231                         String line;
 232                         boolean lastLineEmpty = true;
 233                         int emptyLines = 0;
 234                         while ((line = bufferedReader.readLine()) != null) {
 235                                 if (line.trim().length() == 0) {
 236                                         if (lastLineEmpty) {
 237                                                 continue;
 238                                         }
 239                                         parts.add(new PlainTextPart("\n"));
 240                                         ++emptyLines;
 241                                         lastLineEmpty = emptyLines == 2;
 242                                         continue;
 243                                 }
 244                                 emptyLines = 0;
 245                                 /*
 246                                  * lineComplete tracks whether the block you are parsing is the
 247                                  * first block of the line. this is important because sometimes
 248                                  * you have to add an additional line break.
 249                                  */
 250                                 boolean lineComplete = true;
 251                                 while (line.length() > 0) {
 252                                         Optional<NextLink> nextLink = findNextLink(line);
 253                                         if (!nextLink.isPresent()) {
 254                                                 if (lineComplete && !lastLineEmpty) {
 255                                                         parts.add(new PlainTextPart("\n" + line));
 256                                                 } else {
 257                                                         parts.add(new PlainTextPart(line));
 258                                                 }
 259                                                 break;
 260                                         }
 261                                         LinkType linkType = nextLink.get().getLinkType();
 262                                         int next = nextLink.get().getPosition();
 263
 264                                         /* cut off “freenet:” from before keys. */
 265                                         if (linkType.isFreenetLink() && (next >= 8) && (line.substring(next - 8, next).equals("freenet:"))) {
 266                                                 next -= 8;
 267                                                 line = line.substring(0, next) + line.substring(next + 8);
 268                                         }
 269
 270                                         /* if there is text before the next item, write it out. */
 271                                         if (lineComplete && !lastLineEmpty) {
 272                                                 parts.add(new PlainTextPart("\n"));
 273                                         }
 274                                         if (next > 0) {
 275                                                 parts.add(new PlainTextPart(line.substring(0, next)));
 276                                                 line = line.substring(next);
 277                                         }
 278                                         lineComplete = false;
 279
 280                                         String link = nextLink.get().getLink();
 281                                         logger.log(Level.FINER, String.format("Found link: %s", link));
 282
 283                                         /* if there is no text after the scheme, it’s not a link! */
 284                                         if (link.equals(linkType.getScheme())) {
 285                                                 parts.add(new PlainTextPart(linkType.getScheme()));
 286                                                 line = line.substring(linkType.getScheme().length());
 287                                                 continue;
 288                                         }
 289
 290                                         switch (linkType) {
 291                                                 case SONE:
 292                                                         renderSoneLink(parts, link);
 293                                                         break;
 294                                                 case POST:
 295                                                         renderPostLink(parts, link);
 296                                                         break;
 297                                                 case KSK:
 298                                                 case CHK:
 299                                                 case SSK:
 300                                                 case USK:
 301                                                         renderFreenetLink(parts, link, linkType, context);
 302                                                         break;
 303                                                 case HTTP:
 304                                                 case HTTPS:
 305                                                         renderHttpLink(parts, link, linkType);
 306                                                         break;
 307                                                 case FREEMAIL:
 308                                                         renderFreemailLink(parts, link);
 309                                         }
 310
 311                                         line = nextLink.get().getRemainder();
 312                                 }
 313                                 lastLineEmpty = false;
 314                         }
 315                 } catch (IOException ioe1) {
 316                         // a buffered reader around a string reader should never throw.
 317                         throw new RuntimeException(ioe1);
 318                 }
 319                 for (int partIndex = parts.size() - 1; partIndex >= 0; --partIndex) {
 320                         Part part = parts.get(partIndex);
 321                         if (!(part instanceof PlainTextPart) || !"\n".equals(part.getText())) {
 322                                 break;
 323                         }
 324                         parts.remove(partIndex);
 325                 }
 326                 return parts;
 327         }
 328
 329         public static Optional<NextLink> findNextLink(String line) {
 330                 int earliestLinkPosition = Integer.MAX_VALUE;
 331                 NextLink earliestNextLink = null;
 332                 for (LinkType possibleLinkType : LinkType.values()) {
 333                         Optional<NextLink> nextLink = possibleLinkType.findNext(line);
 334                         if (nextLink.isPresent()) {
 335                                 if (nextLink.get().getPosition() < earliestLinkPosition) {
 336                                         earliestNextLink = nextLink.get();
 337                                 }
 338                         }
 339                 }
 340                 return Optional.fromNullable(earliestNextLink);
 341         }
 342
 343         private void renderSoneLink(List<Part> parts, String line) {
 344                 if (line.length() >= (7 + 43)) {
 345                         String soneId = line.substring(7, 50);
 346                         Optional<Sone> sone = soneProvider.getSone(soneId);
 347                         parts.add(new SonePart(sone.or(new IdOnlySone(soneId))));
 348                 } else {
 349                         parts.add(new PlainTextPart(line));
 350                 }
 351         }
 352
 353         private void renderPostLink(List<Part> parts, String line) {
 354                 if (line.length() >= (7 + 36)) {
 355                         String postId = line.substring(7, 43);
 356                         Optional<Post> post = postProvider.getPost(postId);
 357                         if (post.isPresent()) {
 358                                 parts.add(new PostPart(post.get()));
 359                         } else {
 360                                 parts.add(new PlainTextPart(line.substring(0, 43)));
 361                         }
 362                 } else {
 363                         parts.add(new PlainTextPart(line));
 364                 }
 365         }
 366
 367         private void renderFreenetLink(List<Part> parts, String link, LinkType linkType, @Nullable SoneTextParserContext context) {
 368                 String name = link;
 369                 String linkWithoutParameters = link;
 370                 if (name.indexOf('?') > -1) {
 371                         linkWithoutParameters = name = name.substring(0, name.indexOf('?'));
 372                 }
 373                 if (name.endsWith("/")) {
 374                         name = name.substring(0, name.length() - 1);
 375                 }
 376                 try {
 377                         FreenetURI uri = new FreenetURI(name);
 378                         name = uri.lastMetaString();
 379                         if (name == null) {
 380                                 name = uri.getDocName();
 381                         }
 382                         if (name == null) {
 383                                 name = link.substring(0, Math.min(9, link.length()));
 384                         }
 385                         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());
 386                         parts.add(new FreenetLinkPart(link, name, linkWithoutParameters, fromPostingSone));
 387                 } catch (MalformedURLException mue1) {
 388                         /* not a valid link, insert as plain text. */
 389                         parts.add(new PlainTextPart(link));
 390                 } catch (NullPointerException npe1) {
 391                         /* FreenetURI sometimes throws these, too. */
 392                         parts.add(new PlainTextPart(link));
 393                 } catch (ArrayIndexOutOfBoundsException aioobe1) {
 394                         /* oh, and these, too. */
 395                         parts.add(new PlainTextPart(link));
 396                 }
 397         }
 398
 399         private void renderHttpLink(List<Part> parts, String link, LinkType linkType) {
 400                 String name = link.substring(linkType == LinkType.HTTP ? 7 : 8);
 401                 int firstSlash = name.indexOf('/');
 402                 int lastSlash = name.lastIndexOf('/');
 403                 if ((lastSlash - firstSlash) > 3) {
 404                         name = name.substring(0, firstSlash + 1) + "…" + name.substring(lastSlash);
 405                 }
 406                 if (name.endsWith("/")) {
 407                         name = name.substring(0, name.length() - 1);
 408                 }
 409                 if (((name.indexOf('/') > -1) && (name.indexOf('.') < name.lastIndexOf('.', name.indexOf('/'))) || ((name.indexOf('/') == -1) && (name.indexOf('.') < name.lastIndexOf('.')))) && name.startsWith("www.")) {
 410                         name = name.substring(4);
 411                 }
 412                 if (name.indexOf('?') > -1) {
 413                         name = name.substring(0, name.indexOf('?'));
 414                 }
 415                 parts.add(new LinkPart(link, name));
 416         }
 417
 418         private void renderFreemailLink(List<Part> parts, String line) {
 419                 int separator = line.indexOf('@');
 420                 String freemailId = line.substring(separator + 1, separator + 53);
 421                 String identityId = Base64.encode(Base32.decode(freemailId));
 422                 String emailLocalPart = line.substring(0, separator);
 423                 parts.add(new FreemailPart(emailLocalPart, freemailId, identityId));
 424         }
 425
 426         private static boolean isPunctuation(char character) {
 427                 return (character == '.') || (character == ',') || (character == '!') || (character == '?');
 428         }
 429
 430 }