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