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