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

   1 /*
   2  * Sone - FreenetLinkParser.java - Copyright © 2010–2012 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 net.pterodactylus.sone.core.PostProvider;
  30 import net.pterodactylus.sone.core.SoneProvider;
  31 import net.pterodactylus.sone.data.Post;
  32 import net.pterodactylus.sone.data.Sone;
  33 import net.pterodactylus.sone.database.memory.MemorySone;
  34 import net.pterodactylus.util.io.Closer;
  35 import net.pterodactylus.util.logging.Logging;
  36 import freenet.keys.FreenetURI;
  37
  38 /**
  39  * {@link Parser} implementation that can recognize Freenet URIs.
  40  *
  41  * @author <a href="mailto:bombe@pterodactylus.net">David ‘Bombe’ Roden</a>
  42  */
  43 public class SoneTextParser implements Parser<SoneTextParserContext> {
  44
  45         /** The logger. */
  46         private static final Logger logger = Logging.getLogger(SoneTextParser.class);
  47
  48         /** Pattern to detect whitespace. */
  49         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]");
  50
  51         /**
  52          * Enumeration for all recognized link types.
  53          *
  54          * @author <a href="mailto:bombe@pterodactylus.net">David ‘Bombe’ Roden</a>
  55          */
  56         private enum LinkType {
  57
  58                 /** Link is a KSK. */
  59                 KSK,
  60
  61                 /** Link is a CHK. */
  62                 CHK,
  63
  64                 /** Link is an SSK. */
  65                 SSK,
  66
  67                 /** Link is a USK. */
  68                 USK,
  69
  70                 /** Link is HTTP. */
  71                 HTTP,
  72
  73                 /** Link is HTTPS. */
  74                 HTTPS,
  75
  76                 /** Link is a Sone. */
  77                 SONE,
  78
  79                 /** Link is a post. */
  80                 POST,
  81
  82         }
  83
  84         /** The Sone provider. */
  85         private final SoneProvider soneProvider;
  86
  87         /** The post provider. */
  88         private final PostProvider postProvider;
  89
  90         /**
  91          * Creates a new freenet link parser.
  92          *
  93          * @param soneProvider
  94          *            The Sone provider
  95          * @param postProvider
  96          *            The post provider
  97          */
  98         public SoneTextParser(SoneProvider soneProvider, PostProvider postProvider) {
  99                 this.soneProvider = soneProvider;
 100                 this.postProvider = postProvider;
 101         }
 102
 103         //
 104         // PART METHODS
 105         //
 106
 107         /**
 108          * {@inheritDoc}
 109          */
 110         @Override
 111         public Iterable<Part> parse(SoneTextParserContext context, Reader source) throws IOException {
 112                 PartContainer parts = new PartContainer();
 113                 BufferedReader bufferedReader = (source instanceof BufferedReader) ? (BufferedReader) source : new BufferedReader(source);
 114                 try {
 115                         String line;
 116                         boolean lastLineEmpty = true;
 117                         int emptyLines = 0;
 118                         while ((line = bufferedReader.readLine()) != null) {
 119                                 if (line.trim().length() == 0) {
 120                                         if (lastLineEmpty) {
 121                                                 continue;
 122                                         }
 123                                         parts.add(new PlainTextPart("\n"));
 124                                         ++emptyLines;
 125                                         lastLineEmpty = emptyLines == 2;
 126                                         continue;
 127                                 }
 128                                 emptyLines = 0;
 129                                 /*
 130                                  * lineComplete tracks whether the block you are parsing is the
 131                                  * first block of the line. this is important because sometimes
 132                                  * you have to add an additional line break.
 133                                  */
 134                                 boolean lineComplete = true;
 135                                 while (line.length() > 0) {
 136                                         int nextKsk = line.indexOf("KSK@");
 137                                         int nextChk = line.indexOf("CHK@");
 138                                         int nextSsk = line.indexOf("SSK@");
 139                                         int nextUsk = line.indexOf("USK@");
 140                                         int nextHttp = line.indexOf("http://");
 141                                         int nextHttps = line.indexOf("https://");
 142                                         int nextSone = line.indexOf("sone://");
 143                                         int nextPost = line.indexOf("post://");
 144                                         if ((nextKsk == -1) && (nextChk == -1) && (nextSsk == -1) && (nextUsk == -1) && (nextHttp == -1) && (nextHttps == -1) && (nextSone == -1) && (nextPost == -1)) {
 145                                                 if (lineComplete && !lastLineEmpty) {
 146                                                         parts.add(new PlainTextPart("\n" + line));
 147                                                 } else {
 148                                                         parts.add(new PlainTextPart(line));
 149                                                 }
 150                                                 break;
 151                                         }
 152                                         int next = Integer.MAX_VALUE;
 153                                         LinkType linkType = null;
 154                                         if ((nextKsk > -1) && (nextKsk < next)) {
 155                                                 next = nextKsk;
 156                                                 linkType = LinkType.KSK;
 157                                         }
 158                                         if ((nextChk > -1) && (nextChk < next)) {
 159                                                 next = nextChk;
 160                                                 linkType = LinkType.CHK;
 161                                         }
 162                                         if ((nextSsk > -1) && (nextSsk < next)) {
 163                                                 next = nextSsk;
 164                                                 linkType = LinkType.SSK;
 165                                         }
 166                                         if ((nextUsk > -1) && (nextUsk < next)) {
 167                                                 next = nextUsk;
 168                                                 linkType = LinkType.USK;
 169                                         }
 170                                         if ((nextHttp > -1) && (nextHttp < next)) {
 171                                                 next = nextHttp;
 172                                                 linkType = LinkType.HTTP;
 173                                         }
 174                                         if ((nextHttps > -1) && (nextHttps < next)) {
 175                                                 next = nextHttps;
 176                                                 linkType = LinkType.HTTPS;
 177                                         }
 178                                         if ((nextSone > -1) && (nextSone < next)) {
 179                                                 next = nextSone;
 180                                                 linkType = LinkType.SONE;
 181                                         }
 182                                         if ((nextPost > -1) && (nextPost < next)) {
 183                                                 next = nextPost;
 184                                                 linkType = LinkType.POST;
 185                                         }
 186
 187                                         /* cut off “freenet:” from before keys. */
 188                                         if (((linkType == LinkType.KSK) || (linkType == LinkType.CHK) || (linkType == LinkType.SSK) || (linkType == LinkType.USK)) && (next >= 8) && (line.substring(next - 8, next).equals("freenet:"))) {
 189                                                 next -= 8;
 190                                                 line = line.substring(0, next) + line.substring(next + 8);
 191                                         }
 192
 193                                         /* if there is text before the next item, write it out. */
 194                                         if (lineComplete && !lastLineEmpty) {
 195                                                 parts.add(new PlainTextPart("\n"));
 196                                         }
 197                                         if (next > 0) {
 198                                                 parts.add(new PlainTextPart(line.substring(0, next)));
 199                                                 line = line.substring(next);
 200                                                 next = 0;
 201                                         }
 202                                         lineComplete = false;
 203
 204                                         if (linkType == LinkType.SONE) {
 205                                                 if (line.length() >= (7 + 43)) {
 206                                                         String soneId = line.substring(7, 50);
 207                                                         Sone sone = soneProvider.getSone(soneId, false);
 208                                                         if (sone == null) {
 209                                                                 /*
 210                                                                  * don’t use create=true above, we don’t want
 211                                                                  * the empty shell.
 212                                                                  */
 213                                                                 sone = new MemorySone(soneId, false);
 214                                                         }
 215                                                         parts.add(new SonePart(sone));
 216                                                         line = line.substring(50);
 217                                                 } else {
 218                                                         parts.add(new PlainTextPart(line));
 219                                                         line = "";
 220                                                 }
 221                                                 continue;
 222                                         }
 223                                         if (linkType == LinkType.POST) {
 224                                                 if (line.length() >= (7 + 36)) {
 225                                                         String postId = line.substring(7, 43);
 226                                                         Post post = postProvider.getPost(postId, false);
 227                                                         if ((post != null) && (post.getSone() != null)) {
 228                                                                 parts.add(new PostPart(post));
 229                                                         } else {
 230                                                                 parts.add(new PlainTextPart(line.substring(0, 43)));
 231                                                         }
 232                                                         line = line.substring(43);
 233                                                 } else {
 234                                                         parts.add(new PlainTextPart(line));
 235                                                         line = "";
 236                                                 }
 237                                                 continue;
 238                                         }
 239                                         Matcher matcher = whitespacePattern.matcher(line);
 240                                         int nextSpace = matcher.find(0) ? matcher.start() : line.length();
 241                                         String link = line.substring(0, nextSpace);
 242                                         String name = link;
 243                                         logger.log(Level.FINER, String.format("Found link: %s", link));
 244                                         logger.log(Level.FINEST, String.format("CHK: %d, SSK: %d, USK: %d", nextChk, nextSsk, nextUsk));
 245
 246                                         if ((linkType == LinkType.KSK) || (linkType == LinkType.CHK) || (linkType == LinkType.SSK) || (linkType == LinkType.USK)) {
 247                                                 FreenetURI uri;
 248                                                 if (name.indexOf('?') > -1) {
 249                                                         name = name.substring(0, name.indexOf('?'));
 250                                                 }
 251                                                 if (name.endsWith("/")) {
 252                                                         name = name.substring(0, name.length() - 1);
 253                                                 }
 254                                                 try {
 255                                                         uri = new FreenetURI(name);
 256                                                         name = uri.lastMetaString();
 257                                                         if (name == null) {
 258                                                                 name = uri.getDocName();
 259                                                         }
 260                                                         if (name == null) {
 261                                                                 name = link.substring(0, Math.min(9, link.length()));
 262                                                         }
 263                                                         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());
 264                                                         parts.add(new FreenetLinkPart(link, name, fromPostingSone));
 265                                                 } catch (MalformedURLException mue1) {
 266                                                         /* not a valid link, insert as plain text. */
 267                                                         parts.add(new PlainTextPart(link));
 268                                                 } catch (NullPointerException npe1) {
 269                                                         /* FreenetURI sometimes throws these, too. */
 270                                                         parts.add(new PlainTextPart(link));
 271                                                 } catch (ArrayIndexOutOfBoundsException aioobe1) {
 272                                                         /* oh, and these, too. */
 273                                                         parts.add(new PlainTextPart(link));
 274                                                 }
 275                                         } else if ((linkType == LinkType.HTTP) || (linkType == LinkType.HTTPS)) {
 276                                                 name = link.substring(linkType == LinkType.HTTP ? 7 : 8);
 277                                                 int firstSlash = name.indexOf('/');
 278                                                 int lastSlash = name.lastIndexOf('/');
 279                                                 if ((lastSlash - firstSlash) > 3) {
 280                                                         name = name.substring(0, firstSlash + 1) + "…" + name.substring(lastSlash);
 281                                                 }
 282                                                 if (name.endsWith("/")) {
 283                                                         name = name.substring(0, name.length() - 1);
 284                                                 }
 285                                                 if (((name.indexOf('/') > -1) && (name.indexOf('.') < name.lastIndexOf('.', name.indexOf('/'))) || ((name.indexOf('/') == -1) && (name.indexOf('.') < name.lastIndexOf('.')))) && name.startsWith("www.")) {
 286                                                         name = name.substring(4);
 287                                                 }
 288                                                 if (name.indexOf('?') > -1) {
 289                                                         name = name.substring(0, name.indexOf('?'));
 290                                                 }
 291                                                 parts.add(new LinkPart(link, name));
 292                                         }
 293                                         line = line.substring(nextSpace);
 294                                 }
 295                                 lastLineEmpty = false;
 296                         }
 297                 } finally {
 298                         if (bufferedReader != source) {
 299                                 Closer.close(bufferedReader);
 300                         }
 301                 }
 302                 for (int partIndex = parts.size() - 1; partIndex >= 0; --partIndex) {
 303                         Part part = parts.getPart(partIndex);
 304                         if (!(part instanceof PlainTextPart) || !"\n".equals(((PlainTextPart) part).getText())) {
 305                                 break;
 306                         }
 307                         parts.removePart(partIndex);
 308                 }
 309                 return parts;
 310         }
 311
 312 }