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

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