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.Core;
  30 import net.pterodactylus.sone.data.Post;
  31 import net.pterodactylus.sone.data.Sone;
  32 import net.pterodactylus.util.logging.Logging;
  33 import freenet.keys.FreenetURI;
  34
  35 /**
  36  * {@link Parser} implementation that can recognize Freenet URIs.
  37  *
  38  * @author <a href="mailto:bombe@pterodactylus.net">David ‘Bombe’ Roden</a>
  39  */
  40 public class SoneTextParser implements Parser<SoneTextParserContext> {
  41
  42         /** The logger. */
  43         private static final Logger logger = Logging.getLogger(SoneTextParser.class);
  44
  45         /** Pattern to detect whitespace. */
  46         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]");
  47
  48         /**
  49          * Enumeration for all recognized link types.
  50          *
  51          * @author <a href="mailto:bombe@pterodactylus.net">David ‘Bombe’ Roden</a>
  52          */
  53         private enum LinkType {
  54
  55                 /** Link is a KSK. */
  56                 KSK,
  57
  58                 /** Link is a CHK. */
  59                 CHK,
  60
  61                 /** Link is an SSK. */
  62                 SSK,
  63
  64                 /** Link is a USK. */
  65                 USK,
  66
  67                 /** Link is HTTP. */
  68                 HTTP,
  69
  70                 /** Link is HTTPS. */
  71                 HTTPS,
  72
  73                 /** Link is a Sone. */
  74                 SONE,
  75
  76                 /** Link is a post. */
  77                 POST,
  78
  79         }
  80
  81         /** The core. */
  82         private final Core core;
  83
  84         /**
  85          * Creates a new freenet link parser.
  86          *
  87          * @param core
  88          *            The core
  89          */
  90         public SoneTextParser(Core core) {
  91                 this.core = core;
  92         }
  93
  94         //
  95         // PART METHODS
  96         //
  97
  98         /**
  99          * {@inheritDoc}
 100          */
 101         @Override
 102         public Iterable<Part> parse(SoneTextParserContext context, Reader source) throws IOException {
 103                 PartContainer parts = new PartContainer();
 104                 BufferedReader bufferedReader = (source instanceof BufferedReader) ? (BufferedReader) source : new BufferedReader(source);
 105                 String line;
 106                 boolean lastLineEmpty = true;
 107                 int emptyLines = 0;
 108                 while ((line = bufferedReader.readLine()) != null) {
 109                         if (line.trim().length() == 0) {
 110                                 if (lastLineEmpty) {
 111                                         continue;
 112                                 }
 113                                 parts.add(new PlainTextPart("\n"));
 114                                 ++emptyLines;
 115                                 lastLineEmpty = emptyLines == 2;
 116                                 continue;
 117                         }
 118                         emptyLines = 0;
 119                         boolean lineComplete = true;
 120                         while (line.length() > 0) {
 121                                 int nextKsk = line.indexOf("KSK@");
 122                                 int nextChk = line.indexOf("CHK@");
 123                                 int nextSsk = line.indexOf("SSK@");
 124                                 int nextUsk = line.indexOf("USK@");
 125                                 int nextHttp = line.indexOf("http://");
 126                                 int nextHttps = line.indexOf("https://");
 127                                 int nextSone = line.indexOf("sone://");
 128                                 int nextPost = line.indexOf("post://");
 129                                 if ((nextKsk == -1) && (nextChk == -1) && (nextSsk == -1) && (nextUsk == -1) && (nextHttp == -1) && (nextHttps == -1) && (nextSone == -1) && (nextPost == -1)) {
 130                                         if (lineComplete && !lastLineEmpty) {
 131                                                 parts.add(new PlainTextPart("\n" + line));
 132                                         } else {
 133                                                 parts.add(new PlainTextPart(line));
 134                                         }
 135                                         break;
 136                                 }
 137                                 int next = Integer.MAX_VALUE;
 138                                 LinkType linkType = null;
 139                                 if ((nextKsk > -1) && (nextKsk < next)) {
 140                                         next = nextKsk;
 141                                         linkType = LinkType.KSK;
 142                                 }
 143                                 if ((nextChk > -1) && (nextChk < next)) {
 144                                         next = nextChk;
 145                                         linkType = LinkType.CHK;
 146                                 }
 147                                 if ((nextSsk > -1) && (nextSsk < next)) {
 148                                         next = nextSsk;
 149                                         linkType = LinkType.SSK;
 150                                 }
 151                                 if ((nextUsk > -1) && (nextUsk < next)) {
 152                                         next = nextUsk;
 153                                         linkType = LinkType.USK;
 154                                 }
 155                                 if ((nextHttp > -1) && (nextHttp < next)) {
 156                                         next = nextHttp;
 157                                         linkType = LinkType.HTTP;
 158                                 }
 159                                 if ((nextHttps > -1) && (nextHttps < next)) {
 160                                         next = nextHttps;
 161                                         linkType = LinkType.HTTPS;
 162                                 }
 163                                 if ((nextSone > -1) && (nextSone < next)) {
 164                                         next = nextSone;
 165                                         linkType = LinkType.SONE;
 166                                 }
 167                                 if ((nextPost > -1) && (nextPost < next)) {
 168                                         next = nextPost;
 169                                         linkType = LinkType.POST;
 170                                 }
 171                                 if (linkType == LinkType.SONE) {
 172                                         if (next > 0) {
 173                                                 parts.add(new PlainTextPart(line.substring(0, next)));
 174                                         }
 175                                         if (line.length() >= (next + 7 + 43)) {
 176                                                 String soneId = line.substring(next + 7, next + 50);
 177                                                 Sone sone = core.getSone(soneId, false);
 178                                                 if (sone != null) {
 179                                                         parts.add(new SonePart(sone));
 180                                                 } else {
 181                                                         parts.add(new PlainTextPart(line.substring(next, next + 50)));
 182                                                 }
 183                                                 line = line.substring(next + 50);
 184                                         } else {
 185                                                 parts.add(new PlainTextPart(line.substring(next)));
 186                                                 line = "";
 187                                         }
 188                                         continue;
 189                                 }
 190                                 if (linkType == LinkType.POST) {
 191                                         if (next > 0) {
 192                                                 parts.add(new PlainTextPart(line.substring(0, next)));
 193                                         }
 194                                         if (line.length() >= (next + 7 + 36)) {
 195                                                 String postId = line.substring(next + 7, next + 43);
 196                                                 Post post = core.getPost(postId, false);
 197                                                 if ((post != null) && (post.getSone() != null)) {
 198                                                         String postText = post.getText();
 199                                                         postText = postText.substring(0, Math.min(postText.length(), 20)) + "…";
 200                                                         parts.add(new PostPart(post));
 201                                                 } else {
 202                                                         parts.add(new PlainTextPart(line.substring(next, next + 43)));
 203                                                 }
 204                                                 line = line.substring(next + 43);
 205                                         } else {
 206                                                 parts.add(new PlainTextPart(line.substring(next)));
 207                                                 line = "";
 208                                         }
 209                                         continue;
 210                                 }
 211                                 if ((next >= 8) && (line.substring(next - 8, next).equals("freenet:"))) {
 212                                         next -= 8;
 213                                         line = line.substring(0, next) + line.substring(next + 8);
 214                                 }
 215                                 Matcher matcher = whitespacePattern.matcher(line);
 216                                 int nextSpace = matcher.find(next) ? matcher.start() : line.length();
 217                                 if (nextSpace > (next + 4)) {
 218                                         if (!lastLineEmpty && lineComplete) {
 219                                                 parts.add(new PlainTextPart("\n" + line.substring(0, next)));
 220                                         } else {
 221                                                 parts.add(new PlainTextPart(line.substring(0, next)));
 222                                         }
 223                                         String link = line.substring(next, nextSpace);
 224                                         String name = link;
 225                                         logger.log(Level.FINER, "Found link: %s", link);
 226                                         logger.log(Level.FINEST, "Next: %d, CHK: %d, SSK: %d, USK: %d", new Object[] { next, nextChk, nextSsk, nextUsk });
 227
 228                                         if ((linkType == LinkType.KSK) || (linkType == LinkType.CHK) || (linkType == LinkType.SSK) || (linkType == LinkType.USK)) {
 229                                                 FreenetURI uri;
 230                                                 if (name.indexOf('?') > -1) {
 231                                                         name = name.substring(0, name.indexOf('?'));
 232                                                 }
 233                                                 if (name.endsWith("/")) {
 234                                                         name = name.substring(0, name.length() - 1);
 235                                                 }
 236                                                 try {
 237                                                         uri = new FreenetURI(name);
 238                                                         name = uri.lastMetaString();
 239                                                         if (name == null) {
 240                                                                 name = uri.getDocName();
 241                                                         }
 242                                                         if (name == null) {
 243                                                                 name = link.substring(0, Math.min(9, link.length()));
 244                                                         }
 245                                                         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());
 246                                                         parts.add(new FreenetLinkPart(link, name, fromPostingSone));
 247                                                 } catch (MalformedURLException mue1) {
 248                                                         /* not a valid link, insert as plain text. */
 249                                                         parts.add(new PlainTextPart(link));
 250                                                 } catch (NullPointerException npe1) {
 251                                                         /* FreenetURI sometimes throws these, too. */
 252                                                         parts.add(new PlainTextPart(link));
 253                                                 } catch (ArrayIndexOutOfBoundsException aioobe1) {
 254                                                         /* oh, and these, too. */
 255                                                         parts.add(new PlainTextPart(link));
 256                                                 }
 257                                         } else if ((linkType == LinkType.HTTP) || (linkType == LinkType.HTTPS)) {
 258                                                 name = link.substring(linkType == LinkType.HTTP ? 7 : 8);
 259                                                 int firstSlash = name.indexOf('/');
 260                                                 int lastSlash = name.lastIndexOf('/');
 261                                                 if ((lastSlash - firstSlash) > 3) {
 262                                                         name = name.substring(0, firstSlash + 1) + "…" + name.substring(lastSlash);
 263                                                 }
 264                                                 if (name.endsWith("/")) {
 265                                                         name = name.substring(0, name.length() - 1);
 266                                                 }
 267                                                 if (((name.indexOf('/') > -1) && (name.indexOf('.') < name.lastIndexOf('.', name.indexOf('/'))) || ((name.indexOf('/') == -1) && (name.indexOf('.') < name.lastIndexOf('.')))) && name.startsWith("www.")) {
 268                                                         name = name.substring(4);
 269                                                 }
 270                                                 if (name.indexOf('?') > -1) {
 271                                                         name = name.substring(0, name.indexOf('?'));
 272                                                 }
 273                                                 parts.add(new LinkPart(link, name));
 274                                         }
 275                                         line = line.substring(nextSpace);
 276                                 } else {
 277                                         if (!lastLineEmpty && lineComplete) {
 278                                                 parts.add(new PlainTextPart("\n" + line.substring(0, next + 4)));
 279                                         } else {
 280                                                 parts.add(new PlainTextPart(line.substring(0, next + 4)));
 281                                         }
 282                                         line = line.substring(next + 4);
 283                                 }
 284                                 lineComplete = false;
 285                         }
 286                         lastLineEmpty = false;
 287                 }
 288                 for (int partIndex = parts.size() - 1; partIndex >= 0; --partIndex) {
 289                         Part part = parts.getPart(partIndex);
 290                         if (!(part instanceof PlainTextPart) || !"\n".equals(((PlainTextPart) part).getText())) {
 291                                 break;
 292                         }
 293                         parts.removePart(partIndex);
 294                 }
 295                 return parts;
 296         }
 297
 298 }