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