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