src/main/java/net/pterodactylus/sone/text/FreenetLinkParser.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.io.StringReader;
  24 import java.net.MalformedURLException;
  25 import java.util.logging.Level;
  26 import java.util.logging.Logger;
  27 import java.util.regex.Matcher;
  28 import java.util.regex.Pattern;
  29
  30 import net.pterodactylus.sone.data.Sone;
  31 import net.pterodactylus.util.logging.Logging;
  32 import net.pterodactylus.util.template.TemplateFactory;
  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 FreenetLinkParser implements Parser {
  41
  42         /** The logger. */
  43         private static final Logger logger = Logging.getLogger(FreenetLinkParser.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         }
  74
  75         /** The template factory. */
  76         private final TemplateFactory templateFactory;
  77
  78         /** The Sone that posted the currently parsed text. */
  79         private Sone postingSone;
  80
  81         /**
  82          * Creates a new freenet link parser.
  83          *
  84          * @param templateFactory
  85          *            The template factory
  86          */
  87         public FreenetLinkParser(TemplateFactory templateFactory) {
  88                 this.templateFactory = templateFactory;
  89         }
  90
  91         //
  92         // ACCESSORS
  93         //
  94
  95         /**
  96          * Sets the Sone that posted the text that will be parsed in the next call
  97          * to {@link #parse(Reader)}. You need to synchronize calling this method
  98          * and {@link #parse(Reader)}!
  99          *
 100          * @param sone
 101          *            The Sone that posted the text
 102          */
 103         public void setPostingSone(Sone sone) {
 104                 postingSone = sone;
 105         }
 106
 107         //
 108         // PART METHODS
 109         //
 110
 111         /**
 112          * {@inheritDoc}
 113          */
 114         @Override
 115         public Part parse(Reader source) throws IOException {
 116                 PartContainer parts = new PartContainer();
 117                 BufferedReader bufferedReader = (source instanceof BufferedReader) ? (BufferedReader) source : new BufferedReader(source);
 118                 String line;
 119                 while ((line = bufferedReader.readLine()) != null) {
 120                         line = line.trim() + "\n";
 121                         while (line.length() > 0) {
 122                                 int nextKsk = line.indexOf("KSK@");
 123                                 int nextChk = line.indexOf("CHK@");
 124                                 int nextSsk = line.indexOf("SSK@");
 125                                 int nextUsk = line.indexOf("USK@");
 126                                 int nextHttp = line.indexOf("http://");
 127                                 int nextHttps = line.indexOf("https://");
 128                                 if ((nextKsk == -1) && (nextChk == -1) && (nextSsk == -1) && (nextUsk == -1) && (nextHttp == -1) && (nextHttps == -1)) {
 129                                         parts.add(createPlainTextPart(line));
 130                                         break;
 131                                 }
 132                                 int next = Integer.MAX_VALUE;
 133                                 LinkType linkType = null;
 134                                 if ((nextKsk > -1) && (nextKsk < next)) {
 135                                         next = nextKsk;
 136                                         linkType = LinkType.KSK;
 137                                 }
 138                                 if ((nextChk > -1) && (nextChk < next)) {
 139                                         next = nextChk;
 140                                         linkType = LinkType.CHK;
 141                                 }
 142                                 if ((nextSsk > -1) && (nextSsk < next)) {
 143                                         next = nextSsk;
 144                                         linkType = LinkType.SSK;
 145                                 }
 146                                 if ((nextUsk > -1) && (nextUsk < next)) {
 147                                         next = nextUsk;
 148                                         linkType = LinkType.USK;
 149                                 }
 150                                 if ((nextHttp > -1) && (nextHttp < next)) {
 151                                         next = nextHttp;
 152                                         linkType = LinkType.HTTP;
 153                                 }
 154                                 if ((nextHttps > -1) && (nextHttps < next)) {
 155                                         next = nextHttps;
 156                                         linkType = LinkType.HTTPS;
 157                                 }
 158                                 if ((next >= 8) && (line.substring(next - 8, next).equals("freenet:"))) {
 159                                         next -= 8;
 160                                         line = line.substring(0, next) + line.substring(next + 8);
 161                                 }
 162                                 Matcher matcher = whitespacePattern.matcher(line);
 163                                 int nextSpace = matcher.find(next) ? matcher.start() : line.length();
 164                                 if (nextSpace > (next + 4)) {
 165                                         parts.add(createPlainTextPart(line.substring(0, next)));
 166                                         String link = line.substring(next, nextSpace);
 167                                         String name = link;
 168                                         logger.log(Level.FINER, "Found link: %s", link);
 169                                         logger.log(Level.FINEST, "Next: %d, CHK: %d, SSK: %d, USK: %d", new Object[] { next, nextChk, nextSsk, nextUsk });
 170                                         if (linkType == LinkType.KSK) {
 171                                                 name = link.substring(4);
 172                                         } else if ((linkType == LinkType.CHK) || (linkType == LinkType.SSK) || (linkType == LinkType.USK)) {
 173                                                 if (name.indexOf('/') > -1) {
 174                                                         if (!name.endsWith("/")) {
 175                                                                 name = name.substring(name.lastIndexOf('/') + 1);
 176                                                         } else {
 177                                                                 if (name.indexOf('/') != name.lastIndexOf('/')) {
 178                                                                         name = name.substring(name.lastIndexOf('/', name.lastIndexOf('/') - 1));
 179                                                                 } else {
 180                                                                         /* shorten to 5 chars. */
 181                                                                         name = name.substring(4, Math.min(9, name.length()));
 182                                                                 }
 183                                                         }
 184                                                 }
 185                                                 if (name.indexOf('?') > -1) {
 186                                                         name = name.substring(0, name.indexOf('?'));
 187                                                 }
 188                                                 boolean fromPostingSone = false;
 189                                                 if ((linkType == LinkType.SSK) || (linkType == LinkType.USK)) {
 190                                                         try {
 191                                                                 new FreenetURI(link);
 192                                                                 fromPostingSone = link.substring(4, 47).equals(postingSone.getId());
 193                                                                 parts.add(fromPostingSone ? createTrustedFreenetLinkPart(link, name) : createFreenetLinkPart(link, name));
 194                                                         } catch (MalformedURLException mue1) {
 195                                                                 /* it’s not a valid link. */
 196                                                                 parts.add(createPlainTextPart(link));
 197                                                         }
 198                                                 } else {
 199                                                         parts.add(fromPostingSone ? createTrustedFreenetLinkPart(link, name) : createFreenetLinkPart(link, name));
 200                                                 }
 201                                         } else if ((linkType == LinkType.HTTP) || (linkType == LinkType.HTTPS)) {
 202                                                 name = link.substring(linkType == LinkType.HTTP ? 7 : 8);
 203                                                 int firstSlash = name.indexOf('/');
 204                                                 int lastSlash = name.lastIndexOf('/');
 205                                                 if ((lastSlash - firstSlash) > 3) {
 206                                                         name = name.substring(0, firstSlash + 1) + "…" + name.substring(lastSlash);
 207                                                 }
 208                                                 if (name.endsWith("/")) {
 209                                                         name = name.substring(0, name.length() - 1);
 210                                                 }
 211                                                 if (((name.indexOf('/') > -1) && (name.indexOf('.') < name.lastIndexOf('.', name.indexOf('/'))) || ((name.indexOf('/') == -1) && (name.indexOf('.') < name.lastIndexOf('.')))) && name.startsWith("www.")) {
 212                                                         name = name.substring(4);
 213                                                 }
 214                                                 if (name.indexOf('?') > -1) {
 215                                                         name = name.substring(0, name.indexOf('?'));
 216                                                 }
 217                                                 link = "?_CHECKED_HTTP_=" + link;
 218                                                 parts.add(createInternetLinkPart(link, name));
 219                                         }
 220                                         line = line.substring(nextSpace);
 221                                 } else {
 222                                         parts.add(createPlainTextPart(line.substring(0, next + 4)));
 223                                         line = line.substring(next + 4);
 224                                 }
 225                         }
 226                 }
 227                 return parts;
 228         }
 229
 230         //
 231         // PRIVATE METHODS
 232         //
 233
 234         /**
 235          * Creates a new plain text part based on a template.
 236          *
 237          * @param text
 238          *            The text to display
 239          * @return The part that displays the given text
 240          */
 241         private Part createPlainTextPart(String text) {
 242                 return new TemplatePart(templateFactory.createTemplate(new StringReader("<% text|html>"))).set("text", text);
 243         }
 244
 245         /**
 246          * Creates a new part based on a template that links to a site within the
 247          * normal internet.
 248          *
 249          * @param link
 250          *            The target of the link
 251          * @param name
 252          *            The name of the link
 253          * @return The part that displays the link
 254          */
 255         private Part createInternetLinkPart(String link, String name) {
 256                 return new TemplatePart(templateFactory.createTemplate(new StringReader("<a class=\"internet\" href=\"/<% link|html>\" title=\"<% link|html>\"><% name|html></a>"))).set("link", link).set("name", name);
 257         }
 258
 259         /**
 260          * Creates a new part based on a template that links to a site within
 261          * freenet.
 262          *
 263          * @param link
 264          *            The target of the link
 265          * @param name
 266          *            The name of the link
 267          * @return The part that displays the link
 268          */
 269         private Part createFreenetLinkPart(String link, String name) {
 270                 return new TemplatePart(templateFactory.createTemplate(new StringReader("<a class=\"freenet\" href=\"/<% link|html>\" title=\"<% link|html>\"><% name|html></a>"))).set("link", link).set("name", name);
 271         }
 272
 273         /**
 274          * Creates a new part based on a template that links to a page in the
 275          * poster’s keyspace.
 276          *
 277          * @param link
 278          *            The target of the link
 279          * @param name
 280          *            The name of the link
 281          * @return The part that displays the link
 282          */
 283         private Part createTrustedFreenetLinkPart(String link, String name) {
 284                 return new TemplatePart(templateFactory.createTemplate(new StringReader("<a class=\"freenet-trusted\" href=\"/<% link|html>\" title=\"<% link|html>\"><% name|html></a>"))).set("link", link).set("name", name);
 285         }
 286
 287 }