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.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.Sone;
  30 import net.pterodactylus.util.logging.Logging;
  31 import net.pterodactylus.util.template.TemplateFactory;
  32
  33 /**
  34  * {@link Parser} implementation that can recognize Freenet URIs.
  35  *
  36  * @author <a href="mailto:bombe@pterodactylus.net">David ‘Bombe’ Roden</a>
  37  */
  38 public class FreenetLinkParser implements Parser {
  39
  40         /** The logger. */
  41         private static final Logger logger = Logging.getLogger(FreenetLinkParser.class);
  42
  43         /** Pattern to detect whitespace. */
  44         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]");
  45
  46         /**
  47          * Enumeration for all recognized link types.
  48          *
  49          * @author <a href="mailto:bombe@pterodactylus.net">David ‘Bombe’ Roden</a>
  50          */
  51         private enum LinkType {
  52
  53                 /** Link is a KSK. */
  54                 KSK,
  55
  56                 /** Link is a CHK. */
  57                 CHK,
  58
  59                 /** Link is an SSK. */
  60                 SSK,
  61
  62                 /** Link is a USK. */
  63                 USK,
  64
  65                 /** Link is HTTP. */
  66                 HTTP,
  67
  68                 /** Link is HTTPS. */
  69                 HTTPS;
  70
  71         }
  72
  73         /** The template factory. */
  74         private final TemplateFactory templateFactory;
  75
  76         /** The Sone that posted the currently parsed text. */
  77         private Sone postingSone;
  78
  79         /**
  80          * Creates a new freenet link parser.
  81          *
  82          * @param templateFactory
  83          *            The template factory
  84          */
  85         public FreenetLinkParser(TemplateFactory templateFactory) {
  86                 this.templateFactory = templateFactory;
  87         }
  88
  89         //
  90         // ACCESSORS
  91         //
  92
  93         /**
  94          * Sets the Sone that posted the text that will be parsed in the next call
  95          * to {@link #parse(Reader)}. You need to synchronize calling this method
  96          * and {@link #parse(Reader)}!
  97          *
  98          * @param sone
  99          *            The Sone that posted the text
 100          */
 101         public void setPostingSone(Sone sone) {
 102                 postingSone = sone;
 103         }
 104
 105         //
 106         // PART METHODS
 107         //
 108
 109         /**
 110          * {@inheritDoc}
 111          */
 112         @Override
 113         public Part parse(Reader source) throws IOException {
 114                 PartContainer parts = new PartContainer();
 115                 BufferedReader bufferedReader = (source instanceof BufferedReader) ? (BufferedReader) source : new BufferedReader(source);
 116                 String line;
 117                 while ((line = bufferedReader.readLine()) != null) {
 118                         line = line.trim() + "\n";
 119                         while (line.length() > 0) {
 120                                 int nextKsk = line.indexOf("KSK@");
 121                                 int nextChk = line.indexOf("CHK@");
 122                                 int nextSsk = line.indexOf("SSK@");
 123                                 int nextUsk = line.indexOf("USK@");
 124                                 int nextHttp = line.indexOf("http://");
 125                                 int nextHttps = line.indexOf("https://");
 126                                 if ((nextKsk == -1) && (nextChk == -1) && (nextSsk == -1) && (nextUsk == -1) && (nextHttp == -1) && (nextHttps == -1)) {
 127                                         parts.add(createPlainTextPart(line));
 128                                         break;
 129                                 }
 130                                 int next = Integer.MAX_VALUE;
 131                                 LinkType linkType = null;
 132                                 if ((nextKsk > -1) && (nextKsk < next)) {
 133                                         next = nextKsk;
 134                                         linkType = LinkType.KSK;
 135                                 }
 136                                 if ((nextChk > -1) && (nextChk < next)) {
 137                                         next = nextChk;
 138                                         linkType = LinkType.CHK;
 139                                 }
 140                                 if ((nextSsk > -1) && (nextSsk < next)) {
 141                                         next = nextSsk;
 142                                         linkType = LinkType.SSK;
 143                                 }
 144                                 if ((nextUsk > -1) && (nextUsk < next)) {
 145                                         next = nextUsk;
 146                                         linkType = LinkType.USK;
 147                                 }
 148                                 if ((nextHttp > -1) && (nextHttp < next)) {
 149                                         next = nextHttp;
 150                                         linkType = LinkType.HTTP;
 151                                 }
 152                                 if ((nextHttps > -1) && (nextHttps < next)) {
 153                                         next = nextHttps;
 154                                         linkType = LinkType.HTTPS;
 155                                 }
 156                                 if ((next >= 8) && (line.substring(next - 8, next).equals("freenet:"))) {
 157                                         next -= 8;
 158                                         line = line.substring(0, next) + line.substring(next + 8);
 159                                 }
 160                                 Matcher matcher = whitespacePattern.matcher(line);
 161                                 int nextSpace = matcher.find(next) ? matcher.start() : line.length();
 162                                 if (nextSpace > (next + 4)) {
 163                                         parts.add(createPlainTextPart(line.substring(0, next)));
 164                                         String link = line.substring(next, nextSpace);
 165                                         String name = link;
 166                                         logger.log(Level.FINER, "Found link: %s", link);
 167                                         logger.log(Level.FINEST, "Next: %d, CHK: %d, SSK: %d, USK: %d", new Object[] { next, nextChk, nextSsk, nextUsk });
 168                                         if (linkType == LinkType.KSK) {
 169                                                 name = link.substring(4);
 170                                         } else if ((linkType == LinkType.CHK) || (linkType == LinkType.SSK) || (linkType == LinkType.USK)) {
 171                                                 if (name.indexOf('/') > -1) {
 172                                                         if (!name.endsWith("/")) {
 173                                                                 name = name.substring(name.lastIndexOf('/') + 1);
 174                                                         } else {
 175                                                                 if (name.indexOf('/') != name.lastIndexOf('/')) {
 176                                                                         name = name.substring(name.lastIndexOf('/', name.lastIndexOf('/') - 1));
 177                                                                 } else {
 178                                                                         /* shorten to 5 chars. */
 179                                                                         name = name.substring(4, 9);
 180                                                                 }
 181                                                         }
 182                                                 }
 183                                                 if (name.indexOf('?') > -1) {
 184                                                         name = name.substring(0, name.indexOf('?'));
 185                                                 }
 186                                                 boolean fromPostingSone = false;
 187                                                 if ((linkType == LinkType.SSK) || (linkType == LinkType.USK)) {
 188                                                         fromPostingSone = link.substring(4, 47).equals(postingSone.getId());
 189                                                 }
 190                                                 parts.add(fromPostingSone ? createTrustedFreenetLinkPart(link, name) : createFreenetLinkPart(link, name));
 191                                         } else if ((linkType == LinkType.HTTP) || (linkType == LinkType.HTTPS)) {
 192                                                 name = link.substring(linkType == LinkType.HTTP ? 7 : 8);
 193                                                 int firstSlash = name.indexOf('/');
 194                                                 int lastSlash = name.lastIndexOf('/');
 195                                                 if ((lastSlash - firstSlash) > 3) {
 196                                                         name = name.substring(0, firstSlash + 1) + "…" + name.substring(lastSlash);
 197                                                 }
 198                                                 if (name.endsWith("/")) {
 199                                                         name = name.substring(0, name.length() - 1);
 200                                                 }
 201                                                 if (((name.indexOf('/') > -1) && (name.indexOf('.') < name.lastIndexOf('.', name.indexOf('/'))) || ((name.indexOf('/') == -1) && (name.indexOf('.') < name.lastIndexOf('.')))) && name.startsWith("www.")) {
 202                                                         name = name.substring(4);
 203                                                 }
 204                                                 if (name.indexOf('?') > -1) {
 205                                                         name = name.substring(0, name.indexOf('?'));
 206                                                 }
 207                                                 link = "?_CHECKED_HTTP_=" + link;
 208                                                 parts.add(createInternetLinkPart(link, name));
 209                                         }
 210                                         line = line.substring(nextSpace);
 211                                 } else {
 212                                         parts.add(createPlainTextPart(line.substring(0, next + 4)));
 213                                         line = line.substring(next + 4);
 214                                 }
 215                         }
 216                 }
 217                 return parts;
 218         }
 219
 220         //
 221         // PRIVATE METHODS
 222         //
 223
 224         /**
 225          * Creates a new plain text part based on a template.
 226          *
 227          * @param text
 228          *            The text to display
 229          * @return The part that displays the given text
 230          */
 231         private Part createPlainTextPart(String text) {
 232                 return new TemplatePart(templateFactory.createTemplate(new StringReader("<% text|html>"))).set("text", text);
 233         }
 234
 235         /**
 236          * Creates a new part based on a template that links to a site within the
 237          * normal internet.
 238          *
 239          * @param link
 240          *            The target of the link
 241          * @param name
 242          *            The name of the link
 243          * @return The part that displays the link
 244          */
 245         private Part createInternetLinkPart(String link, String name) {
 246                 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);
 247         }
 248
 249         /**
 250          * Creates a new part based on a template that links to a site within
 251          * freenet.
 252          *
 253          * @param link
 254          *            The target of the link
 255          * @param name
 256          *            The name of the link
 257          * @return The part that displays the link
 258          */
 259         private Part createFreenetLinkPart(String link, String name) {
 260                 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);
 261         }
 262
 263         /**
 264          * Creates a new part based on a template that links to a page in the
 265          * poster’s keyspace.
 266          *
 267          * @param link
 268          *            The target of the link
 269          * @param name
 270          *            The name of the link
 271          * @return The part that displays the link
 272          */
 273         private Part createTrustedFreenetLinkPart(String link, String name) {
 274                 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);
 275         }
 276
 277 }