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.util.logging.Logging;
  31 import net.pterodactylus.util.template.TemplateContextFactory;
  32 import net.pterodactylus.util.template.TemplateParser;
  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<FreenetLinkParserContext> {
  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 TemplateContextFactory templateContextFactory;
  77
  78         /**
  79          * Creates a new freenet link parser.
  80          *
  81          * @param templateContextFactory
  82          *            The template context factory
  83          */
  84         public FreenetLinkParser(TemplateContextFactory templateContextFactory) {
  85                 this.templateContextFactory = templateContextFactory;
  86         }
  87
  88         //
  89         // PART METHODS
  90         //
  91
  92         /**
  93          * {@inheritDoc}
  94          */
  95         @Override
  96         public Part parse(FreenetLinkParserContext context, Reader source) throws IOException {
  97                 PartContainer parts = new PartContainer();
  98                 BufferedReader bufferedReader = (source instanceof BufferedReader) ? (BufferedReader) source : new BufferedReader(source);
  99                 String line;
 100                 boolean lastLineEmpty = true;
 101                 int emptyLines = 0;
 102                 while ((line = bufferedReader.readLine()) != null) {
 103                         if (line.trim().length() == 0) {
 104                                 if (lastLineEmpty) {
 105                                         continue;
 106                                 }
 107                                 parts.add(createPlainTextPart("\n"));
 108                                 ++emptyLines;
 109                                 lastLineEmpty = emptyLines == 2;
 110                                 continue;
 111                         }
 112                         emptyLines = 0;
 113                         boolean lineComplete = true;
 114                         while (line.length() > 0) {
 115                                 int nextKsk = line.indexOf("KSK@");
 116                                 int nextChk = line.indexOf("CHK@");
 117                                 int nextSsk = line.indexOf("SSK@");
 118                                 int nextUsk = line.indexOf("USK@");
 119                                 int nextHttp = line.indexOf("http://");
 120                                 int nextHttps = line.indexOf("https://");
 121                                 if ((nextKsk == -1) && (nextChk == -1) && (nextSsk == -1) && (nextUsk == -1) && (nextHttp == -1) && (nextHttps == -1)) {
 122                                         if (lineComplete && !lastLineEmpty) {
 123                                                 parts.add(createPlainTextPart("\n" + line));
 124                                         } else {
 125                                                 parts.add(createPlainTextPart(line));
 126                                         }
 127                                         break;
 128                                 }
 129                                 int next = Integer.MAX_VALUE;
 130                                 LinkType linkType = null;
 131                                 if ((nextKsk > -1) && (nextKsk < next)) {
 132                                         next = nextKsk;
 133                                         linkType = LinkType.KSK;
 134                                 }
 135                                 if ((nextChk > -1) && (nextChk < next)) {
 136                                         next = nextChk;
 137                                         linkType = LinkType.CHK;
 138                                 }
 139                                 if ((nextSsk > -1) && (nextSsk < next)) {
 140                                         next = nextSsk;
 141                                         linkType = LinkType.SSK;
 142                                 }
 143                                 if ((nextUsk > -1) && (nextUsk < next)) {
 144                                         next = nextUsk;
 145                                         linkType = LinkType.USK;
 146                                 }
 147                                 if ((nextHttp > -1) && (nextHttp < next)) {
 148                                         next = nextHttp;
 149                                         linkType = LinkType.HTTP;
 150                                 }
 151                                 if ((nextHttps > -1) && (nextHttps < next)) {
 152                                         next = nextHttps;
 153                                         linkType = LinkType.HTTPS;
 154                                 }
 155                                 if ((next >= 8) && (line.substring(next - 8, next).equals("freenet:"))) {
 156                                         next -= 8;
 157                                         line = line.substring(0, next) + line.substring(next + 8);
 158                                 }
 159                                 Matcher matcher = whitespacePattern.matcher(line);
 160                                 int nextSpace = matcher.find(next) ? matcher.start() : line.length();
 161                                 if (nextSpace > (next + 4)) {
 162                                         if (!lastLineEmpty && lineComplete) {
 163                                                 parts.add(createPlainTextPart("\n" + line.substring(0, next)));
 164                                         } else {
 165                                                 parts.add(createPlainTextPart(line.substring(0, next)));
 166                                         }
 167                                         String link = line.substring(next, nextSpace);
 168                                         String name = link;
 169                                         logger.log(Level.FINER, "Found link: %s", link);
 170                                         logger.log(Level.FINEST, "Next: %d, CHK: %d, SSK: %d, USK: %d", new Object[] { next, nextChk, nextSsk, nextUsk });
 171
 172                                         if ((linkType == LinkType.KSK) || (linkType == LinkType.CHK) || (linkType == LinkType.SSK) || (linkType == LinkType.USK)) {
 173                                                 FreenetURI uri;
 174                                                 if (name.indexOf('?') > -1) {
 175                                                         name = name.substring(0, name.indexOf('?'));
 176                                                 }
 177                                                 if (name.endsWith("/")) {
 178                                                         name = name.substring(0, name.length() - 1);
 179                                                 }
 180                                                 try {
 181                                                         uri = new FreenetURI(name);
 182                                                         name = uri.lastMetaString();
 183                                                         if (name == null) {
 184                                                                 name = uri.getDocName();
 185                                                         }
 186                                                         if (name == null) {
 187                                                                 name = link.substring(0, Math.min(9, link.length()));
 188                                                         }
 189                                                         boolean fromPostingSone = ((linkType == LinkType.SSK) || (linkType == LinkType.USK)) && link.substring(4, Math.min(link.length(), 47)).equals(context.getPostingSone().getId());
 190                                                         parts.add(fromPostingSone ? createTrustedFreenetLinkPart(link, name) : createFreenetLinkPart(link, name));
 191                                                 } catch (MalformedURLException mue1) {
 192                                                         /* not a valid link, insert as plain text. */
 193                                                         parts.add(createPlainTextPart(link));
 194                                                 } catch (NullPointerException npe1) {
 195                                                         /* FreenetURI sometimes throws these, too. */
 196                                                         parts.add(createPlainTextPart(link));
 197                                                 } catch (ArrayIndexOutOfBoundsException aioobe1) {
 198                                                         /* oh, and these, too. */
 199                                                         parts.add(createPlainTextPart(link));
 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                                         if (!lastLineEmpty && lineComplete) {
 223                                                 parts.add(createPlainTextPart("\n" + line.substring(0, next + 4)));
 224                                         } else {
 225                                                 parts.add(createPlainTextPart(line.substring(0, next + 4)));
 226                                         }
 227                                         line = line.substring(next + 4);
 228                                 }
 229                                 lineComplete = false;
 230                         }
 231                         lastLineEmpty = false;
 232                 }
 233                 for (int partIndex = parts.size() - 1; partIndex >= 0; --partIndex) {
 234                         if (!parts.getPart(partIndex).toString().equals("\n")) {
 235                                 break;
 236                         }
 237                         parts.removePart(partIndex);
 238                 }
 239                 return parts;
 240         }
 241
 242         //
 243         // PRIVATE METHODS
 244         //
 245
 246         /**
 247          * Creates a new plain text part based on a template.
 248          *
 249          * @param text
 250          *            The text to display
 251          * @return The part that displays the given text
 252          */
 253         private Part createPlainTextPart(String text) {
 254                 return new TemplatePart(templateContextFactory, TemplateParser.parse(new StringReader("<% text|html>"))).set("text", text);
 255         }
 256
 257         /**
 258          * Creates a new part based on a template that links to a site within the
 259          * normal internet.
 260          *
 261          * @param link
 262          *            The target of the link
 263          * @param name
 264          *            The name of the link
 265          * @return The part that displays the link
 266          */
 267         private Part createInternetLinkPart(String link, String name) {
 268                 return new TemplatePart(templateContextFactory, TemplateParser.parse(new StringReader("<a class=\"internet\" href=\"/<% link|html>\" title=\"<% link|html>\"><% name|html></a>"))).set("link", link).set("name", name);
 269         }
 270
 271         /**
 272          * Creates a new part based on a template that links to a site within
 273          * freenet.
 274          *
 275          * @param link
 276          *            The target of the link
 277          * @param name
 278          *            The name of the link
 279          * @return The part that displays the link
 280          */
 281         private Part createFreenetLinkPart(String link, String name) {
 282                 return new TemplatePart(templateContextFactory, TemplateParser.parse(new StringReader("<a class=\"freenet\" href=\"/<% link|html>\" title=\"<% link|html>\"><% name|html></a>"))).set("link", link).set("name", name);
 283         }
 284
 285         /**
 286          * Creates a new part based on a template that links to a page in the
 287          * poster’s keyspace.
 288          *
 289          * @param link
 290          *            The target of the link
 291          * @param name
 292          *            The name of the link
 293          * @return The part that displays the link
 294          */
 295         private Part createTrustedFreenetLinkPart(String link, String name) {
 296                 return new TemplatePart(templateContextFactory, TemplateParser.parse(new StringReader("<a class=\"freenet-trusted\" href=\"/<% link|html>\" title=\"<% link|html>\"><% name|html></a>"))).set("link", link).set("name", name);
 297         }
 298
 299 }