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