source/mircoders/localizer/basic/MirBasicProducerAssistantLocalizer.java

   1 /*
   2  * Copyright (C) 2001, 2002 The Mir-coders group
   3  *
   4  * This file is part of Mir.
   5  *
   6  * Mir is free software; you can redistribute it and/or modify
   7  * it under the terms of the GNU General Public License as published by
   8  * the Free Software Foundation; either version 2 of the License, or
   9  * (at your option) any later version.
  10  *
  11  * Mir is distributed in the hope that it will be useful,
  12  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  13  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  14  * GNU General Public License for more details.
  15  *
  16  * You should have received a copy of the GNU General Public License
  17  * along with Mir; if not, write to the Free Software
  18  * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
  19  *
  20  * In addition, as a special exception, The Mir-coders gives permission to link
  21  * the code of this program with  any library licensed under the Apache Software License,
  22  * The Sun (tm) Java Advanced Imaging library (JAI), The Sun JIMI library
  23  * (or with modified versions of the above that use the same license as the above),
  24  * and distribute linked combinations including the two.  You must obey the
  25  * GNU General Public License in all respects for all of the code used other than
  26  * the above mentioned libraries.  If you modify this file, you may extend this
  27  * exception to your version of the file, but you are not obligated to do so.
  28  * If you do not wish to do so, delete this exception statement from your version.
  29  */
  30 package mircoders.localizer.basic;
  31
  32 import java.io.ByteArrayInputStream;
  33 import java.io.IOException;
  34 import java.io.StringWriter;
  35 import java.util.GregorianCalendar;
  36 import java.util.HashMap;
  37 import java.util.Iterator;
  38 import java.util.List;
  39 import java.util.Map;
  40
  41 import gnu.regexp.RE;
  42
  43 import mir.config.MirPropertiesConfiguration;
  44 import mir.entity.adapter.EntityAdapter;
  45 import mir.entity.adapter.EntityIteratorAdapter;
  46 import mir.generator.Generator;
  47 import mir.generator.GeneratorExc;
  48 import mir.generator.GeneratorFailure;
  49 import mir.log.LoggerWrapper;
  50 import mir.misc.StringUtil;
  51 import mir.util.GeneratorDateTimeFunctions;
  52 import mir.util.GeneratorFormatAdapters;
  53 import mir.util.generator.ReflectionGeneratorFunctionsAdapter;
  54 import mir.util.StringRoutines;
  55 import mircoders.global.MirGlobal;
  56 import mircoders.localizer.MirLocalizerExc;
  57 import mircoders.localizer.MirLocalizerFailure;
  58 import mircoders.localizer.MirProducerAssistantLocalizer;
  59
  60 import org.w3c.dom.Document;
  61 import org.w3c.dom.NamedNodeMap;
  62 import org.w3c.dom.Node;
  63 import org.w3c.dom.NodeList;
  64 import org.w3c.tidy.Configuration;
  65 import org.w3c.tidy.Tidy;
  66
  67 public class MirBasicProducerAssistantLocalizer implements MirProducerAssistantLocalizer {
  68   protected LoggerWrapper logger;
  69
  70   private RE regularExpressionLT;
  71   private RE regularExpressionGT;
  72   private RE regularExpressionWhitespace;
  73
  74   public MirBasicProducerAssistantLocalizer() throws MirLocalizerFailure {
  75     try{
  76       regularExpressionLT = new RE("<");
  77       regularExpressionGT = new RE(">");
  78       regularExpressionWhitespace = new RE("\\s+");
  79     }
  80     catch (Throwable t) {
  81       throw new MirLocalizerFailure(t);
  82     }
  83   }
  84
  85   public void initializeGenerationValueSet(Map aValueSet) throws MirLocalizerExc, MirLocalizerFailure  {
  86     try {
  87       Iterator i;
  88
  89       Map configMap = new HashMap();
  90
  91       logger = new LoggerWrapper("Localizer.ProducerAssistant");
  92
  93 // obsolete:
  94       configMap.put("producerDocRoot", MirGlobal.config().getString("Producer.DocRoot"));
  95       configMap.put("storageRoot", MirGlobal.config().getString("Producer.StorageRoot"));
  96       configMap.put("productionHost", MirGlobal.config().getString("Producer.ProductionHost"));
  97       configMap.put("openAction", MirGlobal.config().getString("Producer.OpenAction"));
  98       configMap.put("docRoot", MirGlobal.config().getString("RootUri"));
  99       configMap.put("actionRoot", MirGlobal.config().getString("RootUri") + "/servlet/Mir");
 100       configMap.put("now", new GeneratorFormatAdapters.DateFormatAdapter(new GregorianCalendar().getTime(), MirGlobal.config().getString("Mir.DefaultTimezone")));
 101       configMap.put("videoHost", MirGlobal.config().getString("Producer.Video.Host"));
 102       configMap.put("audioHost", MirGlobal.config().getString("Producer.Audio.Host"));
 103       configMap.put("imageHost", MirGlobal.config().getString("Producer.Image.Host"));
 104       configMap.put("imagePath", MirGlobal.config().getString("Producer.Image.Path"));
 105       configMap.put("mirVersion", MirGlobal.config().getString("Mir.Version"));
 106       configMap.put("defEncoding", MirGlobal.config().getString("Mir.DefaultEncoding"));
 107
 108 // "new":
 109       configMap.putAll(MirPropertiesConfiguration.instance().allSettings());
 110
 111       aValueSet.put("config", configMap);
 112
 113       aValueSet.put("utility", new Utility());
 114
 115       aValueSet.put("languages",
 116         new EntityIteratorAdapter("", "", 20, MirGlobal.localizer().dataModel().adapterModel(), "language"));
 117
 118       aValueSet.put("topics",
 119         new EntityIteratorAdapter("", "", 20, MirGlobal.localizer().dataModel().adapterModel(), "topic"));
 120
 121       Map articleTypeMap = new HashMap();
 122       articleTypeMap.put("openposting", "0");
 123       articleTypeMap.put("newswire", "1");
 124       articleTypeMap.put("feature", "2");
 125       articleTypeMap.put("topicspecial", "3");
 126       articleTypeMap.put("startspecial", "4");
 127
 128       i = new EntityIteratorAdapter("", "", 20, MirGlobal.localizer().dataModel().adapterModel(), "articleType");
 129       while (i.hasNext()) {
 130         EntityAdapter articleType = (EntityAdapter) i.next();
 131
 132         articleTypeMap.put(articleType.get("name"), articleType.get("id"));
 133       }
 134       aValueSet.put("articletype", articleTypeMap);
 135
 136       Map commentStatusMap = new HashMap();
 137       i = new EntityIteratorAdapter("", "", 20, MirGlobal.localizer().dataModel().adapterModel(), "commentStatus");
 138       while (i.hasNext()) {
 139         EntityAdapter commentStatus = (EntityAdapter) i.next();
 140
 141         commentStatusMap.put(commentStatus.get("name"), commentStatus.get("id"));
 142       }
 143       aValueSet.put("commentstatus", commentStatusMap);
 144       aValueSet.put("languageCodeToId", new getLanguageIdFunction());
 145     }
 146     catch (Throwable t) {
 147       logger.error("initializeGenerationValueSet: Exception while collecting comment statuses" + t.getMessage());
 148
 149       throw new MirLocalizerFailure(t);
 150     }
 151
 152   }
 153
 154   public static class getLanguageIdFunction implements Generator.Function {
 155     private Map languageCodeToId;
 156     private String otherLanguageId;
 157     private LoggerWrapper logger = new LoggerWrapper("Localizer.Earth.getLanguageIdFunction");
 158
 159     public getLanguageIdFunction() throws MirLocalizerFailure {
 160       try {
 161         otherLanguageId = "";
 162         languageCodeToId = new HashMap();
 163
 164         Iterator i = new EntityIteratorAdapter("", "", 20, MirGlobal.localizer().dataModel().adapterModel(), "language");
 165         while (i.hasNext()) {
 166           EntityAdapter language = (EntityAdapter) i.next();
 167           if (language.get("code").equals("ot"))
 168             otherLanguageId = (String) language.get("id");
 169
 170           languageCodeToId.put(language.get("code"), language.get("id"));
 171         }
 172       }
 173       catch (Throwable t) {
 174         logger.error(t.toString());
 175
 176         throw new MirLocalizerFailure(t);
 177       }
 178     }
 179
 180     public Object perform(List aParameters) throws GeneratorExc, GeneratorFailure {
 181       try {
 182         if (aParameters.size() != 1)
 183           throw new GeneratorExc("getLanguageIdFunction: 1 parameter expected: language-code");
 184
 185         String result = (String) languageCodeToId.get(aParameters.get(0));
 186         if (result == null)
 187           result = otherLanguageId;
 188
 189         return result;
 190       }
 191       catch (GeneratorExc e) {
 192         throw e;
 193       }
 194       catch (Throwable t) {
 195         throw new GeneratorFailure("getLanguageIdFunction: " + t.getMessage(), t);
 196       }
 197     }
 198   }
 199
 200
 201   public String filterNonHTMLText(String aText) {
 202
 203     logger.debug("about to filter non HTML Text of length " + aText.length());
 204     try {
 205       String result =
 206           StringUtil.createHTML(
 207           StringUtil.removeHTMLTags(aText),
 208           MirGlobal.config().getString("Producer.ImageRoot"),
 209           MirGlobal.config().getString("Producer.MailLinkName"),
 210           MirGlobal.config().getString("Producer.ExtLinkName"),
 211           MirGlobal.config().getString("Producer.IntLinkName")
 212           );
 213       logger.debug("done filtering non-HTML text ");
 214       return result;
 215     }
 216     catch (Throwable t) {
 217       logger.error("error while filtering non-HTML text: " + t.toString());
 218
 219       throw new RuntimeException(t.toString());
 220     }
 221   }
 222   public String filterHTMLText(String aText) {
 223     try {
 224       StringWriter out = new StringWriter();
 225       Tidy tidy = new Tidy();
 226       ByteArrayInputStream in = new ByteArrayInputStream(aText.getBytes("UTF8"));
 227       tidy.setMakeClean(true);
 228       tidy.setCharEncoding(Configuration.UTF8);
 229       tidy.setErrout(logger.asPrintWriter(LoggerWrapper.DEBUG_MESSAGE));
 230       print(tidy.parseDOM(in, null), out);
 231
 232       return out.toString();
 233     }
 234     catch (IOException e) {
 235       return e.getMessage();
 236     }
 237   }
 238
 239
 240   private String[] badAttributeValuePrefixes= {"javascript","vbscript","about","wysiwyg","data","view-source","ms-its","mhtml","shell","lynxexec","lynxcgi","hcp","ms-help","help","disk","vnd.ms.radio","opera","res","resource","chrome","mocha","livescript"};
 241
 242
 243   private String[] badAttributes = {"onabort", "onblur",  "onchange", "onclick", "ondblclick", "onerror", "onfocus", "onkeydown", "onKeypress", "onkeyup", "onload", "onmousedown", "onmousemove", "onmouseout", "onmouseover", "onmouseup", "onreset", "onselect", "onsubmit", "onunload","onload","onclick","onfocus","onblur","FSCommand","onAbort","onActivate","onAfterPrint","onAfterUpdate","onBeforeActivate","onBeforeCopy","onBeforeCut","onBeforeDeactivate","onBeforeEditFocus","onBeforePaste","onBeforePrint","onBeforeUnload","onBegin","onBlur","onBounce","onCellChange","onChange","onClick","onContextMenu","onControlSelect","onCopy","onCut","onDataAvailible","onDataSetChanged","onDataSetComplete","onDblClick","onDeactivate","onDrag","onDragEnd","onDragLeave","onDragEnter","onDragOver","onDragDrop","onDrop","onEnd","onError","onErrorUpdate","onExit","onFilterChange","onFinish","onFocus","onFocusIn","onFocusOut","onHelp","onKeyDown","onKeyPress","onKeyUp","onLayoutComplete","onLoad","onLoseCapture","onMediaComplete","onMediaError","onMouseDown","onMouseEnter","onMouseLeave","onMouseMove","onMouseOut","onMouseOver","onMouseUp","onMouseWheel","onMove","onMoveEnd","onMoveStart","onOutOfSync","onPaste","onPause","onProgress","onPropertyChange","onReadyStateChange","onRepeat","onReset","onResize","onResizeEnd","onResizeStart","onResume","onReverse","onRowEnter","onRowExit","onRowDelete","onRowInserted","onScroll","onSeek","onSelect","onSelectionChange","onSelectStart","onStart","onStop","onSynchRestored","onSubmit","onTimeError","onTrackChange","onUnload","onURLFlip","seekSegmentTime","style","height","width"};
 244
 245   private boolean isBadAttr(String attrName){
 246     for (int i=0;i<badAttributes.length;i++){
 247       if (badAttributes[i].toLowerCase().equals(attrName.toLowerCase()))
 248         return true;
 249       }
 250     return false;
 251   }
 252
 253   private String stripWhitespace(String aString){
 254     try{
 255       return regularExpressionWhitespace.substituteAll(aString, "");
 256      }
 257     catch (Throwable t){
 258       return "";
 259     }
 260   }
 261
 262   private boolean checkAttr(String attrName) {
 263     if (isBadAttr(attrName)){
 264         return false;
 265     }
 266     return true;
 267
 268   }
 269
 270   private boolean checkAttrValue(String attrValue) {
 271     for (int i=0;i<badAttributeValuePrefixes.length;i++){
 272       if ((stripWhitespace(attrValue.toLowerCase())).startsWith(badAttributeValuePrefixes[i].toLowerCase()+":")){
 273         return false;
 274       }
 275     }
 276     return true;
 277   }
 278
 279
 280   private boolean checkNode(String nodeName) {
 281     List languages =  StringRoutines.splitString(MirGlobal.config().getString("Localizer.HTML.Whitelist"), ";");
 282
 283     Iterator i = languages.iterator();
 284     while (i.hasNext()) {
 285       if (nodeName.equals(i.next()))
 286         return true;
 287     }
 288     return false;
 289   }
 290
 291   private void print(Node node, StringWriter out) throws IOException {
 292     if (node == null) {
 293       return;
 294     }
 295     int type = node.getNodeType();
 296     boolean canOutput = checkNode(node.getNodeName());
 297
 298     switch (type) {
 299
 300       case Node.DOCUMENT_NODE:
 301
 302         print(((Document) node).getDocumentElement(), out);
 303         out.flush();
 304         break;
 305
 306       case Node.ELEMENT_NODE:
 307         if (canOutput) {
 308           out.write('<');
 309
 310           out.write(node.getNodeName());
 311           NamedNodeMap attrs = node.getAttributes();
 312
 313           for (int i = 0; i < attrs.getLength(); i++) {
 314             String attrName = attrs.item(i).getNodeName();
 315             String attrValue = attrs.item(i).getNodeValue();
 316             if (checkAttr(attrName) && checkAttrValue(attrValue)) {
 317               out.write(' ');
 318               out.write(attrs.item(i).getNodeName());
 319               out.write("=\"");
 320
 321               out.write(attrs.item(i).getNodeValue());
 322               out.write('"');
 323             }
 324           }
 325
 326           if (node.getChildNodes()==null || node.getChildNodes().getLength()==0) {
 327             out.write("/");
 328           }
 329           out.write('>');
 330         }
 331         NodeList children = node.getChildNodes();
 332         if (children != null) {
 333           int len = children.getLength();
 334           for (int i = 0; i < len; i++) {
 335             print(children.item(i), out);
 336           }
 337         }
 338         break;
 339
 340       case Node.TEXT_NODE:
 341         String value=node.getNodeValue();
 342         try{
 343           value=regularExpressionLT.substituteAll(value, "&lt;");
 344           value=regularExpressionGT.substituteAll(value, "&gt;");
 345         }
 346         catch (Throwable t){
 347           value="";
 348         }
 349         out.write(value);
 350
 351         break;
 352
 353     }
 354
 355     if (type == Node.ELEMENT_NODE && canOutput && node.getChildNodes()!=null && node.getChildNodes().getLength()>0) {
 356       out.write("</");
 357       out.write(node.getNodeName());
 358       out.write('>');
 359     }
 360
 361     out.flush();
 362   }
 363
 364   public static class Utility extends ReflectionGeneratorFunctionsAdapter {
 365     public Utility () {
 366       super(new MirBasicUtilityFunctions());
 367     }
 368
 369     public Object getDatetime() {
 370       return new GeneratorDateTimeFunctions.DateTimeFunctions(
 371         MirPropertiesConfiguration.instance().getString("Mir.DefaultTimezone"));
 372     }
 373
 374     public Object getCompressWhitespace() {
 375       return new freemarker.template.utility.CompressWhitespace();
 376     }
 377   }
 378 }