src/cx/fbn/nevernote/xml/XMLNoteRepair.java

   1 /*
   2  * This file is part of NeverNote
   3  * Copyright 2009 Randy Baumgarte
   4  *
   5  * This file may be licensed under the terms of of the
   6  * GNU General Public License Version 2 (the ``GPL'').
   7  *
   8  * Software distributed under the License is distributed
   9  * on an ``AS IS'' basis, WITHOUT WARRANTY OF ANY KIND, either
  10  * express or implied. See the GPL for the specific language
  11  * governing rights and limitations.
  12  *
  13  * You should have received a copy of the GPL along with this
  14  * program. If not, go to http://www.gnu.org/licenses/gpl.html
  15  * or write to the Free Software Foundation, Inc.,
  16  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  17  *
  18 */
  19 package cx.fbn.nevernote.xml;
  20
  21 import java.io.File;
  22 import java.io.IOException;
  23 import java.io.StringReader;
  24
  25 import javax.xml.parsers.ParserConfigurationException;
  26 import javax.xml.parsers.SAXParser;
  27 import javax.xml.parsers.SAXParserFactory;
  28
  29 import org.xml.sax.InputSource;
  30 import org.xml.sax.SAXException;
  31 import org.xml.sax.SAXParseException;
  32
  33 import cx.fbn.nevernote.Global;
  34
  35 public class XMLNoteRepair {
  36         public boolean saveInvalidXML;
  37
  38         public String parse(String xmlData, boolean validate) {
  39                 saveInvalidXML = false;
  40                 InputSource is = new InputSource();
  41                 is.setCharacterStream(new StringReader(xmlData));
  42                 XMLNoteRepairHandler handler = new XMLNoteRepairHandler();
  43
  44                 // Replace DTD with local copy in case we are not connected
  45                 File dtdFile = Global.getFileManager().getXMLDirFile("enml2.dtd");
  46                 String dtd = dtdFile.toURI().toString();
  47                 xmlData = xmlData.replace("<!DOCTYPE en-note SYSTEM \'http://xml.evernote.com/pub/enml2.dtd'>",
  48                                 "<!DOCTYPE en-note SYSTEM \"" +dtd +"\">");
  49                 xmlData = xmlData.replace("<!DOCTYPE en-note SYSTEM \"http://xml.evernote.com/pub/enml2.dtd\">",
  50                                 "<!DOCTYPE en-note SYSTEM \"" +dtd +"\">");
  51
  52                 handler.setXml(xmlData);
  53                 is.setCharacterStream(new StringReader(handler.getXml()));
  54
  55                 boolean fixed = false;
  56                 int i=0;
  57                 int max = 10;
  58                 if (validate)
  59                         max = 10000;
  60                 while (!fixed && i<max) {
  61                         try {
  62                                 i++;
  63                                 SAXParserFactory factory = SAXParserFactory.newInstance();
  64                                 factory.setValidating(validate);
  65                                 SAXParser parser = factory.newSAXParser();
  66                                 parser.parse(is, handler);
  67                                 fixed = true;
  68                         } catch (EnmlException e) {
  69                                 String message = e.getMessage();
  70                                 saveInvalidXML = true;
  71                                 //System.out.println("ENML Exception: " +message);
  72                                 boolean found = false;
  73                                 int endAttribute = message.indexOf(" must be declared for element type ");
  74                                 if (message.startsWith("Attribute ") && endAttribute > -1) {
  75                                         String attribute = message.substring(11, endAttribute-1);
  76                                         String element = message.substring(message.indexOf("\"", endAttribute+3));
  77                                         element = element.replace("\"", "");
  78                                         element = element.substring(0,element.length()-1);
  79                                         Global.addInvalidAttribute(element, attribute);
  80                                         handler.stripAttribute(attribute, e.getLineNumber(), e.getColumnNumber());
  81                                         is.setCharacterStream(new StringReader(handler.getXml()));
  82                                         found = true;
  83                                 }
  84                                 int endElement = message.indexOf(" must be declared.");
  85                                 if (message.startsWith("Element type") && endElement > -1) {
  86                                         String element = message.substring(14,endElement-1);
  87                                         Global.addInvalidElement(element);
  88                                         handler.renameElement(element, e.getLineNumber(), e.getColumnNumber());
  89                                         is.setCharacterStream(new StringReader(handler.getXml()));
  90                                         found = true;
  91                                 }
  92                                 if (!found)
  93                                         System.err.println("New enml validation error: " +e.getMessage() +" Line:" +e.getLineNumber() +" Column:" +e.getColumnNumber());
  94                         } catch (SAXParseException e) {
  95                                 System.err.println("SAXParse Exception - Attempt #"+i +" "+e.getMessage());
  96                                 handler.repair(e.getLineNumber(), e.getColumnNumber());
  97                                 is.setCharacterStream(new StringReader(handler.getXml()));
  98                                 if (validate) {
  99                                         System.err.println("Error validating ENML2 DTD");
 100                                         return null;
 101                                 }
 102                         } catch (SAXException e) {
 103                                 System.err.append("SAXException");
 104                                 e.printStackTrace();
 105                         } catch (ParserConfigurationException e) {
 106                                 System.err.println("Parser Config Error");
 107                                 e.printStackTrace();
 108                         } catch (IOException e) {
 109                                 System.err.println("IO Exception");
 110                                 e.printStackTrace();
 111                         }
 112                 }
 113                 if (!fixed)
 114                         return null;
 115                 else {
 116                         // Replace DTD with online copy
 117                         xmlData = handler.getXml();
 118                         xmlData = xmlData.replace("<!DOCTYPE en-note SYSTEM \"" +dtd +"\">", "<!DOCTYPE en-note SYSTEM \'http://xml.evernote.com/pub/enml2.dtd'>");
 119                         return xmlData;
 120                 }
 121         }
 122 }