src/cx/fbn/nevernote/evernote/EnmlConverter.java

   1 /*\r
   2  * This file is part of NeverNote \r
   3  * Copyright 2009 Randy Baumgarte\r
   4  * \r
   5  * This file may be licensed under the terms of of the\r
   6  * GNU General Public License Version 2 (the ``GPL'').\r
   7  *\r
   8  * Software distributed under the License is distributed\r
   9  * on an ``AS IS'' basis, WITHOUT WARRANTY OF ANY KIND, either\r
  10  * express or implied. See the GPL for the specific language\r
  11  * governing rights and limitations.\r
  12  *\r
  13  * You should have received a copy of the GPL along with this\r
  14  * program. If not, go to http://www.gnu.org/licenses/gpl.html\r
  15  * or write to the Free Software Foundation, Inc.,\r
  16  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.\r
  17  *\r
  18 */\r
  19 package cx.fbn.nevernote.evernote;\r
  20 \r
  21 import java.io.File;\r
  22 import java.util.ArrayList;\r
  23 import java.util.List;\r
  24 \r
  25 import cx.fbn.nevernote.Global;\r
  26 import cx.fbn.nevernote.utilities.ApplicationLogger;\r
  27 import cx.fbn.nevernote.xml.XMLCleanup;\r
  28 import cx.fbn.nevernote.xml.XMLNoteRepair;\r
  29 \r
  30 public class EnmlConverter {\r
  31         private final ApplicationLogger logger;\r
  32         private List<String>                    resources;\r
  33         public boolean saveInvalidXML;\r
  34         \r
  35         public EnmlConverter(ApplicationLogger l) {\r
  36                 logger = l;\r
  37 //              conn = c;\r
  38                 saveInvalidXML = false;\r
  39                 resources = new ArrayList<String>();\r
  40         }\r
  41 \r
  42         public List<String> getResources() {\r
  43                 return resources;\r
  44         }\r
  45         public String convert(String noteGuid, String content) {\r
  46                 logger.log(logger.HIGH, "Entering DBRunner.convertToEnml");\r
  47                 logger.log(logger.EXTREME, "Note Text:" +content);\r
  48                 \r
  49                 // Replace the en-note tags with body tags in case we came from \r
  50                 // someplace other than the editor (for example, if we are merging notes).\r
  51                 content = content.replace("<en-note>", "<body>");\r
  52                 content = content.replace("</en-note>", "</body>");\r
  53                 // Start removing stuff we don't need or want\r
  54                 int br = content.lastIndexOf("</body>");\r
  55                 if (br > 0)\r
  56                         content = new String(content.substring(0,br));\r
  57                 String newContent;\r
  58                 int k = content.indexOf("<body");\r
  59                 if (k>-1)\r
  60                         newContent = new String(content.substring(k));\r
  61                 else\r
  62                         newContent = "<body>"+content;\r
  63 \r
  64                 \r
  65                 // Check that we have a vaild header.  Normally we should not\r
  66                 // but sometimes it seems that we can.  I don't see how, but it is\r
  67                 // easy enough to check.\r
  68                 if (!newContent.startsWith("<?xml"))\r
  69                         newContent = "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n" \r
  70                                 +"<!DOCTYPE en-note SYSTEM \"http://xml.evernote.com/pub/enml2.dtd\">\n"\r
  71                                 +newContent \r
  72                                 +"</body>";\r
  73                 \r
  74 \r
  75                 // Fix the more common XML problems that Webkit creates, but are not considered \r
  76                 // valid XML.\r
  77                 newContent = fixStupidXMLProblems(newContent);\r
  78                 \r
  79                 \r
  80                 // Change the contents to have enml instead of body tags or\r
  81                 // we'll fail validation later.\r
  82                 newContent = newContent.replace("<body", "<en-note");\r
  83                 newContent = newContent.replace("</body>", "</en-note>");\r
  84                 \r
  85                 // First pass through the data.  The goal of this pass is to \r
  86                 // validate that we have a good XML document and to repair\r
  87                 // any problems found.\r
  88                 \r
  89                 XMLNoteRepair repair = new XMLNoteRepair();\r
  90                 logger.log(logger.HIGH, "Checking XML Structure");\r
  91                 newContent = repair.parse(newContent, false);\r
  92                 logger.log(logger.HIGH, "Check complete");\r
  93                 \r
  94 \r
  95                 // If the repair above returned null, then the XML is foobar.\r
  96                 // We are done here.\r
  97                 if (newContent == null) {\r
  98                         // Houston, we've had a problem.\r
  99                         logger.log(logger.LOW, "Parse error when converting to ENML");\r
 100                         logger.log(logger.LOW, "Start of unmodified note HTML");\r
 101                         logger.log(logger.LOW, content);\r
 102                         logger.log(logger.LOW, "End of unmodified note HTML");\r
 103                         logger.log(logger.LOW, "Start of modified note HTML");\r
 104                         logger.log(logger.LOW, newContent);\r
 105                         logger.log(logger.LOW, "End of modified note HTML");\r
 106 //                              logger.log(logger.LOW, result.errorMessage);\r
 107 //                              logger.log(logger.LOW, "Error Line:Column "+result.errorLine+":" +result.errorColumn);\r
 108                         System.exit(16);\r
 109 \r
 110 \r
 111                 }\r
 112                 \r
 113                 // Second pass through the data.  The goal of this pass is to \r
 114                 // remove any things we added in NeverNote that do not match\r
 115                 // the ENML schema\r
 116                 XMLCleanup v = new XMLCleanup();\r
 117                 v.setValue(newContent);\r
 118                 logger.log(logger.HIGH, "Beginning ENML Cleanup");\r
 119                 v.validate();\r
 120                 logger.log(logger.HIGH, "Cleanup complete.");\r
 121                 \r
 122         \r
 123                         \r
 124                 // Final pass through the data.  In this one we\r
 125                 // remove any invalid attributes and to save the\r
 126                 // new resources.\r
 127                 logger.log(logger.EXTREME, "Rebuilt ENML:");\r
 128                 logger.log(logger.EXTREME, v.getValue());       \r
 129                 logger.log(logger.EXTREME, "End Of Rebuilt ENML:");\r
 130                 resources = v.getResources();\r
 131 \r
 132                 \r
 133                 // The XML has the dtd to validate set against Evernote's web\r
 134                 // address.  We change it to a local one because otherwise it would\r
 135                 // fail if the user doesn't have internet connectivity.  The local copy\r
 136                 // also contains the 3 other PUBLIC definitions at the beginning of the dtd.\r
 137                 newContent = v.getValue();\r
 138                 File dtdFile = new File(Global.getDirectoryPath()+"xml/enml2.dtd");\r
 139                 String dtd = dtdFile.toURI().toString();\r
 140                 newContent = newContent.replace("<!DOCTYPE en-note SYSTEM \'http://xml.evernote.com/pub/enml2.dtd'>", \r
 141                                 "<!DOCTYPE en-note SYSTEM \"" +dtd +"\">");\r
 142                 \r
 143                 logger.log(logger.HIGH, "Validating ENML");\r
 144                 newContent = repair.parse(newContent, true);\r
 145                 logger.log(logger.HIGH, "Validation complete");\r
 146                 saveInvalidXML = repair.saveInvalidXML;\r
 147                 \r
 148                 // Restore the correct XML header.\r
 149                 newContent = newContent.replace("<!DOCTYPE en-note SYSTEM \"" +dtd +"\">", \r
 150                                 "<!DOCTYPE en-note SYSTEM 'http://xml.evernote.com/pub/enml2.dtd'>");\r
 151                 \r
 152                 \r
 153                 \r
 154                 return newContent;\r
 155         }\r
 156         \r
 157         \r
 158         // Fix XML problems that Qt can't deal with\r
 159         public String fixStupidXMLProblems(String content) {\r
 160                 logger.log(logger.HIGH, "Entering DBRunner.fixStupidXMLProblems");\r
 161 \r
 162                 // Fix the problem that the document body isn't properly closed\r
 163                 String newContent = new String(content);\r
 164                 logger.log(logger.MEDIUM, "Inside fixStupidXMLProblems.  Old content:");\r
 165                 logger.log(logger.MEDIUM, content);\r
 166                 \r
 167                 // Fix the problem that the img tag isn't properly closed\r
 168                 int endPos;\r
 169                 logger.log(logger.MEDIUM, "Checking img tags");\r
 170                 for (int i=newContent.indexOf("<img"); i>0; i = newContent.indexOf("<img",i+1)) {\r
 171                         endPos = newContent.indexOf(">",i+1);\r
 172                         String end = newContent.substring(endPos+1);\r
 173                         newContent = newContent.subSequence(0,endPos) +"/>"+end;\r
 174                 }\r
 175                 \r
 176                 // Fix the problem that the input tag isn't properly closed\r
 177                 logger.log(logger.MEDIUM, "Checking input tags");\r
 178                 for (int i=newContent.indexOf("<input"); i>0; i = newContent.indexOf("<input",i+1)) {\r
 179                         endPos = newContent.indexOf(">",i+1);\r
 180                         String end = newContent.substring(endPos+1);\r
 181                         newContent = newContent.subSequence(0,endPos) +"/>"+end;\r
 182                 }\r
 183                 \r
 184                 \r
 185                 // Fix the problem that the <br> tag isn't properly closed\r
 186                 logger.log(logger.MEDIUM, "Checking br tags");\r
 187                 for (int i=newContent.indexOf("<br"); i>0; i = newContent.indexOf("<br",i+1)) {\r
 188                         endPos = newContent.indexOf(">",i+1);\r
 189                         String end = newContent.substring(endPos+1);\r
 190                         newContent = newContent.subSequence(0,endPos) +"/>"+end;\r
 191                 }\r
 192                         \r
 193                 // Fix the problem that the <hr> tag isn't properly closed\r
 194                 logger.log(logger.MEDIUM, "Checking hr tags");\r
 195                 for (int i=newContent.indexOf("<hr"); i>0; i = newContent.indexOf("<hr",i+1)) {\r
 196                         endPos = newContent.indexOf(">",i+1);\r
 197                         String end = newContent.substring(endPos+1);\r
 198                         newContent = newContent.subSequence(0,endPos) +"/>"+end;\r
 199                 }\r
 200                 \r
 201                 logger.log(logger.MEDIUM, "Leaving fixStupidXMLProblems");\r
 202                 logger.log(logger.HIGH, "Leaving DBRunner.fixStupidXMLProblems");\r
 203                 return newContent.toString();\r
 204         }\r
 205 \r
 206 \r
 207         // Fix XML that Evernote thinks is invalid\r
 208         public String fixEnXMLCrap(String note) {\r
 209                 int pos;\r
 210                 StringBuffer buffer = new StringBuffer(note);\r
 211                 \r
 212                 // change all <b/> to <b></b> because Evernote hates them if they happen in <span>\r
 213                 pos = buffer.indexOf("<b/>");\r
 214                 for (; pos>-1; ) {\r
 215                         buffer.replace(pos, pos+4, "<b></b>");\r
 216                         pos = buffer.indexOf("<b/>",pos);\r
 217                 }\r
 218                 // change all <br/> to <br></br> because Evernote hates them if they happen in <span>\r
 219                 pos = buffer.indexOf("<br/>");\r
 220                 for (; pos>-1; ) {\r
 221                         buffer.replace(pos, pos+5, "<br></br>");\r
 222                         pos = buffer.indexOf("<br/>",pos);\r
 223                 }\r
 224                 \r
 225                 // change all <span> elements in lists because Evernote hates them if they happen \r
 226                 int endPos = 0;\r
 227                 int spanPos;\r
 228                 pos = buffer.indexOf("<li>");\r
 229                 spanPos = buffer.indexOf("<span>");\r
 230 /*              for (; pos>-1 && spanPos >-1;) {\r
 231                         endPos = buffer.indexOf("</li>",pos);\r
 232                         if (spanPos > pos && spanPos < endPos) {\r
 233                                 buffer.replace(spanPos,spanPos+6,"");\r
 234                                 spanPos = buffer.indexOf("</span>");                            \r
 235                                 buffer.replace(spanPos,spanPos+7,"");\r
 236                         }\r
 237                         pos=buffer.indexOf("<li>",pos+1);\r
 238                         spanPos = buffer.indexOf("<span>",spanPos);\r
 239                 }\r
 240 */              \r
 241                 // Get rid of empty spans in <li> elements\r
 242                 pos = buffer.indexOf("<li>");\r
 243                 spanPos = buffer.indexOf("<span/>");\r
 244                 for (; pos>-1 && spanPos >-1;) {\r
 245                         endPos = buffer.indexOf("</li>",pos);\r
 246                         if (spanPos > pos && spanPos < endPos) {\r
 247                                 buffer.replace(spanPos,spanPos+7,"");\r
 248                         }\r
 249                         pos=buffer.indexOf("<li>",pos+1);\r
 250                         spanPos = buffer.indexOf("<span/>",spanPos);\r
 251                 }\r
 252                 \r
 253                 return buffer.toString();\r
 254         }\r
 255         \r
 256         // Fix stupid en-media problems\r
 257         public String fixEnMediaCrap(String note) {\r
 258                 if (note == null)\r
 259                         return null;\r
 260                 \r
 261                 StringBuffer buffer = new StringBuffer(note);\r
 262                 // get rid of any </en-media> tags since they shouldn't exist.\r
 263                 int pos = buffer.indexOf("</en-media>");\r
 264                 for (; pos>-1; ) {\r
 265                         buffer.replace(pos, pos+11, "");\r
 266                         pos = buffer.indexOf("</en-media>",pos);\r
 267                 }\r
 268                 \r
 269                 \r
 270                 // Make sure we have a proper /> ending the en-media tag\r
 271                 pos = buffer.indexOf("<en-media");\r
 272                 for (; pos>-1; ) {\r
 273                         pos=buffer.indexOf(">", pos);\r
 274                         if (!buffer.substring(pos-1,pos).equals("/"))\r
 275                         buffer.replace(pos, pos+1, " />");\r
 276                         pos = buffer.indexOf("<en-media",pos);\r
 277                 }\r
 278                 \r
 279                 return buffer.toString();\r
 280         }\r
 281 }\r