XMLDocShredder.cpp

   1 /**********************************************************
   2  * XMLDocShredder.cpp
   3  * ---------------------
   4  * Implementation of the class that receives events from the XML parser and
   5  * invokes corresponding construction methods of the storage interface.
   6  *
   7  * Author: Greg Leighton
   8  * Date: 02/11/08
   9  * Changes:
  10  *              05/11/08 -- Fixed bug related to parsing empty elements
  11  *                               -- Set parser properties to automatically resolve
  12  *                                      entity references and load external DTD if present
  13  *                               -- Modified processEndDocument() by adding a nodeFinished()
  14  *                                      call to the storage interface to close off the
  15  *                                      document node
  16  *
  17  */
  18
  19 #include <iostream>
  20 #include "XMLDocShredder.h"
  21 #include "SXSIStorageInterface.h"
  22 #include <libxml++/exceptions/parse_error.h>
  23 #include "Utils.h"
  24
  25 using namespace Glib;
  26
  27 void XMLDocShredder::setProperties(){
  28   /* instruct the parser to expand entity references and report as
  29    * regular PCDATA
  30    */
  31   reader_->set_parser_property(
  32                                TextReader::SubstEntities, true);
  33
  34   /* instruct parser to read external DTD, if present.  This is
  35          * needed to obtain any entity definitions in the DTD
  36          */
  37   reader_->set_parser_property(
  38                                TextReader::LoadDtd, true);
  39
  40
  41   /*
  42    */
  43   reader_->set_parser_property(
  44                                TextReader::DefaultAttrs, true);
  45
  46
  47   /* but we don't want to do validation since it would slow us down
  48    */
  49
  50
  51   reader_->set_parser_property(
  52                                TextReader::Validate, false);
  53
  54 }
  55 XMLDocShredder::XMLDocShredder(const unsigned char * data,
  56                                TextReader::size_type size)
  57 {
  58   last_text = false;
  59   reader_ = new TextReader(data,size,"");
  60   setProperties();
  61   storageIfc_ = new SXSIStorageInterface();
  62 }
  63
  64 XMLDocShredder::XMLDocShredder(const string inFileName)
  65 {
  66   last_text = false;
  67   reader_ = new TextReader(inFileName);
  68   setProperties();
  69   storageIfc_ = new SXSIStorageInterface();
  70
  71 }
  72
  73 XMLDocShredder::~XMLDocShredder()
  74 {
  75         delete reader_;
  76         delete storageIfc_;
  77
  78 }
  79
  80
  81 void XMLDocShredder::processStartElement()
  82 {
  83         // fetch element name; this will be the full qualified name
  84         ustring name = reader_->get_name();
  85         bool empty = false;
  86
  87         if (!last_text)
  88           storageIfc_->newText(""); //prevText
  89         last_text = false;
  90
  91         storageIfc_->newChild(name);
  92
  93         /* We must be really carefull here. calling process attributes moves
  94            the document pointer on the last attribute, hence calling reader_->is_empty
  95            afterwards will yield the wrong result. It is better to call it while we are
  96            on the element and generate a nodeFinished() call at the end */
  97         empty = reader_->is_empty_element();
  98
  99
 100         // now, process attributes
 101         if (reader_->has_attributes())
 102           {
 103             processAttributes();
 104           };
 105
 106
 107         if (empty){
 108             storageIfc_->newText("");  //myText
 109             storageIfc_->nodeFinished(name);
 110             storageIfc_->newText("");  //nextText
 111         };
 112
 113
 114
 115
 116
 117 }
 118
 119 void XMLDocShredder::processEndElement()
 120 {
 121   // tell the storage interface that the current node has been completely processed
 122   if (!last_text)
 123     storageIfc_->newText(""); //nextText of previous node
 124   last_text = false;
 125   storageIfc_->nodeFinished(reader_->get_name());
 126 }
 127
 128 void XMLDocShredder::processPCDATA()
 129 {
 130         // send the content of this PCDATA node to the storage interface as a text node
 131
 132         if (reader_->has_value())
 133         {
 134           storageIfc_->newChild("<$>");
 135           storageIfc_->newText(reader_->get_value());
 136           last_text = true;
 137         }
 138         else
 139           storageIfc_->newText("");
 140 }
 141
 142 void XMLDocShredder::processAttributes()
 143 {
 144         reader_->move_to_first_attribute();
 145
 146         string nspaceStr = "xmlns";
 147         storageIfc_->newChild("<@>");
 148         do
 149         {
 150                 ustring name = reader_->get_name();
 151                 ustring value = reader_->get_value();
 152
 153                 /* the libxml++ TextReader treats the xmlns attribute like an ordinary attribute,
 154                 * so we have to extract it and build a namespace uri node out of it before
 155                 * passing to the storage interface */
 156
 157                 if ((name.find(nspaceStr.c_str(), 0, 5)) == 0)
 158                 {
 159                         storageIfc_->newChild(":" + value);
 160                         storageIfc_->nodeFinished(":" + value);
 161                 }
 162
 163                 /* otherwise, this is an ordinary attribute, so we construct a new child node of the
 164                  * parent element to store the attribute name, possessing a child text node storing the
 165                  * attribute value.  Then, we close off the attribute node with a call to nodeFinished()
 166                  */
 167
 168                 else
 169                 {
 170                         storageIfc_->newChild(name);
 171                         storageIfc_->newChild("<$>");
 172                         storageIfc_->newText(value);
 173                         storageIfc_->nodeFinished("<$>");
 174                 }
 175         }
 176         while (reader_->move_to_next_attribute());
 177         storageIfc_->nodeFinished("<@>");
 178 }
 179
 180 void XMLDocShredder::processSignificantWhitespace()
 181 {
 182         ustring value = reader_->get_value();
 183
 184         // each significant whitespace sequence constructs a text node
 185         storageIfc_->newChild("<$>");
 186         storageIfc_->newText(value);
 187
 188
 189 }
 190
 191 void XMLDocShredder::processStartDocument(const string docName)
 192 {
 193   // tell storage interface to construct the document name
 194   storageIfc_->newChild("");
 195
 196 }
 197
 198 void XMLDocShredder::processEndDocument()
 199 {
 200         /* tell the storage interface that document parsing has finished, and structures
 201          * can now be written to disk. */
 202   storageIfc_->nodeFinished("");
 203   storageIfc_->parsingFinished();
 204 }
 205
 206 void XMLDocShredder::processComment()
 207 {
 208   //storageIfc_->newChild("!" + reader_->get_value());
 209   //storageIfc_->nodeFinished();
 210 }
 211
 212 void XMLDocShredder::processProcessingInstruction()
 213 {
 214         ustring name = reader_->get_name();
 215         ustring value = reader_->get_value();
 216
 217         /* Create a child node to store the target of the PI, append a text node to it to store
 218          * the PI data, send to the storage interface.  Close off the PI node with a call to
 219          * nodeFinished
 220          */
 221
 222         // storageIfc_->newChild("?" + name);
 223         // storageIfc_->newText(value);
 224         // storageIfc_->nodeFinished();
 225 }
 226
 227 void XMLDocShredder::processDocTypeDeclaration()
 228 {
 229         /* We currently ignore the DOCTYPE declaration, but we'll provide this method skeleton
 230          * in case we do want to process it in the future.
 231         */
 232 }
 233
 234 void XMLDocShredder::processCDATASection()
 235 {
 236         /* Currently, we don't preserve CDATA sections since they aren't part of the XPath data
 237          * model.  Instead, we simply pass the converted text value to the storage interface as
 238          * a text node attached to the current context node.
 239          */
 240
 241         ustring value = reader_->get_value();
 242         storageIfc_->newChild("<$>");
 243         storageIfc_->newText(value);
 244         last_text = true;
 245         //      storageIfc_->nodeFinished();
 246
 247 }
 248
 249 void XMLDocShredder::processUnknownNodeType()
 250 {
 251         cout << "unknown token encountered during parsing" << endl;
 252         throw xmlpp::parse_error("unknown token encountered during parsing");
 253
 254 }
 255
 256 void XMLDocShredder::parse()
 257 {
 258         while (reader_->read() && (reader_->get_read_state() != TextReader::Error))
 259         {
 260                 switch (reader_->get_node_type())
 261                 {
 262                         case TextReader::Element:
 263                                 processStartElement();
 264                                 break;
 265
 266                         case TextReader::Text:
 267                                 processPCDATA();
 268                                 break;
 269
 270                         case TextReader::EndElement:
 271                                 processEndElement();
 272                                 break;
 273
 274                         case TextReader::SignificantWhitespace:
 275                                 processSignificantWhitespace();
 276                                 break;
 277
 278                         case TextReader::Comment:
 279                                 processComment();
 280                                 break;
 281
 282                         case TextReader::DocumentType:
 283                                 processDocTypeDeclaration();
 284                                 break;
 285
 286                         case TextReader::ProcessingInstruction:
 287                                 processProcessingInstruction();
 288                                 break;
 289
 290                         case TextReader::CDATA:
 291                                 processCDATASection();
 292                                 break;
 293
 294                         case TextReader::None:
 295                                 processUnknownNodeType();
 296                                 break;
 297
 298                         default:
 299                                 int type = reader_->get_node_type();
 300                                 cout << "  Node type: " << type << endl;
 301                                 break;
 302
 303                 }
 304         }
 305 }