void multiTxtDocument::loadFile ( string fileName, bool caseOn, bool noPunct, bool debugMode, bool noTxtIds, bool tercomLike )
{
if ( multiTxtDocumentParams.debugMode )
{
cerr << "DEBUG tercpp : multiTxtDocument::loadFile : loading files " << endl << fileName << endl << "END DEBUG" << endl;
cerr << "DEBUG tercpp : multiTxtDocument::loadFile : testing params " << endl << Tools::printParams ( multiTxtDocumentParams ) << endl << "END DEBUG" << endl;
cerr << "DEBUG tercpp : multiTxtDocument::loadFile : testing others params " << endl << "caseOn : " << caseOn << endl << "noPunct : " << noPunct << endl << "debugMode : " << debugMode << endl << "noTxtIds : " << noTxtIds << endl << "tercomLike : " << tercomLike << endl << "END DEBUG" << endl;
}
ifstream fichierLoad ( fileName.c_str(), ios::in );
string line="";
documentStructure l_doc;
stringstream l_stream;
l_doc.setFileName(fileName);
l_stream.str ( "" );
l_stream << ( int ) documents.size();
l_doc.setDocId ( l_stream.str() );
if ( fichierLoad )
{
int l_ids = 1;
l_stream.str ( "" );
string l_key="";
string line_mod="";
while ( getline ( fichierLoad, line ) )
{
l_key="";
line_mod="";
l_stream.str ( "" );
if ( noTxtIds )
{
l_stream << l_ids;
l_key = l_stream.str();
line_mod = line;
l_ids++;
}
else
{
if ((int)line.rfind ( "(" )==-1)
{
cerr << "ERROR : multiTxtDocument::loadFile : Id not found, maybe you should use the --noTxtIds Option ? " << endl;
exit ( 0 );
}
l_key = line.substr ( line.rfind ( "(" ), line.size() - 1 );
line_mod = line.substr ( 0, line.rfind ( "(" ) - 1 );
}
if ( multiTxtDocumentParams.debugMode )
{
cerr << "DEBUG multiTxtDocument::loadFile : line NOT tokenized |" << line_mod << "|" << endl << "END DEBUG" << endl;
}
if ( !tercomLike )
{
if ( multiTxtDocumentParams.debugMode )
{
cerr << "DEBUG tercpp : multiTxtDocument::loadFile : " << endl << "TERCOM AT FALSE " << endl << "END DEBUG" << endl;
}
// line_mod = tokenizePunct ( line_mod );
}
if ( !caseOn )
{
if ( multiTxtDocumentParams.debugMode )
{
cerr << "DEBUG tercpp : multiTxtDocument::loadFile : " << endl << "CASEON AT FALSE " << endl << "END DEBUG" << endl;
}
line_mod = lowerCase ( line_mod );
}
if ( noPunct )
{
if ( multiTxtDocumentParams.debugMode )
{
cerr << "DEBUG tercpp : multiTxtDocument::loadFile : " << endl << "NOPUNCT AT TRUE " << endl << "END DEBUG" << endl;
}
if ( !tercomLike )
{
line_mod = removePunctTercom ( line_mod );
}
else
{
line_mod = removePunct ( line_mod );
}
}
if ( multiTxtDocumentParams.debugMode )
{
cerr << "DEBUG multiTxtDocument::loadFile : line tokenized |" << line_mod << "|" << endl << "END DEBUG" << endl;
}
vector<string> vecDocLine = stringToVector ( line_mod, " " );
// string l_key;
// hashHypothesis.addValue(l_key,vecDocLine);
// l_key=(string)vecDocLine.at((int)vecDocLine.size()-1);
// vecDocLine.pop_back();
if ( multiTxtDocumentParams.debugMode )
{
cerr << "DEBUG tercpp multiTxtDocument::loadFile : " << l_key << "|" << vectorToString ( vecDocLine ) << "|" << endl << "Vector Size : " << vecDocLine.size() << endl << "Line length : " << ( int ) line_mod.length() << endl << "END DEBUG" << endl;
}
// hashHypothesis.addValue(l_key,vecDocLine);
segmentStructure l_seg ( l_key, vecDocLine, l_doc.getDocId() );
l_doc.addSegments ( l_seg );
}
// Ref=line;
// getline ( fichierHyp, line );
// Hyp=line;
fichierLoad.close(); // on ferme le fichier
addDocument ( l_doc );
if ( multiTxtDocumentParams.debugMode )
{
cerr << "DEBUG multiTxtDocument::loadFile : document " << l_doc.getDocId() << " added !!!" << endl << "END DEBUG" << endl;
}
l_key.erase();
line_mod.erase();
l_stream.str("");
}
else // sinon
{
cerr << "ERROR : multiTxtDocument::loadFile : can't open file : " + fileName + " !" << endl;
exit ( 0 );
}
}
void xmlStructure::copy_to_SGMLDocument ( SGMLDocument* sgmlDoc, TiXmlNode* pParent, unsigned int indent )
{
if ( !pParent )
return;
TiXmlNode* pChild;
TiXmlText* pText;
int t = pParent->Type();
// printf ( "%s", getIndent ( indent ) );
// int num;
string elementValue;
bool unknownTag=false;
if ( xmlParams.debugMode )
{
cerr << "DEBUG tercpp : xmlStructure::copy_to_SGMLDocument : " << endl << " TiXmlNode: " << t << endl << "END DEBUG" << endl;
cerr << "DEBUG tercpp : xmlStructure::copy_to_SGMLDocument : " << endl << " indent: " << indent << endl << "END DEBUG" << endl;
}
switch ( t )
{
case TiXmlNode::DOCUMENT:
if ( xmlParams.debugMode )
{
cerr << "DEBUG tercpp : xmlStructure::copy_to_SGMLDocument : " << endl << " Document" << endl << "END DEBUG" << endl;
}// printf ( "Document" );
break;
case TiXmlNode::ELEMENT:
// printf ( "Element [%s]", pParent->Value() );
elementValue = Tools::lowerCase ( pParent->Value() );
if ( xmlParams.debugMode )
{
cerr << "DEBUG tercpp : xmlStructure::copy_to_SGMLDocument : " << endl << " elementValue: " << elementValue << endl << "END DEBUG" << endl;
}
if ( ( ( int ) elementValue.compare ( "refset" ) == 0 ) || ( ( int ) elementValue.compare ( "tstset" ) == 0 ) )
{
sgmlDoc->setDocType ( elementValue );
}
else
if (( ( int ) elementValue.compare ( "doc" ) == 0 ) || ( ( int ) elementValue.compare ( "DOC" ) == 0 ))
{
documentStructure tmp_doc;
sgmlDoc->addDocument ( tmp_doc );
}
else
if ( ( ( int ) elementValue.compare ( "seg" ) == 0 ) || ( ( int ) elementValue.compare ( "SEG" ) == 0 ) )
{
segmentStructure tmp_seg;
( sgmlDoc->getLastDocument() )->addSegments ( tmp_seg );
}
else
{
unknownTag=true;
}
if ( xmlParams.debugMode )
{
cerr << "DEBUG tercpp : xmlStructure::copy_to_SGMLDocument : " << endl << " Calling dump_attribs_to_SGMLDocuments with indent :" << indent + 1 << endl << "END DEBUG" << endl;
}
if (!unknownTag)
{
dump_attribs_to_SGMLDocuments ( sgmlDoc, pParent->ToElement(), indent + 1 );
}
// num = dump_attribs_to_stdout ( pParent->ToElement(), indent + 1 );
// switch ( num )
// {
// case 0:
// printf ( " (No attributes)" );
// break;
// case 1:
// printf ( "%s1 attribute", getIndentAlt ( indent ) );
// break;
// default:
// printf ( "%s%d attributes", getIndentAlt ( indent ), num );
// break;
// }
break;
// case TiXmlNode::COMMENT:
// printf ( "Comment: [%s]", pParent->Value() );
// break;
//
// case TiXmlNode::UNKNOWN:
// printf ( "Unknown" );
// break;
case TiXmlNode::TEXT:
pText = pParent->ToText();
// printf ( "Text: [%s]", pText->Value() );
if ( indent >= 2 )
{
documentStructure * l_tmp_doc = sgmlDoc->getLastDocument();
segmentStructure * l_tmp_seg = l_tmp_doc->getLastSegments();
string l_text = pText->Value();
string line_mod = l_text;
if ( !xmlParams.tercomLike )
{
if ( xmlParams.debugMode )
{
cerr << "DEBUG xmlStructure::copy_to_SGMLDocument : line NOT tokenized |" << line_mod << "|" << endl << "END DEBUG" << endl;
}
if ( xmlParams.debugMode )
{
cerr << "DEBUG tercpp : xmlStructure::copy_to_SGMLDocument : " << endl << "TERCOM AT FALSE " << endl << "END DEBUG" << endl;
}
line_mod = tokenizePunct ( line_mod );
}
if ( !xmlParams.caseOn )
{
if ( xmlParams.debugMode )
{
cerr << "DEBUG tercpp : xmlStructure::copy_to_SGMLDocument : " << endl << "CASEON AT FALSE " << endl << "END DEBUG" << endl;
}
line_mod = lowerCase ( line_mod );
}
if ( xmlParams.noPunct )
{
if ( xmlParams.debugMode )
{
cerr << "DEBUG tercpp : xmlStructure::copy_to_SGMLDocument : " << endl << "NOPUNCT AT TRUE " << endl << "END DEBUG" << endl;
}
if ( !xmlParams.tercomLike )
{
line_mod = removePunctTercom ( line_mod );
}
else
{
line_mod = removePunct ( line_mod );
}
}
if ( xmlParams.debugMode )
{
cerr << "DEBUG xmlStructure::copy_to_SGMLDocument : line tokenized |" << line_mod << "|" << endl << "END DEBUG" << endl;
}
l_tmp_seg->addContent ( line_mod );
}
break;
// case TiXmlNode::DECLARATION:
// printf ( "Declaration" );
// break;
default:
cerr << "DEBUG tercpp : xmlStructure::copy_to_SGMLDocument : " << endl << "Default TiXmlNode: " << t << endl << "END DEBUG" << endl;
break;
}
// printf ( "\n" );
for ( pChild = pParent->FirstChild(); pChild != 0; pChild = pChild->NextSibling() )
{
copy_to_SGMLDocument ( sgmlDoc, pChild, indent + 1 );
}
}