//
// setString():
//
void MadelineTable::setString(std::string inString){
_readString(inString);
_getLineCount();
_allocateLineInformationVector();
_assignLinePointers();
_determineNumberOfBlocks();
_assignRecordTypes();
_checkDataBlockRectangularity();
_calculateColumnFillRates();
_determineNumberOfColumns();
_determineFieldBoundaries();
_parseHeader();
//
// Now Handled downstream in DataTable instead:
//_classifyColumns();
// This is basically obsoleted now because DataTable now handles classification:
//_resolveDeclaredVersusDiscoveredColumnTypes();
//
_fillVectors();
// DEBUG:
//display();
}
///
/// readFile: Reads a local file on disk
///
void Parser::readFile(const std::string &fileName){
_fileName = fileName;
//////////////////////////////////////////////////////////////////////////////
//
// First find out what kind of file it is using the fileTypeClassifier class:
//
//////////////////////////////////////////////////////////////////////////////
//
// FileTypeClassifier is now a local instance only:
//
FileTypeClassifier fileTypeClassifier;
if(!fileTypeClassifier.setFileToTest(_fileName.c_str())){
Warning("In Parser::readFile(), File %s cannot be located or read and will be ignored",
_fileName.c_str() );
return;
}
_fileType = fileTypeClassifier.classify();
_fileTypeName = fileTypeClassifier.classifyByName();
std::string fileData;
Compressor newCompressor;
//
// Message for user:
//
std::cout << "Parser::readFile(): Opening a(n) " << _fileTypeName << " file ...\n";
//
// If the file is compressed, decompress it into a std::string.
// Otherwise, read it into a std::string.
//
if( _fileType==FileTypeClassifier::PKZIP ||
_fileType==FileTypeClassifier::GZIP ||
_fileType==FileTypeClassifier::BZIP2
)
{
//
// Decompress into a std::string :
//
std::cout << "Parser::readFile(): Decompressing " << _fileTypeName << " file ...\n";
fileData = newCompressor.decompress(_fileName);
}
else
{
//
// Read into a std::string :
//
fileData = _stringify(_fileName);
}
////////////////////////////////////////
//
// Now process the fileData std::string:
//
////////////////////////////////////////
if( _xmlTagManager.discoverFormat(fileData) )
{
//
// fileData are in a recognized XML format:
//
std::cout << "Reading file data in " << _xmlTagManager.getFormatName() << " format ...\n";
_readXML(fileData);
}
else
{
//
// If it is not an XML file, then it is a flat file.
// Two flat file formats are recognized:
//
// (1) Original Madeline format (2) Delimited format
//
// If there is only 1 block, then it is a tab delimited file.
// If there are 2 blocks, it is in Madeline format:
//
int blockCount = _determineNumberOfBlocks( fileData );
if( blockCount == 1)
{
std::cout << "Reading file data in delimited format ...\n";
_readDelimited(fileData);
}
else if( blockCount == 2)
{
std::cout << "Reading file data in Madeline flat file format ...\n";
_readMadeline(fileData);
}
else
{
Warning("Parser::readFile(): Sorry, the format of the file %s could not be determined and will be ignored.",
_fileName.c_str()
);
}
}
}