QgsDelimitedTextProvider::QgsDelimitedTextProvider( QString uri )
: QgsVectorDataProvider( uri )
, mDelimiter( "," )
, mDelimiterType( "plain" )
, mFieldCount( 0 )
, mXFieldIndex( -1 )
, mYFieldIndex( -1 )
, mWktFieldIndex( -1 )
, mWktHasZM( false )
, mWktZMRegexp( "\\s+(?:z|m|zm)(?=\\s*\\()", Qt::CaseInsensitive )
, mWktCrdRegexp( "(\\-?\\d+(?:\\.\\d*)?\\s+\\-?\\d+(?:\\.\\d*)?)\\s[\\s\\d\\.\\-]+" )
, mFile( 0 )
, mStream( 0 )
, mSkipLines( 0 )
, mFirstDataLine( 0 )
, mShowInvalidLines( false )
, mCrs()
, mWkbType( QGis::WKBUnknown )
{
QUrl url = QUrl::fromEncoded( uri.toAscii() );
// Extract the provider definition from the url
mFileName = url.toLocalFile();
QString wktField;
QString xField;
QString yField;
if ( url.hasQueryItem( "delimiter" ) )
mDelimiter = url.queryItemValue( "delimiter" );
if ( url.hasQueryItem( "delimiterType" ) )
mDelimiterType = url.queryItemValue( "delimiterType" );
if ( url.hasQueryItem( "wktField" ) )
wktField = url.queryItemValue( "wktField" );
if ( url.hasQueryItem( "xField" ) )
xField = url.queryItemValue( "xField" );
if ( url.hasQueryItem( "yField" ) )
yField = url.queryItemValue( "yField" );
if ( url.hasQueryItem( "skipLines" ) )
mSkipLines = url.queryItemValue( "skipLines" ).toInt();
if ( url.hasQueryItem( "crs" ) )
mCrs.createFromString( url.queryItemValue( "crs" ) );
if ( url.hasQueryItem( "decimalPoint" ) )
mDecimalPoint = url.queryItemValue( "decimalPoint" );
QgsDebugMsg( "Data source uri is " + uri );
QgsDebugMsg( "Delimited text file is: " + mFileName );
QgsDebugMsg( "Delimiter is: " + mDelimiter );
QgsDebugMsg( "Delimiter type is: " + mDelimiterType );
QgsDebugMsg( "wktField is: " + wktField );
QgsDebugMsg( "xField is: " + xField );
QgsDebugMsg( "yField is: " + yField );
QgsDebugMsg( "skipLines is: " + QString::number( mSkipLines ) );
// if delimiter contains some special characters, convert them
if ( mDelimiterType != "regexp" )
mDelimiter.replace( "\\t", "\t" ); // replace "\t" with a real tabulator
// Set the selection rectangle to null
mSelectionRectangle = QgsRectangle();
// assume the layer is invalid until proven otherwise
mValid = false;
if ( mFileName.isEmpty() || mDelimiter.isEmpty() )
{
// uri is invalid so the layer must be too...
QgsDebugMsg( "Data source is invalid" );
return;
}
// check to see that the file exists and perform some sanity checks
if ( !QFile::exists( mFileName ) )
{
QgsDebugMsg( "Data source " + dataSourceUri() + " doesn't exist" );
return;
}
// Open the file and get number of rows, etc. We assume that the
// file has a header row and process accordingly. Caller should make
// sure that the delimited file is properly formed.
mFile = new QFile( mFileName );
if ( !mFile->open( QIODevice::ReadOnly ) )
{
QgsDebugMsg( "Data source " + dataSourceUri() + " could not be opened" );
delete mFile;
mFile = 0;
return;
}
// now we have the file opened and ready for parsing
// set the initial extent
mExtent = QgsRectangle();
QMap<int, bool> couldBeInt;
QMap<int, bool> couldBeDouble;
mStream = new QTextStream( mFile );
QString line;
mNumberFeatures = 0;
int lineNumber = 0;
bool hasFields = false;
while ( !mStream->atEnd() )
{
lineNumber++;
line = readLine( mStream ); // line of text excluding '\n', default local 8 bit encoding.
if ( lineNumber < mSkipLines + 1 )
continue;
if ( line.isEmpty() )
continue;
if ( !hasFields )
{
// Get the fields from the header row and store them in the
// fields vector
QStringList fieldList = splitLine( line );
mFieldCount = fieldList.count();
// We don't know anything about a text based field other
// than its name. All fields are assumed to be text
int fieldPos = 0;
for ( int column = 0; column < mFieldCount; column++ )
{
QString field = fieldList[column];
if (( field.left( 1 ) == "'" || field.left( 1 ) == "\"" ) &&
field.left( 1 ) == field.right( 1 ) )
// eat quotes
field = field.mid( 1, field.length() - 2 );
if ( field.length() == 0 )
// skip empty field names
continue;
// check to see if this field matches either the x or y field
if ( !wktField.isEmpty() && wktField == field )
{
QgsDebugMsg( "Found wkt field: " + ( field ) );
mWktFieldIndex = column;
}
else if ( !xField.isEmpty() && xField == field )
{
QgsDebugMsg( "Found x field: " + ( field ) );
mXFieldIndex = column;
}
else if ( !yField.isEmpty() && yField == field )
{
QgsDebugMsg( "Found y field: " + ( field ) );
mYFieldIndex = column;
}
// WKT geometry field won't be displayed in attribute tables
if ( column == mWktFieldIndex )
continue;
QgsDebugMsg( "Adding field: " + ( field ) );
// assume that the field could be integer or double
// for now, let's set field type as text
attributeColumns.append( column );
attributeFields[fieldPos] = QgsField( field, QVariant::String, "Text" );
couldBeInt.insert( fieldPos, true );
couldBeDouble.insert( fieldPos, true );
fieldPos++;
}
if ( mWktFieldIndex >= 0 )
{
mXFieldIndex = -1;
mYFieldIndex = -1;
}
QgsDebugMsg( "wktfield index: " + QString::number( mWktFieldIndex ) );
QgsDebugMsg( "xfield index: " + QString::number( mXFieldIndex ) );
QgsDebugMsg( "yfield index: " + QString::number( mYFieldIndex ) );
QgsDebugMsg( "Field count for the delimited text file is " + QString::number( attributeFields.size() ) );
hasFields = true;
}
else // hasFields == true - field names already read
{
if ( mFirstDataLine == 0 )
mFirstDataLine = lineNumber;
// split the line on the delimiter
QStringList parts = splitLine( line );
// Ensure that the input has at least the required number of fields (mainly to tolerate
// missed blank strings at end of row)
while ( parts.size() < mFieldCount )
parts.append( QString::null );
if ( mWktFieldIndex >= 0 )
{
// Get the wkt - confirm it is valid, get the type, and
// if compatible with the rest of file, add to the extents
QString sWkt = parts[mWktFieldIndex];
QgsGeometry *geom = 0;
try
{
if ( !mWktHasZM && sWkt.indexOf( mWktZMRegexp ) >= 0 )
mWktHasZM = true;
if ( mWktHasZM )
{
sWkt.remove( mWktZMRegexp ).replace( mWktCrdRegexp, "\\1" );
}
geom = QgsGeometry::fromWkt( sWkt );
}
catch ( ... )
{
mInvalidLines << line;
geom = 0;
}
if ( geom )
{
QGis::WkbType type = geom->wkbType();
if ( type != QGis::WKBNoGeometry )
{
if ( mNumberFeatures == 0 )
{
mNumberFeatures++;
mWkbType = type;
mExtent = geom->boundingBox();
}
else if ( type == mWkbType )
{
mNumberFeatures++;
QgsRectangle bbox( geom->boundingBox() );
mExtent.combineExtentWith( &bbox );
}
}
delete geom;
}
}
else if ( mWktFieldIndex == -1 && mXFieldIndex >= 0 && mYFieldIndex >= 0 )
{
// Get the x and y values, first checking to make sure they
// aren't null.
QString sX = parts[mXFieldIndex];
QString sY = parts[mYFieldIndex];
if ( !mDecimalPoint.isEmpty() )
{
sX.replace( mDecimalPoint, "." );
sY.replace( mDecimalPoint, "." );
}
bool xOk = false;
bool yOk = false;
double x = sX.toDouble( &xOk );
double y = sY.toDouble( &yOk );
if ( xOk && yOk )
{
if ( mNumberFeatures > 0 )
{
mExtent.combineExtentWith( x, y );
}
else
{
// Extent for the first point is just the first point
mExtent.set( x, y, x, y );
mWkbType = QGis::WKBPoint;
}
mNumberFeatures++;
}
else
{
mInvalidLines << line;
}
}
else
{
mWkbType = QGis::WKBNoGeometry;
mNumberFeatures++;
}
for ( int i = 0; i < attributeFields.size(); i++ )
{
QString &value = parts[attributeColumns[i]];
if ( value.isEmpty() )
continue;
// try to convert attribute values to integer and double
if ( couldBeInt[i] )
{
value.toInt( &couldBeInt[i] );
}
if ( couldBeDouble[i] )
{
value.toDouble( &couldBeDouble[i] );
}
}
}
}
QgsDebugMsg( "geometry type is: " + QString::number( mWkbType ) );
QgsDebugMsg( "feature count is: " + QString::number( mNumberFeatures ) );
// now it's time to decide the types for the fields
for ( QgsFieldMap::iterator it = attributeFields.begin(); it != attributeFields.end(); ++it )
{
if ( couldBeInt[it.key()] )
{
it->setType( QVariant::Int );
it->setTypeName( "integer" );
}
else if ( couldBeDouble[it.key()] )
{
it->setType( QVariant::Double );
it->setTypeName( "double" );
}
}
mValid = mWkbType != QGis::WKBUnknown;
}
QgsDelimitedTextProvider::QgsDelimitedTextProvider( QString uri )
: QgsVectorDataProvider( uri ),
mXFieldIndex( -1 ), mYFieldIndex( -1 ),
mShowInvalidLines( true )
{
// Get the file name and mDelimiter out of the uri
mFileName = uri.left( uri.indexOf( "?" ) );
// split the string up on & to get the individual parameters
QStringList parameters = uri.mid( uri.indexOf( "?" ) ).split( "&", QString::SkipEmptyParts );
QgsDebugMsg( "Parameter count after split on &" + QString::number( parameters.size() ) );
// get the individual parameters and assign values
QStringList temp = parameters.filter( "delimiter=" );
mDelimiter = temp.size() ? temp[0].mid( temp[0].indexOf( "=" ) + 1 ) : "";
temp = parameters.filter( "delimiterType=" );
mDelimiterType = temp.size() ? temp[0].mid( temp[0].indexOf( "=" ) + 1 ) : "";
temp = parameters.filter( "xField=" );
QString xField = temp.size() ? temp[0].mid( temp[0].indexOf( "=" ) + 1 ) : "";
temp = parameters.filter( "yField=" );
QString yField = temp.size() ? temp[0].mid( temp[0].indexOf( "=" ) + 1 ) : "";
// Decode the parts of the uri. Good if someone entered '=' as a delimiter, for instance.
mFileName = QUrl::fromPercentEncoding( mFileName.toUtf8() );
mDelimiter = QUrl::fromPercentEncoding( mDelimiter.toUtf8() );
mDelimiterType = QUrl::fromPercentEncoding( mDelimiterType.toUtf8() );
xField = QUrl::fromPercentEncoding( xField.toUtf8() );
yField = QUrl::fromPercentEncoding( yField.toUtf8() );
QgsDebugMsg( "Data source uri is " + uri );
QgsDebugMsg( "Delimited text file is: " + mFileName );
QgsDebugMsg( "Delimiter is: " + mDelimiter );
QgsDebugMsg( "Delimiter type is: " + mDelimiterType );
QgsDebugMsg( "xField is: " + xField );
QgsDebugMsg( "yField is: " + yField );
// if delimiter contains some special characters, convert them
if ( mDelimiterType == "regexp" )
mDelimiterRegexp = QRegExp( mDelimiter );
else
mDelimiter.replace( "\\t", "\t" ); // replace "\t" with a real tabulator
// Set the selection rectangle to null
mSelectionRectangle = QgsRectangle();
// assume the layer is invalid until proven otherwise
mValid = false;
if ( mFileName.isEmpty() || mDelimiter.isEmpty() || xField.isEmpty() || yField.isEmpty() )
{
// uri is invalid so the layer must be too...
QString( "Data source is invalid" );
return;
}
// check to see that the file exists and perform some sanity checks
if ( !QFile::exists( mFileName ) )
{
QgsDebugMsg( "Data source " + dataSourceUri() + " doesn't exist" );
return;
}
// Open the file and get number of rows, etc. We assume that the
// file has a header row and process accordingly. Caller should make
// sure the the delimited file is properly formed.
mFile = new QFile( mFileName );
if ( !mFile->open( QIODevice::ReadOnly ) )
{
QgsDebugMsg( "Data source " + dataSourceUri() + " could not be opened" );
delete mFile;
return;
}
// now we have the file opened and ready for parsing
// set the initial extent
mExtent = QgsRectangle();
QMap<int, bool> couldBeInt;
QMap<int, bool> couldBeDouble;
mStream = new QTextStream( mFile );
QString line;
mNumberFeatures = 0;
int lineNumber = 0;
bool firstPoint = true;
bool hasFields = false;
while ( !mStream->atEnd() )
{
lineNumber++;
line = mStream->readLine(); // line of text excluding '\n', default local 8 bit encoding.
if ( !hasFields )
{
// Get the fields from the header row and store them in the
// fields vector
QgsDebugMsg( "Attempting to split the input line: " + line + " using delimiter " + mDelimiter );
QStringList fieldList;
if ( mDelimiterType == "regexp" )
fieldList = line.split( mDelimiterRegexp );
else
fieldList = line.split( mDelimiter );
QgsDebugMsg( "Split line into " + QString::number( fieldList.size() ) + " parts" );
// We don't know anything about a text based field other
// than its name. All fields are assumed to be text
int fieldPos = 0;
for ( QStringList::Iterator it = fieldList.begin();
it != fieldList.end(); ++it )
{
QString field = *it;
if ( field.length() > 0 )
{
// for now, let's set field type as text
attributeFields[fieldPos] = QgsField( *it, QVariant::String, "Text" );
// check to see if this field matches either the x or y field
if ( xField == *it )
{
QgsDebugMsg( "Found x field: " + ( *it ) );
mXFieldIndex = fieldPos;
}
else if ( yField == *it )
{
QgsDebugMsg( "Found y field: " + ( *it ) );
mYFieldIndex = fieldPos;
}
QgsDebugMsg( "Adding field: " + ( *it ) );
// assume that the field could be integer or double
couldBeInt.insert( fieldPos, true );
couldBeDouble.insert( fieldPos, true );
fieldPos++;
}
}
QgsDebugMsg( "Field count for the delimited text file is " + QString::number( attributeFields.size() ) );
hasFields = true;
}
else if ( mXFieldIndex != -1 && mYFieldIndex != -1 )
{
mNumberFeatures++;
// split the line on the delimiter
QStringList parts;
if ( mDelimiterType == "regexp" )
parts = line.split( mDelimiterRegexp );
else
parts = line.split( mDelimiter );
// Skip malformed lines silently. Report line number with nextFeature()
if ( attributeFields.size() != parts.size() )
{
continue;
}
// Get the x and y values, first checking to make sure they
// aren't null.
QString sX = parts[mXFieldIndex];
QString sY = parts[mYFieldIndex];
bool xOk = true;
bool yOk = true;
double x = sX.toDouble( &xOk );
double y = sY.toDouble( &yOk );
if ( xOk && yOk )
{
if ( !firstPoint )
{
mExtent.combineExtentWith( x, y );
}
else
{ // Extent for the first point is just the first point
mExtent.set( x, y, x, y );
firstPoint = false;
}
}
int i = 0;
for ( QStringList::iterator it = parts.begin(); it != parts.end(); ++it, ++i )
{
// try to convert attribute values to integer and double
if ( couldBeInt[i] )
{
it->toInt( &couldBeInt[i] );
}
if ( couldBeDouble[i] )
{
it->toDouble( &couldBeDouble[i] );
}
}
}
}
// now it's time to decide the types for the fields
for ( QgsFieldMap::iterator it = attributeFields.begin(); it != attributeFields.end(); ++it )
{
if ( couldBeInt[it.key()] )
{
it->setType( QVariant::Int );
it->setTypeName( "integer" );
}
else if ( couldBeDouble[it.key()] )
{
it->setType( QVariant::Double );
it->setTypeName( "double" );
}
}
if ( mXFieldIndex != -1 && mYFieldIndex != -1 )
{
QgsDebugMsg( "Data store is valid" );
QgsDebugMsg( "Number of features " + QString::number( mNumberFeatures ) );
QgsDebugMsg( "Extents " + mExtent.toString() );
mValid = true;
}
else
{
QgsDebugMsg( "Data store is invalid. Specified x,y fields do not match those in the database" );
}
QgsDebugMsg( "Done checking validity" );
}