public void mReadMap(RecordSetMD aRecordSetMD, JniPMMLItem aJniPMML, PrintWriter pw, Boolean bFillDictionaryNames)
//throws com.WDataSci.WDS.WDSException
{
try {
int i = -1;
int ii = -1;
int j = -1;
int jj = -1;
int k = -1;
int kk = -1;
//Java Boolean bUsingPMML = (aJniPMML != null && aJniPMML.PMMLMatter.Doc != null);
//C#
Boolean bUsingPMML = false;
String[] lFieldStringNames = null;
int nDataFieldNames = 0;
if (bUsingPMML)
{
lFieldStringNames = aJniPMML.PMMLDataFieldStringNames();
nDataFieldNames = lFieldStringNames.Length;
}
if (pw != null)
{
pw.printf("In RecordSetMD constructor\n");
}
if (pw != null)
{
pw.flush();
}
//point to file or memory
DBB buffer = this.Header.Buffer;
buffer.position(0, 0, 0);
try {
//get compound dataset information
long nColumns = buffer.nRecords;
long nBlockMaxStringByteLength = this.Header.MaxStringByteLength;
int nBlockCoreColumnSize = (int)(buffer.nRecordFLenBytes);
int nBlockAllocatedSize = (int)(buffer.nDBBRequiredBytes);
if (nBlockAllocatedSize > this.Header.Buffer.Length)
{
throw new com.WDataSci.WDS.WDSException("Error, HeaderBuffer capacity, " + this.Header.Buffer.Length
+ ", is less then what should be BlockAllocatedSize, " + nBlockAllocatedSize
);
}
int nBlockCoreSize = (int)(buffer.nDBBLeadingBytes + buffer.nDBBFLenBytes);
if (pw != null)
{
pw.printf("In RecordSetMD constructor, nColumns=%d\n", nColumns);
}
if (pw != null)
{
pw.flush();
}
byte[] namebuffer = new byte[(int)(nBlockMaxStringByteLength)];
//iterate through columns (dataset members)
aRecordSetMD.Column = new FieldMD[(int)(nColumns)];
int bptr = 0;
for (ii = 0; ii < nColumns; ii++, bptr += (int)nBlockCoreColumnSize)
{
buffer.position(buffer.ptr, bptr, buffer.vlenptr);
aRecordSetMD.Column[ii] = new FieldMD();
FieldMD col = aRecordSetMD.Column[ii];
//the first two fields are always names and taken to be variable length
col.Name = buffer.GetLayerVLenString(1, nBlockMaxStringByteLength);
//Check for PMML DataFieldName map
//If not mapped externally, the next VLenString pointer will be 0 and will come back as empty
String tmpname = buffer.GetLayerVLenString(1, nBlockMaxStringByteLength);
//Search for PMML DataFieldName map, take input supplied map first, then the usual search
/* Java >>> *
* Boolean found = false;
* if ( bUsingPMML && tmpname.length() > 0 ) {
* for (j = 0; !found && j < nDataFieldNames; j++) {
* if ( tmpname.equals(lFieldStringNames[j]) ) {
* col.MapToMapKey(lFieldStringNames[j]);
* found = true;
* break;
* }
* }
* }
* if ( bUsingPMML && !found ) {
* for (j = 0; !found && j < nDataFieldNames; j++) {
* if ( col.Name.equals(lFieldStringNames[j]) ) {
* col.MapToMapKey(lFieldNames[j]);
* found = true;
* break;
* }
* }
* }
* /* <<< Java */
if (!bUsingPMML && bFillDictionaryNames)
{
col.MapToMapKey(col.Name);
}
//Java col.DTyp = FieldMDEnums.eDTyp.FromInt(buffer.GetLayerInt(1));
//C#
col.DTyp = FieldMDExt.eDTyp_FromInt((int)buffer.GetLayerInt(1));
//See notes above on Date and DateTime types
//Since the header block is being passed in, Date and DateTime types are provided.
col.ByteMemLength = (long)buffer.GetLayerLong(1);
col.ByteMaxLength = (long)buffer.GetLayerLong(1);
if (col.DTyp.bIn(FieldMDEnums.eDTyp.VLS, FieldMDEnums.eDTyp.Str))
{
col.StringMaxLength = (int)(col.ByteMaxLength / 2);
}
//no longer using the last byte for bIsMappedToPMMLFieldName, so there is filler space at the end
}
}
catch (Exception e) {
throw new com.WDataSci.WDS.WDSException("Error in RecordSetMD processing of Header DBB:", e);
}
if (pw != null)
{
pw.printf("leaving RecordSetMD constructor\n");
}
if (pw != null)
{
pw.flush();
}
}
catch (Exception e) {
throw new com.WDataSci.WDS.WDSException("Error in ProcessInputMapFor", e);
}
}
public void mReadMapFor(RecordSetMD aRecordSetMD, JniPMMLItem aJniPMML, PrintWriter pw, Boolean bFillDictionaryNames)
//throws com.WDataSci.WDS.WDSException
{
try {
int ii = -1;
int j = -1;
int k = -1;
//are we using a JniPMML object (as when called from C# and does it have PMMLMatter
Boolean bUsingJniPMML = (aJniPMML != null);
//Java Boolean bCheckingAgainstPMML = (aJniPMML != null && aJniPMML.PMMLMatter.Doc != null);
//C#
Boolean bCheckingAgainstPMML = false;
//Java org.dmg.pmml.DataField[] lDataFields = null;
String[] lFieldStringNames = null;
int nDataFieldNames = 0;
if (bCheckingAgainstPMML)
{
/* Java >>> *
* lDataFields = aJniPMML.PMMLDataFields();
* nDataFieldNames = lDataFields.length;
* lFieldStringNames = new String[nDataFieldNames];
* for (i = 0; i < nDataFieldNames; i++)
* lFieldStringNames[i] = lDataFields[i].getName().getValue();
* /* <<< Java */
}
String aPathAndName = null;
aPathAndName = com.WDataSci.WDS.Util.PathAndName(this.Path, this.FileName);
/* Java >>> *
* this.__CSVParser = new CSVParser(new FileReader(aPathAndName), CSVFormat.EXCEL);
* this.__CSVParserIterator = this.__CSVParser.iterator();
* CSVRecord inputLine = this.__CSVParserIterator.next();
* /* <<< Java */
/* C# >>> */
this.__StreamReader = new StreamReader(aPathAndName);
this.__CSV = new CsvParser(this.__StreamReader);
String[] inputLine = this.__CSV.Read();
/* <<< C# */
//Java int xnlLength = inputLine.size();
//C#
int xnlLength = inputLine.Length;
aRecordSetMD.Column = new FieldMD[xnlLength];
//for when the length is packed into the XSD type for limiting strings
int[] typl = new int[1];
for (ii = 0; ii < xnlLength; ii++)
{
aRecordSetMD.Column[ii] = new FieldMD();
FieldMD cm = aRecordSetMD.Column[ii];
//Java cm.Name = inputLine.get(ii);
//C#
cm.Name = inputLine[ii];
Boolean found = false;
if (bCheckingAgainstPMML)
{
//Search for PMML DataFieldName map
for (j = 0; !found && j < nDataFieldNames; j++)
{
if (cm.Name.equals(lFieldStringNames[j]))
{
found = true;
cm.MapToMapKey(lFieldStringNames[j]);
break;
}
}
/* Java >>> *
* // if found in PMML, check for PMML DataType
* if ( found ) {
* org.dmg.pmml.DataType ofdtyp = lDataFields[j].getDataType();
* switch ( ofdtyp ) {
* case FLOAT:
* case DOUBLE:
* cm.DTyp = FieldMDEnums.eDTyp.Dbl;
* break;
* case INTEGER:
* if ( cm.Name.endsWith("ID") )
* cm.DTyp = FieldMDEnums.eDTyp.Lng;
* else
* cm.DTyp = FieldMDEnums.eDTyp.Int;
* break;
* case STRING:
* cm.DTyp = FieldMDEnums.eDTyp.VLS;
* break;
* case DATE:
* cm.DTyp = FieldMDEnums.eDTyp.Dte;
* break;
* case DATE_TIME:
* case TIME:
* cm.DTyp = FieldMDEnums.eDTyp.DTm;
* break;
* case BOOLEAN:
* cm.DTyp = FieldMDEnums.eDTyp.Bln;
* break;
* case DATE_DAYS_SINCE_0:
* case DATE_DAYS_SINCE_1960:
* case DATE_DAYS_SINCE_1970:
* case DATE_DAYS_SINCE_1980:
* case TIME_SECONDS:
* case DATE_TIME_SECONDS_SINCE_0:
* case DATE_TIME_SECONDS_SINCE_1960:
* case DATE_TIME_SECONDS_SINCE_1970:
* case DATE_TIME_SECONDS_SINCE_1980:
* default:
* cm.DTyp = FieldMDEnums.eDTyp.Int;
* break;
* }
* }
* /* <<< Java */
}
if (!found)
{
if (bFillDictionaryNames)
{
cm.MapToMapKey(cm.Name);
}
String s = cm.Name;
String cls = null;
String mod = null;
String rep = null;
String postmod = null;
//We are assuming a modified three part naming convention
//The core three part naming convention is assumed to be CamelCase and of the form
// Class[Modifier]Representation
//The Representation defines the data type (we will assume all strings are VLen).
//
//We allow a _PostModifier (preceded by an underscore). This implies that all
//fields with a root name are related and we are not embedding obfuscation.
//
//Examples based on consumer loan data are:
// Long form Short form
// PrincipalBalance PrinBal
// PaymentAmount PmtAmt
// PaymentPrincipalAmount PmtPrinAmt
// PaymentPrincipalAmount_Lag1 PmtPrinAmt_Lag1
//
//See documentation on WDataSci naming conventions.
k = s.lastIndexOf("_");
if (k > 0)
{
postmod = s.substring(k + 1);
s = s.substring(0, k - 1);
}
String sl = s.toLowerCase();
//recognize Date and DateTime first
if (s.endsWith("Date"))
{
cm.DTyp = FieldMDEnums.eDTyp.Dte;
}
else if (s.endsWith("Time"))
{
cm.DTyp = FieldMDEnums.eDTyp.DTm;
}
//longs
else if (cm.Name.endsWith("ID"))
{
cm.DTyp = FieldMDEnums.eDTyp.Lng;
}
else if (cm.Name.endsWith("Nbr") && (sl.startsWith("acc") || sl.startsWith("loan")))
{
cm.DTyp = FieldMDEnums.eDTyp.Lng;
}
else if (cm.Name.endsWith("Nbr"))
{
cm.DTyp = FieldMDEnums.eDTyp.Int;
}
else if (cm.Name.endsWith("Count"))
{
cm.DTyp = FieldMDEnums.eDTyp.Int;
}
else if (cm.Name.endsWith("Status"))
{
cm.DTyp = FieldMDEnums.eDTyp.VLS;
}
else if (cm.Name.length() > 3 &&
bIn(cm.Name.substring(cm.Name.length() - 3)
, "Bal", "Amt", "Pct"))
{
cm.DTyp = FieldMDEnums.eDTyp.Dbl;
}
else if (cm.Name.length() > 3 &&
bIn(cm.Name.substring(cm.Name.length() - 3)
, "Mos"))
{
cm.DTyp = FieldMDEnums.eDTyp.Int;
}
else if (cm.Name.length() > 4 &&
bIn(cm.Name.substring(cm.Name.length() - 4)
, "Rate"))
{
cm.DTyp = FieldMDEnums.eDTyp.Dbl;
}
else if (cm.Name.length() > 4 &&
bIn(cm.Name.substring(cm.Name.length() - 4)
, "Name", "Code", "Stat", "Flag"))
{
cm.DTyp = FieldMDEnums.eDTyp.VLS;
}
else
{
cm.DTyp = FieldMDEnums.eDTyp.VLS;
}
}
if (cm.DTyp.bIn(FieldMDEnums.eDTyp.VLS, FieldMDEnums.eDTyp.Str))
{
cm.StringMaxLength = FieldMD.Default.StringMaxLength;
}
}
}
catch (Exception e) {
throw new com.WDataSci.WDS.WDSException("Error mapping input columns:", e);
}
}
//Java public void mReadMapFor(Document aDoc, RecordSetMD aRecordSetMD, JniPMMLItem aJniPMMLItem, PrintWriter pw, Boolean bFillDictionaryNames)
//C#
public void mReadMapFor(XmlDocument aDoc, RecordSetMD aRecordSetMD, JniPMMLItem aJniPMMLItem, PrintWriter pw, Boolean bFillDictionaryNames)
//throws com.WDataSci.WDS.WDSException
{
try {
int jj = -1;
int j = -1;
//are we using a JniPMML object (as when called from C# and does it have PMMLMatter
Boolean bUsingJniPMML = (aJniPMMLItem != null);
Boolean bCheckingAgainstPMML = (aJniPMMLItem != null && aJniPMMLItem.PMMLMatter.Doc != null);
String[] lFieldStringNames = null;
int nDataFieldNames = 0;
if (bCheckingAgainstPMML)
{
lFieldStringNames = aJniPMMLItem.PMMLDataFieldStringNames();
nDataFieldNames = lFieldStringNames.Length;
}
if (aRecordSetMD.SchemaMatter == null)
{
throw new com.WDataSci.WDS.WDSException("Error, aRecordSetMD.SchemaMatter not populated");
}
String rns = aRecordSetMD.SchemaMatter.RecordSetElementName;
String rn = aRecordSetMD.SchemaMatter.RecordElementName;
if (rns == null || rns.isEmpty())
{
rns = Util.RecordSetElementName(aRecordSetMD.SchemaMatter.InputSchema);
}
if (rn == null || rn.isEmpty())
{
rn = Util.RecordSingleName(rns);
}
if (rns.equals(rn))
{
throw new com.WDataSci.WDS.WDSException("Error, RecordSetMD needs a valid RecordSetElementName, it's singular version, and they cannot be equal");
}
if (aRecordSetMD.SchemaMatter.RecordSetElementName == null)
{
aRecordSetMD.SchemaMatter.RecordSetElementName = rns;
}
if (aRecordSetMD.SchemaMatter.RecordElementName == null)
{
aRecordSetMD.SchemaMatter.RecordElementName = rn;
}
String xPathQ = "/node()/child::*[local-name()='element' and @name='" + rns + "']" +
"//*[local-name()='element' and @name='" + rn + "']" +
"//*[local-name()='element']";
//Java XPath xPath = XPathFactory.newInstance().newXPath();
//Java NodeList xnl = (NodeList) xPath.evaluate(xPathQ, aDoc, XPathConstants.NODESET);
//C#
XmlNodeList xnl = aDoc.SelectNodes(xPathQ);
int xnlLength = xnl.getLength();
aRecordSetMD.Column = new FieldMD[xnlLength];
//for when the Length is packed into the XSD type for limiting strings
int[] typl = new int[1];
for (jj = 0; jj < xnlLength; jj++)
{
aRecordSetMD.Column[jj] = new FieldMD();
FieldMD cm = aRecordSetMD.Column[jj];
//Get the XML/XSD element
//Java Element xn = (Element) xnl.item(jj);
//C#
XmlNode xn = xnl[jj];
//Get the name and initialize the mapped names
//Java cm.Name = xn.getAttribute("name");
//C#
cm.Name = xn.Attributes.GetNamedItem("name").Value;
//Get the XML/XSD - type
//Java String typ = xn.getAttribute("type");
//C#
String typ = xn.Attributes.GetNamedItem("type").Value;
//Java cm.DTyp = FieldMDEnums.eDTyp.FromAlias(typ, typl);
//C#
cm.DTyp = FieldMDExt.eDTyp_FromAlias(typ, ref typl);
if (typl[0] > 0)
{
cm.StringMaxLength = typl[0];
}
else if (typl[0] < 0)
{
//C#
XmlNode wds_StringMaxLength = xn.Attributes.GetNamedItem("wds:StringMaxLength");
//Java String wds_StringMaxLength = xn.getAttribute("wds:StringMaxLength");
if (wds_StringMaxLength != null)
{
try {
//C#
cm.StringMaxLength = int.Parse(wds_StringMaxLength.Value);
//Java cm.StringMaxLength = Integer.parseInt(wds_StringMaxLength);
} catch (Exception e) {
cm.StringMaxLength = FieldMD.Default.StringMaxLength;
}
}
else
{
/* Java >>> *
* if (xn.hasAttributes()) {
* for ( int ani = 0; ani < xn.getAttributes().getLength(); ani++ ) {
* if ( xn.getAttributes().item(ani).getNodeName().toLowerCase().endsWith("maxlength") ) {
* cm.StringMaxLength = Integer.parseInt(xn.getAttributes().item(ani).getNodeValue());
* typl[0] = 0;
* break;
* }
* }
* }
* /* <<< Java */
/* C# >>> */
for (int ani = 0; ani < xn.Attributes.Count; ani++)
{
if (xn.Attributes.Item(ani).LocalName.toLowerCase().endsWith("maxlength"))
{
cm.StringMaxLength = int.Parse(xn.Attributes.Item(ani).Value);
typl[0] = 0;
break;
}
}
/* <<< C# */
if (typl[0] < 0)
{
xPathQ = "//*[local-name()='simpleType' and @name='" + typ + "']//*[local-name()='maxLength']";
//Java NodeList xnr = (NodeList) xPath.evaluate(xPathQ, aDoc, XPathConstants.NODESET);
//C#
XmlNodeList xnr = aDoc.SelectNodes(xPathQ);
if (xnr.getLength() == 1)
{
try {
//Java cm.StringMaxLength = Integer.parseInt(((Element) xnr.item(0)).getAttribute("value"));
//Cs
cm.StringMaxLength = int.Parse(xnr[0].Attributes.GetNamedItem("value").Value);
}
catch (Exception e) {
cm.StringMaxLength = FieldMD.Default.StringMaxLength;
}
}
else
{
cm.StringMaxLength = FieldMD.Default.StringMaxLength;
}
}
}
}
if (bCheckingAgainstPMML)
{
//Search for PMML DataFieldName map
for (j = 0; j < nDataFieldNames; j++)
{
if (cm.Name.equals(lFieldStringNames[j]))
{
cm.MapToMapKey(lFieldStringNames[j]);
break;
}
}
}
else if (bFillDictionaryNames)
{
cm.MapToMapKey(cm.Name);
}
}
}
catch (Exception e) {
throw new com.WDataSci.WDS.WDSException("Error mapping input columns:", e);
}
}