/// <summary>
/// fully set arff instance, used by ArffLoader
/// </summary>
/// <param name="headers"></param>
/// <param name="datasets"></param>
/// <param name="skipIntCheck"></param>
public ArffInstance(Library headers, Library datasets, bool skipIntCheck = false)
: this()
{
if (headers == null || datasets == null)
throw new CekaException("can not pass null for library values of an ArffInstance");
this.Headers = headers;
this.Datasets = datasets;
Headers.CleanUp();
Datasets.CleanUp();
this.Relation = Headers.getRelationName(); //grab name out of it
Headers.removeRelation(); //and delete it afterwards, we dont need it, as we want clean Libraries
//running the int-check
if (!skipIntCheck)
this.integrityCheck();
}
/// <summary>
/// returns a deep copy of this library
/// </summary>
/// <returns></returns>
public Library toCopy()
{
Library nl = new Library();
Story[] cc = new Story[this.content.Count];
this.content.CopyTo(cc);
nl.content = new List<Story>(cc);
return nl;
}
/// <summary>
/// checking if the size of a single story-dataset differs from the size of the defined headers
/// </summary>
/// <param name="header"></param>
/// <returns></returns>
public bool integretyCheck(Library header)
{
int hs = header.headerSize();
foreach (Story s in this.content)
{
if (s.getDatasetSize() != hs)
{
Helpers.Utils.Debug("size (" + s.getDatasetSize() + ") of dataset l." + s.arffFileLine + " differs from header size (" + hs + ").");
return false;
}
}
return true;
}
/// <summary>
/// checking if the values of each single story-dataset element are allowed as they have to be defined in the headers
/// </summary>
/// <param name="header"></param>
/// <returns></returns>
public bool deepIntegretyCheck(Library header)
{
int sti = 0;
//run through all stories
foreach (Story st in this.content)
{
//run through every single story-dataset element
for (int i = 0; i < st.Data.Length; i++)
{
//check if the appropriat header attribute definition contains the value of the dataset:
if (header.content[i].isNumeric)
continue; //skip this entry because it belongs to an numeric attribute type
if(!header.content[i].Values.Contains(st.Data[i]))
{
if (string.CompareOrdinal(st.Data[i], ArffFile.ATT_EMPTY_VALUE.ToString()) == 0)
{
Helpers.Utils.Debug(String.Format("ignored row {0} column {1} as the value was empty={2}.", sti, i, ArffFile.ATT_EMPTY_VALUE));
continue;
}
Helpers.Utils.Debug("dataset(" + i + ") element (" + st.Data[i] + ") does not fit to the defined range of attribute (" + header.content[i].Name + ").");
return false;
}
}
sti++;
}
return true;
}