/// <summary>
/// constructor for a faster init, that also enables to set the values of the dataset
/// </summary>
/// <param name="name"></param>
/// <param name="set"></param>
/// <param name="type"></param>
public Story(string name, string[] set, EArffTypes type)
{
this.isNumeric = false;
this.name = name;
this.type = type;
if (type == EArffTypes.ATTRIBUTE)
{
this.values = set;
}
else if (type == EArffTypes.DATA)
{
this.data = set;
}
else
{
throw new CekaException("can not set any values or dataset(elements) for the type " + type.ToString());
}
}
/// <summary>
/// constructor for a faster init, that also enables to set the values of the dataset
/// </summary>
/// <param name="name"></param>
/// <param name="set"></param>
/// <param name="type"></param>
public Story(string name, string[] set, EArffTypes type)
{
this.isNumeric = false;
this.name = name;
this.type = type;
if (type == EArffTypes.ATTRIBUTE)
this.values = set;
else if (type == EArffTypes.DATA)
this.data = set;
else
throw new CekaException("can not set any values or dataset(elements) for the type " + type.ToString());
}
/// <summary>
/// this function takes care of the definition of the ARFF File line (that is passed as parameter) and sets the value(s) and type of the Story
/// </summary>
/// <param name="al"></param>
private void parseType(string al)
{
//making sure we sort out the empty or new lines first
if (string.IsNullOrEmpty(al) || string.IsNullOrWhiteSpace(al) || al.Length <= 2)
{
this.value = "";
this.type = EArffTypes.NONE;
return;
}
//next up is a fast check if the story might be a higher type
if (al[0].Equals(ArffFile.ATT_PRE))
{
//find out what kind of type it is
string[] _data = al.Split(ArffFile.ARFF_SPACE);
if (_data.Length <= 0)
throw new CekaException("this story (" + al + ") seems to be an attribute"
+ " but there is something wrong with the whitespaces in your file.");
//switch on behalf of the attribute's type
//(just make sure everything is put in place here, the rest is done by the library later on)
switch(_data[0])
{
case ArffFile.ARFF_RELATION:
this.type = EArffTypes.RELATION;
//a relation just ends on the 2n index
this.name = _data[1];
this.value = "relation-type";
break;
case ArffFile.ARFF_ATTRIBUTE:
this.type = EArffTypes.ATTRIBUTE;
al = al.Replace(_data[0], "");
if (testForNumericContent(al))
{
//value indicates that this attribute is a numeric or REAL type, parse it!
parseNumeric(al);
}
else
{
//value and name have to be splitted
filterNameAndValue(al);
//to make sure we even drill down into the single elements of attributes values
furtherSplitOfValue();
}
break;
case ArffFile.ARFF_DATA:
this.type = EArffTypes.DATA_ATT; //beware of the difference between DATA_ATT and DATA!
//nothing to do here, its just the end of the header section == @data
this.value = "data-type";
this.name = "none";
break;
default:
throw new CekaException("this story (" + al + ") seemed to be alright, but the attribute "
+ "type (" + _data[0] + ") is unknown.");
}
}
else
{
//it is no higher type, its a clean data-line
if (al.Contains(ArffFile.ARFF_SPACE))
throw new CekaException("this story (data) (" + al + ") is including whitespaces.");
//simply add the data of the line to this story
this.data = al.Split(ArffFile.STORY_DELIMITTER);
if (this.data.Length <= 0)
throw new CekaException("this story (" + al + ") seems to be a "
+ "dataset, although its not using any '" + ArffFile.STORY_DELIMITTER + "'.");
StringBuilder sb = null;
//an remove any "/'" s (should be clean of whitespaces already)
for (int i = 0; i < this.data.Length; i++)
{
sb = new StringBuilder();
foreach (char c in data[i])
{
if (!c.Equals(ArffFile.ATT_NAME_COMB))
sb.Append(c);
}
data[i] = sb.ToString();
}
this.type = EArffTypes.DATA;
this.value = "data-value";
}
}
/// <summary>
/// this function takes care of the definition of the ARFF File line (that is passed as parameter) and sets the value(s) and type of the Story
/// </summary>
/// <param name="al"></param>
private void parseType(string al)
{
//making sure we sort out the empty or new lines first
if (string.IsNullOrEmpty(al) || string.IsNullOrWhiteSpace(al) || al.Length <= 2)
{
this.value = "";
this.type = EArffTypes.NONE;
return;
}
//next up is a fast check if the story might be a higher type
if (al[0].Equals(ArffFile.ATT_PRE))
{
//find out what kind of type it is
string[] _data = al.Split(ArffFile.ARFF_SPACE);
if (_data.Length <= 0)
{
throw new CekaException("this story (" + al + ") seems to be an attribute"
+ " but there is something wrong with the whitespaces in your file.");
}
//switch on behalf of the attribute's type
//(just make sure everything is put in place here, the rest is done by the library later on)
switch (_data[0])
{
case ArffFile.ARFF_RELATION:
this.type = EArffTypes.RELATION;
//a relation just ends on the 2n index
this.name = _data[1];
this.value = "relation-type";
break;
case ArffFile.ARFF_ATTRIBUTE:
this.type = EArffTypes.ATTRIBUTE;
al = al.Replace(_data[0], "");
if (testForNumericContent(al))
{
//value indicates that this attribute is a numeric or REAL type, parse it!
parseNumeric(al);
}
else
{
//value and name have to be splitted
filterNameAndValue(al);
//to make sure we even drill down into the single elements of attributes values
furtherSplitOfValue();
}
break;
case ArffFile.ARFF_DATA:
this.type = EArffTypes.DATA_ATT; //beware of the difference between DATA_ATT and DATA!
//nothing to do here, its just the end of the header section == @data
this.value = "data-type";
this.name = "none";
break;
default:
throw new CekaException("this story (" + al + ") seemed to be alright, but the attribute "
+ "type (" + _data[0] + ") is unknown.");
}
}
else
{
//it is no higher type, its a clean data-line
if (al.Contains(ArffFile.ARFF_SPACE))
{
throw new CekaException("this story (data) (" + al + ") is including whitespaces.");
}
//simply add the data of the line to this story
this.data = al.Split(ArffFile.STORY_DELIMITTER);
if (this.data.Length <= 0)
{
throw new CekaException("this story (" + al + ") seems to be a "
+ "dataset, although its not using any '" + ArffFile.STORY_DELIMITTER + "'.");
}
StringBuilder sb = null;
//an remove any "/'" s (should be clean of whitespaces already)
for (int i = 0; i < this.data.Length; i++)
{
sb = new StringBuilder();
foreach (char c in data[i])
{
if (!c.Equals(ArffFile.ATT_NAME_COMB))
{
sb.Append(c);
}
}
data[i] = sb.ToString();
}
this.type = EArffTypes.DATA;
this.value = "data-value";
}
}
