public static Example exampleFromString(string data, DataSetSpecification dataSetSpec, string separator)
{
IRegularExpression splitter = TextFactory.CreateRegularExpression(separator);
IMap <string, IAttribute> attributes = CollectionFactory.CreateInsertionOrderedMap <string, IAttribute>();
ICollection <string> attributeValues = CollectionFactory.CreateQueue <string>(splitter.Split(data));
if (dataSetSpec.isValid(attributeValues))
{
ICollection <string> names = dataSetSpec.getAttributeNames();
int min = names.Size() > attributes.Size() ? names.Size() : attributes.Size();
for (int i = 0; i < min; ++i)
{
string name = names.Get(i);
IAttributeSpecification attributeSpec = dataSetSpec.getAttributeSpecFor(name);
IAttribute attribute = attributeSpec.CreateAttribute(attributeValues.Get(i));
attributes.Put(name, attribute);
}
string targetAttributeName = dataSetSpec.getTarget();
return(new Example(attributes, attributes.Get(targetAttributeName)));
}
else
{
throw new RuntimeException("Unable to construct Example from " + data);
}
}
// string split constructor
public Rule(string lhs, string rhs, float probability)
{
this.lhs = CollectionFactory.CreateQueue <string>();
this.rhs = CollectionFactory.CreateQueue <string>();
IRegularExpression regex = TextFactory.CreateRegularExpression("\\s*,\\s*");
if (!string.IsNullOrEmpty(lhs))
{
this.lhs = CollectionFactory.CreateQueue <string>();
foreach (string input in regex.Split(lhs))
{
if (!string.IsNullOrEmpty(input))
{
this.lhs.Add(input);
}
}
}
if (!string.IsNullOrEmpty(rhs))
{
foreach (string input in regex.Split(rhs))
{
if (!string.IsNullOrEmpty(input))
{
this.rhs.Add(input);
}
}
}
this.PROB = validateProb(probability);
}
// TODO
// Make more intelligent link search
public ICollection <string> getOutlinks(Page page)
{
string content = page.getContent();
ICollection <string> outLinks = CollectionFactory.CreateQueue <string>();
// search content for all href="x" outlinks
ICollection <string> allMatches = CollectionFactory.CreateQueue <string>();
IRegularExpression m = TextFactory.CreateRegularExpression("href=\"(/wiki/.*?)\"");
foreach (string ma in m.Matches(content))
{
allMatches.Add(ma);
}
for (int i = 0; i < allMatches.Size(); ++i)
{
string match = allMatches.Get(i);
string[] tokens = TextFactory.CreateRegularExpression("\"").Split(match);
string location = tokens[1].ToLower(); // also, tokens[0] = the
// text before the first
// quote,
// and tokens[2] is the
// text after the second
// quote
outLinks.Add(location);
}
return(outLinks);
}
private ICollection <double> exampleFromString(string line, string separator)
{
// assumes all values for inout and target are doubles
ICollection <double> rexample = CollectionFactory.CreateQueue <double>();
IRegularExpression regex = TextFactory.CreateRegularExpression(separator);
ICollection <string> attributeValues = CollectionFactory.CreateQueue <string>(regex.Split(line));
foreach (string valString in attributeValues)
{
rexample.Add(double.Parse(valString,
System.Globalization.NumberStyles.Any,
System.Globalization.CultureInfo.InvariantCulture));
}
return(rexample);
}
