/// <summary>
/// Creates a new instance of <see cref="NodeInfo"/>.
/// </summary>
/// <param name="element">The XML element.</param>
/// <param name="parentSignature">The signature of the parent XML node.</param>
/// <param name="predicate">If true, the <see cref="XElement"/> will not be loaded.</param>
/// <param name="schemaRegistry">Stores the schema of the XML node.</param>
private NodeInfo(XElement element, string parentSignature, Func <XElement, bool> predicate, SchemaRegistry schemaRegistry)
{
var name = element.Name.LocalName;
var signature = parentSignature != null ? $"{parentSignature}.{name}" : name;
var children = element.Elements().Where(predicate).Select(childElement => new NodeInfo(childElement, signature, predicate, schemaRegistry));
var properties = new Dictionary <string, Bigram>();
var text = (element.FirstNode as XText)?.Value;
foreach (var attribute in element.Attributes())
{
properties[attribute.Name.LocalName] = new Bigram(attribute.Value);
}
if (text != null)
{
properties["&text"] = new Bigram(text);
}
schemaRegistry.AddPropertyNames(signature, properties.Keys);
Children = children;
Element = element;
Properties = properties;
Signature = signature;
}
/// <summary>
/// Formats the specified properties to match the current schema.
/// </summary>
/// <param name="properties">The properties to be formatted.</param>
/// <param name="id">The ID of the current schema.</param>
/// <returns>The formatted properties.</returns>
internal ImmutableArray <Bigram> Format(IReadOnlyDictionary <string, Bigram> properties, out int id)
{
id = Id;
var count = 0;
var values = new Bigram[PropertyNames.Count];
foreach (var propertyName in PropertyNames)
{
values[count++] = properties.TryGetValue(propertyName, out var value) ? value : Bigram.Empty;
}
return(values.ToImmutableArray());
}
private static float Compare(Bigram a, Bigram b)
{
if (a.Tokens.IsEmpty && b.Tokens.IsEmpty)
{
return(1.0f);
}
if (a.Tokens.IsEmpty || b.Tokens.IsEmpty)
{
return(0.0f);
}
var enumeratorA = a.Tokens.GetEnumerator();
var enumeratorB = b.Tokens.GetEnumerator();
var moveNextA = enumeratorA.MoveNext();
var moveNextB = enumeratorB.MoveNext();
var count = 0;
while (moveNextA && moveNextB)
{
var comparison = enumeratorA.Current.CompareTo(enumeratorB.Current);
if (comparison < 0)
{
moveNextA = enumeratorA.MoveNext();
}
else if (comparison > 0)
{
moveNextB = enumeratorB.MoveNext();
}
else
{
count++;
moveNextA = enumeratorA.MoveNext();
moveNextB = enumeratorB.MoveNext();
}
}
return((float)count / Math.Max(a.Tokens.Length, b.Tokens.Length));
}
/// <summary>
/// Compares the current bigram to the specified bigram using the Sørensen–Dice coefficient.
/// </summary>
/// <param name="other">The other bigram.</param>
/// <returns>A value between 0 and 1, where 1 is a perfect match.</returns>
internal float CompareTo(Bigram other)
{
return(Compare(this, other));
}