public static StringDistribution EmptyOrStartsWith(ImmutableDiscreteChar charsInMainString, ImmutableDiscreteChar startsWith)
{
// TODO: fix equality and then use factory methods to create this
var result = new StringAutomaton.Builder();
result.Start.SetEndWeight(Weight.One);
var otherState = result.Start.AddTransition(startsWith, Weight.FromLogValue(-startsWith.GetLogAverageOf(startsWith)));
otherState.AddSelfTransition(charsInMainString, Weight.FromLogValue(-charsInMainString.GetLogAverageOf(charsInMainString)));
otherState.SetEndWeight(Weight.One);
return(StringDistribution.FromWeightFunction(result.GetAutomaton()));
}
/// <summary>
/// Creates a uniform distribution over any string starting and ending with a non-word character.
/// Characters other than the first and the last are restricted to be non-zero probability characters
/// from a given distribution.
/// </summary>
/// <param name="allowedChars">The distribution representing allowed characters.</param>
/// <param name="nonWordCharacter">The word separating characters.</param>
/// <returns>The created distribution.</returns>
public static StringDistribution WordMiddle(ImmutableDiscreteChar allowedChars, ImmutableDiscreteChar?nonWordCharacter = null)
{
// TODO: fix equality and then use factory methods to create this
nonWordCharacter = nonWordCharacter ?? NonWordCharacter;
var result = new StringAutomaton.Builder();
var otherState1 = result.Start.AddTransition(
Option.FromNullable(nonWordCharacter),
Weight.FromLogValue(-nonWordCharacter.Value.GetLogAverageOf(nonWordCharacter.Value)));
otherState1.SetEndWeight(Weight.One);
var otherState2 = otherState1.AddEpsilonTransition(Weight.One)
.AddSelfTransition(allowedChars, Weight.FromLogValue(-allowedChars.GetLogAverageOf(allowedChars))).AddTransition(
Option.FromNullable(nonWordCharacter),
Weight.FromLogValue(-nonWordCharacter.Value.GetLogAverageOf(nonWordCharacter.Value)));
otherState2.SetEndWeight(Weight.One);
return(StringDistribution.FromWeightFunction(result.GetAutomaton()));
}