public static MarkovChain ReadFromStream(StreamReader reader)
{
MarkovChain chain = null;
while (!reader.EndOfStream)
{
var prefixline = reader.ReadLine();
var prefixes = prefixline.Split(',');
if (chain == null)
{
chain = new MarkovChain(prefixes.Length);
}
var prefix = new MarkovPrefix(prefixes.Select(x => x[0] == '"' ? x.Substring(1, x.Length - 2) : null));
List <string> Postfixes = new List <string>();
var nextLine = reader.ReadLine();
while (nextLine != "-")
{
Postfixes.Add(nextLine);
nextLine = reader.ReadLine();
}
chain.Segments.Add(prefix, MarkovChainSegment.GenerateSegment(Postfixes));
}
return(chain);
}
public MarkovChainSegment Clone()
{
MarkovChainSegment tmp = new MarkovChainSegment("");
tmp.Postfixes.Clear();
foreach (var Postfix in Postfixes)
{
tmp.Postfixes.Add(Postfix.Key, Postfix.Value);
}
return(tmp);
}
public static MarkovChainSegment GenerateSegment(List <string> postfixes)
{
var tuples = postfixes.Select(x => new Tuple <int, string>(int.Parse(x.Substring(0, x.IndexOf(' '))), x.Substring(x.IndexOf(' ') + 1))).ToArray();
MarkovChainSegment segment = new MarkovChainSegment(tuples[0].Item2);
segment.Postfixes.Clear();
foreach (var tup in tuples)
{
segment.Postfixes.Add(tup.Item2, tup.Item1);
}
return(segment);
}
public void Extend(MarkovChain otherChain)
{
foreach (var seg in otherChain.Segments)
{
if (Segments.ContainsKey(seg.Key))
{
MarkovChainSegment.Merge(Segments[seg.Key], seg.Value);
}
else
{
Segments.Add(seg.Key, seg.Value);
}
}
}
public static MarkovChainSegment Merge(MarkovChainSegment mergeInto, MarkovChainSegment data)
{
foreach (var seg in data.Postfixes)
{
if (mergeInto.Postfixes.ContainsKey(seg.Key))
{
mergeInto.Postfixes[seg.Key] += seg.Value;
}
else
{
mergeInto.Postfixes.Add(seg.Key, seg.Value);
}
}
return(mergeInto);
}
public static double ComparePercentage(MarkovChainSegment test, MarkovChainSegment source)
{
var perTest = 1.0 / test.Postfixes.Sum(x => x.Value);
var total = 0;
foreach (string postfix in test.Postfixes.Keys)
{
if (source.Postfixes.ContainsKey(postfix))
{
total += test.Postfixes[postfix];
}
}
return(perTest * total);
}
public static double PercentageMatch(MarkovChain test, MarkovChain source)
{
int totalSegments = test.Segments.Count;
double perSegment = 1.0 / totalSegments;
var total = 0.0;
foreach (var segment in test.Segments)
{
var prefix = segment.Key;
if (source.Segments.ContainsKey(prefix))
{
var sourceSegment = source.Segments[prefix];
total += perSegment * MarkovChainSegment.ComparePercentage(sourceSegment, source.Segments[prefix]);
}
}
return(total);
}
public IEnumerable <MarkovChain> BuildSubChains()
{
var skip = 1;
for (var length = this.MaxChainLength - 1; length >= 1; length--)
{
MarkovChain tempChain = new MarkovChain(length);
var grp = this.Segments.GroupBy((kv) => new MarkovPrefix(kv.Key.Prefix.Skip(skip)));
foreach (var SegmentGroup in grp)
{
var prefix = new MarkovPrefix(SegmentGroup.First().Key.Prefix.Skip(skip));
MarkovChainSegment newSegment = new MarkovChainSegment("");
newSegment.Postfixes.Clear();
foreach (var Segment in SegmentGroup)
{
newSegment = MarkovChainSegment.Merge(newSegment, Segment.Value);
}
tempChain.Segments.Add(prefix, newSegment);
}
yield return(tempChain);
skip++;
}
}
