private DString wrap(string s)
{
// Presence is a tallymark-like system for counting character occurrences in a string, regardless of position
ulong presence = 0L;
Dictionary <char, CCount> counts = CCount.Decompose(s);
foreach (CCount cbits in bits)
{
ushort marked = 0;
CCount c;
if (counts.TryGetValue(cbits.Chr, out c))
{
marked = c.Count;
}
ushort upTo = cbits.Count;
for (ushort i = 0; i < upTo; i++)
{
presence = presence << 1;
if (marked > 0)
{
presence |= 1;
marked--;
}
}
}
return(new DString(s, presence));
}
public FuzzyMatcher Build()
{
Dictionary <char, Queue <uint> > costs = computeCosts();
// Setup a priority queue in largest-first order, with the first cost value for each character
var q = new SortedSet <WeightedChar>();
foreach (KeyValuePair <char, Queue <uint> > e in costs)
{
q.Add(new WeightedChar(e.Key, e.Value.Dequeue()));
}
var bitCounts = new Dictionary <char, ushort>();
int iterations = 64; // bits in a Long
while (iterations > 0 && q.Count != 0)
{
WeightedChar wc = q.Max;
q.Remove(wc);
// Increment bit count for the character
ushort bitsCurrent;
if (!bitCounts.TryGetValue(wc.Chr, out bitsCurrent))
{
bitsCurrent = 0;
}
bitsCurrent++;
bitCounts[wc.Chr] = bitsCurrent;
Queue <uint> vals = costs[wc.Chr];
if (vals != null && vals.Count != 0)
{
// Costs queue still has values to go through, put it back into the priority queue
q.Add(new WeightedChar(wc.Chr, vals.Dequeue()));
}
iterations--;
}
// Flatten the bitCounts map into an array of (character, bits)
CCount[] bits = new CCount[bitCounts.Count];
int i = 0;
foreach (KeyValuePair <char, ushort> e in bitCounts)
{
bits[i] = new CCount(e.Key, e.Value);
i++;
}
Array.Sort(bits);
return(new FuzzyMatcher(bits));
}
public void Update(string str)
{
if (str.Length == 0)
{
return;
}
foreach (CCount cc in CCount.Decompose(str).Values)
{
List <uint> vals;
if (!counts.TryGetValue(cc.Chr, out vals))
{
vals = new List <uint>();
counts.Add(cc.Chr, vals);
}
while (vals.Count < cc.Count)
{
// Insert zeros until the array of counts can support the value we want to insert
vals.Add(0);
}
// We can't have a count of 0, so count positions need to be offset by 1
vals[cc.Count - 1]++;
}
total++;
}
