public static Dawg CreateFromCorpus(string corpusLocation)
{
var builder = new PartitionedGraphBuilder();
using var stream = File.OpenRead(corpusLocation);
using var reader = new StreamReader(stream);
string?line;
while ((line = reader.ReadLine()) != null)
{
var lineTokens = line.Split(' ');
if (lineTokens.Length != 2)
{
continue;
}
if (!ulong.TryParse(lineTokens[1], out var count))
{
continue;
}
builder.Insert(lineTokens[0], count);
}
using var compressedGraph = builder.AsCompressedSparseRows();
compressedGraph.Save(TempAugPath);
using var dawgStream = File.OpenRead(TempAugPath);
return(new Dawg(dawgStream));
}
private static CompressedSparseRowGraph BuildGraph(string corpusPath)
{
var builder = new PartitionedGraphBuilder();
using (var stream = File.OpenRead(corpusPath))
{
if (stream == null)
{
throw new InvalidOperationException();
}
using var reader = new StreamReader(stream);
string?line;
while ((line = reader.ReadLine()) != null)
{
var lineTokens = line.Split(' ');
if (lineTokens == null || lineTokens.Length != 2)
{
continue;
}
if (!ulong.TryParse(lineTokens[1], out var count))
{
continue;
}
builder.Insert(lineTokens[0], count);
}
}
var result = builder.AsCompressedSparseRows();
return(result);
}
public static Dawg Create(params string[] words)
{
var builder = new PartitionedGraphBuilder();
foreach (var word in words.OrderBy(x => x))
{
builder.Insert(word, 0);
}
using var compressed = builder.AsCompressedSparseRows();
compressed.Save(TempAugPath);
using var read = File.OpenRead(TempAugPath);
return(new Dawg(read));
}
// ReSharper disable once UnusedMember.Global
public static Dawg CreateDictionary(string corpusPath, string savePath)
{
var builder = new PartitionedGraphBuilder();
using (var stream = File.OpenRead(corpusPath))
{
if (stream == null)
{
throw new InvalidOperationException();
}
using var reader = new StreamReader(stream);
string?line;
while ((line = reader.ReadLine()) != null)
{
var lineTokens = line.Split(' ');
if (lineTokens == null || lineTokens.Length != 2)
{
continue;
}
if (!ulong.TryParse(lineTokens[1], out var count))
{
continue;
}
builder.Insert(lineTokens[0], count);
}
}
using (var compressedGraph = builder.AsCompressedSparseRows())
{
compressedGraph.Save(savePath);
}
using var dawgStream = File.OpenRead(savePath);
return(new Dawg(dawgStream));
}
