public void SimpleConstructionWithStringOutputs()
{
var inputs = new string[]
{
"a banana",
"a lemon",
"a mandarine",
"a mango",
"an apple",
"an orange",
};
var outputs = new string[]
{
"one",
"two",
"three",
"minusone",
"minustwo",
"minuseight"
};
var fst = new FSTBuilder <string>(FSTStringOutput.Instance).FromList(inputs, outputs);
Verify(fst, inputs, outputs);
var fst1 = FST <string> .FromBytes(fst.GetBytes(), FSTStringOutput.Instance);
Verify(fst1, inputs, outputs);
var fst2 = FST <string> .FromBytesCompressed(fst.GetBytesCompressed(), FSTStringOutput.Instance);
Verify(fst2, inputs, outputs);
}
public void SimpleConstructionWithIntOutputs()
{
var inputs = new string[]
{
"a banana",
"a lemon",
"a mandarine",
"a mango",
"an apple",
"an orange",
};
var outputs = new int[]
{
1,
2,
3,
-2,
15,
8
};
var fst = new FSTBuilder <int>(FSTVarIntOutput.Instance).FromList(inputs, outputs);
Verify(fst, inputs, outputs);
var fst1 = FST <int> .FromBytes(fst.GetBytes(), FSTVarIntOutput.Instance);
Verify(fst1, inputs, outputs);
var fst2 = FST <int> .FromBytesCompressed(fst.GetBytesCompressed(), FSTVarIntOutput.Instance);
Verify(fst2, inputs, outputs);
}
public void ConstructionWithIntOutputs()
{
var inputs = new string[]
{
"a",
"ab",
"abilities",
"ability",
};
var outputs = new int[]
{
4,
3134,
7488,
1580,
};
var fst = new FSTBuilder <int>(FSTVarIntOutput.Instance).FromList(inputs, outputs);
Verify(fst, inputs, outputs);
var fst1 = FST <int> .FromBytes(fst.GetBytes(), FSTVarIntOutput.Instance);
Verify(fst1, inputs, outputs);
var fst2 = FST <int> .FromBytesCompressed(fst.GetBytesCompressed(), FSTVarIntOutput.Instance);
Verify(fst2, inputs, outputs);
}
public void WildcardMatchingTest()
{
var inputs = new string[] {
"Albert Schweitzer Ziekenhuis. Locatie Amstelwijck Heliport",
"Amsterdam Airfield",
"Amsterdam Airport",
"Amsterdam Airport Schiphol",
"Amsterdam Heliport",
"Chafei Amsei Airport",
"New Amsterdam Airport",
"Schwarzheide/Schipkau Airport"
};
var outputs = new int[] {
43711,
23465,
41198,
2513,
43207,
5873,
41521,
29065
};
var fst = new FSTBuilder <int>(FSTVarIntOutput.Instance).FromList(inputs, outputs);
Verify(fst, inputs, outputs);
var expectedTerms = new string[]
{
"Amsterdam Airport Schiphol",
"Schwarzheide/Schipkau Airport"
};
var expectedOutputs = new int[]
{
2513,
29065
};
var expectedIndex = 0;
Assert.Equal(expectedOutputs.Length, expectedTerms.Length);
foreach (var term in fst.Match(new WildcardMatcher("*Schip*", 255)))
{
Assert.Equal(expectedTerms[expectedIndex], term);
Assert.True(fst.TryMatch(term, out int value));
Assert.Equal(expectedOutputs[expectedIndex], value);
++expectedIndex;
}
Assert.Equal(expectedOutputs.Length, expectedIndex);
}
public void Dispose()
{
if (input != null)
{
var fst = new FSTBuilder <int>(FSTVarIntOutput.Instance).FromList(input.ToArray(), output.ToArray());
Validate(fst); //TODO: Optional
{
var fstData = fst.GetBytesCompressed();
storage.WriteAll(0, fstData, 0, fstData.Length);
}
input = null;
output = null;
}
}
public void AcceptanceTestWithIntegerOutput()
{
var inputs = new string[] {
"Albert Schweitzer Ziekenhuis. Locatie Amstelwijck Heliport",
"Amsterdam Airfield",
"Amsterdam Airport",
"Amsterdam Airport Schiphol",
"Amsterdam Heliport",
"Chafei Amsei Airport",
"New Amsterdam Airport",
"Schwarzheide/Schipkau Airport"
};
var outputs = new int[] {
43711,
23465,
41198,
2513,
43207,
5873,
41521,
29065
};
var fst = new FSTBuilder <int>(FSTVarIntOutput.Instance).FromList(inputs, outputs);
Verify(fst, inputs, outputs);
var fst1 = FST <int> .FromBytes(fst.GetBytes(), FSTVarIntOutput.Instance);
Verify(fst1, inputs, outputs);
var fst2 = FST <int> .FromBytesCompressed(fst.GetBytesCompressed(), FSTVarIntOutput.Instance);
Verify(fst2, inputs, outputs);
}
private static int DoBuild(BuildOptions opts)
{
var timer = Stopwatch.StartNew();
if (File.Exists(opts.OutputFile))
{
File.Delete(opts.OutputFile);
}
timer.Restart();
var terms = 0;
using (var outputFile = new FileStorage(opts.OutputFile))
{
using (var fstBuilder = new FSTBuilder <int>(outputType, opts.CacheSize, outputFile))
{
fstBuilder.Begin();
foreach (var(term, score) in ParseFromOptions(opts))
{
fstBuilder.Add(term, score);
++terms;
}
fstBuilder.End();
PrintConsole(ConsoleColor.White, $"FST constructed time: {timer.Elapsed}, terms: {terms}, cache size: {opts.CacheSize}, Memory: {Process.GetCurrentProcess().WorkingSet64}, output size: {outputFile.Length}");
}
}
using (var outputFile = new FileStorage(opts.OutputFile))
{
if (outputFile.Length < 64 * 1024 * 1024)
{
timer.Restart();
var data = new byte[outputFile.Length];
outputFile.ReadAll(0, data, 0, data.Length);
var fst = FST <int> .FromBytesCompressed(data, outputType);
foreach (var(term, score) in ParseFromOptions(opts))
{
if (!fst.TryMatch(term, out var value) || value != score)
{
throw new Exception($"Bug at term {term}: {value} != {score}");
}
}
PrintConsole(ConsoleColor.White, $"FST (memory) verification time: {timer.Elapsed}");
}
}
timer.Restart();
using (var outputFile = new FileStorage(opts.OutputFile))
{
using (var fst = new PersistentFST <int>(outputType, outputFile))
{
foreach (var(term, score) in ParseFromOptions(opts))
{
if (!fst.TryMatch(term, out var value) || value != score)
{
throw new Exception($"Bug at term {term}: {value} != {score}");
}
}
}
}
PrintConsole(ConsoleColor.White, $"FST (file) verification time: {timer.Elapsed}");
return(0);
}
private static int DoBuild(BuildOptions opts)
{
var timer = Stopwatch.StartNew();
var input = File.ReadAllLines(opts.InputFile).OrderBy(x => x.Split("->")[0], StringComparer.Ordinal).ToArray();
var terms = new string[input.Length];
var outputs = new int[input.Length];
for (int i = 0; i < input.Length; ++i)
{
var s = input[i].Split("->");
terms[i] = s[0];
outputs[i] = int.Parse(s[1]);
// Console.WriteLine($"{terms[i]}->{outputs[i]}");
}
PrintConsole(ConsoleColor.White, $"Input read term: {terms.Length}, time: {timer.Elapsed}");
timer.Restart();
var fst = new FSTBuilder <int>(outputType).FromList(terms, outputs);
PrintConsole(ConsoleColor.White, $"FST constructed time: {timer.Elapsed}");
timer.Restart();
for (int i = 0; i < terms.Length; ++i)
{
if (!fst.TryMatch(terms[i], out var value) || value != outputs[i])
{
throw new Exception($"Bug at term {terms[i]}: {value} != {outputs[i]}");
}
}
PrintConsole(ConsoleColor.White, $"FST verification time: {timer.Elapsed}");
var size = 0;
timer.Restart();
if (opts.Format == "Default")
{
var fstBytes = fst.GetBytes();
var data = new byte[6 + 1 + fstBytes.Length];
data[0] = (byte)'F';
data[1] = (byte)'S';
data[2] = (byte)'T';
data[3] = (byte)'-';
data[4] = (byte)'0';
data[5] = (byte)'1';
data[6] = (byte)'D';
Array.Copy(fstBytes, 0, data, 7, fstBytes.Length);
File.WriteAllBytes(opts.OutputFile, data);
size = data.Length;
}
else if (opts.Format == "Compressed")
{
var fstBytes = fst.GetBytesCompressed();
var data = new byte[6 + 1 + fstBytes.Length];
data[0] = (byte)'F';
data[1] = (byte)'S';
data[2] = (byte)'T';
data[3] = (byte)'-';
data[4] = (byte)'0';
data[5] = (byte)'1';
data[6] = (byte)'C';
Array.Copy(fstBytes, 0, data, 7, fstBytes.Length);
File.WriteAllBytes(opts.OutputFile, data);
size = data.Length;
}
else if (opts.Format == "Dot")
{
throw new NotImplementedException();
}
PrintConsole(ConsoleColor.White, $"FST written to the output file: {opts.OutputFile}, size: {size}, time: {timer.Elapsed}");
return(0);
}
