public static void Build(DictionaryFormat format,
string inputDirname,
string outputDirname,
string encoding,
bool normalizeEntry)
{
Console.WriteLine("building tokeninfo dict...");
TokenInfoDictionaryBuilder tokenInfoBuilder = new TokenInfoDictionaryBuilder(format, encoding, normalizeEntry);
TokenInfoDictionaryWriter tokenInfoDictionary = tokenInfoBuilder.Build(inputDirname);
tokenInfoDictionary.Write(outputDirname);
//tokenInfoDictionary = null; // LUCENENET: IDE0059: Remove unnecessary value assignment
//tokenInfoBuilder = null; // LUCENENET: IDE0059: Remove unnecessary value assignment
Console.WriteLine("done");
Console.WriteLine("building unknown word dict...");
UnknownDictionaryBuilder unkBuilder = new UnknownDictionaryBuilder(encoding);
UnknownDictionaryWriter unkDictionary = unkBuilder.Build(inputDirname);
unkDictionary.Write(outputDirname);
//unkDictionary = null; // LUCENENET: IDE0059: Remove unnecessary value assignment
//unkBuilder = null; // LUCENENET: IDE0059: Remove unnecessary value assignment
Console.WriteLine("done");
Console.WriteLine("building connection costs...");
ConnectionCostsWriter connectionCosts
= ConnectionCostsBuilder.Build(inputDirname + System.IO.Path.DirectorySeparatorChar + "matrix.def");
connectionCosts.Write(outputDirname);
Console.WriteLine("done");
}
public virtual TokenInfoDictionaryWriter BuildDictionary(IList <string> csvFiles)
{
TokenInfoDictionaryWriter dictionary = new TokenInfoDictionaryWriter(10 * 1024 * 1024);
// all lines in the file
Console.WriteLine(" parse...");
List <string[]> lines = new List <string[]>(400000);
foreach (string file in csvFiles)
{
using (Stream inputStream = new FileStream(file, FileMode.Open, FileAccess.Read))
{
Encoding decoder = Encoding.GetEncoding(encoding);
TextReader reader = new StreamReader(inputStream, decoder);
string line = null;
while ((line = reader.ReadLine()) != null)
{
string[] entry = CSVUtil.Parse(line);
if (entry.Length < 13)
{
Console.WriteLine("Entry in CSV is not valid: " + line);
continue;
}
string[] formatted = FormatEntry(entry);
lines.Add(formatted);
// NFKC normalize dictionary entry
if (normalizeEntries)
{
//if (normalizer.isNormalized(entry[0])){
if (entry[0].IsNormalized(NormalizationForm.FormKC))
{
continue;
}
string[] normalizedEntry = new string[entry.Length];
for (int i = 0; i < entry.Length; i++)
{
//normalizedEntry[i] = normalizer.normalize(entry[i]);
normalizedEntry[i] = entry[i].Normalize(NormalizationForm.FormKC);
}
formatted = FormatEntry(normalizedEntry);
lines.Add(formatted);
}
}
}
}
Console.WriteLine(" sort...");
// sort by term: we sorted the files already and use a stable sort.
lines.Sort(new ComparerAnonymousHelper());
Console.WriteLine(" encode...");
PositiveInt32Outputs fstOutput = PositiveInt32Outputs.Singleton;
Builder <long?> fstBuilder = new Builder <long?>(Lucene.Net.Util.Fst.FST.INPUT_TYPE.BYTE2, 0, 0, true, true, int.MaxValue, fstOutput, null, true, PackedInt32s.DEFAULT, true, 15);
Int32sRef scratch = new Int32sRef();
long ord = -1; // first ord will be 0
string lastValue = null;
// build tokeninfo dictionary
foreach (string[] entry in lines)
{
int next = dictionary.Put(entry);
if (next == offset)
{
Console.WriteLine("Failed to process line: " + Collections.ToString(entry));
continue;
}
string token = entry[0];
if (!token.Equals(lastValue, StringComparison.Ordinal))
{
// new word to add to fst
ord++;
lastValue = token;
scratch.Grow(token.Length);
scratch.Length = token.Length;
for (int i = 0; i < token.Length; i++)
{
scratch.Int32s[i] = (int)token[i];
}
fstBuilder.Add(scratch, ord);
}
dictionary.AddMapping((int)ord, offset);
offset = next;
}
FST <long?> fst = fstBuilder.Finish();
Console.WriteLine(" " + fst.NodeCount + " nodes, " + fst.ArcCount + " arcs, " + fst.GetSizeInBytes() + " bytes... ");
dictionary.SetFST(fst);
Console.WriteLine(" done");
return(dictionary);
}
