public static Dawg <FormInterpretations> CreateDAWG(MRDFileReader mrdFile)
{
DateTime start = DateTime.Now;
Console.WriteLine("Inserting forms in DAWG... Please wait...");
DawgSharp.DawgBuilder <FormInterpretations> dawgBuilder = new DawgBuilder <FormInterpretations>();
UInt64 cntForms = 0;
foreach (WordForm f in mrdFile.AllForms)
{
string word = f.Prefix + f.Flexia.Prefix + f.Lemma.Base + f.Flexia.Flexion;
FormInterpretations payload = null;
dawgBuilder.TryGetValue(word, out payload);
if (payload == null)
{
payload = new FormInterpretations();
dawgBuilder.Insert(word, payload);
}
payload.Add(f);
cntForms++;
}
Console.WriteLine("All forms count: " + cntForms);
Console.WriteLine("Building... please wait...");
Dawg <FormInterpretations> dawg = dawgBuilder.BuildDawg();
Console.WriteLine("DAWG create time: {0}", DateTime.Now - start);
return(dawg);
}
public FormInterpretations Lookup(string word)
{
word = word.ToUpperInvariant();
FormInterpretations forms = _dawg[word];
return(forms);
}
private FormInterpretations ReadPayload(BinaryReader binaryReader)
{
// cnt
// <acc1> <ancode-list1>
// <acc2> <ancode-list2>
//
// <ancode-list> := <cnt-byte> <ushort>
var p = new FormInterpretations();
int count = binaryReader.ReadByte();
while (count > 0)
{
int accent = binaryReader.ReadByte();
int listCnt = binaryReader.ReadByte();
List <ushort> list = new List <ushort>();
while (listCnt > 0)
{
list.Add(binaryReader.ReadUInt16());
listCnt--;
}
p.AccentToAncodes.Add(accent, list);
count--;
}
return(p);
}
private void WritePayload(BinaryWriter binaryWriter, FormInterpretations formInterpretations)
{
// cnt
// <acc1> <ancode-list1>
// <acc2> <ancode-list2>
//
// <ancode-list> := <cnt-byte> <ushort>
binaryWriter.Write((byte)formInterpretations.AccentToAncodes.Count);
foreach (var kvp in formInterpretations.AccentToAncodes)
{
binaryWriter.Write((byte)kvp.Key);
binaryWriter.Write((byte)kvp.Value.Count);
foreach (ushort ancodeNo in kvp.Value)
{
binaryWriter.Write((ushort)ancodeNo);
}
}
}
public void PrintLookup(string word)
{
FormInterpretations forms = Lookup(word);
if (forms != null)
{
foreach (var kvp in forms.AccentToAncodes)
{
Console.WriteLine("Accent {0} {1}", kvp.Key, AccentHelper.SetAccent(word, kvp.Key));
foreach (ushort ancodeNo in kvp.Value)
{
Console.WriteLine(" {0}", _gramtab.LookupByNo(ancodeNo));
}
}
}
else
{
Console.WriteLine(word + " not found");
}
}
public virtual void ProcessWord()
{
Console.WriteLine(_currentWord);
FormInterpretations forms = _morph.Lookup(_currentWord);
if (forms != null)
{
if (forms.AccentToAncodes.Count == 1)
{
SingleAccentWords.Add(AccentHelper.SetAccent(_currentWord, forms.AccentToAncodes.Keys.First()));
}
else
{
ManyAccentWords.Add(_currentWord);
}
}
else
{
NotFoundWords.Add(_currentWord);
}
}
