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 void EmptyKey()
{
var db = new DawgBuilder <int>();
Assert.False(db.TryGetValue("", out int n));
Assert.AreEqual(0, n);
}
private static void Increment(DawgBuilder <int> db, string key)
{
int n;
db.TryGetValue(key, out n);
db.Insert(key, n + 1);
}
public void EmptyKey()
{
var db = new DawgBuilder<int>();
int n;
Assert.AreEqual(true, db.TryGetValue("", out n));
Assert.AreEqual(0, n);
}
public void TryGetValueOnPartialKey()
{
var builder = new DawgBuilder <bool>();
builder.Insert("dates", true);
bool b = builder.TryGetValue("date", out var v);
Assert.False(v);
Assert.False(b);
}
/// <summary>
/// Добавляет новую лексему к модели.
/// </summary>
/// <param name="lexem">Лексема.</param>
public void AddLexem(WordForm lexem)
{
ulong tagOut = (ulong)Tag.NoWord;
ulong tag = (ulong)lexem.Tag;
if (builder.TryGetValue(lexem.Word, out tagOut))
{
tag |= tagOut;
}
builder.Insert(lexem.Word, tag);
}
public void TryGetValueUnexpectedBehaviour()
{
var builder = new DawgBuilder <string>();
builder.Insert("dates", "dates");
bool b = builder.TryGetValue("date", out var v);
Assert.True(b);
Assert.Null(v);
}
public void IncrementValue()
{
var db = new DawgBuilder<int>();
Increment(db, "test");
Increment(db, "test");
Increment(db, "test");
int n;
Assert.AreEqual(true, db.TryGetValue("test", out n));
Assert.AreEqual(3, n);
}
public void IncrementValue()
{
var db = new DawgBuilder <int>();
Increment(db, "test");
Increment(db, "test");
Increment(db, "test");
int n;
Assert.AreEqual(true, db.TryGetValue("test", out n));
Assert.AreEqual(3, n);
}
public void IndexTerm(DocumentId id, string word, int position)
{
builder.TryGetValue(word, out var blockList);
if (blockList == null)
{
blockList = new RangePostingsList();
builder.Insert(word, blockList);
}
blockList.Add(id);
allDocuments.Add(id);
}
private static void Increment(DawgBuilder<int> db, string key)
{
int n;
db.TryGetValue(key, out n);
db.Insert(key, n + 1);
}
