public void Can_find_exact()
{
var tree = new LcrsTrie('\0', false);
Word word;
Assert.False(tree.HasWord("xxx", out word));
tree.Add("xxx");
Assert.True(tree.HasWord("xxx", out word));
Assert.False(tree.HasWord("baby", out word));
Assert.False(tree.HasWord("dad", out word));
tree.Add("baby");
Assert.True(tree.HasWord("xxx", out word));
Assert.True(tree.HasWord("baby", out word));
Assert.False(tree.HasWord("dad", out word));
tree.Add("dad");
Assert.True(tree.HasWord("xxx", out word));
Assert.True(tree.HasWord("baby", out word));
Assert.True(tree.HasWord("dad", out word));
}
public void Can_append()
{
var root = new LcrsTrie('\0', false);
root.Add("baby");
root.Add("bad");
Assert.That(root.LeftChild.Value, Is.EqualTo('b'));
Assert.That(root.LeftChild.LeftChild.Value, Is.EqualTo('a'));
Assert.That(root.LeftChild.LeftChild.LeftChild.Value, Is.EqualTo('d'));
Assert.That(root.LeftChild.LeftChild.LeftChild.RightSibling.Value, Is.EqualTo('b'));
Assert.That(root.LeftChild.LeftChild.LeftChild.RightSibling.LeftChild.Value, Is.EqualTo('y'));
Assert.True(root.HasWord("baby"));
Assert.True(root.HasWord("bad"));
}
public void Can_append()
{
var root = new LcrsTrie('\0', false);
root.Add("baby");
root.Add("bad");
Word word;
Assert.AreEqual('b', root.LeftChild.Value);
Assert.AreEqual('a', root.LeftChild.LeftChild.Value);
Assert.AreEqual('d', root.LeftChild.LeftChild.LeftChild.RightSibling.Value);
Assert.AreEqual('b', root.LeftChild.LeftChild.LeftChild.Value);
Assert.AreEqual('y', root.LeftChild.LeftChild.LeftChild.LeftChild.Value);
Assert.IsTrue(root.HasWord("baby", out word));
Assert.IsTrue(root.HasWord("bad", out word));
}
public void Commit()
{
var deleteSet = new LcrsTrie();
foreach (var value in _values)
{
var hashString = value.ToHash().ToString(CultureInfo.InvariantCulture);
deleteSet.Add(hashString);
}
foreach (var ix in _ixs)
{
var docHashFileName = Path.Combine(_directory, string.Format("{0}.{1}", ix.VersionId, "pk"));
var tmpDocHashFileName = Path.Combine(_directory, string.Format("{0}.{1}", ix.VersionId, "pk.tmp"));
var tmpIxFileName = Path.Combine(_directory, ix.VersionId + ".ix.tmp");
var ixFileName = Path.Combine(_directory, ix.VersionId + ".ix");
var deleted = 0;
using (var stream = new FileStream(tmpDocHashFileName, FileMode.Create, FileAccess.Write, FileShare.None))
{
foreach (var document in Serializer.DeserializeDocHashes(docHashFileName))
{
Word found;
var hash = document.Hash.ToString(CultureInfo.InvariantCulture);
if (deleteSet.HasWord(hash, out found))
{
if (!document.IsObsolete)
{
document.IsObsolete = true;
deleted++;
}
}
document.Serialize(stream);
}
}
if (deleted > 0)
{
ix.DocumentCount -= deleted;
ix.Serialize(tmpIxFileName);
File.Copy(tmpIxFileName, ixFileName, overwrite: true);
File.Copy(tmpDocHashFileName, docHashFileName, overwrite: true);
File.Delete(tmpIxFileName);
File.Delete(tmpDocHashFileName);
}
}
}
public void Can_build_two_legs()
{
var root = new LcrsTrie('\0', false);
root.Add("baby");
root.Add("dad");
Word word;
Assert.That(root.LeftChild.RightSibling.Value, Is.EqualTo('d'));
Assert.That(root.LeftChild.LeftChild.Value, Is.EqualTo('a'));
Assert.That(root.LeftChild.RightSibling.LeftChild.LeftChild.Value, Is.EqualTo('d'));
Assert.That(root.LeftChild.Value, Is.EqualTo('b'));
Assert.That(root.LeftChild.RightSibling.LeftChild.Value, Is.EqualTo('a'));
Assert.That(root.LeftChild.LeftChild.LeftChild.Value, Is.EqualTo('b'));
Assert.That(root.LeftChild.LeftChild.LeftChild.LeftChild.Value, Is.EqualTo('y'));
Assert.True(root.HasWord("baby", out word));
Assert.True(root.HasWord("dad", out word));
}
public void Can_deserialize_whole_file()
{
var fileName = Path.Combine(Dir, "Can_serialize_whole_file.tri");
var tree = new LcrsTrie('\0', false);
tree.Add("baby");
tree.Add("bad");
tree.Add("bank");
tree.Add("box");
tree.Add("dad");
tree.Add("dance");
Word found;
Assert.IsTrue(tree.HasWord("baby", out found));
Assert.IsTrue(tree.HasWord("bad", out found));
Assert.IsTrue(tree.HasWord("bank", out found));
Assert.IsTrue(tree.HasWord("box", out found));
Assert.IsTrue(tree.HasWord("dad", out found));
Assert.IsTrue(tree.HasWord("dance", out found));
tree.Serialize(fileName);
var recreated = Serializer.DeserializeTrie(Dir, new FileInfo(fileName).Name);
Assert.IsTrue(recreated.HasWord("baby", out found));
Assert.IsTrue(recreated.HasWord("bad", out found));
Assert.IsTrue(recreated.HasWord("bank", out found));
Assert.IsTrue(recreated.HasWord("box", out found));
Assert.IsTrue(recreated.HasWord("dad", out found));
Assert.IsTrue(recreated.HasWord("dance", out found));
}
public void Can_append_tries()
{
var one = new LcrsTrie('\0', false);
one.Add("ape");
one.Add("app");
one.Add("banana");
var two = new LcrsTrie('\0', false);
two.Add("apple");
two.Add("banana");
one.Merge(two);
Word found;
Assert.IsTrue(one.HasWord("ape", out found));
Assert.IsTrue(one.HasWord("app", out found));
Assert.IsTrue(one.HasWord("apple", out found));
Assert.IsTrue(one.HasWord("banana", out found));
}
public void Can_build_one_leg()
{
var tree = new LcrsTrie('\0', false);
tree.Add("baby");
Assert.That(tree.LeftChild.Value, Is.EqualTo('b'));
Assert.That(tree.LeftChild.LeftChild.Value, Is.EqualTo('a'));
Assert.That(tree.LeftChild.LeftChild.LeftChild.Value, Is.EqualTo('b'));
Assert.That(tree.LeftChild.LeftChild.LeftChild.LeftChild.Value, Is.EqualTo('y'));
Assert.True(tree.HasWord("baby"));
}
public void Can_build_one_leg()
{
var tree = new LcrsTrie('\0', false);
Word word;
tree.Add("baby");
Assert.AreEqual('b', tree.LeftChild.Value);
Assert.AreEqual('a', tree.LeftChild.LeftChild.Value);
Assert.AreEqual('b', tree.LeftChild.LeftChild.LeftChild.Value);
Assert.AreEqual('y', tree.LeftChild.LeftChild.LeftChild.LeftChild.Value);
Assert.IsTrue(tree.HasWord("baby", out word));
}
