public Row[] Next()
{
if (!HasNext())
throw new Exception("no more elements");
var rows = new Row[_columnRowCounters.Length];
for (int i = 0; i < _columnRowCounters.Length; i++)
{
rows[i] = _enclosingInstance.Columns[i].Rows[_columnRowCounters[i]];
}
IncrementColumnRowCounters();
return rows;
}
/// <summary>
/// Loads one column from the token stream.
///
/// When the last token is read from the token stream it will column.setLast(true);
/// </summary>
/// <returns>true if it manage to read one more column from the input token stream</returns>
private bool ReadColumn()
{
Token token;
if (_readColumnBuf != null)
{
token = _readColumnBuf;
_readColumnBuf = null;
}
else
{
token = GetNextInputToken(new Token());
}
if (token == null)
return false;
var currentReaderColumn = new Column(Matrix);
var currentReaderRow = new Row(currentReaderColumn);
currentReaderRow.Tokens.AddLast(token);
TokenPositioner tokenPositioner;
while ((_readColumnBuf = GetNextInputToken(new Token())) != null &&
(tokenPositioner = _settingsCodec.GetTokenPositioner(_readColumnBuf)) != TokenPositioner.NewColumn)
{
if (tokenPositioner == TokenPositioner.SameRow)
{
currentReaderRow.Tokens.AddLast(_readColumnBuf);
}
else
{
currentReaderRow = new Row(currentReaderColumn);
currentReaderRow.Tokens.AddLast(_readColumnBuf);
}
_readColumnBuf = null;
}
if (_readColumnBuf == null)
{
_readColumnBuf = GetNextInputToken(new Token());
if (_readColumnBuf == null)
currentReaderColumn.IsLast = true;
}
return true;
}
public Column(Token token, Analysis.Shingle.Matrix.Matrix enclosingInstance)
: this(enclosingInstance)
{
var row = new Row(this);
row.Tokens.AddLast(token);
}
public void TestMatrix()
{
// some other tests set this to null.
// set it here in case tests are run out of the usual order.
ShingleMatrixFilter.DefaultSettingsCodec = new SimpleThreeDimensionalTokenSettingsCodec();
var matrix = new Matrix.Matrix();
new Column(TokenFactory("no", 1), matrix);
new Column(TokenFactory("surprise", 1), matrix);
new Column(TokenFactory("to", 1), matrix);
new Column(TokenFactory("see", 1), matrix);
new Column(TokenFactory("england", 1), matrix);
new Column(TokenFactory("manager", 1), matrix);
var col = new Column(matrix);
// sven göran eriksson is a multi token synonym to svennis
new Row(col).Tokens.AddLast(TokenFactory("svennis", 1));
var row = new Row(col);
row.Tokens.AddLast(TokenFactory("sven", 1));
row.Tokens.AddLast(TokenFactory("göran", 1));
row.Tokens.AddLast(TokenFactory("eriksson", 1));
new Column(TokenFactory("in", 1), matrix);
new Column(TokenFactory("the", 1), matrix);
new Column(TokenFactory("croud", 1), matrix);
TokenStream ts = new ShingleMatrixFilter(matrix, 2, 4, '_', true,
new SimpleThreeDimensionalTokenSettingsCodec());
// for (Token token = ts.next(new Token()); token != null; token = ts.next(token)) {
// System.out.println("assertNext(ts, \"" + token.term() + "\", " + token.getPositionIncrement() + ", " + (token.getPayload() == null ? "1.0" : PayloadHelper.decodeFloat(token.getPayload().getData())) + "f, " + token.startOffset() + ", " + token.endOffset() + ");");
// token.clear();
// }
AssertNext(ts, "no_surprise", 1, 1.4142135f, 0, 0);
AssertNext(ts, "no_surprise_to", 1, 1.7320508f, 0, 0);
AssertNext(ts, "no_surprise_to_see", 1, 2.0f, 0, 0);
AssertNext(ts, "surprise_to", 1, 1.4142135f, 0, 0);
AssertNext(ts, "surprise_to_see", 1, 1.7320508f, 0, 0);
AssertNext(ts, "surprise_to_see_england", 1, 2.0f, 0, 0);
AssertNext(ts, "to_see", 1, 1.4142135f, 0, 0);
AssertNext(ts, "to_see_england", 1, 1.7320508f, 0, 0);
AssertNext(ts, "to_see_england_manager", 1, 2.0f, 0, 0);
AssertNext(ts, "see_england", 1, 1.4142135f, 0, 0);
AssertNext(ts, "see_england_manager", 1, 1.7320508f, 0, 0);
AssertNext(ts, "see_england_manager_svennis", 1, 2.0f, 0, 0);
AssertNext(ts, "england_manager", 1, 1.4142135f, 0, 0);
AssertNext(ts, "england_manager_svennis", 1, 1.7320508f, 0, 0);
AssertNext(ts, "england_manager_svennis_in", 1, 2.0f, 0, 0);
AssertNext(ts, "manager_svennis", 1, 1.4142135f, 0, 0);
AssertNext(ts, "manager_svennis_in", 1, 1.7320508f, 0, 0);
AssertNext(ts, "manager_svennis_in_the", 1, 2.0f, 0, 0);
AssertNext(ts, "svennis_in", 1, 1.4142135f, 0, 0);
AssertNext(ts, "svennis_in_the", 1, 1.7320508f, 0, 0);
AssertNext(ts, "svennis_in_the_croud", 1, 2.0f, 0, 0);
AssertNext(ts, "in_the", 1, 1.4142135f, 0, 0);
AssertNext(ts, "in_the_croud", 1, 1.7320508f, 0, 0);
AssertNext(ts, "the_croud", 1, 1.4142135f, 0, 0);
AssertNext(ts, "see_england_manager_sven", 1, 2.0f, 0, 0);
AssertNext(ts, "england_manager_sven", 1, 1.7320508f, 0, 0);
AssertNext(ts, "england_manager_sven_göran", 1, 2.0f, 0, 0);
AssertNext(ts, "manager_sven", 1, 1.4142135f, 0, 0);
AssertNext(ts, "manager_sven_göran", 1, 1.7320508f, 0, 0);
AssertNext(ts, "manager_sven_göran_eriksson", 1, 2.0f, 0, 0);
AssertNext(ts, "sven_göran", 1, 1.4142135f, 0, 0);
AssertNext(ts, "sven_göran_eriksson", 1, 1.7320508f, 0, 0);
AssertNext(ts, "sven_göran_eriksson_in", 1, 2.0f, 0, 0);
AssertNext(ts, "göran_eriksson", 1, 1.4142135f, 0, 0);
AssertNext(ts, "göran_eriksson_in", 1, 1.7320508f, 0, 0);
AssertNext(ts, "göran_eriksson_in_the", 1, 2.0f, 0, 0);
AssertNext(ts, "eriksson_in", 1, 1.4142135f, 0, 0);
AssertNext(ts, "eriksson_in_the", 1, 1.7320508f, 0, 0);
AssertNext(ts, "eriksson_in_the_croud", 1, 2.0f, 0, 0);
Assert.IsFalse(ts.IncrementToken());
}