public void EscapeStateTokenizeNonWhitespace()
{
var lexer = new Lexer();
var tokens = lexer.Tokenize("\\S");
var expected = new NonWhitespaceToken();
Assert.AreEqual(tokens.Last(), expected);
}
public void EscapeStateTokenizeNumeric()
{
var lexer = new Lexer();
var tokens = lexer.Tokenize("\\d");
var expected = new NumericToken();
Assert.AreEqual(tokens.Last(), expected);
}
public void BeginSetStateTokenizeNot()
{
var lexer = new Lexer();
var tokens = lexer.Tokenize("[^");
var expected = new NotToken();
Assert.AreEqual(tokens.Last(), expected);
}
public void EscapeStateTokenizeLiteral()
{
var lexer = new Lexer();
var tokens = lexer.Tokenize("\\z");
var expected = new LiteralToken { Character = 'z' };
Assert.AreEqual(tokens.Last(), expected);
}
public void EscapeStateTokenizeWord()
{
var lexer = new Lexer();
var tokens = lexer.Tokenize("\\w");
var expected = new WordToken();
Assert.AreEqual(tokens.Last(), expected);
}
public void SetStateTokenizeRange()
{
var lexer = new Lexer();
var tokens = lexer.Tokenize("[a-");
var expected = new RangeToken();
Assert.AreEqual(tokens.Last(), expected);
}
public void SetStateTokenizeBracketRight()
{
var lexer = new Lexer();
var tokens = lexer.Tokenize("[a]");
var expected = new BracketRightToken();
Assert.AreEqual(tokens.Last(), expected);
}
public void RepetitionStateTokenizeToManyCommas()
{
var lexer = new Lexer();
Assert.Throws<ArgumentException>(() => lexer.Tokenize("{1,,5}").ToList());
}
public void LiteralStateTokenizeBracketLeft()
{
var lexer = new Lexer();
var tokens = lexer.Tokenize("[");
var expected = new BracketLeftToken();
Assert.AreEqual(tokens.First(), expected);
}
public void RepetitionStateTokenizeMissingValues()
{
var lexer = new Lexer();
Assert.Throws<ArgumentException>(() => lexer.Tokenize("{}").ToList());
}
public void RepetitionStateTokenizeInvalid()
{
var lexer = new Lexer();
Assert.Throws<ArgumentException>(() => lexer.Tokenize("{1,a}").ToList());
}
public void LiteralStateTokenizeZeroOrOne()
{
var lexer = new Lexer();
var tokens = lexer.Tokenize("?");
var expected = new RepetitionToken { MinOccurs = 0, MaxOccurs = 1 };
Assert.AreEqual(tokens.First(), expected);
}
public void LiteralStateTokenizeParenthesisRight()
{
var lexer = new Lexer();
var tokens = lexer.Tokenize(")");
var expected = new ParenthesisRightToken();
Assert.AreEqual(tokens.First(), expected);
}
public void LiteralStateTokenizeLiterals()
{
var lexer = new Lexer();
var tokens = lexer.Tokenize("abc");
var expected = new List<IToken>
{
new LiteralToken {Character = 'a'}
, new LiteralToken {Character = 'b'}
, new LiteralToken {Character = 'c'}
};
Assert.IsTrue(expected.SequenceEqual(tokens));
}
public void RepetitionStateTokenizeShort()
{
var lexer = new Lexer();
var tokens = lexer.Tokenize("{7}");
var expected = new RepetitionToken { MinOccurs = 7, MaxOccurs = 7 };
Assert.AreEqual(tokens.First(), expected);
}
public void LiteralStateTokenizeAny()
{
var lexer = new Lexer();
var tokens = lexer.Tokenize(".");
var expected = new AnyToken();
Assert.AreEqual(tokens.First(), expected);
}
public INode Parse(string expression)
{
var tokens = new Lexer().Tokenize(expression);
return Parse(tokens);
}