protected override object VisitBackreferenceBlock(BackreferenceBlock block)
{
string text;
if (!_groupValues.TryGetValue(block.GroupIndex, out text))
throw new ArgumentException("Reference to unspecified group", "block");
if (CanUseText(_position, text))
return new[] { _position + text.Length };
return _errorResult;
}
public void GetLinesFromBackreferenceBlockTest()
{
Dictionary<int, string> groupValues = new Dictionary<int,string>();
var input = "\0\0'";
int groupId = 1;
var visitor1 = new BuilderRegexVisitor(0, input, groupValues, null);
groupValues[groupId] = "ab";
var backreferenceBlock = new BackreferenceBlock(groupId);
string[] expectedLines = new string[] { "ab" };
CheckVisitorResult(visitor1, backreferenceBlock, expectedLines);
groupValues[groupId] = "abc";
var visitor2 = new BuilderRegexVisitor(0, input, groupValues, null);
CheckVisitorResult(visitor2, backreferenceBlock, Enumerable.Empty<string>());
}
public void GetPositionsFromBackreferenceBlockTest()
{
var groupValues = new Dictionary<int, string>();
string input = "\0\0";
int groupId = 1;
var visitor1 = new CheckerRegexVisitor(0, input, groupValues);
groupValues[groupId] = "ab";
var backreferenceBlock = new BackreferenceBlock(groupId);
var positions1 = visitor1.GetPositions(backreferenceBlock);
positions1.Should().BeEquivalentTo(new[] { 2 });
groupValues[groupId] = "abc";
var visitor2 = new CheckerRegexVisitor(0, input, groupValues);
var positions2 = visitor2.GetPositions(backreferenceBlock);
positions2.Should().BeEmpty();
}
public void ParseTest11()
{
var parser = new RegexParser();
string pattern = @".+(.)(.)(.)(.)\4\3\2\1.*";
// Expected regular expression
var anyCharacterBlock = new AnyCharacterBlock();
var oneOrMoreBlock = new OneOrMoreBlock(anyCharacterBlock);
var zeroOrOneBlock = new ZeroOrMoreBlock(anyCharacterBlock);
var andGroupBlock = new AndGroupBlock(new [] { anyCharacterBlock });
var b4 = new BackreferenceBlock(4);
var b3 = new BackreferenceBlock(3);
var b2 = new BackreferenceBlock(2);
var b1 = new BackreferenceBlock(1);
var groupBlock = new AndGroupBlock(new RegexBlock[] { oneOrMoreBlock, andGroupBlock, andGroupBlock, andGroupBlock, andGroupBlock, b4, b3, b2, b1, zeroOrOneBlock });
var expected = new RegularExpression(groupBlock);
var actual = parser.Parse(pattern);
actual.ShouldBeEquivalentTo(expected, options => options.IncludingAllRuntimeProperties());
}
public void ParseTest11()
{
var parser = new RegexParser();
string pattern = @".+(.)(.)(.)(.)\4\3\2\1.*";
// Expected regular expression
var anyCharacterBlock = new AnyCharacterBlock();
var oneOrMoreBlock = new OneOrMoreBlock(anyCharacterBlock);
var zeroOrOneBlock = new ZeroOrMoreBlock(anyCharacterBlock);
var andGroupBlock = new AndGroupBlock(new [] { anyCharacterBlock });
var b4 = new BackreferenceBlock(4);
var b3 = new BackreferenceBlock(3);
var b2 = new BackreferenceBlock(2);
var b1 = new BackreferenceBlock(1);
var groupBlock = new AndGroupBlock(new RegexBlock[] { oneOrMoreBlock, andGroupBlock, andGroupBlock, andGroupBlock, andGroupBlock, b4, b3, b2, b1, zeroOrOneBlock });
var expected = new RegularExpression(groupBlock);
var actual = parser.Parse(pattern);
Assert.IsTrue(expected.Equals(actual), "Pattern was parsed incorrectly");
}
protected virtual object VisitBackreferenceBlock(BackreferenceBlock block)
{
return block;
}
public RegularExpression Parse(string pattern)
{
var containers = new Stack<BlockContainer>();
containers.Push(new BlockContainer { Type = BlockContainerType.AndContainer });
bool expectGoupId = false;
int position = 0;
bool startNewBlock = false;
while (position < pattern.Length)
{
char c = pattern[position];
var currentContainer = containers.Peek();
if (currentContainer == null)
throw new ArgumentException("Invalid position for character" + c, "pattern");
if (char.IsLetterOrDigit(c))
{
if (expectGoupId)
{
var backreferenceBlock = new BackreferenceBlock(int.Parse(c.ToString(CultureInfo.InvariantCulture)));
currentContainer.PushBlock(backreferenceBlock);
expectGoupId = false;
}
else
{
if (startNewBlock || !currentContainer.Blocks.Any() || !IsCharacterContainer(currentContainer.PeekBlock()))
{
currentContainer.PushBlock(new TextBlock(string.Empty));
}
var characterContainer = currentContainer.PopBlock();
currentContainer.PushBlock(AddCharacter(characterContainer, c));
}
startNewBlock = false;
}
else
{
startNewBlock = false;
switch (c)
{
case '.':
currentContainer.PushBlock(new AnyCharacterBlock());
break;
case '+':
{
SplitTextBlock(currentContainer);
var previousBlock = currentContainer.PopBlock();
currentContainer.PushBlock(new OneOrMoreBlock(previousBlock));
}
break;
case '*':
{
SplitTextBlock(currentContainer);
var previousBlock = currentContainer.PopBlock();
currentContainer.PushBlock(new ZeroOrMoreBlock(previousBlock));
}
break;
case '?':
{
SplitTextBlock(currentContainer);
var previousBlock = currentContainer.PopBlock();
currentContainer.PushBlock(new ZeroOrOneBlock(previousBlock));
}
break;
case '[':
currentContainer.PushBlock(new InclusiveSetBlock(string.Empty));
break;
case ']':
startNewBlock = true;
break;
case '^':
{
var inclusiveSetBlock = currentContainer.PopBlock() as InclusiveSetBlock;
if (inclusiveSetBlock == null || inclusiveSetBlock.Characters.Any()) throw new ArgumentException("Invalid position for character ^", "pattern");
currentContainer.PushBlock(new ExclusiveSetBlock(inclusiveSetBlock.Characters));
}
break;
case '(':
containers.Push(new BlockContainer());
break;
case ')':
{
if (currentContainer.Type == BlockContainerType.Undefined) currentContainer.Type = BlockContainerType.AndContainer;
GroupBlock containerBlock = ConvertToBlock(currentContainer);
containers.Pop();
var outerBlockContainer = containers.Peek();
if (outerBlockContainer == null) throw new ArgumentException("Invalid position for character ]", "pattern");
outerBlockContainer.PushBlock(containerBlock);
}
break;
case '|':
startNewBlock = true;
currentContainer.Type = BlockContainerType.OrContainer;
break;
case '\\':
expectGoupId = true;
break;
default:
throw new ArgumentException("Unrecognized symbol in pattern", "pattern");
}
}
position++;
}
var container = containers.Pop();
GroupBlock block = ConvertToBlock(container);
// Remove redundant group group if any.
if (block.InnerBlocks.Count() == 1)
{
var orGroupBlock = block.InnerBlocks.First() as OrGroupBlock;
if (orGroupBlock != null)
block = orGroupBlock;
}
return new RegularExpression(block);
}
