building regex parser

This commit is contained in:
Youmu Liu
2026-02-05 18:35:42 +08:00
parent f22e90af55
commit 1bfb8b67c4
6 changed files with 342 additions and 316 deletions
@@ -38,17 +38,87 @@ namespace StgSharp.RegularAnalysis.TextRegex
public class RegexInterpreter
{
public static AbstractSyntaxTree<RegexAstNode, RegexElementLabel> Analyze(
string regex
private AbstractSyntaxTree<RegexAstNode, RegexElementLabel> _tree = new();
private CompileStack<RegexAstNode, RegexElementLabel> _stack = new();
public AbstractSyntaxTree<RegexAstNode, RegexElementLabel> Analyze(
string regex
)
{
RegexTokenReader reader = new RegexTokenReader(regex);
TokenParser<RegexElementLabel, RegexElementLabel> lexer = reader.Pipe(() => new RegexTokenParser());
while (lexer.TryReadToken(out Token<RegexElementLabel> token)) { }
while (lexer.TryReadToken(out Token<RegexElementLabel> token))
{
if ((token.Flag & RegexElementLabel.SEQUENCE) != 0)
{
// char or charset
_stack.PushOperand(new RegexAstNode(token));
} else if ((token.Flag & RegexElementLabel.GROUP_BEGIN) != 0)
{
_stack.IncreaseDepth(0);
} else if ((token.Flag & RegexElementLabel.GROUP_END) != 0)
{
// TODO: close all nodes here
while (_stack.TryPopOperator(out RegexAstNode? op))
{
ProcessOperator(op);
}
if (_stack.OperandInDepthCount != 1) {
throw new InvalidOperationException();
}
_stack.DecreaseDepth();
} else if ((token.Flag & RegexElementLabel.OPERATOR) != 0) { }
}
}
private static void
private void ProcessOperator(
RegexAstNode op
)
{
if (op.EqualityTypeConvert == RegexElementLabel.CONCAT)
{
if (_stack.TryPopOperand(out RegexAstNode? _1) &&
_stack.TryPopOperand(out RegexAstNode? _2))
{
if (RegexAstNode.TryMergeUnit(_2, _1, out RegexAstNode? _3))
{
_stack.PushOperand(_3);
} else
{
op.Left = _2;
op.Right = _1;
_stack.PushOperand(op);
}
} else
{
throw new InvalidOperationException("Insufficient operands for CONCAT operator.");
}
} else if (op.EqualityTypeConvert == RegexElementLabel.COUNT)
{
if (_stack.TryPopOperand(out RegexAstNode? _1))
{
op.Left = _1;
_stack.PushOperand(op);
} else
{
throw new InvalidOperationException("Insufficient operands for CONCAT operator.");
}
} else if (op.EqualityTypeConvert == RegexElementLabel.ALT)
{
if (_stack.TryPopOperand(out RegexAstNode? _1) &&
_stack.TryPopOperand(out RegexAstNode? _2))
{
op.Left = _1;
op.Right = _2;
_stack.PushOperand(op);
} else
{
throw new InvalidOperationException("Insufficient operands for CONCAT operator.");
}
}
}
}
}