youmuliu/stgsharp
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

scanning min scan length

Browse Source
This commit is contained in:
Youmu Liu
2026-07-08 18:25:05 +08:00
parent 439128798e
commit 29b24191e1
8 changed files with 125 additions and 93 deletions
Download Patch File Download Diff File Expand all files Collapse all files
StgSharp.RegularAnalysis/Text/RegexAnalyzer.IRGenerate.cs
+54 -27
View File
@@ -24,10 +24,11 @@ namespace StgSharp.RegularAnalysis.Text
{
internal static List<RegexIR> GenerateIR(
AbstractSyntaxTree<RegexAstNode, RegexElementLabel> tree
AbstractSyntaxTree<RegexAstNode, RegexElementLabel> tree,
ref RegexInfo info
)
{
return GenerateIRRecursion(tree.Root, 0).ToList();
return GenerateIRRecursion(tree.Root, 0, ref info).ToList();
}
[MethodImpl(MethodImplOptions.AggressiveInlining)]
@@ -37,7 +38,8 @@ namespace StgSharp.RegularAnalysis.Text
int min,
int max,
bool isGreedy,
RegexAstNode right
RegexAstNode right,
ref RegexInfo info
)
{
bool is_operator = !RegexAstNode.IsNullOrEmpty(right) &&
@@ -45,7 +47,7 @@ namespace StgSharp.RegularAnalysis.Text
if (is_operator) // seq to be count is complex
{
RegexIRGenerator right_ir = GenerateIRRecursion(right, curDepth + 1);
RegexIRGenerator right_ir = GenerateIRRecursion(right, curDepth + 1, ref info);
ir.EmitCountComplex(min, max, isGreedy, right_ir.Count);
ir.EmitIRStream(right_ir);
} else
@@ -70,7 +72,8 @@ namespace StgSharp.RegularAnalysis.Text
private static RegexIRGenerator GenerateIRRecursion(
RegexAstNode node,
int depth
int depth,
ref RegexInfo info
)
{
#pragma warning disable IDE0010
@@ -99,7 +102,7 @@ namespace StgSharp.RegularAnalysis.Text
string.Empty; /* unnamed group (?:) */
RegexIRGenerator groupIR = new();
rightIR = GenerateIRRecursion(right, curDepth + 1);
rightIR = GenerateIRRecursion(right, curDepth + 1, ref info);
groupIR.EmitTry(rightIR.Count);
groupIR.EmitIRStream(rightIR);
groupIR.EmitCondition(2, 1);
@@ -109,31 +112,19 @@ namespace StgSharp.RegularAnalysis.Text
case RegexElementLabel.COUNT:
string count_source = token.Value;
bool isGreedy = !(count_source.Length > 1 && count_source.EndsWith('?'));
int seq_mask = (RegexAstNode.IsNullOrEmpty(left) ? 0 : 2) + (RegexAstNode.IsNullOrEmpty(right) ?
0 :
1);
RegexAstNode seq = seq_mask switch
{
// no seq, invalid count operator
0 => throw new InvalidOperationException("Count operator must have a sequence to operate on"),
1 => right,// right seq
2 => left,// left seq
// both sides, invalid count operator
3 => throw new InvalidOperationException("Count operator cannot have two sequences to operate on"),
_ => throw new InvalidOperationException("Unexpected sequence mask for count operator"),
};
RegexAstNode seq = right;
ReadOnlySpan<char> count_source_span = count_source;
RegexIRGenerator count_ir = new();
if (count_source_span[0] == '*') // infinite count
{
EmitCount(curDepth, count_ir, 0, -1, isGreedy, seq);
EmitCount(curDepth, count_ir, 0, -1, isGreedy, seq, ref info);
} else if (count_source_span[0] == '+') // at least one
{
EmitCount(curDepth, count_ir, 1, -1, isGreedy, seq);
EmitCount(curDepth, count_ir, 1, -1, isGreedy, seq, ref info);
} else if (count_source_span[0] == '?') // zero or one
{
EmitCount(curDepth, count_ir, 0, 1, isGreedy, seq);
EmitCount(curDepth, count_ir, 0, 1, isGreedy, seq, ref info);
} else if (count_source_span[0] == '{') // zero or one, non-greedy
{
int comma_pos = count_source_span.IndexOf(',');
@@ -141,7 +132,8 @@ namespace StgSharp.RegularAnalysis.Text
{
if (int.TryParse(count_source[1..(isGreedy ? ^2 : ^1)], out int count))
{
EmitCount(curDepth, count_ir, count, count, isGreedy, seq);
EmitCount(curDepth, count_ir, count, count, isGreedy, seq,
ref info);
} else
{
throw new InvalidCastException($"Invalid count format: {count_source}");
@@ -151,7 +143,7 @@ namespace StgSharp.RegularAnalysis.Text
if (int.TryParse(count_source[1..comma_pos], out int min) &&
int.TryParse(count_source[(comma_pos + 1)..(isGreedy ? ^1 : ^2)], out int max))
{
EmitCount(curDepth, count_ir, min, max, isGreedy, seq);
EmitCount(curDepth, count_ir, min, max, isGreedy, seq, ref info);
} else
{
throw new InvalidCastException($"Invalid count format: {count_source}");
@@ -187,7 +179,7 @@ namespace StgSharp.RegularAnalysis.Text
}
} else if ((left.Source.Flag & RegexElementLabel.VAST_OPERATOR) != 0)
{
leftIR = GenerateIRRecursion(left, curDepth + 1);
leftIR = GenerateIRRecursion(left, curDepth + 1, ref info);
concat.EmitIRStream(leftIR);
}
@@ -212,7 +204,7 @@ namespace StgSharp.RegularAnalysis.Text
}
} else if ((right.Source.Flag & RegexElementLabel.VAST_OPERATOR) != 0)
{
rightIR = GenerateIRRecursion(right, curDepth + 1);
rightIR = GenerateIRRecursion(right, curDepth + 1, ref info);
concat.EmitIRStream(rightIR);
}
return concat;
@@ -223,7 +215,7 @@ namespace StgSharp.RegularAnalysis.Text
int ir_count = 0;
while (!RegexAstNode.IsNullOrEmpty(cur_node))
{
RegexIRGenerator _case = GenerateIRRecursion(cur_node, curDepth + 1);
RegexIRGenerator _case = GenerateIRRecursion(cur_node, curDepth + 1, ref info);
casesIRList.Add(_case);
ir_count += _case.Count;
cur_node = cur_node.Next;
@@ -255,6 +247,41 @@ namespace StgSharp.RegularAnalysis.Text
return null!;
}
#pragma warning restore IDE0010
// info.MinPredictLength = ComputeMinLength(tree.Root);
static int ComputeMinLength(
RegexAstNode node
)
{
if (node.EqualityTypeConvert is RegexElementLabel.UNIT_SPAN or
RegexElementLabel.UNIT)
{
return node.Value.Length;
} else if (node.EqualityTypeConvert == RegexElementLabel.CONCAT)
{
return ComputeMinLength(node.Left) + ComputeMinLength(node.Right);
} else if (node.EqualityTypeConvert == RegexElementLabel.ALT)
{
RegexAstNode case_0 = node.Right;
int length = int.MaxValue;
do
{
length = int.Min(ComputeMinLength(case_0), length);
case_0 = case_0.Next;
} while (!RegexAstNode.IsNullOrEmpty(case_0));
return length;
} else if (node.EqualityTypeConvert == RegexElementLabel.COUNT)
{
string source = node.Source.Value;
int single_count = ComputeMinLength(node.Right);
return 0;
} else
{
return 0;
}
}
}
}