|
private EmitFragment ( int nodetype, |
||
| nodetype | int | |
| node | ||
| curIndex | int | |
| return | void | |
private void EmitFragment(int nodetype, RegexNode node, int curIndex)
{
int bits = 0;
if (nodetype <= RegexNode.Ref)
{
if (node.UseOptionR())
bits |= RegexCode.Rtl;
if ((node._options & RegexOptions.IgnoreCase) != 0)
bits |= RegexCode.Ci;
}
switch (nodetype)
{
case RegexNode.Concatenate | BeforeChild:
case RegexNode.Concatenate | AfterChild:
case RegexNode.Empty:
break;
case RegexNode.Alternate | BeforeChild:
if (curIndex < node._children.Count - 1)
{
PushInt(CurPos());
Emit(RegexCode.Lazybranch, 0);
}
break;
case RegexNode.Alternate | AfterChild:
{
if (curIndex < node._children.Count - 1)
{
int LBPos = PopInt();
PushInt(CurPos());
Emit(RegexCode.Goto, 0);
PatchJump(LBPos, CurPos());
}
else
{
int I;
for (I = 0; I < curIndex; I++)
{
PatchJump(PopInt(), CurPos());
}
}
break;
}
case RegexNode.Testref | BeforeChild:
switch (curIndex)
{
case 0:
Emit(RegexCode.Setjump);
PushInt(CurPos());
Emit(RegexCode.Lazybranch, 0);
Emit(RegexCode.Testref, MapCapnum(node._m));
Emit(RegexCode.Forejump);
break;
}
break;
case RegexNode.Testref | AfterChild:
switch (curIndex)
{
case 0:
{
int Branchpos = PopInt();
PushInt(CurPos());
Emit(RegexCode.Goto, 0);
PatchJump(Branchpos, CurPos());
Emit(RegexCode.Forejump);
if (node._children.Count > 1)
break;
// else fallthrough
goto case 1;
}
case 1:
PatchJump(PopInt(), CurPos());
break;
}
break;
case RegexNode.Testgroup | BeforeChild:
switch (curIndex)
{
case 0:
Emit(RegexCode.Setjump);
Emit(RegexCode.Setmark);
PushInt(CurPos());
Emit(RegexCode.Lazybranch, 0);
break;
}
break;
case RegexNode.Testgroup | AfterChild:
switch (curIndex)
{
case 0:
Emit(RegexCode.Getmark);
Emit(RegexCode.Forejump);
break;
case 1:
int Branchpos = PopInt();
PushInt(CurPos());
Emit(RegexCode.Goto, 0);
PatchJump(Branchpos, CurPos());
Emit(RegexCode.Getmark);
Emit(RegexCode.Forejump);
if (node._children.Count > 2)
break;
// else fallthrough
goto case 2;
case 2:
PatchJump(PopInt(), CurPos());
break;
}
break;
case RegexNode.Loop | BeforeChild:
case RegexNode.Lazyloop | BeforeChild:
if (node._n < int.MaxValue || node._m > 1)
Emit(node._m == 0 ? RegexCode.Nullcount : RegexCode.Setcount, node._m == 0 ? 0 : 1 - node._m);
else
Emit(node._m == 0 ? RegexCode.Nullmark : RegexCode.Setmark);
if (node._m == 0)
{
PushInt(CurPos());
Emit(RegexCode.Goto, 0);
}
PushInt(CurPos());
break;
case RegexNode.Loop | AfterChild:
case RegexNode.Lazyloop | AfterChild:
{
int StartJumpPos = CurPos();
int Lazy = (nodetype - (RegexNode.Loop | AfterChild));
if (node._n < int.MaxValue || node._m > 1)
Emit(RegexCode.Branchcount + Lazy, PopInt(), node._n == int.MaxValue ? int.MaxValue : node._n - node._m);
else
Emit(RegexCode.Branchmark + Lazy, PopInt());
if (node._m == 0)
PatchJump(PopInt(), StartJumpPos);
}
break;
case RegexNode.Group | BeforeChild:
case RegexNode.Group | AfterChild:
break;
case RegexNode.Capture | BeforeChild:
Emit(RegexCode.Setmark);
break;
case RegexNode.Capture | AfterChild:
Emit(RegexCode.Capturemark, MapCapnum(node._m), MapCapnum(node._n));
break;
case RegexNode.Require | BeforeChild:
// NOTE: the following line causes lookahead/lookbehind to be
// NON-BACKTRACKING. It can be commented out with (*)
Emit(RegexCode.Setjump);
Emit(RegexCode.Setmark);
break;
case RegexNode.Require | AfterChild:
Emit(RegexCode.Getmark);
// NOTE: the following line causes lookahead/lookbehind to be
// NON-BACKTRACKING. It can be commented out with (*)
Emit(RegexCode.Forejump);
break;
case RegexNode.Prevent | BeforeChild:
Emit(RegexCode.Setjump);
PushInt(CurPos());
Emit(RegexCode.Lazybranch, 0);
break;
case RegexNode.Prevent | AfterChild:
Emit(RegexCode.Backjump);
PatchJump(PopInt(), CurPos());
Emit(RegexCode.Forejump);
break;
case RegexNode.Greedy | BeforeChild:
Emit(RegexCode.Setjump);
break;
case RegexNode.Greedy | AfterChild:
Emit(RegexCode.Forejump);
break;
case RegexNode.One:
case RegexNode.Notone:
Emit(node._type | bits, node._ch);
break;
case RegexNode.Notoneloop:
case RegexNode.Notonelazy:
case RegexNode.Oneloop:
case RegexNode.Onelazy:
if (node._m > 0)
Emit(((node._type == RegexNode.Oneloop || node._type == RegexNode.Onelazy) ?
RegexCode.Onerep : RegexCode.Notonerep) | bits, node._ch, node._m);
if (node._n > node._m)
Emit(node._type | bits, node._ch, node._n == int.MaxValue ?
int.MaxValue : node._n - node._m);
break;
case RegexNode.Setloop:
case RegexNode.Setlazy:
if (node._m > 0)
Emit(RegexCode.Setrep | bits, StringCode(node._str), node._m);
if (node._n > node._m)
Emit(node._type | bits, StringCode(node._str),
(node._n == int.MaxValue) ? int.MaxValue : node._n - node._m);
break;
case RegexNode.Multi:
Emit(node._type | bits, StringCode(node._str));
break;
case RegexNode.Set:
Emit(node._type | bits, StringCode(node._str));
break;
case RegexNode.Ref:
Emit(node._type | bits, MapCapnum(node._m));
break;
case RegexNode.Nothing:
case RegexNode.Bol:
case RegexNode.Eol:
case RegexNode.Boundary:
case RegexNode.Nonboundary:
case RegexNode.ECMABoundary:
case RegexNode.NonECMABoundary:
case RegexNode.Beginning:
case RegexNode.Start:
case RegexNode.EndZ:
case RegexNode.End:
Emit(node._type);
break;
default:
throw new ArgumentException(SR.Format(SR.UnexpectedOpcode, nodetype.ToString(CultureInfo.CurrentCulture)));
}
}
/// <summary>
/// The top level RegexCode generator. It does a depth-first walk
/// through the tree and calls EmitFragment to emits code before
/// and after each child of an interior node, and at each leaf.
/// </summary>
public RegexCode RegexCodeFromRegexTree(RegexTree tree)
{
Span <int> emittedSpan = stackalloc int[EmittedSize];
Span <int> intStackSpan = stackalloc int[IntStackSize];
RegexWriter writer = new RegexWriter(emittedSpan, intStackSpan);
// construct sparse capnum mapping if some numbers are unused
int capsize;
if (tree._capnumlist == null || tree._captop == tree._capnumlist.Length)
{
capsize = tree._captop;
writer._caps = null;
}
else
{
capsize = tree._capnumlist.Length;
writer._caps = tree._caps;
for (int i = 0; i < tree._capnumlist.Length; i++)
{
writer._caps[tree._capnumlist[i]] = i;
}
}
RegexNode curNode = tree._root;
int curChild = 0;
writer.Emit(RegexCode.Lazybranch, 0);
for (; ;)
{
if (curNode._children == null)
{
writer.EmitFragment(curNode._type, curNode, 0);
}
else if (curChild < curNode._children.Count)
{
writer.EmitFragment(curNode._type | BeforeChild, curNode, curChild);
curNode = curNode._children[curChild];
writer._intStack.Append(curChild);
curChild = 0;
continue;
}
if (writer._intStack.Length == 0)
{
break;
}
curChild = writer._intStack.Pop();
curNode = curNode._next;
writer.EmitFragment(curNode._type | AfterChild, curNode, curChild);
curChild++;
}
writer.PatchJump(0, writer._emitted.Length);
writer.Emit(RegexCode.Stop);
RegexPrefix fcPrefix = RegexFCD.FirstChars(tree);
RegexPrefix prefix = RegexFCD.Prefix(tree);
bool rtl = ((tree._options & RegexOptions.RightToLeft) != 0);
CultureInfo culture = (tree._options & RegexOptions.CultureInvariant) != 0 ? CultureInfo.InvariantCulture : CultureInfo.CurrentCulture;
RegexBoyerMoore bmPrefix;
if (prefix != null && prefix.Prefix.Length > 0)
{
bmPrefix = new RegexBoyerMoore(prefix.Prefix, prefix.CaseInsensitive, rtl, culture);
}
else
{
bmPrefix = null;
}
int anchors = RegexFCD.Anchors(tree);
int[] emitted = writer._emitted.AsReadOnlySpan().ToArray();
// Cleaning up and returning the borrowed arrays
writer._emitted.Dispose();
writer._intStack.Dispose();
return(new RegexCode(emitted, writer._stringTable, writer._trackCount, writer._caps, capsize, bmPrefix, fcPrefix, anchors, rtl));
}
