/*
* Since RegexReplacement shares the same parser as Regex,
* the constructor takes a RegexNode which is a concatenation
* of constant strings and backreferences.
*/
internal RegexReplacement(String rep, RegexNode concat, Hashtable _caps)
{
StringBuilder sb;
ArrayList strings;
ArrayList rules;
int slot;
_rep = rep;
if (concat.Type() != RegexNode.Concatenate)
{
throw new ArgumentException(SR.GetString(SR.ReplacementError));
}
sb = new StringBuilder();
strings = new ArrayList();
rules = new ArrayList();
for (int i = 0; i < concat.ChildCount(); i++)
{
RegexNode child = concat.Child(i);
switch (child.Type())
{
case RegexNode.Multi:
sb.Append(child._str);
break;
case RegexNode.One:
sb.Append(child._ch);
break;
case RegexNode.Ref:
if (sb.Length > 0)
{
rules.Add(strings.Count);
strings.Add(sb.ToString());
sb.Length = 0;
}
slot = child._m;
if (_caps != null && slot >= 0)
{
slot = (int)_caps[slot];
}
rules.Add(-Specials - 1 - slot);
break;
default:
throw new ArgumentException(SR.GetString(SR.ReplacementError));
}
}
if (sb.Length > 0)
{
rules.Add(strings.Count);
strings.Add(sb.ToString());
}
_strings = strings;
_rules = rules;
}
/*
* This is a related computation: it takes a RegexTree and computes the
* leading substring if it see one. It's quite trivial and gives up easily.
*/
internal static RegexPrefix Prefix(RegexTree tree)
{
RegexNode curNode;
RegexNode concatNode = null;
int nextChild = 0;
curNode = tree._root;
for (;;)
{
switch (curNode._type)
{
case RegexNode.Concatenate:
if (curNode.ChildCount() > 0)
{
concatNode = curNode;
nextChild = 0;
}
break;
case RegexNode.Greedy:
case RegexNode.Capture:
curNode = curNode.Child(0);
concatNode = null;
continue;
case RegexNode.Oneloop:
case RegexNode.Onelazy:
if (curNode._m > 0)
{
string pref = String.Empty.PadRight(curNode._m, curNode._ch);
return(new RegexPrefix(pref, 0 != (curNode._options & RegexOptions.IgnoreCase)));
}
else
{
return(RegexPrefix.Empty);
}
case RegexNode.One:
return(new RegexPrefix(curNode._ch.ToString(CultureInfo.InvariantCulture), 0 != (curNode._options & RegexOptions.IgnoreCase)));
case RegexNode.Multi:
return(new RegexPrefix(curNode._str, 0 != (curNode._options & RegexOptions.IgnoreCase)));
case RegexNode.Bol:
case RegexNode.Eol:
case RegexNode.Boundary:
case RegexNode.ECMABoundary:
case RegexNode.Beginning:
case RegexNode.Start:
case RegexNode.EndZ:
case RegexNode.End:
case RegexNode.Empty:
case RegexNode.Require:
case RegexNode.Prevent:
break;
default:
return(RegexPrefix.Empty);
}
if (concatNode == null || nextChild >= concatNode.ChildCount())
{
return(RegexPrefix.Empty);
}
curNode = concatNode.Child(nextChild++);
}
}
/*
* The main 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.
*/
internal 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 < Int32.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 < Int32.MaxValue || node._m > 1)
{
Emit(RegexCode.Branchcount + Lazy, PopInt(), node._n == Int32.MaxValue ? Int32.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, (int)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, (int)node._ch, node._m);
}
if (node._n > node._m)
{
Emit(node._type | bits, (int)node._ch, node._n == Int32.MaxValue ?
Int32.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 == Int32.MaxValue) ? Int32.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 MakeException(SR.GetString(SR.UnexpectedOpcode, nodetype.ToString(CultureInfo.CurrentCulture)));
}
}
/*
* FC computation and shortcut cases for each node type
*/
private void CalculateFC(int NodeType, RegexNode node, int CurIndex)
{
bool ci = false;
bool rtl = false;
if (NodeType <= RegexNode.Ref)
{
if ((node._options & RegexOptions.IgnoreCase) != 0)
{
ci = true;
}
if ((node._options & RegexOptions.RightToLeft) != 0)
{
rtl = true;
}
}
switch (NodeType)
{
case RegexNode.Concatenate | BeforeChild:
case RegexNode.Alternate | BeforeChild:
case RegexNode.Testref | BeforeChild:
case RegexNode.Loop | BeforeChild:
case RegexNode.Lazyloop | BeforeChild:
break;
case RegexNode.Testgroup | BeforeChild:
if (CurIndex == 0)
{
SkipChild();
}
break;
case RegexNode.Empty:
PushFC(new RegexFC(true));
break;
case RegexNode.Concatenate | AfterChild:
if (CurIndex != 0)
{
RegexFC child = PopFC();
RegexFC cumul = TopFC();
_failed = !cumul.AddFC(child, true);
}
if (!TopFC()._nullable)
{
_skipAllChildren = true;
}
break;
case RegexNode.Testgroup | AfterChild:
if (CurIndex > 1)
{
RegexFC child = PopFC();
RegexFC cumul = TopFC();
_failed = !cumul.AddFC(child, false);
}
break;
case RegexNode.Alternate | AfterChild:
case RegexNode.Testref | AfterChild:
if (CurIndex != 0)
{
RegexFC child = PopFC();
RegexFC cumul = TopFC();
_failed = !cumul.AddFC(child, false);
}
break;
case RegexNode.Loop | AfterChild:
case RegexNode.Lazyloop | AfterChild:
if (node._m == 0)
{
TopFC()._nullable = true;
}
break;
case RegexNode.Group | BeforeChild:
case RegexNode.Group | AfterChild:
case RegexNode.Capture | BeforeChild:
case RegexNode.Capture | AfterChild:
case RegexNode.Greedy | BeforeChild:
case RegexNode.Greedy | AfterChild:
break;
case RegexNode.Require | BeforeChild:
case RegexNode.Prevent | BeforeChild:
SkipChild();
PushFC(new RegexFC(true));
break;
case RegexNode.Require | AfterChild:
case RegexNode.Prevent | AfterChild:
break;
case RegexNode.One:
case RegexNode.Notone:
PushFC(new RegexFC(node._ch, NodeType == RegexNode.Notone, false, ci));
break;
case RegexNode.Oneloop:
case RegexNode.Onelazy:
PushFC(new RegexFC(node._ch, false, node._m == 0, ci));
break;
case RegexNode.Notoneloop:
case RegexNode.Notonelazy:
PushFC(new RegexFC(node._ch, true, node._m == 0, ci));
break;
case RegexNode.Multi:
if (node._str.Length == 0)
{
PushFC(new RegexFC(true));
}
else if (!rtl)
{
PushFC(new RegexFC(node._str[0], false, false, ci));
}
else
{
PushFC(new RegexFC(node._str[node._str.Length - 1], false, false, ci));
}
break;
case RegexNode.Set:
PushFC(new RegexFC(node._str, false, ci));
break;
case RegexNode.Setloop:
case RegexNode.Setlazy:
PushFC(new RegexFC(node._str, node._m == 0, ci));
break;
case RegexNode.Ref:
PushFC(new RegexFC(RegexCharClass.AnyClass, true, false));
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:
PushFC(new RegexFC(true));
break;
default:
throw new ArgumentException(SR.GetString(SR.UnexpectedOpcode, NodeType.ToString(CultureInfo.CurrentCulture)));
}
}
