internal RegexTree(RegexNode root, Dictionary<int, int> caps, int[] capnumlist, int captop, Dictionary<string, int> capnames, string[] capslist, RegexOptions opts)
{
_root = root;
_caps = caps;
_capnumlist = capnumlist;
_capnames = capnames;
_capslist = capslist;
_captop = captop;
_options = opts;
}
private readonly List<int> _rules; // negative -> group #, positive -> string #
/// <summary>
/// Since RegexReplacement shares the same parser as Regex,
/// the constructor takes a RegexNode which is a concatenation
/// of constant strings and backreferences.
/// </summary>
internal RegexReplacement(string rep, RegexNode concat, Dictionary<int, int> _caps)
{
if (concat.Type() != RegexNode.Concatenate)
throw new ArgumentException(SR.ReplacementError);
StringBuilder sb = new StringBuilder();
List<string> strings = new List<string>();
List<int> rules = new List<int>();
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;
}
int slot = child._m;
if (_caps != null && slot >= 0)
slot = (int)_caps[slot];
rules.Add(-Specials - 1 - slot);
break;
default:
throw new ArgumentException(SR.ReplacementError);
}
}
if (sb.Length > 0)
{
rules.Add(strings.Count);
strings.Add(sb.ToString());
}
_rep = rep;
_strings = strings;
_rules = rules;
}
internal void AddChild(RegexNode newChild)
{
RegexNode reducedChild;
if (_children == null)
{
_children = new List <RegexNode>(4);
}
reducedChild = newChild.Reduce();
_children.Add(reducedChild);
reducedChild._next = this;
}
/*
* Finish the current quantifiable (when a quantifier is not found or is not possible)
*/
internal void AddConcatenate()
{
// The first (| inside a Testgroup group goes directly to the group
_concatenation.AddChild(_unit);
_unit = null;
}
/*
* Finish the current concatenation (in response to a |)
*/
internal void AddAlternate()
{
// The | parts inside a Testgroup group go directly to the group
if (_group.Type() == RegexNode.Testgroup || _group.Type() == RegexNode.Testref)
{
_group.AddChild(_concatenation.ReverseLeft());
}
else
{
_alternation.AddChild(_concatenation.ReverseLeft());
}
_concatenation = new RegexNode(RegexNode.Concatenate, _options);
}
/*
* Start a new round for the parser state (in response to an open paren or string start)
*/
internal void StartGroup(RegexNode openGroup)
{
_group = openGroup;
_alternation = new RegexNode(RegexNode.Alternate, _options);
_concatenation = new RegexNode(RegexNode.Concatenate, _options);
}
/*
* 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.Format(SR.UnexpectedOpcode, NodeType.ToString(CultureInfo.CurrentCulture)));
}
}
/*
* Resets parsing to the beginning of the pattern.
*/
internal void Reset(RegexOptions topopts)
{
_currentPos = 0;
_autocap = 1;
_ignoreNextParen = false;
if (_optionsStack.Count > 0)
_optionsStack.RemoveRange(0, _optionsStack.Count - 1);
_options = topopts;
_stack = null;
}
/*
* Sets the current unit to an assertion of the specified type
*/
internal void AddUnitType(int type)
{
_unit = new RegexNode(type, _options);
}
/*
* Sets the current unit to a single set node
*/
internal void AddUnitSet(string cc)
{
_unit = new RegexNode(RegexNode.Set, _options, cc);
}
/// <summary>
/// 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.
/// </summary>
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)));
}
}
/*
* 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(), 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++);
}
}
/*
* 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:
case RegexNode.ResetMatchStart:
PushFC(new RegexFC(true));
break;
default:
throw new ArgumentException(SR.Format(SR.UnexpectedOpcode, NodeType.ToString(CultureInfo.CurrentCulture)));
}
}
internal void AddChild(RegexNode newChild)
{
RegexNode reducedChild;
if (_children == null)
_children = new List<RegexNode>(4);
reducedChild = newChild.Reduce();
_children.Add(reducedChild);
reducedChild._next = this;
}
internal RegexNode MakeQuantifier(bool lazy, int min, int max)
{
RegexNode result;
if (min == 0 && max == 0)
return new RegexNode(Empty, _options);
if (min == 1 && max == 1)
return this;
switch (_type)
{
case One:
case Notone:
case Set:
MakeRep(lazy ? Onelazy : Oneloop, min, max);
return this;
default:
result = new RegexNode(lazy ? Lazyloop : Loop, _options, min, max);
result.AddChild(this);
return result;
}
}
/*
* Finish the current quantifiable (when a quantifier is found)
*/
internal void AddConcatenate(bool lazy, int min, int max)
{
_concatenation.AddChild(_unit.MakeQuantifier(lazy, min, max));
_unit = null;
}
/*
* Sets the current unit to a single inverse-char node
*/
internal void AddUnitNotone(char ch)
{
if (UseOptionI())
ch = _culture.TextInfo.ToLower(ch);
_unit = new RegexNode(RegexNode.Notone, _options, ch);
}
/*
* Add a string to the last concatenate.
*/
internal void AddConcatenate(int pos, int cch, bool isReplacement)
{
RegexNode node;
if (cch == 0)
return;
if (cch > 1)
{
string str = _pattern.Substring(pos, cch);
if (UseOptionI() && !isReplacement)
{
// We do the ToLower character by character for consistency. With surrogate chars, doing
// a ToLower on the entire string could actually change the surrogate pair. This is more correct
// linguistically, but since Regex doesn't support surrogates, it's more important to be
// consistent.
var sb = new StringBuilder(str.Length);
for (int i = 0; i < str.Length; i++)
sb.Append(_culture.TextInfo.ToLower(str[i]));
str = sb.ToString();
}
node = new RegexNode(RegexNode.Multi, _options, str);
}
else
{
char ch = _pattern[pos];
if (UseOptionI() && !isReplacement)
ch = _culture.TextInfo.ToLower(ch);
node = new RegexNode(RegexNode.One, _options, ch);
}
_concatenation.AddChild(node);
}
/*
* Sets the current unit to a subtree
*/
internal void AddUnitNode(RegexNode node)
{
_unit = node;
}
/*
* Push the parser state (in response to an open paren)
*/
internal void PushGroup()
{
_group._next = _stack;
_alternation._next = _group;
_concatenation._next = _alternation;
_stack = _concatenation;
}
/*
* Finish the current group (in response to a ')' or end)
*/
internal void AddGroup()
{
if (_group.Type() == RegexNode.Testgroup || _group.Type() == RegexNode.Testref)
{
_group.AddChild(_concatenation.ReverseLeft());
if (_group.Type() == RegexNode.Testref && _group.ChildCount() > 2 || _group.ChildCount() > 3)
throw MakeException(SR.TooManyAlternates);
}
else
{
_alternation.AddChild(_concatenation.ReverseLeft());
_group.AddChild(_alternation);
}
_unit = _group;
}
/*
* Remember the pushed state (in response to a ')')
*/
internal void PopGroup()
{
_concatenation = _stack;
_alternation = _concatenation._next;
_group = _alternation._next;
_stack = _group._next;
// The first () inside a Testgroup group goes directly to the group
if (_group.Type() == RegexNode.Testgroup && _group.ChildCount() == 0)
{
if (_unit == null)
throw MakeException(SR.IllegalCondition);
_group.AddChild(_unit);
_unit = null;
}
}
/*
* Simple parsing for replacement patterns
*/
internal RegexNode ScanReplacement()
{
int c;
int startpos;
_concatenation = new RegexNode(RegexNode.Concatenate, _options);
for (;;)
{
c = CharsRight();
if (c == 0)
break;
startpos = Textpos();
while (c > 0 && RightChar() != '$')
{
MoveRight();
c--;
}
AddConcatenate(startpos, Textpos() - startpos, true);
if (c > 0)
{
if (MoveRightGetChar() == '$')
AddUnitNode(ScanDollar());
AddConcatenate();
}
}
return _concatenation;
}
/// <summary>
/// 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.
/// </summary>
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)));
}
}