private void SaveGroups(ref RxContext cxt, int start, int end,
bool clear, out RxSavedGroups groups)
{
groups.Start = start;
groups.LastOpenCapture = cxt.LastOpenCapture;
groups.LastClosedCapture = cxt.LastClosedCapture;
groups.Data = new RxCapture[end - start];
for (int i = start; i < end; ++i)
{
if (clear || cxt.Captures.Length <= i)
groups.Data[i - start].Start = groups.Data[i - start].End = -1;
else
{
groups.Data[i - start] = cxt.Captures[i];
}
}
}
private bool MatchAt(Runtime runtime, string str, int pos,
out RxResult res)
{
int len = str.Length;
res.Matched = false;
res.Start = res.End = -1;
res.Captures = null;
res.StringCaptures = null;
RxContext cxt;
cxt.Pos = pos;
cxt.Groups = new List<RxGroup>();
cxt.States = new List<RxState>();
cxt.StateBacktrack = new List<int>();
if (Captures > 0)
cxt.Captures = new RxCapture[Captures];
else
cxt.Captures = null;
if (Saved > 0)
cxt.Saved = new int[Saved];
else
cxt.Saved = null;
cxt.LastOpenCapture = cxt.LastClosedCapture = -1;
for (int index = 0; index >= 0;)
{
switch (Ops[index].Number)
{
case Opcode.OpNumber.OP_RX_START_MATCH:
// do nothing for now
++index;
break;
case Opcode.OpNumber.OP_RX_EXACT:
case Opcode.OpNumber.OP_RX_EXACT_I:
{
var s = Exact[Ops[index].Index];
bool case_insensitive = Ops[index].Number == Opcode.OpNumber.OP_RX_EXACT_I;
if ( cxt.Pos + s.Length > len
|| string.Compare(str, cxt.Pos, s, 0, s.Length,
case_insensitive) != 0)
index = Backtrack(ref cxt);
else
{
cxt.Pos += s.Length;
++index;
}
break;
}
case Opcode.OpNumber.OP_RX_SAVE_POS:
{
cxt.Saved[Ops[index].Index] = cxt.Pos;
++index;
break;
}
case Opcode.OpNumber.OP_RX_RESTORE_POS:
{
cxt.Pos = cxt.Saved[Ops[index].Index];
++index;
break;
}
case Opcode.OpNumber.OP_RX_ANY_NONEWLINE:
{
if (cxt.Pos == len || str[cxt.Pos] == '\n')
index = Backtrack(ref cxt);
else
{
++cxt.Pos;
++index;
}
break;
}
case Opcode.OpNumber.OP_RX_ANY:
{
if (cxt.Pos == len)
index = Backtrack(ref cxt);
else
{
++cxt.Pos;
++index;
}
break;
}
case Opcode.OpNumber.OP_RX_CLASS:
{
var c = Classes[Ops[index].Index];
if (cxt.Pos < len)
{
char ch = str[cxt.Pos];
++cxt.Pos;
++index;
if (c.Exact.IndexOf(ch) >= 0)
break;
if ((c.Flags & Opcode.RX_CLASS_WORDS) != 0 && IsWord(ch))
break;
if ((c.Flags & Opcode.RX_CLASS_NOT_WORDS) != 0 && !IsWord(ch))
break;
if ((c.Flags & Opcode.RX_CLASS_DIGITS) != 0 && char.IsDigit(ch))
break;
if ((c.Flags & Opcode.RX_CLASS_NOT_DIGITS) != 0 && !char.IsDigit(ch))
break;
if ((c.Flags & Opcode.RX_CLASS_SPACES) != 0 && char.IsWhiteSpace(ch))
break;
if ((c.Flags & Opcode.RX_CLASS_NOT_SPACES) != 0 && !char.IsWhiteSpace(ch))
break;
}
index = Backtrack(ref cxt);
break;
}
case Opcode.OpNumber.OP_RX_BEGINNING:
{
if (cxt.Pos != 0)
index = Backtrack(ref cxt);
else
++index;
break;
}
case Opcode.OpNumber.OP_RX_START_GROUP:
{
var grp = new RxGroup(-1);
cxt.Groups.Add(grp);
cxt.StateBacktrack.Add(cxt.States.Count);
index = Targets[Ops[index].Index];
break;
}
case Opcode.OpNumber.OP_RX_BACKTRACK:
case Opcode.OpNumber.OP_RX_TRY:
{
var st = new RxState(cxt.Pos, Targets[Ops[index].Index],
cxt.Groups.Count);
cxt.States.Add(st);
++index;
break;
}
case Opcode.OpNumber.OP_RX_POP_STATE:
{
cxt.States.RemoveAt(cxt.States.Count - 1);
++index;
break;
}
case Opcode.OpNumber.OP_RX_FAIL:
{
index = Backtrack(ref cxt);
break;
}
case Opcode.OpNumber.OP_RX_QUANTIFIER:
{
var group = cxt.Groups[cxt.Groups.Count - 1];
var quant = Quantifiers[Ops[index].Index];
int lastMatch = group.LastMatch;
++index;
group.Count += 1;
group.LastMatch = cxt.Pos;
if (group.Count > 0 && quant.Group >= 0)
EndCapture(ref cxt, quant.Group);
// max repeat count
if (group.Count == quant.MaxCount)
{
cxt.Groups.RemoveAt(cxt.Groups.Count - 1);
break;
}
// zero-length match
if (cxt.Pos == lastMatch)
{
break;
}
RxSavedGroups gr;
if (group.Count == 0 || group.Count >= quant.MinCount)
SaveGroups(ref cxt, quant.SubgroupStart, quant.SubgroupEnd,
group.Count == 0,
out gr);
else
gr = new RxSavedGroups();
if (group.Count == 0 && quant.MinCount > 0)
{
// force failure on backtrack
var st = new RxState(cxt.Pos, -1, -1);
st.Groups = gr;
cxt.States.Add(st);
}
else if (quant.IsGreedy && group.Count >= quant.MinCount)
{
var st = new RxState(cxt.Pos, index, cxt.Groups.Count - 1);
st.Groups = gr;
cxt.States.Add(st);
}
cxt.Groups[cxt.Groups.Count - 1] = group;
// if nongreedy, match at least min
if (!quant.IsGreedy && group.Count >= quant.MinCount)
{
var st = new RxState(cxt.Pos, Targets[quant.To], cxt.Groups.Count);
st.Groups = gr;
cxt.States.Add(st);
break;
}
if (quant.Group >= 0)
StartCapture(ref cxt, quant.Group);
index = Targets[quant.To];
break;
}
case Opcode.OpNumber.OP_RX_CAPTURE_START:
{
StartCapture(ref cxt, Ops[index].Index);
++index;
break;
}
case Opcode.OpNumber.OP_RX_CAPTURE_END:
{
EndCapture(ref cxt, Ops[index].Index);
++index;
break;
}
case Opcode.OpNumber.OP_RX_ACCEPT:
{
for (int i = cxt.LastOpenCapture + 1; i < Ops[index].Index; ++i)
cxt.Captures[i].Start = cxt.Captures[i].End = -1;
if (cxt.Captures != null)
{
res.StringCaptures = new string[cxt.Captures.Length];
for (int i = 0; i < cxt.Captures.Length; ++i)
if (cxt.Captures[i].End != -1)
res.StringCaptures[i] =
str.Substring(cxt.Captures[i].Start,
cxt.Captures[i].End - cxt.Captures[i].Start);
}
res.Start = pos;
res.End = cxt.Pos;
res.Captures = cxt.Captures;
res.Matched = true;
index = -1;
break;
}
case Opcode.OpNumber.OP_JUMP:
index = Targets[Ops[index].Index];
break;
default:
throw new System.Exception("PANIC: unhandled opcode " + Ops[index].Number);
}
}
return res.Matched;
}
private void RestoreGroups(ref RxContext cxt, ref RxSavedGroups groups)
{
if (groups.Data == null)
return;
for (int i = 0; i < groups.Data.Length; ++i)
cxt.Captures[i + groups.Start] = groups.Data[i];
cxt.LastOpenCapture = groups.LastOpenCapture;
cxt.LastClosedCapture = groups.LastClosedCapture;
}
public RxState(int pos, int target, int group)
{
Pos = pos;
Target = target;
Group = group;
Groups = new RxSavedGroups();
}