public static RegexNode operator %(RegexNode node1, RegexNode node2)
{
var left = node1.Exp as CharSetExpression;
var right = node2.Exp as CharSetExpression;
Debug.Assert(left != null && right != null && !left.IsReverse && !right.IsReverse);
var expression = new CharSetExpression();
expression.IsReverse = false;
foreach (var r in left.Ranges)
{
expression.Ranges.Add(r);
}
foreach (var r in right.Ranges)
{
if (!expression.AddRangeWithConflict(r.Begin, r.End))
{
Debug.Assert(false, "Failed");
}
}
return(new RegexNode(expression));
}
public static RegexNode GetCharSetExpression(char a, char b)
{
var expression = new CharSetExpression();
expression.AddRangeWithConflict(a, b);
return(new RegexNode(expression));
}
public static RegexNode operator %(RegexNode node1, RegexNode node2)
{
var left = node1.Exp as CharSetExpression;
var right = node2.Exp as CharSetExpression;
Debug.Assert(left != null && right != null && !left.IsReverse && !right.IsReverse);
var expression = new CharSetExpression();
expression.IsReverse = false;
foreach (var r in left.Ranges)
{
expression.Ranges.Add(r);
}
foreach(var r in right.Ranges)
{
if(!expression.AddRangeWithConflict(r.Begin,r.End))
{
Debug.Assert(false, "Failed");
}
}
return new RegexNode(expression);
}
//be responsible for ^ $ . \b(etc.) [a-z] and any character
public Expression ParseCharSet()
{
if (_sourceWindow.PeekChar() == SlidingTextWindow.InvalidCharacter)
{
return(null);
}
if (_sourceWindow.AdvanceIfMatches('^'))
{
return(new BeginExpression());
}
else if (_sourceWindow.AdvanceIfMatches('$'))
{
return(new EndExpression());
}
else if (_sourceWindow.AdvanceIfMatches('.'))
{
var ret = new CharSetExpression();
ret.Add((char)1, char.MaxValue);
return(ret);
}
else if (_sourceWindow.AdvanceIfMatches('\\') || _sourceWindow.AdvanceIfMatches('/'))
{
//\d etc.
var exp = new CharSetExpression();
var c2 = _sourceWindow.PeekChar();
switch (c2)
{
case 'r':
exp.Add('\r');
break;
case 'n':
exp.Add('\n');
break;
case 't':
exp.Add('\t');
break;
//需要转义的字符在这里
case '\\':
case '/':
case '(':
case ')':
case '+':
case '*':
case '?':
case '|':
case '{':
case '}':
case '[':
case ']':
case '<':
case '>':
case '^':
case '$':
case '!':
case '=':
case '.':
exp.Add(c2);
break;
case 'S':
exp.IsReverse = true;
goto case 's';
case 's':
//spaces
exp.Add(' ');
exp.Add('\r');
exp.Add('\n');
exp.Add('\t');
break;
case 'D':
exp.IsReverse = true;
goto case 'd';
case 'd':
exp.Add('0', '9');
break;
case 'L':
exp.IsReverse = true;
goto case 'l';
case 'l':
exp.Add('_');
exp.Add('A', 'Z');
exp.Add('a', 'z');
break;
case 'W':
exp.IsReverse = true;
goto case 'w';
case 'w':
exp.Add('0', '9');
exp.Add('_');
exp.Add('A', 'Z');
exp.Add('a', 'z');
break;
default:
throw new ArgumentException("Error character after \\");
}
_sourceWindow.AdvanceChar();
return(exp);
}
//stuff like [a-z]
else if (_sourceWindow.AdvanceIfMatches('['))
{
var exp = new CharSetExpression();
exp.IsReverse = _sourceWindow.AdvanceIfMatches('^');
bool midState = false;
char lhs = default(char), rhs = default(char);
while (true)
{
if (_sourceWindow.AdvanceIfMatches('\\') || _sourceWindow.AdvanceIfMatches('/'))
{
var c = _sourceWindow.PeekChar();
char tmp = default(char);
switch (c)
{
case 'r':
tmp = '\r';
break;
case 'n':
tmp = '\n';
break;
case 't':
tmp = '\t';
break;
//需要转义的字符在这里
case '\\':
case '/':
case '(':
case ')':
case '+':
case '*':
case '?':
case '|':
case '{':
case '}':
case '[':
case ']':
case '<':
case '>':
case '^':
case '$':
case '!':
case '=':
case '.':
tmp = c;
break;
default:
throw new ArgumentException("Error syntax in []");
}
_sourceWindow.AdvanceChar();
if (midState)
{
rhs = c;
}
else
{
lhs = c;
}
midState = !midState;
}
else if (_sourceWindow.AdvanceIfMatches("-]"))
{
throw new ArgumentException("-] occurred.");
}
else
{
var c2 = _sourceWindow.NextChar();
if (_sourceWindow.IsValid(c2))
{
if (midState)
{
rhs = c2;
}
else
{
lhs = c2;
}
midState = !midState;
}
else
{
throw new ArgumentException("Error in []");
}
}
if (_sourceWindow.AdvanceIfMatches(']'))
{
if (midState)
{
rhs = lhs;
}
if (!exp.AddRangeWithConflict(lhs, rhs))
{
throw new ArgumentException();
}
break;
}
else if (_sourceWindow.AdvanceIfMatches('-'))
{
if (!midState)
{
throw new ArgumentException("Invalid - in []");
}
}
else
{
var c2 = _sourceWindow.NextChar();
if (_sourceWindow.IsValid(c2))
{
if (midState)
{
rhs = lhs;
}
if (exp.AddRangeWithConflict(lhs, rhs))
{
midState = false;
}
else
{
throw new ArgumentException();
}
}
}
}
return(exp);
}
else
{
char c = default(char);
if (_sourceWindow.AdvanceIfOneOf("()+*?{}|", out c))
{
_sourceWindow.AdvanceChar(-1);
return(null);
}
else
{
//the character itself
var exp = new CharSetExpression();
exp.IsReverse = false;
exp.Add(_sourceWindow.NextChar());
return(exp);
}
}
}
//be responsible for ^ $ . \b(etc.) [a-z] and any character
public Expression ParseCharSet()
{
if (_sourceWindow.PeekChar() == SlidingTextWindow.InvalidCharacter)
return null;
if (_sourceWindow.AdvanceIfMatches('^'))
return new BeginExpression();
else if (_sourceWindow.AdvanceIfMatches('$'))
return new EndExpression();
else if (_sourceWindow.AdvanceIfMatches('.'))
{
var ret = new CharSetExpression();
ret.Add((char)1,char.MaxValue);
return ret;
}
else if (_sourceWindow.AdvanceIfMatches('\\') || _sourceWindow.AdvanceIfMatches('/'))
{
//\d etc.
var exp = new CharSetExpression();
var c2 = _sourceWindow.PeekChar();
switch (c2)
{
case 'r':
exp.Add('\r');
break;
case 'n':
exp.Add('\n');
break;
case 't':
exp.Add('\t');
break;
//需要转义的字符在这里
case '\\':
case '/':
case '(':
case ')':
case '+':
case '*':
case '?':
case '|':
case '{':
case '}':
case '[':
case ']':
case '<':
case '>':
case '^':
case '$':
case '!':
case '=':
case '.':
exp.Add(c2);
break;
case 'S':
exp.IsReverse = true;
goto case 's';
case 's':
//spaces
exp.Add(' ');
exp.Add('\r');
exp.Add('\n');
exp.Add('\t');
break;
case 'D':
exp.IsReverse = true;
goto case 'd';
case 'd':
exp.Add('0', '9');
break;
case 'L':
exp.IsReverse = true;
goto case 'l';
case 'l':
exp.Add('_');
exp.Add('A', 'Z');
exp.Add('a', 'z');
break;
case 'W':
exp.IsReverse = true;
goto case 'w';
case 'w':
exp.Add('0', '9');
exp.Add('_');
exp.Add('A', 'Z');
exp.Add('a', 'z');
break;
default:
throw new ArgumentException("Error character after \\");
}
_sourceWindow.AdvanceChar();
return exp;
}
//stuff like [a-z]
else if (_sourceWindow.AdvanceIfMatches('['))
{
var exp = new CharSetExpression();
exp.IsReverse = _sourceWindow.AdvanceIfMatches('^');
bool midState = false;
char lhs = default(char), rhs = default(char);
while (true)
{
if (_sourceWindow.AdvanceIfMatches('\\') || _sourceWindow.AdvanceIfMatches('/'))
{
var c = _sourceWindow.PeekChar();
char tmp = default(char);
switch (c)
{
case 'r':
tmp = '\r';
break;
case 'n':
tmp = '\n';
break;
case 't':
tmp = '\t';
break;
//需要转义的字符在这里
case '\\':
case '/':
case '(':
case ')':
case '+':
case '*':
case '?':
case '|':
case '{':
case '}':
case '[':
case ']':
case '<':
case '>':
case '^':
case '$':
case '!':
case '=':
case '.':
tmp = c;
break;
default:
throw new ArgumentException("Error syntax in []");
}
_sourceWindow.AdvanceChar();
if (midState)
rhs = c;
else
lhs = c;
midState = !midState;
}
else if (_sourceWindow.AdvanceIfMatches("-]"))
throw new ArgumentException("-] occurred.");
else
{
var c2 = _sourceWindow.NextChar();
if (_sourceWindow.IsValid(c2))
{
if (midState)
rhs = c2;
else
lhs = c2;
midState = !midState;
}
else throw new ArgumentException("Error in []");
}
if (_sourceWindow.AdvanceIfMatches(']'))
{
if (midState)
rhs = lhs;
if (!exp.AddRangeWithConflict(lhs, rhs))
throw new ArgumentException();
break;
}
else if (_sourceWindow.AdvanceIfMatches('-'))
{
if (!midState)
throw new ArgumentException("Invalid - in []");
}
else
{
var c2 = _sourceWindow.NextChar();
if (_sourceWindow.IsValid(c2))
{
if (midState)
{
rhs = lhs;
}
if (exp.AddRangeWithConflict(lhs, rhs))
midState = false;
else throw new ArgumentException();
}
}
}
return exp;
}
else
{
char c = default(char);
if (_sourceWindow.AdvanceIfOneOf("()+*?{}|",out c))
{
_sourceWindow.AdvanceChar(-1);
return null;
}
else
{
//the character itself
var exp = new CharSetExpression();
exp.IsReverse = false;
exp.Add(_sourceWindow.NextChar());
return exp;
}
}
}
public static RegexNode GetCharSetExpression(char a,char b)
{
var expression = new CharSetExpression();
expression.AddRangeWithConflict(a, b);
return new RegexNode(expression);
}
