/// <summary>
/// Initializes a new instance of the <see cref = "RegExp" /> class from a string.
/// </summary>
/// <param name = "s">A string with the regular expression.</param>
/// <param name = "syntaxFlags">Boolean 'or' of optional syntax constructs to be enabled.</param>
public RegExp(string s, RegExpSyntaxOptions syntaxFlags)
{
this.b = s;
this.flags = syntaxFlags;
RegExp e;
if (s.Length == 0)
{
e = RegExp.MakeString(string.Empty);
}
else
{
e = this.ParseUnionExp();
if (this.pos < b.Length)
{
throw new ArgumentException("end-of-string expected at position " + this.pos);
}
}
this.Kind = e.Kind;
this.Expr1 = e.Expr1;
this.Expr2 = e.Expr2;
this.SourceRegExpr = e.SourceRegExpr;
this.Char = e.Char;
this.Min = e.Min;
this.Max = e.Max;
this.Digits = e.Digits;
this.FromChar = e.FromChar;
this.ToChar = e.ToChar;
this.b = null;
}
/// <summary>
/// Initializes a new instance of the <see cref = "RegExp" /> class from a string.
/// </summary>
/// <param name = "s">A string with the regular expression.</param>
/// <param name = "anyCharAlphabet">A string with the regular expression.</param>
/// <param name = "syntaxFlags">Boolean 'or' of optional syntax constructs to be enabled.</param>
public RegExp(string s, string anyCharAlphabet, RegExpSyntaxOptions syntaxFlags)
{
this.b = s;
this.flags = syntaxFlags;
RegExp e;
if (anyCharAlphabet != null)
{
this.anyCharAlphabet = anyCharAlphabet;
}
if (s.Length == 0)
{
e = RegExp.MakeString(string.Empty);
}
else
{
e = this.ParseUnionExp();
if (this.pos < b.Length)
{
throw new ArgumentException("end-of-string expected at position " + this.pos);
}
}
this.kind = e.kind;
this.exp1 = e.exp1;
this.exp2 = e.exp2;
this.s = e.s;
this.c = e.c;
this.min = e.min;
this.max = e.max;
this.digits = e.digits;
this.from = e.from;
this.to = e.to;
this.b = null;
}
private static RegExp MakeConcatenation(RegExp exp1, RegExp exp2)
{
if ((exp1.kind == Kind.RegexpChar || exp1.kind == Kind.RegexpString) &&
(exp2.kind == Kind.RegexpChar || exp2.kind == Kind.RegexpString))
{
return(RegExp.MakeString(exp1, exp2));
}
var r = new RegExp();
r.kind = Kind.RegexpConcatenation;
if (exp1.kind == Kind.RegexpConcatenation &&
(exp1.exp2.kind == Kind.RegexpChar || exp1.exp2.kind == Kind.RegexpString) &&
(exp2.kind == Kind.RegexpChar || exp2.kind == Kind.RegexpString))
{
r.exp1 = exp1.exp1;
r.exp2 = RegExp.MakeString(exp1.exp2, exp2);
}
else if ((exp1.kind == Kind.RegexpChar || exp1.kind == Kind.RegexpString) &&
exp2.kind == Kind.RegexpConcatenation &&
(exp2.exp1.kind == Kind.RegexpChar || exp2.exp1.kind == Kind.RegexpString))
{
r.exp1 = RegExp.MakeString(exp1, exp2.exp1);
r.exp2 = exp2.exp2;
}
else
{
r.exp1 = exp1;
r.exp2 = exp2;
}
return(r);
}
/// <summary>
/// Initializes a new instance of the <see cref = "RegExp" /> class from a string.
/// </summary>
/// <param name = "s">A string with the regular expression.</param>
/// <param name = "syntaxFlags">Boolean 'or' of optional syntax constructs to be enabled.</param>
internal RegExp(string s, RegExpSyntaxOptions syntaxFlags)
{
this.b = RegExp.ReplaceShorthandCharacterClasses(s);
this.flags = syntaxFlags;
RegExp e;
if (s.Length == 0)
{
e = RegExp.MakeString(string.Empty);
}
else
{
e = this.ParseUnionExp();
if (this.pos < this.b.Length)
{
throw new ArgumentException("end-of-string expected at position " + this.pos);
}
}
this.kind = e.kind;
this.exp1 = e.exp1;
this.exp2 = e.exp2;
this.s = e.s;
this.c = e.c;
this.min = e.min;
this.max = e.max;
this.digits = e.digits;
this.from = e.from;
this.to = e.to;
this.b = null;
}
private static RegExp MakeString(RegExp exp1, RegExp exp2)
{
var sb = new StringBuilder();
if (exp1.kind == Kind.RegexpString)
{
sb.Append(exp1.s);
}
else
{
sb.Append(exp1.c);
}
if (exp2.kind == Kind.RegexpString)
{
sb.Append(exp2.s);
}
else
{
sb.Append(exp2.c);
}
return(RegExp.MakeString(sb.ToString()));
}
private static RegExp MakeString(RegExp exp1, RegExp exp2)
{
var sb = new StringBuilder();
if (exp1.Kind == Kind.RegexpString)
{
sb.Append(exp1.SourceRegExpr);
}
else
{
sb.Append(exp1.Char);
}
if (exp2.Kind == Kind.RegexpString)
{
sb.Append(exp2.SourceRegExpr);
}
else
{
sb.Append(exp2.Char);
}
return(RegExp.MakeString(sb.ToString()));
}
private RegExp ParseSimpleExp()
{
if (this.Match('.'))
{
return(MakeAnyPrintableASCIIChar(this.anyCharAlphabet));
}
if (this.Check(RegExpSyntaxOptions.Empty) && this.Match('#'))
{
return(RegExp.MakeEmpty());
}
if (this.Check(RegExpSyntaxOptions.Anystring) && this.Match('@'))
{
return(RegExp.MakeAnyString());
}
if (this.Match('"'))
{
int start = pos;
while (this.More() && !this.Peek("\""))
{
this.Next();
}
if (!this.Match('"'))
{
throw new ArgumentException("expected '\"' at position " + pos);
}
return(RegExp.MakeString(b.Substring(start, ((pos - 1) - start))));
}
if (this.Match('('))
{
if (this.Match('?'))
{
this.SkipNonCapturingSubpatternExp();
}
if (this.Match(')'))
{
return(RegExp.MakeString(string.Empty));
}
RegExp e = this.ParseUnionExp();
if (!this.Match(')'))
{
throw new ArgumentException("expected ')' at position " + pos);
}
return(e);
}
if ((this.Check(RegExpSyntaxOptions.Automaton) || this.Check(RegExpSyntaxOptions.Interval)) && this.Match('<'))
{
int start = pos;
while (this.More() && !this.Peek(">"))
{
this.Next();
}
if (!this.Match('>'))
{
throw new ArgumentException("expected '>' at position " + pos);
}
string str = b.Substring(start, ((pos - 1) - start));
int i = str.IndexOf('-');
if (i == -1)
{
if (!this.Check(RegExpSyntaxOptions.Automaton))
{
throw new ArgumentException("interval syntax error at position " + (pos - 1));
}
return(RegExp.MakeAutomaton(str));
}
if (!this.Check(RegExpSyntaxOptions.Interval))
{
throw new ArgumentException("illegal identifier at position " + (pos - 1));
}
try
{
if (i == 0 || i == str.Length - 1 || i != str.LastIndexOf('-'))
{
throw new FormatException();
}
string smin = str.Substring(0, i - 0);
string smax = str.Substring(i + 1, (str.Length - (i + 1)));
int imin = int.Parse(smin);
int imax = int.Parse(smax);
int numdigits = smin.Length == smax.Length ? smin.Length : 0;
if (imin > imax)
{
int t = imin;
imin = imax;
imax = t;
}
return(RegExp.MakeInterval(imin, imax, numdigits));
}
catch (FormatException)
{
throw new ArgumentException("interval syntax error at position " + (pos - 1));
}
}
if (this.Match('\\'))
{
// Escaped '\' character.
if (this.Match('\\'))
{
return(MakeChar('\\'));
}
bool inclusion;
// Digits.
if ((inclusion = this.Match('d')) || this.Match('D'))
{
RegExp digitChars = MakeCharRange('0', '9');
return(inclusion
? this.anyCharAlphabet == null
? digitChars
: MakeAnyPrintableASCIIChar(new string((char[])this.anyCharAlphabet.ToCharArray().Where(char.IsDigit).ToArray()))
: ExcludeChars(digitChars, MakeAnyPrintableASCIIChar(this.anyCharAlphabet)));
}
// Whitespace chars only.
if ((inclusion = this.Match('s')) || this.Match('S'))
{
RegExp whitespaceChars = MakeUnion(MakeChar(' '), MakeChar('\t'));
return(inclusion
? this.anyCharAlphabet == null
? whitespaceChars
: MakeAnyPrintableASCIIChar(new string((char[])this.anyCharAlphabet.ToCharArray().Where(char.IsWhiteSpace).ToArray()))
: ExcludeChars(whitespaceChars, MakeAnyPrintableASCIIChar(this.anyCharAlphabet)));
}
// Word character. Range is [A-Za-z0-9_]
if ((inclusion = this.Match('w')) || this.Match('W'))
{
var ranges = new[] { MakeCharRange('A', 'Z'), MakeCharRange('a', 'z'), MakeCharRange('0', '9') };
RegExp wordChars = ranges.Aggregate(MakeChar('_'), MakeUnion);
return(inclusion
? this.anyCharAlphabet == null
? wordChars
: MakeAnyPrintableASCIIChar(new string((char[])this.anyCharAlphabet.ToCharArray()
.Where(x => char.IsLetter(x) || char.IsDigit(x) || x == '_').ToArray()))
: ExcludeChars(wordChars, MakeAnyPrintableASCIIChar(this.anyCharAlphabet)));
}
}
return(RegExp.MakeChar(this.ParseCharExp()));
}
private RegExp ParseSimpleExp()
{
if (this.Match('.'))
{
return(MakeAnyPrintableASCIIChar());
}
/* Issue 32, https://github.com/moodmosaic/Fare/issues/32
* The intent of the original code is a little unclear. The comment for the Empty value in the
* enum is 'Enables empty language.' Using '#' as token seems non-standard, and caused
* unhandled exception in some cases. The best option at this point is to remove handling of the
* Empty option until a proper implementation is proposed.
*/
// if (this.Check(RegExpSyntaxOptions.Empty) && this.Match('#'))
// {
// return RegExp.MakeEmpty();
// }
if (this.Check(RegExpSyntaxOptions.Anystring) && this.Match('@'))
{
return(RegExp.MakeAnyString());
}
if (this.Match('"'))
{
int start = pos;
while (this.More() && !this.Peek("\""))
{
this.Next();
}
if (!this.Match('"'))
{
throw new ArgumentException("expected '\"' at position " + pos);
}
return(RegExp.MakeString(b.Substring(start, ((pos - 1) - start))));
}
if (this.Match('('))
{
if (this.Match('?'))
{
this.SkipNonCapturingSubpatternExp();
}
if (this.Match(')'))
{
return(RegExp.MakeString(string.Empty));
}
RegExp e = this.ParseUnionExp();
if (!this.Match(')'))
{
throw new ArgumentException("expected ')' at position " + pos);
}
return(e);
}
if ((this.Check(RegExpSyntaxOptions.Automaton) || this.Check(RegExpSyntaxOptions.Interval)) && this.Match('<'))
{
int start = pos;
while (this.More() && !this.Peek(">"))
{
this.Next();
}
if (!this.Match('>'))
{
throw new ArgumentException("expected '>' at position " + pos);
}
string str = b.Substring(start, ((pos - 1) - start));
int i = str.IndexOf('-');
if (i == -1)
{
if (!this.Check(RegExpSyntaxOptions.Automaton))
{
throw new ArgumentException("interval syntax error at position " + (pos - 1));
}
return(RegExp.MakeAutomaton(str));
}
if (!this.Check(RegExpSyntaxOptions.Interval))
{
throw new ArgumentException("illegal identifier at position " + (pos - 1));
}
try
{
if (i == 0 || i == str.Length - 1 || i != str.LastIndexOf('-'))
{
throw new FormatException();
}
string smin = str.Substring(0, i - 0);
string smax = str.Substring(i + 1, (str.Length - (i + 1)));
int imin = int.Parse(smin, System.Globalization.NumberFormatInfo.CurrentInfo);
int imax = int.Parse(smax, System.Globalization.NumberFormatInfo.CurrentInfo);
int numdigits = smin.Length == smax.Length ? smin.Length : 0;
if (imin > imax)
{
int t = imin;
imin = imax;
imax = t;
}
return(RegExp.MakeInterval(imin, imax, numdigits));
}
catch (FormatException)
{
throw new ArgumentException("interval syntax error at position " + (pos - 1));
}
}
if (this.Match('\\'))
{
// Escaped '\' character.
if (this.Match('\\'))
{
return(MakeChar('\\'));
}
bool inclusion;
// Digits.
if ((inclusion = this.Match('d')) || this.Match('D'))
{
RegExp digitChars = MakeCharRange('0', '9');
return(inclusion ? digitChars : ExcludeChars(digitChars, MakeAnyPrintableASCIIChar()));
}
// Whitespace chars only.
if ((inclusion = this.Match('s')) || this.Match('S'))
{
// Do not add line breaks, as usually RegExp is single line.
RegExp whitespaceChars = MakeUnion(MakeChar(' '), MakeChar('\t'));
return(inclusion ? whitespaceChars : ExcludeChars(whitespaceChars, MakeAnyPrintableASCIIChar()));
}
// Word character. Range is [A-Za-z0-9_]
if ((inclusion = this.Match('w')) || this.Match('W'))
{
var ranges = new[] { MakeCharRange('A', 'Z'), MakeCharRange('a', 'z'), MakeCharRange('0', '9') };
RegExp wordChars = ranges.Aggregate(MakeChar('_'), MakeUnion);
return(inclusion ? wordChars : ExcludeChars(wordChars, MakeAnyPrintableASCIIChar()));
}
}
return(RegExp.MakeChar(this.ParseCharExp()));
}
private RegExp ParseSimpleExp()
{
if (this.Match('.'))
{
return(RegExp.MakeAnyChar());
}
if (this.Check(RegExpSyntaxOptions.Empty) && this.Match('#'))
{
return(RegExp.MakeEmpty());
}
if (this.Check(RegExpSyntaxOptions.Anystring) && this.Match('@'))
{
return(RegExp.MakeAnyString());
}
if (this.Match('"'))
{
int start = this.pos;
while (this.More() && !this.Peek("\""))
{
this.Next();
}
if (!this.Match('"'))
{
throw new ArgumentException("expected '\"' at position " + this.pos);
}
return(RegExp.MakeString(this.b.Substring(start, ((this.pos - 1) - start))));
}
if (this.Match('('))
{
if (this.Match('?'))
{
this.SkipNonCapturingSubpatternExp();
}
if (this.Match(')'))
{
return(RegExp.MakeString(string.Empty));
}
RegExp e = this.ParseUnionExp();
if (!this.Match(')'))
{
throw new ArgumentException("expected ')' at position " + this.pos);
}
return(e);
}
if ((this.Check(RegExpSyntaxOptions.Automaton) || this.Check(RegExpSyntaxOptions.Interval)) && this.Match('<'))
{
int start = this.pos;
while (this.More() && !this.Peek(">"))
{
this.Next();
}
if (!this.Match('>'))
{
throw new ArgumentException("expected '>' at position " + this.pos);
}
string str = this.b.Substring(start, ((this.pos - 1) - start));
int i = str.IndexOf('-');
if (i == -1)
{
if (!this.Check(RegExpSyntaxOptions.Automaton))
{
throw new ArgumentException("interval syntax error at position " + (this.pos - 1));
}
return(RegExp.MakeAutomaton(str));
}
if (!this.Check(RegExpSyntaxOptions.Interval))
{
throw new ArgumentException("illegal identifier at position " + (this.pos - 1));
}
try
{
if (i == 0 || i == str.Length - 1 || i != str.LastIndexOf('-'))
{
throw new FormatException();
}
string smin = str.Substring(0, i - 0);
string smax = str.Substring(i + 1, (str.Length - (i + 1)));
int imin = int.Parse(smin, CultureInfo.CurrentCulture);
int imax = int.Parse(smax, CultureInfo.CurrentCulture);
int numdigits = smin.Length == smax.Length ? smin.Length : 0;
if (imin > imax)
{
int t = imin;
imin = imax;
imax = t;
}
return(RegExp.MakeInterval(imin, imax, numdigits));
}
catch (FormatException)
{
throw new ArgumentException("interval syntax error at position " + (this.pos - 1));
}
}
return(RegExp.MakeChar(this.ParseCharExp()));
}
