private static RegExp MakeIntersection(RegExp exp1, RegExp exp2)
{
var r = new RegExp();
r.kind = Kind.RegexpIntersection;
r.exp1 = exp1;
r.exp2 = exp2;
return(r);
}
private static RegExp MakeRepeat(RegExp exp, int min)
{
var r = new RegExp();
r.kind = Kind.RegexpRepeatMin;
r.exp1 = exp;
r.min = min;
return(r);
}
private RegExp ParseComplExp()
{
if (this.Check(RegExpSyntaxOptions.Complement) && this.Match('~'))
{
return(RegExp.MakeComplement(this.ParseComplExp()));
}
return(this.ParseCharClassExp());
}
private static RegExp MakeCharRange(char from, char to)
{
var r = new RegExp();
r.kind = Kind.RegexpCharRange;
r.from = from;
r.to = to;
return(r);
}
private static RegExp MakeRepeat(RegExp exp, int min, int max)
{
var r = new RegExp();
r.Kind = Kind.RegexpRepeatMinMax;
r.Expr1 = exp;
r.Min = min;
r.Max = max;
return(r);
}
private static RegExp MakeInterval(int min, int max, int digits)
{
var r = new RegExp();
r.kind = Kind.RegexpInterval;
r.min = min;
r.max = max;
r.digits = digits;
return(r);
}
private RegExp ParseCharClasses()
{
RegExp e = this.ParseCharClass();
while (this.More() && !this.Peek("]"))
{
e = RegExp.MakeUnion(e, this.ParseCharClass());
}
return(e);
}
private RegExp ParseConcatExp()
{
RegExp e = this.ParseRepeatExp();
if (this.More() && !this.Peek(")|") && (!this.Check(RegExpSyntaxOptions.Intersection) || !this.Peek("&")))
{
e = RegExp.MakeConcatenation(e, this.ParseConcatExp());
}
return(e);
}
private RegExp ParseUnionExp()
{
RegExp e = this.ParseInterExp();
if (this.Match('|'))
{
e = RegExp.MakeUnion(e, this.ParseUnionExp());
}
return(e);
}
private RegExp ParseInterExp()
{
RegExp e = this.ParseConcatExp();
if (this.Check(RegExpSyntaxOptions.Intersection) && this.Match('&'))
{
e = RegExp.MakeIntersection(e, this.ParseInterExp());
}
return(e);
}
public void DumpRegEx()
{
string rawRegEx = @"\d{3}\d3}";
Fare.RegExp e = new Fare.RegExp(rawRegEx);
_testOutput.WriteLine(e.ToString());
// fail test to get output
Assert.True(false);
}
public void DumpDot()
{
string rawRegEx = @"\d{3}-\d{2}-(a|q)J";
rawRegEx = @"0x[0-9A-Fa-f]{4}";
Fare.RegExp e = new Fare.RegExp(rawRegEx);
var a = e.ToAutomaton();
_testOutput.WriteLine($"/* {rawRegEx} */");
_testOutput.WriteLine(Fare.DotFormatter.ToDot(a));
// fail test to get output
Assert.True(false);
}
private RegExp ParseCharClass()
{
char @char = this.ParseCharExp();
if (this.Match('-'))
{
if (this.Peek("]"))
{
return(RegExp.MakeUnion(RegExp.MakeChar(@char), RegExp.MakeChar('-')));
}
return(RegExp.MakeCharRange(@char, this.ParseCharExp()));
}
return(RegExp.MakeChar(@char));
}
private void FindLeaves(
RegExp exp,
Kind regExpKind,
IList <Automaton> list,
IDictionary <String, Automaton> automata,
IAutomatonProvider automatonProvider,
bool minimize)
{
if (exp.kind == regExpKind)
{
this.FindLeaves(exp.exp1, regExpKind, list, automata, automatonProvider, minimize);
this.FindLeaves(exp.exp2, regExpKind, list, automata, automatonProvider, minimize);
}
else
{
list.Add(exp.ToAutomaton(automata, automatonProvider, minimize));
}
}
private Automaton ToAutomatonAllowMutate(
IDictionary <string, Automaton> automata,
IAutomatonProvider automatonProvider,
bool minimize)
{
bool @bool = false;
if (allowMutation)
{
@bool = RegExp.SetAllowMutate(true); // This is not thead safe.
}
Automaton a = this.ToAutomaton(automata, automatonProvider, minimize);
if (allowMutation)
{
RegExp.SetAllowMutate(@bool);
}
return(a);
}
public void DumpAST()
{
string rawRegEx = @"\d{3}\d3}";
Fare.RegExp e = new Fare.RegExp(rawRegEx);
var a = e.ToAutomaton();
_testOutput.WriteLine($"expr: {rawRegEx} has {a.NumberOfStates} states and {a.NumberOfTransitions} transistions");
foreach (var state in a.GetStates())
{
_testOutput.WriteLine($"\tState (Id:{state.Id}, Number:{state.Number}, Accept: {state.Accept}) has {state.Transitions.Count} transitions");
foreach (var trans in state.GetSortedTransitions(false))
{
_testOutput.WriteLine($"\t\tTransition to: {trans.To.Id} [{trans.Min} - {trans.Max}]");
}
}
// fail test to get output
Assert.True(false);
}
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()));
}
/// <summary>
/// Initializes a new instance of the <see cref="Xeger"/> class.
/// </summary>
/// <param name="regex">The regex.</param>
/// <param name="random">The random.</param>
/// <param name="anyCharAlphabet">The list of characters used for computing the possible values for classes "." "\s", "\d", "\w" (and "\S", "\D", "\W"). It does not check explicitly defined chars in regexp.</param>
public Xeger(string regex, Random random, string anyCharAlphabet = null)
{
if (string.IsNullOrEmpty(regex))
{
throw new ArgumentNullException("regex");
}
if (random == null)
{
throw new ArgumentNullException("random");
}
if (anyCharAlphabet != null)
{
this.anyCharAlphabet = anyCharAlphabet;
}
regex = RemoveStartEndMarkers(regex);
var rx = new RegExp(regex, anyCharAlphabet, AllExceptAnyString);
this.UsedAlphabet = rx.UsedAlphabet();
this.automaton = rx.ToAutomaton();
this.random = random;
}
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 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()));
}
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 static RegExp MakeAnyFromCharset(string charset)
{
var rmajor = new RegExp($"[{charset}]", null);
return(rmajor);
}
private static RegExp ExcludeChars(RegExp exclusion, RegExp allChars)
{
return(MakeIntersection(allChars, MakeComplement(exclusion)));
}
private RegExp ParseRepeatExp()
{
RegExp e = this.ParseComplExp();
while (this.Peek("?*+{"))
{
if (this.Match('?'))
{
e = RegExp.MakeOptional(e);
}
else if (this.Match('*'))
{
e = RegExp.MakeRepeat(e);
}
else if (this.Match('+'))
{
e = RegExp.MakeRepeat(e, 1);
}
else if (this.Match('{'))
{
int start = pos;
while (this.Peek("0123456789"))
{
this.Next();
}
if (start == pos)
{
throw new ArgumentException("integer expected at position " + pos);
}
int n = int.Parse(b.Substring(start, pos - start));
int m = -1;
if (this.Match(','))
{
start = pos;
while (this.Peek("0123456789"))
{
this.Next();
}
if (start != pos)
{
m = int.Parse(b.Substring(start, pos - start));
}
}
else
{
m = n;
}
if (!this.Match('}'))
{
throw new ArgumentException("expected '}' at position " + pos);
}
e = m == -1 ? RegExp.MakeRepeat(e, n) : RegExp.MakeRepeat(e, n, m);
}
}
return(e);
}
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()));
}