public bool contains(IntCharSet set)
{
/*
* IntCharSet test = set.copy();
*
* test.sub(this);
*
* return (test.numIntervalls() == 0);
* /*/ int i = 0;
int j = 0;
while (j < set.intervalls.Count)
{
Interval x = (Interval)intervalls[i];
Interval y = (Interval)set.intervalls[j];
if (x.contains(y))
{
j++;
}
if (x.start > y.end)
{
return(false);
}
if (x.end < y.start)
{
i++;
}
}
return(true); /* */
}
/**
* Returns an array of all CharClassIntervalls in this
* char class collection.
*
* The array is ordered by char code, i.e.
* <code>result[i+1].start = result[i].end+1</code>
*
* Each CharClassInterval contains the number of the
* char class it belongs to.
*/
public CharClassInterval [] getIntervalls()
{
int i, c;
int size = classes.Count;
int numIntervalls = 0;
for (i = 0; i < size; i++)
{
numIntervalls += ((IntCharSet)classes[i]).numIntervalls();
}
CharClassInterval [] result = new CharClassInterval[numIntervalls];
i = 0;
c = 0;
while (i < numIntervalls)
{
int code = getClassCode((char)c);
IntCharSet set = (IntCharSet)classes[code];
Interval iv = set.getNext();
result[i++] = new CharClassInterval(iv.start, iv.end, code);
c = iv.end + 1;
}
return(result);
}
/**
* Check consistency of the stored classes [debug].
*
* all classes must be disjoint, checks if all characters
* have a class assigned.
*/
public void check()
{
for (int i = 0; i < classes.Count; i++)
{
for (int j = i + 1; j < classes.Count; j++)
{
IntCharSet x = (IntCharSet)classes[i];
IntCharSet y = (IntCharSet)classes[j];
if (x.and(y).containsElements())
{
Console.WriteLine("Error: non disjoint char classes {0} and {1}", i, j);
Console.WriteLine("class {0}: {1}", i, x);
Console.WriteLine("class {0}: {1}", j, y);
}
}
}
// check if each character has a classcode
// (= if getClassCode terminates)
for (char c = (char)0; c < maxChar; c++)
{
getClassCode(c);
if (c % 100 == 0)
{
Console.Write(".");
}
}
getClassCode(maxChar);
}
public IntCharSet add(IntCharSet set)
{
for (int i = 0; i < set.intervalls.Count; i++)
{
add((Interval)set.intervalls[i]);
}
return(this);
}
/* intersection */
public IntCharSet and(IntCharSet set)
{
if (DEBUG)
{
Out.dump("intersection");
Out.dump("this : " + this);
Out.dump("other : " + set);
}
IntCharSet result = new IntCharSet();
int i = 0; // index in this.intervalls
int j = 0; // index in set.intervalls
int size = intervalls.Count;
int setSize = set.intervalls.Count;
while (i < size && j < setSize)
{
Interval x = (Interval)this.intervalls[i];
Interval y = (Interval)set.intervalls[j];
if (x.end < y.start)
{
i++;
continue;
}
if (y.end < x.start)
{
j++;
continue;
}
result.intervalls.Add(
new Interval(
max(x.start, y.start),
min(x.end, y.end)
)
);
if (x.end >= y.end)
{
j++;
}
if (y.end >= x.end)
{
i++;
}
}
if (DEBUG)
{
Out.dump("result: " + result);
}
return(result);
}
/**
* Updates the current partition, so that the specified set of characters
* gets a new character class.
*
* Characters that are elements of <code>set</code> are not in the same
* equivalence class with characters that are not elements of <code>set</code>.
*
* @param set the set of characters to distinguish from the rest
* @param caseless if true upper/lower/title case are considered equivalent
*/
public void makeClass(IntCharSet set, bool caseless)
{
if (caseless)
{
set = set.getCaseless();
}
if (DEBUG)
{
Out.dump("makeClass(" + set + ")");
dump();
}
try
{
int oldSize = classes.Count;
for (int i = 0; i < oldSize; i++)
{
IntCharSet x = (IntCharSet)classes[i];
if (x.Equals(set))
{
return;
}
IntCharSet and = x.and(set);
if (and.containsElements())
{
if (x.Equals(and))
{
set.sub(and);
continue;
}
else if (set.Equals(and))
{
x.sub(and);
classes.Add(and);
return;
}
set.sub(and);
x.sub(and);
classes.Add(and);
}
}
}
finally
{
if (DEBUG)
{
Out.dump("makeClass(..) finished");
dump();
}
}
}
/**
* Returns the last set of the expression
*
* (the last-charater-projection of the language)
*/
private static IntCharSet last(RegExp re)
{
RegExp2 r;
switch (re.type)
{
case sym.BAR:
r = (RegExp2)re;
return(last(r.r1).add(last(r.r2)));
case sym.CONCAT:
r = (RegExp2)re;
if (containsEpsilon(r.r2))
{
return(last(r.r1).add(last(r.r2)));
}
else
{
return(last(r.r2));
}
case sym.STAR:
case sym.PLUS:
case sym.QUESTION:
return(last((RegExp)((RegExp1)re).content));
case sym.CCLASS:
return(new IntCharSet((ArrayList)((RegExp1)re).content));
case sym.CCLASSNOT:
IntCharSet all = new IntCharSet(new Interval((char)0, maxChar));
IntCharSet set = new IntCharSet((ArrayList)((RegExp1)re).content);
all.sub(set);
return(all);
case sym.CHAR:
return(new IntCharSet((char)((RegExp1)re).content));
case sym.STRING:
String content = (String)((RegExp1)re).content;
if (content.Length > 0)
{
return(new IntCharSet(content[content.Length - 1]));
}
else
{
return(new IntCharSet());
}
case sym.MACROUSE:
return(last(macros.getDefinition((String)((RegExp1)re).content)));
}
throw new Exception("Unkown expression type " + re.type + " in " + re); //$NON-NLS-1$ //$NON-NLS-2$
}
/**
* Return a (deep) copy of this char set
*
* @return the copy
*/
public IntCharSet copy()
{
IntCharSet result = new IntCharSet();
int size = intervalls.Count;
for (int i = 0; i < size; i++)
{
Interval iv = ((Interval)intervalls[i]).copy();
result.intervalls.Add(iv);
}
return(result);
}
/**
* Returns the code of the character class the specified character belongs to.
*/
public int getClassCode(char letter)
{
int i = -1;
while (true)
{
IntCharSet x = (IntCharSet)classes[++i];
if (x.contains(letter))
{
return(i);
}
}
}
/**
* Returns an array that contains the character class codes of all characters
* in the specified set of input characters.
*/
private int [] getClassCodes(IntCharSet set, bool negate)
{
if (DEBUG)
{
Out.dump("getting class codes for " + set);
if (negate)
{
Out.dump("[negated]");
}
}
int size = classes.Count;
// [fixme: optimize]
int[] temp = new int [size];
int length = 0;
for (int i = 0; i < size; i++)
{
IntCharSet x = (IntCharSet)classes[i];
if (negate)
{
if (!set.and(x).containsElements())
{
temp[length++] = i;
if (DEBUG)
{
Out.dump("code " + i);
}
}
}
else
{
if (set.and(x).containsElements())
{
temp[length++] = i;
if (DEBUG)
{
Out.dump("code " + i);
}
}
}
}
int[] result = new int [length];
Array.Copy(temp, 0, result, 0, length);
return(result);
}
/**
* Create a caseless version of this charset.
* <p>
* The caseless version contains all characters of this char set,
* and additionally all lower/upper/title case variants of the
* characters in this set.
*
* @return a caseless copy of this set
*/
public IntCharSet getCaseless()
{
IntCharSet n = copy();
int size = intervalls.Count;
for (int i = 0; i < size; i++)
{
Interval elem = (Interval)intervalls[i];
for (char c = elem.start; c <= elem.end; c++)
{
n.add(char.ToLower(c));
n.add(char.ToUpper(c));
//n.add(char.toTitleCase(c));
}
}
return(n);
}
/**
* o instanceof Interval
*/
public override bool Equals(Object o)
{
IntCharSet set = (IntCharSet)o;
if (intervalls.Count != set.intervalls.Count)
{
return(false);
}
for (int i = 0; i < intervalls.Count; i++)
{
if (!intervalls[i].Equals(set.intervalls[i]))
{
return(false);
}
}
return(true);
}
/* complement */
/* prec: this.contains(set), set != null */
public void sub(IntCharSet set)
{
if (DEBUG)
{
Out.dump("complement");
Out.dump("this : " + this);
Out.dump("other : " + set);
}
int i = 0; // index in this.intervalls
int j = 0; // index in set.intervalls
int setSize = set.intervalls.Count;
while (i < intervalls.Count && j < setSize)
{
Interval x = (Interval)this.intervalls[i];
Interval y = (Interval)set.intervalls[j];
if (DEBUG)
{
Out.dump("this : " + this);
Out.dump("this [" + i + "] : " + x);
Out.dump("other [" + j + "] : " + y);
}
if (x.end < y.start)
{
i++;
continue;
}
if (y.end < x.start)
{
j++;
continue;
}
// x.end >= y.start && y.end >= x.start ->
// x.end <= y.end && x.start >= y.start (prec)
if (x.start == y.start && x.end == y.end)
{
intervalls.RemoveAt(i);
j++;
continue;
}
// x.end <= y.end && x.start >= y.start &&
// (x.end < y.end || x.start > y.start) ->
// x.start < x.end
if (x.start == y.start)
{
x.start = (char)(y.end + 1);
j++;
continue;
}
if (x.end == y.end)
{
x.end = (char)(y.start - 1);
i++;
j++;
continue;
}
intervalls.Insert(i, new Interval(x.start, (char)(y.start - 1)));
x.start = (char)(y.end + 1);
i++;
j++;
}
if (DEBUG)
{
Out.dump("result: " + this);
}
}
