SyntaxTree _parse(char[] expression) throws MalformedPatternException {
SyntaxTree tree;
__openParen = __closeParen = 0;
__regularExpression = expression;
__bytesRead = 0;
__expressionLength = expression.length;
__inCharacterClass = false;
__position = 0;
__match(__lookahead); // Call match to read first input.
if (__lookahead == '^') {
__beginAnchor = true;
__match(__lookahead);
}
if (__expressionLength > 0 && expression[__expressionLength - 1] == '$') {
--__expressionLength;
__endAnchor = true;
}
if (__expressionLength > 1 || (__expressionLength == 1 && !__beginAnchor)) {
CatNode root;
root = new CatNode();
root._left = __regex();
// end marker
root._right = new TokenNode((char) LeafNode._END_MARKER_TOKEN, __position++);
tree = new SyntaxTree(root, __position);
} else tree = new SyntaxTree(new TokenNode((char) LeafNode._END_MARKER_TOKEN, 0), 1);
tree._computeFollowPositions();
return tree;
}
private SyntaxNode __branch() throws MalformedPatternException {
CatNode current;
SyntaxNode left, root;
left = __piece();
if (__lookahead == ')') {
if (__openParen > __closeParen) return left;
else
throw new MalformedPatternException(
"Parse error: close parenthesis"
+ " without matching open parenthesis at position "
+ __bytesRead);
} else if (__lookahead == '|' || __lookahead == _END_OF_INPUT) return left;
root = current = new CatNode();
current._left = left;
while (true) {
left = __piece();
if (__lookahead == ')') {
if (__openParen > __closeParen) {
current._right = left;
break;
} else
throw new MalformedPatternException(
"Parse error: close parenthesis"
+ " without matching open parenthesis at position "
+ __bytesRead);
} else if (__lookahead == '|' || __lookahead == _END_OF_INPUT) {
current._right = left;
break;
}
current._right = new CatNode();
current = (CatNode) current._right;
current._left = left;
}
return root;
}
private SyntaxNode __repetition(SyntaxNode atom) throws MalformedPatternException {
int min, max, startPosition[];
SyntaxNode root = null;
CatNode catNode;
__match('{');
min = __parseUnsignedInteger(10, 1, Integer.MAX_VALUE);
startPosition = new int[1];
startPosition[0] = __position;
if (__lookahead == '}') {
// Match exactly min times. Concatenate the atom min times.
__match('}');
if (min == 0)
throw new MalformedPatternException(
"Parse error: Superfluous interval specified at position "
+ __bytesRead
+ ". Number of occurences was set to zero.");
if (min == 1) return atom;
root = catNode = new CatNode();
catNode._left = atom;
while (--min > 1) {
atom = atom._clone(startPosition);
catNode._right = new CatNode();
catNode = (CatNode) catNode._right;
catNode._left = atom;
}
catNode._right = atom._clone(startPosition);
} else if (__lookahead == ',') {
__match(',');
if (__lookahead == '}') {
// match at least min times
__match('}');
if (min == 0) return new StarNode(atom);
if (min == 1) return new PlusNode(atom);
root = catNode = new CatNode();
catNode._left = atom;
while (--min > 0) {
atom = atom._clone(startPosition);
catNode._right = new CatNode();
catNode = (CatNode) catNode._right;
catNode._left = atom;
}
catNode._right = new StarNode(atom._clone(startPosition));
} else {
// match at least min times and at most max times
max = __parseUnsignedInteger(10, 1, Integer.MAX_VALUE);
__match('}');
if (max < min)
throw new MalformedPatternException(
"Parse error: invalid interval; "
+ max
+ " is less than "
+ min
+ " at position "
+ __bytesRead);
if (max == 0)
throw new MalformedPatternException(
"Parse error: Superfluous interval specified at position "
+ __bytesRead
+ ". Number of occurences was set to zero.");
if (min == 0) {
if (max == 1) return new QuestionNode(atom);
root = catNode = new CatNode();
atom = new QuestionNode(atom);
catNode._left = atom;
while (--max > 1) {
atom = atom._clone(startPosition);
catNode._right = new CatNode();
catNode = (CatNode) catNode._right;
catNode._left = atom;
}
catNode._right = atom._clone(startPosition);
} else if (min == max) {
if (min == 1) return atom;
root = catNode = new CatNode();
catNode._left = atom;
while (--min > 1) {
atom = atom._clone(startPosition);
catNode._right = new CatNode();
catNode = (CatNode) catNode._right;
catNode._left = atom;
}
catNode._right = atom._clone(startPosition);
} else {
int count;
root = catNode = new CatNode();
catNode._left = atom;
for (count = 1; count < min; count++) {
atom = atom._clone(startPosition);
catNode._right = new CatNode();
catNode = (CatNode) catNode._right;
catNode._left = atom;
}
atom = new QuestionNode(atom._clone(startPosition));
count = max - min;
if (count == 1) catNode._right = atom;
else {
catNode._right = new CatNode();
catNode = (CatNode) catNode._right;
catNode._left = atom;
while (--count > 1) {
atom = atom._clone(startPosition);
catNode._right = new CatNode();
catNode = (CatNode) catNode._right;
catNode._left = atom;
}
catNode._right = atom._clone(startPosition);
}
}
}
} else
throw new MalformedPatternException(
"Parse error: unexpected character "
+ __lookahead
+ " in interval at position "
+ __bytesRead);
__position = startPosition[0];
return root;
}
