To iterate over code points, use a loop like this: *
* UnicodeSetIterator it(set);
* while (set.next()) {
* if (set.codepoint != UnicodeSetIterator::IS_STRING) {
* processCodepoint(set.codepoint);
* } else {
* processString(set.string);
* }
* }
*
*
* To iterate over code point ranges, use a loop like this: *
* UnicodeSetIterator it(set);
* while (set.nextRange()) {
* if (set.codepoint != UnicodeSetIterator::IS_STRING) {
* processCodepointRange(set.codepoint, set.codepointEnd);
* } else {
* processString(set.string);
* }
* }
*
* @author M. Davis
* @stable ICU 2.0
*/
public class UnicodeSetIterator {
/**
* Value of codepoint if the iterator points to a string.
* If codepoint == IS_STRING, then examine
* string for the current iteration result.
* @stable ICU 2.0
*/
public static int IS_STRING = -1;
/**
* Current code point, or the special value IS_STRING, if
* the iterator points to a string.
* @stable ICU 2.0
*/
public int codepoint;
/**
* When iterating over ranges using nextRange(),
* codepointEnd contains the inclusive end of the
* iteration range, if codepoint != IS_STRING. If
* iterating over code points using next(), or if
* codepoint == IS_STRING, then the value of
* codepointEnd is undefined.
* @stable ICU 2.0
*/
public int codepointEnd;
/**
* If codepoint == IS_STRING, then string points
* to the current string. If codepoint != IS_STRING, the
* value of string is undefined.
* @stable ICU 2.0
*/
public String string;
/**
* Create an iterator over the given set.
* @param set set to iterate over
* @stable ICU 2.0
*/
public UnicodeSetIterator(UnicodeSet set) {
reset(set);
}
/**
* Returns the next element in the set, either a code point range
* or a string. If there are no more elements in the set, return
* false. If codepoint == IS_STRING, the value is a
* string in the string field. Otherwise the value is a
* range of one or more code points from codepoint to
* codepointeEnd inclusive.
*
* The order of iteration is all code points ranges in sorted * order, followed by all strings sorted order. Ranges are * disjoint and non-contiguous. string is undefined * unless codepoint == IS_STRING. Do not mix calls to * next() and nextRange() without calling * reset() between them. The results of doing so are * undefined. * * @return true if there was another element in the set and this * object contains the element. * @stable ICU 2.0 */ public boolean nextRange() { if (nextElement <= endElement) { codepointEnd = endElement; codepoint = nextElement; nextElement = endElement+1; return true; } if (range < endRange) { loadRange(++range); codepointEnd = endElement; codepoint = nextElement; nextElement = endElement+1; return true; } // stringIterator == null iff there are no string elements remaining if (stringIterator == null) return false; codepoint = IS_STRING; // signal that value is actually a string string = (String)stringIterator.next(); if (!stringIterator.hasNext()) stringIterator = null; return true; } /** * Sets this iterator to visit the elements of the given set and * resets it to the start of that set. The iterator is valid only * so long as set is valid. * @param set the set to iterate over. * @stable ICU 2.0 */ public void reset(UnicodeSet uset) { set = uset; reset(); } /** * Resets this iterator to the start of the set. * @stable ICU 2.0 */ public void reset() { endRange = set.getRangeCount() - 1; range = 0; endElement = -1; nextElement = 0; if (endRange >= 0) { loadRange(range); } stringIterator = null; if (set.strings != null) { stringIterator = set.strings.iterator(); if (!stringIterator.hasNext()) stringIterator = null; } } // ======================= PRIVATES =========================== private UnicodeSet set; private int endRange = 0; private int range = 0; /** * @internal */ protected int endElement; /** * @internal */ protected int nextElement; private Iterator stringIterator = null; /** * Invariant: stringIterator is null when there are no (more) strings remaining */ /** * @internal */ protected void loadRange(int aRange) { nextElement = set.getRangeStart(aRange); endElement = set.getRangeEnd(aRange); } }