To iterate over code points, use a loop like this: + *
+ * UnicodeSetIterator it(set);
+ * while (set.next()) {
+ * if (set.isString()) {
+ * processString(set.getString());
+ * } else {
+ * processCodepoint(set.getCodepoint());
+ * }
+ * }
+ *
+ *
+ * To iterate over code point ranges, use a loop like this: + *
+ * UnicodeSetIterator it(set);
+ * while (it.nextRange()) {
+ * if (it.isString()) {
+ * processString(it.getString());
+ * } else {
+ * processCodepointRange(it.getCodepoint(), it.getCodepointEnd());
+ * }
+ * }
+ *
+ * @author M. Davis
+ * @stable ICU 2.4
+ */
+class U_COMMON_API UnicodeSetIterator : public UObject {
+
+ protected:
+
+ /**
+ * Value of codepoint if the iterator points to a string.
+ * If codepoint == IS_STRING, then examine
+ * string for the current iteration result.
+ * @stable ICU 2.4
+ */
+ enum { IS_STRING = -1 };
+
+ /**
+ * Current code point, or the special value IS_STRING, if
+ * the iterator points to a string.
+ * @stable ICU 2.4
+ */
+ UChar32 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 2.4
+ */
+ UChar32 codepointEnd;
+
+ /**
+ * If codepoint == IS_STRING, then string points
+ * to the current string. If codepoint != IS_STRING, the
+ * value of string is undefined.
+ * @stable ICU 2.4
+ */
+ const UnicodeString* string;
+
+ public:
+
+ /**
+ * Create an iterator over the given set. The iterator is valid
+ * only so long as set is valid.
+ * @param set set to iterate over
+ * @stable ICU 2.4
+ */
+ UnicodeSetIterator(const UnicodeSet& set);
+
+ /**
+ * Create an iterator over nothing. next() and
+ * nextRange() return false. This is a convenience
+ * constructor allowing the target to be set later.
+ * @stable ICU 2.4
+ */
+ UnicodeSetIterator();
+
+ /**
+ * Destructor.
+ * @stable ICU 2.4
+ */
+ virtual ~UnicodeSetIterator();
+
+ /**
+ * Returns true if the current element is a string. If so, the
+ * caller can retrieve it with getString(). If this
+ * method returns false, the current element is a code point or
+ * code point range, depending on whether next() or
+ * nextRange() was called, and the caller can retrieve it
+ * with getCodepoint() and, for a range,
+ * getCodepointEnd().
+ * @stable ICU 2.4
+ */
+ inline UBool isString() const;
+
+ /**
+ * Returns the current code point, if isString() returned
+ * false. Otherwise returns an undefined result.
+ * @stable ICU 2.4
+ */
+ inline UChar32 getCodepoint() const;
+
+ /**
+ * Returns the end of the current code point range, if
+ * isString() returned false and nextRange() was
+ * called. Otherwise returns an undefined result.
+ * @stable ICU 2.4
+ */
+ inline UChar32 getCodepointEnd() const;
+
+ /**
+ * Returns the current string, if isString() returned
+ * true. Otherwise returns an undefined result.
+ * @stable ICU 2.4
+ */
+ inline const UnicodeString& getString() const;
+
+ /**
+ * Returns the next element in the set, either a single code point
+ * 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
+ * single code point in the codepoint field.
+ *
+ * The order of iteration is all code points in sorted order, + * followed by all strings sorted order. codepointEnd is + * undefined after calling this method. 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 2.4 + */ + UBool next(); + + /** + * 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 2.4 + */ + UBool nextRange(); + + /** + * 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 2.4 + */ + void reset(const UnicodeSet& set); + + /** + * Resets this iterator to the start of the set. + * @stable ICU 2.4 + */ + void reset(); + + /** + * ICU "poor man's RTTI", returns a UClassID for this class. + * + * @stable ICU 2.4 + */ + static UClassID U_EXPORT2 getStaticClassID(); + + /** + * ICU "poor man's RTTI", returns a UClassID for the actual class. + * + * @stable ICU 2.4 + */ + virtual UClassID getDynamicClassID() const; + + // ======================= PRIVATES =========================== + + protected: + + // endElement and nextElements are really UChar32's, but we keep + // them as signed int32_t's so we can do comparisons with + // endElement set to -1. Leave them as int32_t's. + /** The set + * @stable ICU 2.4 + */ + const UnicodeSet* set; + /** End range + * @stable ICU 2.4 + */ + int32_t endRange; + /** Range + * @stable ICU 2.4 + */ + int32_t range; + /** End element + * @stable ICU 2.4 + */ + int32_t endElement; + /** Next element + * @stable ICU 2.4 + */ + int32_t nextElement; + //UBool abbreviated; + /** Next string + * @stable ICU 2.4 + */ + int32_t nextString; + /** String count + * @stable ICU 2.4 + */ + int32_t stringCount; + + /** Copy constructor. Disallowed. + * @stable ICU 2.4 + */ + UnicodeSetIterator(const UnicodeSetIterator&); // disallow + + /** Assignment operator. Disallowed. + * @stable ICU 2.4 + */ + UnicodeSetIterator& operator=(const UnicodeSetIterator&); // disallow + + /** Load range + * @stable ICU 2.4 + */ + virtual void loadRange(int32_t range); + +}; + +inline UBool UnicodeSetIterator::isString() const { + return codepoint == (UChar32)IS_STRING; +} + +inline UChar32 UnicodeSetIterator::getCodepoint() const { + return codepoint; +} + +inline UChar32 UnicodeSetIterator::getCodepointEnd() const { + return codepointEnd; +} + +inline const UnicodeString& UnicodeSetIterator::getString() const { + return *string; +} + +U_NAMESPACE_END + +#endif