56 * }
57 * </pre>
58 *
59 * <p>To iterate over code point ranges, use a loop like this:
60 * <pre>
61 * UnicodeSetIterator it(set);
62 * while (set.nextRange()) {
63 * if (set.codepoint != UnicodeSetIterator::IS_STRING) {
64 * processCodepointRange(set.codepoint, set.codepointEnd);
65 * } else {
66 * processString(set.string);
67 * }
68 * }
69 * </pre>
70 * @author M. Davis
71 * @stable ICU 2.0
72 */
73 public class UnicodeSetIterator {
74
75 /**
76 * Value of <tt>codepoint</tt> if the iterator points to a string.
77 * If <tt>codepoint == IS_STRING</tt>, then examine
78 * <tt>string</tt> for the current iteration result.
79 * @stable ICU 2.0
80 */
81 public static int IS_STRING = -1;
82
83 /**
84 * Current code point, or the special value <tt>IS_STRING</tt>, if
85 * the iterator points to a string.
86 * @stable ICU 2.0
87 */
88 public int codepoint;
89
90 /**
91 * When iterating over ranges using <tt>nextRange()</tt>,
92 * <tt>codepointEnd</tt> contains the inclusive end of the
93 * iteration range, if <tt>codepoint != IS_STRING</tt>. If
94 * iterating over code points using <tt>next()</tt>, or if
95 * <tt>codepoint == IS_STRING</tt>, then the value of
96 * <tt>codepointEnd</tt> is undefined.
97 * @stable ICU 2.0
98 */
99 public int codepointEnd;
100
101 /**
102 * If <tt>codepoint == IS_STRING</tt>, then <tt>string</tt> points
103 * to the current string. If <tt>codepoint != IS_STRING</tt>, the
104 * value of <tt>string</tt> is undefined.
105 * @stable ICU 2.0
106 */
107 public String string;
108
109 /**
110 * Create an iterator over the given set.
111 * @param set set to iterate over
112 * @stable ICU 2.0
113 */
114 public UnicodeSetIterator(UnicodeSet set) {
115 reset(set);
116 }
117
118 /**
119 * Returns the next element in the set, either a code point range
120 * or a string. If there are no more elements in the set, return
121 * false. If <tt>codepoint == IS_STRING</tt>, the value is a
122 * string in the <tt>string</tt> field. Otherwise the value is a
123 * range of one or more code points from <tt>codepoint</tt> to
124 * <tt>codepointeEnd</tt> inclusive.
125 *
126 * <p>The order of iteration is all code points ranges in sorted
127 * order, followed by all strings sorted order. Ranges are
128 * disjoint and non-contiguous. <tt>string</tt> is undefined
129 * unless <tt>codepoint == IS_STRING</tt>. Do not mix calls to
130 * <tt>next()</tt> and <tt>nextRange()</tt> without calling
131 * <tt>reset()</tt> between them. The results of doing so are
132 * undefined.
133 *
134 * @return true if there was another element in the set and this
135 * object contains the element.
136 * @stable ICU 2.0
137 */
138 public boolean nextRange() {
139 if (nextElement <= endElement) {
140 codepointEnd = endElement;
141 codepoint = nextElement;
142 nextElement = endElement+1;
143 return true;
144 }
145 if (range < endRange) {
146 loadRange(++range);
147 codepointEnd = endElement;
148 codepoint = nextElement;
149 nextElement = endElement+1;
150 return true;
151 }
152
153 // stringIterator == null iff there are no string elements remaining
154
155 if (stringIterator == null) return false;
156 codepoint = IS_STRING; // signal that value is actually a string
157 string = stringIterator.next();
158 if (!stringIterator.hasNext()) stringIterator = null;
159 return true;
160 }
161
162 /**
163 * Sets this iterator to visit the elements of the given set and
164 * resets it to the start of that set. The iterator is valid only
165 * so long as <tt>set</tt> is valid.
166 * @param set the set to iterate over.
167 * @stable ICU 2.0
168 */
169 public void reset(UnicodeSet uset) {
170 set = uset;
171 reset();
172 }
173
174 /**
175 * Resets this iterator to the start of the set.
176 * @stable ICU 2.0
177 */
178 public void reset() {
179 endRange = set.getRangeCount() - 1;
180 range = 0;
181 endElement = -1;
182 nextElement = 0;
183 if (endRange >= 0) {
184 loadRange(range);
185 }
186 stringIterator = null;
|
56 * }
57 * </pre>
58 *
59 * <p>To iterate over code point ranges, use a loop like this:
60 * <pre>
61 * UnicodeSetIterator it(set);
62 * while (set.nextRange()) {
63 * if (set.codepoint != UnicodeSetIterator::IS_STRING) {
64 * processCodepointRange(set.codepoint, set.codepointEnd);
65 * } else {
66 * processString(set.string);
67 * }
68 * }
69 * </pre>
70 * @author M. Davis
71 * @stable ICU 2.0
72 */
73 public class UnicodeSetIterator {
74
75 /**
76 * Value of {@code codepoint} if the iterator points to a string.
77 * If {@code codepoint == IS_STRING}, then examine
78 * {@code string} for the current iteration result.
79 * @stable ICU 2.0
80 */
81 public static int IS_STRING = -1;
82
83 /**
84 * Current code point, or the special value {@code IS_STRING}, if
85 * the iterator points to a string.
86 * @stable ICU 2.0
87 */
88 public int codepoint;
89
90 /**
91 * When iterating over ranges using {@code nextRange()},
92 * {@code codepointEnd} contains the inclusive end of the
93 * iteration range, if {@code codepoint != IS_STRING}. If
94 * iterating over code points using {@code next()}, or if
95 * {@code codepoint == IS_STRING}, then the value of
96 * {@code codepointEnd} is undefined.
97 * @stable ICU 2.0
98 */
99 public int codepointEnd;
100
101 /**
102 * If {@code codepoint == IS_STRING}, then {@code string} points
103 * to the current string. If {@code codepoint != IS_STRING}, the
104 * value of {@code string} is undefined.
105 * @stable ICU 2.0
106 */
107 public String string;
108
109 /**
110 * Create an iterator over the given set.
111 * @param set set to iterate over
112 * @stable ICU 2.0
113 */
114 public UnicodeSetIterator(UnicodeSet set) {
115 reset(set);
116 }
117
118 /**
119 * Returns the next element in the set, either a code point range
120 * or a string. If there are no more elements in the set, return
121 * false. If {@code codepoint == IS_STRING}, the value is a
122 * string in the {@code string} field. Otherwise the value is a
123 * range of one or more code points from {@code codepoint} to
124 * {@code codepointeEnd} inclusive.
125 *
126 * <p>The order of iteration is all code points ranges in sorted
127 * order, followed by all strings sorted order. Ranges are
128 * disjoint and non-contiguous. {@code string} is undefined
129 * unless {@code codepoint == IS_STRING}. Do not mix calls to
130 * {@code next()} and {@code nextRange()} without calling
131 * {@code reset()} between them. The results of doing so are
132 * undefined.
133 *
134 * @return true if there was another element in the set and this
135 * object contains the element.
136 * @stable ICU 2.0
137 */
138 public boolean nextRange() {
139 if (nextElement <= endElement) {
140 codepointEnd = endElement;
141 codepoint = nextElement;
142 nextElement = endElement+1;
143 return true;
144 }
145 if (range < endRange) {
146 loadRange(++range);
147 codepointEnd = endElement;
148 codepoint = nextElement;
149 nextElement = endElement+1;
150 return true;
151 }
152
153 // stringIterator == null iff there are no string elements remaining
154
155 if (stringIterator == null) return false;
156 codepoint = IS_STRING; // signal that value is actually a string
157 string = stringIterator.next();
158 if (!stringIterator.hasNext()) stringIterator = null;
159 return true;
160 }
161
162 /**
163 * Sets this iterator to visit the elements of the given set and
164 * resets it to the start of that set. The iterator is valid only
165 * so long as {@code set} is valid.
166 * @param uset the set to iterate over.
167 * @stable ICU 2.0
168 */
169 public void reset(UnicodeSet uset) {
170 set = uset;
171 reset();
172 }
173
174 /**
175 * Resets this iterator to the start of the set.
176 * @stable ICU 2.0
177 */
178 public void reset() {
179 endRange = set.getRangeCount() - 1;
180 range = 0;
181 endElement = -1;
182 nextElement = 0;
183 if (endRange >= 0) {
184 loadRange(range);
185 }
186 stringIterator = null;
|