com.ibm.icu.text.UnicodeSetIterator Maven / Gradle / Ivy
Show all versions of icu4j Show documentation
/*
*******************************************************************************
* Copyright (C) 1996-2010, International Business Machines Corporation and *
* others. All Rights Reserved. *
*******************************************************************************
*/
package com.ibm.icu.text;
import java.util.Iterator;
/**
* UnicodeSetIterator iterates over the contents of a UnicodeSet. It
* iterates over either code points or code point ranges. After all
* code points or ranges have been returned, it returns the
* multicharacter strings of the UnicodSet, if any.
*
* To iterate over code points and multicharacter strings,
* use a loop like this:
*
* for (UnicodeSetIterator it = new UnicodeSetIterator(set); it.next();) {
* processString(it.getString());
* }
*
*
* To iterate over code point ranges, use a loop like this:
*
* for (UnicodeSetIterator it = new UnicodeSetIterator(set); it.nextRange();) {
* if (it.codepoint != UnicodeSetIterator.IS_STRING) {
* processCodepointRange(it.codepoint, it.codepointEnd);
* } else {
* processString(it.getString());
* }
* }
*
* @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);
}
/**
* Create an iterator over nothing. next() and
* nextRange() return false. This is a convenience
* constructor allowing the target to be set later.
* @stable ICU 2.0
*/
public UnicodeSetIterator() {
reset(new UnicodeSet());
}
/**
* 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 2.0
*/
public boolean next() {
if (nextElement <= endElement) {
codepoint = codepointEnd = nextElement++;
return true;
}
if (range < endRange) {
loadRange(++range);
codepoint = codepointEnd = nextElement++;
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 = stringIterator.next();
if (!stringIterator.hasNext()) {
stringIterator = null;
}
return true;
}
/**
* 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 = 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 uset 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;
}
}
}
/**
* Gets the current string from the iterator. Only use after calling next(), not nextRange().
* @stable ICU 4.0
*/
public String getString() {
if (codepoint != IS_STRING) {
return UTF16.valueOf(codepoint);
}
return string;
}
// ======================= PRIVATES ===========================
private UnicodeSet set;
private int endRange = 0;
private int range = 0;
/**
* @internal
* @deprecated This API is ICU internal only.
*/
public UnicodeSet getSet() {
return set;
}
/**
* @internal
* @deprecated This API is ICU internal only.
*/
protected int endElement;
/**
* @internal
* @deprecated This API is ICU internal only.
*/
protected int nextElement;
private Iterator stringIterator = null;
/**
* Invariant: stringIterator is null when there are no (more) strings remaining
*/
/**
* @internal
* @deprecated This API is ICU internal only.
*/
protected void loadRange(int aRange) {
nextElement = set.getRangeStart(aRange);
endElement = set.getRangeEnd(aRange);
}
}