The UnicodeSet class is not designed to be subclassed. * *
UnicodeSet supports two APIs. The first is the
* operand API that allows the caller to modify the value of
* a UnicodeSet object. It conforms to Java 2's
* java.util.Set interface, although
* UnicodeSet does not actually implement that
* interface. All methods of Set are supported, with the
* modification that they take a character range or single character
* instead of an Object, and they take a
* UnicodeSet instead of a Collection. The
* operand API may be thought of in terms of boolean logic: a boolean
* OR is implemented by add, a boolean AND is implemented
* by retain, a boolean XOR is implemented by
* complement taking an argument, and a boolean NOT is
* implemented by complement with no argument. In terms
* of traditional set theory function names, add is a
* union, retain is an intersection, remove
* is an asymmetric difference, and complement with no
* argument is a set complement with respect to the superset range
* MIN_VALUE-MAX_VALUE
*
*
The second API is the
* applyPattern()/toPattern() API from the
* java.text.Format-derived classes. Unlike the
* methods that add characters, add categories, and control the logic
* of the set, the method applyPattern() sets all
* attributes of a UnicodeSet at once, based on a
* string pattern.
*
*
Pattern syntax
* * Patterns are accepted by the constructors and the *applyPattern() methods and returned by the
* toPattern() method. These patterns follow a syntax
* similar to that employed by version 8 regular expression character
* classes. Here are some simple examples:
*
* ** * Any character may be preceded by a backslash in order to remove any special * meaning. White space characters, as defined by the Unicode Pattern_White_Space property, are * ignored, unless they are escaped. * **
** * []No characters ** * [a]The character 'a' ** ** [ae]The characters 'a' and 'e' ** ** [a-e]The characters 'a' through 'e' inclusive, in Unicode code * point order ** ** [\\u4E01]The character U+4E01 ** ** [a{ab}{ac}]The character 'a' and the multicharacter strings "ab" and * "ac" ** ** [\p{Lu}]All characters in the general category Uppercase Letter *
Property patterns specify a set of characters having a certain * property as defined by the Unicode standard. Both the POSIX-like * "[:Lu:]" and the Perl-like syntax "\p{Lu}" are recognized. For a * complete list of supported property patterns, see the User's Guide * for UnicodeSet at * * http://www.icu-project.org/userguide/unicodeSet.html. * Actual determination of property data is defined by the underlying * Unicode database as implemented by UCharacter. * *
Patterns specify individual characters, ranges of characters, and * Unicode property sets. When elements are concatenated, they * specify their union. To complement a set, place a '^' immediately * after the opening '['. Property patterns are inverted by modifying * their delimiters; "[:^foo]" and "\P{foo}". In any other location, * '^' has no special meaning. * *
Ranges are indicated by placing two a '-' between two * characters, as in "a-z". This specifies the range of all * characters from the left to the right, in Unicode order. If the * left character is greater than or equal to the * right character it is a syntax error. If a '-' occurs as the first * character after the opening '[' or '[^', or if it occurs as the * last character before the closing ']', then it is taken as a * literal. Thus "[a\\-b]", "[-ab]", and "[ab-]" all indicate the same * set of three characters, 'a', 'b', and '-'. * *
Sets may be intersected using the {@literal '&'} operator or the asymmetric * set difference may be taken using the '-' operator, for example, * "{@code [[:L:]&[\\u0000-\\u0FFF]]}" indicates the set of all Unicode letters * with values less than 4096. Operators ({@literal '&'} and '|') have equal * precedence and bind left-to-right. Thus * "[[:L:]-[a-z]-[\\u0100-\\u01FF]]" is equivalent to * "[[[:L:]-[a-z]]-[\\u0100-\\u01FF]]". This only really matters for * difference; intersection is commutative. * *
[a] | The set containing 'a' * |
[a-z] | The set containing 'a' * through 'z' and all letters in between, in Unicode order * |
[^a-z] | The set containing * all characters but 'a' through 'z', * that is, U+0000 through 'a'-1 and 'z'+1 through U+10FFFF * |
[[pat1][pat2]]
* | The union of sets specified by pat1 and pat2 * |
[[pat1]&[pat2]]
* | The intersection of sets specified by pat1 and pat2 * |
[[pat1]-[pat2]]
* | The asymmetric difference of sets specified by pat1 and * pat2 * |
[:Lu:] or \p{Lu}
* | The set of characters having the specified * Unicode property; in * this case, Unicode uppercase letters * |
[:^Lu:] or \P{Lu}
* | The set of characters not having the given * Unicode property * |
Warning: you cannot add an empty string ("") to a UnicodeSet.
* *Formal syntax
* ****
** ** pattern :=* ('[' '^'? item* ']') | * property* ** item :=* char | (char '-' char) | pattern-expr
** ** pattern-expr :=* pattern | pattern-expr pattern | * pattern-expr op pattern
** ** op :=* '&' | '-'
** ** special :=* '[' | ']' | '-'
** ** char :=any character that is not *specialany character
* | ('\\')
* | ('\u' hex hex hex hex)
** ** hex :=any character for which * *Character.digit(c, 16)* returns a non-negative result* ** property :=a Unicode property set pattern *
**
** *Legend: **
** ** a := b* * amay be replaced byb* ** a?* zero or one instance of *a
** ** a** one or more instances of *a
** ** a | b* either *aorb
** ** 'a'* the literal string between the quotes *
To iterate over contents of UnicodeSet, the following are available: *
To replace, count elements, or delete spans, see {@link com.ibm.icu.text.UnicodeSetSpanner UnicodeSetSpanner}.
*
* @author Alan Liu
* @stable ICU 2.0
*/
public class UnicodeSet {
private static final int LOW = 0x000000; // LOW <= all valid values. ZERO for codepoints
private static final int HIGH = 0x110000; // HIGH > all valid values. 10000 for code units.
// 110000 for codepoints
/**
* Minimum value that can be stored in a UnicodeSet.
* @stable ICU 2.0
*/
public static final int MIN_VALUE = LOW;
/**
* Maximum value that can be stored in a UnicodeSet.
* @stable ICU 2.0
*/
public static final int MAX_VALUE = HIGH - 1;
private int len; // length used; list may be longer to minimize reallocs
private int[] list; // MUST be terminated with HIGH
private int[] rangeList; // internal buffer
private int[] buffer; // internal buffer
// NOTE: normally the field should be of type SortedSet; but that is missing a public clone!!
// is not private so that UnicodeSetIterator can get access
TreeSet To replace, count elements, or delete spans, see {@link com.ibm.icu.text.UnicodeSetSpanner UnicodeSetSpanner}.
* @param s The string to be spanned
* @param spanCondition The span condition
* @return the length of the span
* @stable ICU 4.4
*/
public int span(CharSequence s, SpanCondition spanCondition) {
return span(s, 0, spanCondition);
}
/**
* Span a string using this UnicodeSet.
* If the start index is less than 0, span will start from 0.
* If the start index is greater than the string length, span returns the string length.
* To replace, count elements, or delete spans, see {@link com.ibm.icu.text.UnicodeSetSpanner UnicodeSetSpanner}.
* @param s The string to be spanned
* @param start The start index that the span begins
* @param spanCondition The span condition
* @return the string index which ends the span (i.e. exclusive)
* @stable ICU 4.4
*/
public int span(CharSequence s, int start, SpanCondition spanCondition) {
int end = s.length();
if (start < 0) {
start = 0;
} else if (start >= end) {
return end;
}
if (bmpSet != null) {
// Frozen set without strings, or no string is relevant for span().
return bmpSet.span(s, start, spanCondition, null);
}
if (stringSpan != null) {
return stringSpan.span(s, start, spanCondition);
} else if (!strings.isEmpty()) {
int which = spanCondition == SpanCondition.NOT_CONTAINED ? UnicodeSetStringSpan.FWD_UTF16_NOT_CONTAINED
: UnicodeSetStringSpan.FWD_UTF16_CONTAINED;
UnicodeSetStringSpan strSpan = new UnicodeSetStringSpan(this, new ArrayList To replace, count elements, or delete spans, see {@link com.ibm.icu.text.UnicodeSetSpanner UnicodeSetSpanner}.
* @param outCount An output-only object (must not be null) for returning the count.
* @return the limit (exclusive end) of the span
*/
public int spanAndCount(CharSequence s, int start, SpanCondition spanCondition, OutputInt outCount) {
if (outCount == null) {
throw new IllegalArgumentException("outCount must not be null");
}
int end = s.length();
if (start < 0) {
start = 0;
} else if (start >= end) {
return end;
}
if (stringSpan != null) {
// We might also have bmpSet != null,
// but fully-contained strings are relevant for counting elements.
return stringSpan.spanAndCount(s, start, spanCondition, outCount);
} else if (bmpSet != null) {
return bmpSet.span(s, start, spanCondition, outCount);
} else if (!strings.isEmpty()) {
int which = spanCondition == SpanCondition.NOT_CONTAINED ? UnicodeSetStringSpan.FWD_UTF16_NOT_CONTAINED
: UnicodeSetStringSpan.FWD_UTF16_CONTAINED;
which |= UnicodeSetStringSpan.WITH_COUNT;
UnicodeSetStringSpan strSpan = new UnicodeSetStringSpan(this, new ArrayList To replace, count elements, or delete spans, see {@link com.ibm.icu.text.UnicodeSetSpanner UnicodeSetSpanner}.
* @param s The string to be spanned
* @param fromIndex The index of the char (exclusive) that the string should be spanned backwards
* @param spanCondition The span condition
* @return The string index which starts the span (i.e. inclusive).
* @stable ICU 4.4
*/
public int spanBack(CharSequence s, int fromIndex, SpanCondition spanCondition) {
if (fromIndex <= 0) {
return 0;
}
if (fromIndex > s.length()) {
fromIndex = s.length();
}
if (bmpSet != null) {
// Frozen set without strings, or no string is relevant for spanBack().
return bmpSet.spanBack(s, fromIndex, spanCondition);
}
if (stringSpan != null) {
return stringSpan.spanBack(s, fromIndex, spanCondition);
} else if (!strings.isEmpty()) {
int which = (spanCondition == SpanCondition.NOT_CONTAINED)
? UnicodeSetStringSpan.BACK_UTF16_NOT_CONTAINED
: UnicodeSetStringSpan.BACK_UTF16_CONTAINED;
UnicodeSetStringSpan strSpan = new UnicodeSetStringSpan(this, new ArrayList
* The functionality is straightforward for sets with only single code points, without strings (which is the common
* case):
*
* Note: Unpaired surrogates are treated like surrogate code points. Similarly, set strings match only on code point
* boundaries, never in the middle of a surrogate pair.
*
* @stable ICU 4.4
*/
public enum SpanCondition {
/**
* Continues a span() while there is no set element at the current position.
* Increments by one code point at a time.
* Stops before the first set element (character or string).
* (For code points only, this is like while contains(current)==false).
*
* When span() returns, the substring between where it started and the position it returned consists only of
* characters that are not in the set, and none of its strings overlap with the span.
*
* @stable ICU 4.4
*/
NOT_CONTAINED,
/**
* Spans the longest substring that is a concatenation of set elements (characters or strings).
* (For characters only, this is like while contains(current)==true).
*
* When span() returns, the substring between where it started and the position it returned consists only of set
* elements (characters or strings) that are in the set.
*
* If a set contains strings, then the span will be the longest substring for which there
* exists at least one non-overlapping concatenation of set elements (characters or strings).
* This is equivalent to a POSIX regular expression for
* When span() returns, the substring between where it started and the position it returned consists only of set
* elements (characters or strings) that are in the set.
*
* If a set only contains single characters, then this is the same as CONTAINED.
*
* If a set contains strings, then the span will be the longest substring with a match at each position with the
* longest single set element (character or string).
*
* Use this span condition together with other longest-match algorithms, such as ICU converters
* (ucnv_getUnicodeSet()).
*
* @stable ICU 4.4
*/
SIMPLE,
}
}
end >
* start then an empty set is created.
*
* @param start first character, inclusive, of range
* @param end last character, inclusive, of range
* @stable ICU 2.0
*/
public UnicodeSet(int start, int end) {
this();
complement(start, end);
}
/**
* Constructs a set from the given pattern. See the class description
* for the syntax of the pattern language. Whitespace is ignored.
* @param pattern a string specifying what characters are in the set
* @exception java.lang.IllegalArgumentException if the pattern contains
* a syntax error.
* @stable ICU 2.0
*/
public UnicodeSet(String pattern) {
this();
applyPattern(pattern, null);
}
/**
* Make this object represent the same set as other.
* @param other a UnicodeSet whose value will be
* copied to this object
* @stable ICU 2.0
*/
public UnicodeSet set(UnicodeSet other) {
checkFrozen();
list = other.list.clone();
len = other.len;
strings = new TreeSet
Warning: you cannot add an empty string ("") to a UnicodeSet.
* @param s the source string
* @return this object, for chaining
* @stable ICU 2.0
*/
public final UnicodeSet add(CharSequence s) {
checkFrozen();
int cp = getSingleCP(s);
if (cp < 0) {
strings.add(s.toString());
} else {
add_unchecked(cp, cp);
}
return this;
}
/**
* Utility for getting code point from single code point CharSequence.
* See the public UTF16.getSingleCodePoint()
* @return a code point IF the string consists of a single one.
* otherwise returns -1.
* @param s to test
*/
private static int getSingleCP(CharSequence s) {
if (s.length() < 1) {
throw new IllegalArgumentException("Can't use zero-length strings in UnicodeSet");
}
if (s.length() > 2) return -1;
if (s.length() == 1) return s.charAt(0);
// at this point, len = 2
int cp = UTF16.charAt(s, 0);
if (cp > 0xFFFF) { // is surrogate pair
return cp;
}
return -1;
}
/**
* Complements the specified range in this set. Any character in
* the range will be removed if it is in this set, or will be
* added if it is not in this set. If {@code end > start}
* then an empty range is complemented, leaving the set unchanged.
*
* @param start first character, inclusive, of range to be removed
* from this set.
* @param end last character, inclusive, of range to be removed
* from this set.
* @stable ICU 2.0
*/
public UnicodeSet complement(int start, int end) {
checkFrozen();
if (start < MIN_VALUE || start > MAX_VALUE) {
throw new IllegalArgumentException("Invalid code point U+" + Utility.hex(start, 6));
}
if (end < MIN_VALUE || end > MAX_VALUE) {
throw new IllegalArgumentException("Invalid code point U+" + Utility.hex(end, 6));
}
if (start <= end) {
xor(range(start, end), 2, 0);
}
return this;
}
/**
* Returns true if this set contains the given character.
* @param c character to be checked for containment
* @return true if the test condition is met
* @stable ICU 2.0
*/
public boolean contains(int c) {
if (c < MIN_VALUE || c > MAX_VALUE) {
throw new IllegalArgumentException("Invalid code point U+" + Utility.hex(c, 6));
}
if (bmpSet != null) {
return bmpSet.contains(c);
}
if (stringSpan != null) {
return stringSpan.contains(c);
}
/*
// Set i to the index of the start item greater than ch
// We know we will terminate without length test!
int i = -1;
while (true) {
if (c < list[++i]) break;
}
*/
int i = findCodePoint(c);
return ((i & 1) != 0); // return true if odd
}
/**
* Returns the smallest value i such that c < list[i]. Caller
* must ensure that c is a legal value or this method will enter
* an infinite loop. This method performs a binary search.
* @param c a character in the range MIN_VALUE..MAX_VALUE
* inclusive
* @return the smallest integer i in the range 0..len-1,
* inclusive, such that c < list[i]
*/
private final int findCodePoint(int c) {
/* Examples:
findCodePoint(c)
set list[] c=0 1 3 4 7 8
=== ============== ===========
[] [110000] 0 0 0 0 0 0
[\u0000-\u0003] [0, 4, 110000] 1 1 1 2 2 2
[\u0004-\u0007] [4, 8, 110000] 0 0 0 1 1 2
[:all:] [0, 110000] 1 1 1 1 1 1
*/
// Return the smallest i such that c < list[i]. Assume
// list[len - 1] == HIGH and that c is legal (0..HIGH-1).
if (c < list[0]) return 0;
// High runner test. c is often after the last range, so an
// initial check for this condition pays off.
if (len >= 2 && c >= list[len-2]) return len-1;
int lo = 0;
int hi = len - 1;
// invariant: c >= list[lo]
// invariant: c < list[hi]
for (;;) {
int i = (lo + hi) >>> 1;
if (i == lo) return hi;
if (c < list[i]) {
hi = i;
} else {
lo = i;
}
}
}
/**
* Retains only the elements in this set that are contained in the
* specified set. In other words, removes from this set all of
* its elements that are not contained in the specified set. This
* operation effectively modifies this set so that its value is
* the intersection of the two sets.
*
* @param c set that defines which elements this set will retain.
* @stable ICU 2.0
*/
public UnicodeSet retainAll(UnicodeSet c) {
checkFrozen();
retain(c.list, c.len, 0);
strings.retainAll(c.strings);
return this;
}
/**
* Removes all of the elements from this set. This set will be
* empty after this call returns.
* @stable ICU 2.0
*/
public UnicodeSet clear() {
checkFrozen();
list[0] = HIGH;
len = 1;
strings.clear();
return this;
}
/**
* Iteration method that returns the number of ranges contained in
* this set.
* @see #getRangeStart
* @see #getRangeEnd
* @stable ICU 2.0
*/
public int getRangeCount() {
return len/2;
}
/**
* Iteration method that returns the first character in the
* specified range of this set.
* @exception ArrayIndexOutOfBoundsException if index is outside
* the range 0..getRangeCount()-1
* @see #getRangeCount
* @see #getRangeEnd
* @stable ICU 2.0
*/
public int getRangeStart(int index) {
return list[index*2];
}
/**
* Iteration method that returns the last character in the
* specified range of this set.
* @exception ArrayIndexOutOfBoundsException if index is outside
* the range 0..getRangeCount()-1
* @see #getRangeStart
* @see #getRangeEnd
* @stable ICU 2.0
*/
public int getRangeEnd(int index) {
return (list[index*2 + 1] - 1);
}
//----------------------------------------------------------------
// Implementation: Pattern parsing
//----------------------------------------------------------------
/**
* Parses the given pattern, starting at the given position. The character
* at pattern.charAt(pos.getIndex()) must be '[', or the parse fails.
* Parsing continues until the corresponding closing ']'. If a syntax error
* is encountered between the opening and closing brace, the parse fails.
* Upon return from a successful parse, the ParsePosition is updated to
* point to the character following the closing ']', and an inversion
* list for the parsed pattern is returned. This method
* calls itself recursively to parse embedded subpatterns.
*
* @param pattern the string containing the pattern to be parsed. The
* portion of the string from pos.getIndex(), which must be a '[', to the
* corresponding closing ']', is parsed.
* @param pos upon entry, the position at which to being parsing. The
* character at pattern.charAt(pos.getIndex()) must be a '['. Upon return
* from a successful parse, pos.getIndex() is either the character after the
* closing ']' of the parsed pattern, or pattern.length() if the closing ']'
* is the last character of the pattern string.
* @return an inversion list for the parsed substring
* of pattern
* @exception java.lang.IllegalArgumentException if the parse fails.
*/
private UnicodeSet applyPattern(String pattern,
ParsePosition pos) {
if ("[:age=3.2:]".equals(pattern)) {
checkFrozen();
VersionInfo version = VersionInfo.getInstance("3.2");
applyFilter(new VersionFilter(version), UCharacterProperty.SRC_PROPSVEC);
} else {
throw new IllegalStateException("UnicodeSet.applyPattern(unexpected pattern "
+ pattern + ")");
}
return this;
}
//----------------------------------------------------------------
// Implementation: Utility methods
//----------------------------------------------------------------
private void ensureCapacity(int newLen) {
if (newLen <= list.length) return;
int[] temp = new int[newLen + GROW_EXTRA];
System.arraycopy(list, 0, temp, 0, len);
list = temp;
}
private void ensureBufferCapacity(int newLen) {
if (buffer != null && newLen <= buffer.length) return;
buffer = new int[newLen + GROW_EXTRA];
}
/**
* Assumes start <= end.
*/
private int[] range(int start, int end) {
if (rangeList == null) {
rangeList = new int[] { start, end+1, HIGH };
} else {
rangeList[0] = start;
rangeList[1] = end+1;
}
return rangeList;
}
//----------------------------------------------------------------
// Implementation: Fundamental operations
//----------------------------------------------------------------
// polarity = 0, 3 is normal: x xor y
// polarity = 1, 2: x xor ~y == x === y
private UnicodeSet xor(int[] other, int otherLen, int polarity) {
ensureBufferCapacity(len + otherLen);
int i = 0, j = 0, k = 0;
int a = list[i++];
int b;
if (polarity == 1 || polarity == 2) {
b = LOW;
if (other[j] == LOW) { // skip base if already LOW
++j;
b = other[j];
}
} else {
b = other[j++];
}
// simplest of all the routines
// sort the values, discarding identicals!
while (true) {
if (a < b) {
buffer[k++] = a;
a = list[i++];
} else if (b < a) {
buffer[k++] = b;
b = other[j++];
} else if (a != HIGH) { // at this point, a == b
// discard both values!
a = list[i++];
b = other[j++];
} else { // DONE!
buffer[k++] = HIGH;
len = k;
break;
}
}
// swap list and buffer
int[] temp = list;
list = buffer;
buffer = temp;
return this;
}
// polarity = 0 is normal: x union y
// polarity = 2: x union ~y
// polarity = 1: ~x union y
// polarity = 3: ~x union ~y
private UnicodeSet add(int[] other, int otherLen, int polarity) {
ensureBufferCapacity(len + otherLen);
int i = 0, j = 0, k = 0;
int a = list[i++];
int b = other[j++];
// change from xor is that we have to check overlapping pairs
// polarity bit 1 means a is second, bit 2 means b is.
main:
while (true) {
switch (polarity) {
case 0: // both first; take lower if unequal
if (a < b) { // take a
// Back up over overlapping ranges in buffer[]
if (k > 0 && a <= buffer[k-1]) {
// Pick latter end value in buffer[] vs. list[]
a = max(list[i], buffer[--k]);
} else {
// No overlap
buffer[k++] = a;
a = list[i];
}
i++; // Common if/else code factored out
polarity ^= 1;
} else if (b < a) { // take b
if (k > 0 && b <= buffer[k-1]) {
b = max(other[j], buffer[--k]);
} else {
buffer[k++] = b;
b = other[j];
}
j++;
polarity ^= 2;
} else { // a == b, take a, drop b
if (a == HIGH) break main;
// This is symmetrical; it doesn't matter if
// we backtrack with a or b. - liu
if (k > 0 && a <= buffer[k-1]) {
a = max(list[i], buffer[--k]);
} else {
// No overlap
buffer[k++] = a;
a = list[i];
}
i++;
polarity ^= 1;
b = other[j++]; polarity ^= 2;
}
break;
case 3: // both second; take higher if unequal, and drop other
if (b <= a) { // take a
if (a == HIGH) break main;
buffer[k++] = a;
} else { // take b
if (b == HIGH) break main;
buffer[k++] = b;
}
a = list[i++]; polarity ^= 1; // factored common code
b = other[j++]; polarity ^= 2;
break;
case 1: // a second, b first; if b < a, overlap
if (a < b) { // no overlap, take a
buffer[k++] = a; a = list[i++]; polarity ^= 1;
} else if (b < a) { // OVERLAP, drop b
b = other[j++]; polarity ^= 2;
} else { // a == b, drop both!
if (a == HIGH) break main;
a = list[i++]; polarity ^= 1;
b = other[j++]; polarity ^= 2;
}
break;
case 2: // a first, b second; if a < b, overlap
if (b < a) { // no overlap, take b
buffer[k++] = b; b = other[j++]; polarity ^= 2;
} else if (a < b) { // OVERLAP, drop a
a = list[i++]; polarity ^= 1;
} else { // a == b, drop both!
if (a == HIGH) break main;
a = list[i++]; polarity ^= 1;
b = other[j++]; polarity ^= 2;
}
break;
}
}
buffer[k++] = HIGH; // terminate
len = k;
// swap list and buffer
int[] temp = list;
list = buffer;
buffer = temp;
return this;
}
// polarity = 0 is normal: x intersect y
// polarity = 2: x intersect ~y == set-minus
// polarity = 1: ~x intersect y
// polarity = 3: ~x intersect ~y
private UnicodeSet retain(int[] other, int otherLen, int polarity) {
ensureBufferCapacity(len + otherLen);
int i = 0, j = 0, k = 0;
int a = list[i++];
int b = other[j++];
// change from xor is that we have to check overlapping pairs
// polarity bit 1 means a is second, bit 2 means b is.
main:
while (true) {
switch (polarity) {
case 0: // both first; drop the smaller
if (a < b) { // drop a
a = list[i++]; polarity ^= 1;
} else if (b < a) { // drop b
b = other[j++]; polarity ^= 2;
} else { // a == b, take one, drop other
if (a == HIGH) break main;
buffer[k++] = a; a = list[i++]; polarity ^= 1;
b = other[j++]; polarity ^= 2;
}
break;
case 3: // both second; take lower if unequal
if (a < b) { // take a
buffer[k++] = a; a = list[i++]; polarity ^= 1;
} else if (b < a) { // take b
buffer[k++] = b; b = other[j++]; polarity ^= 2;
} else { // a == b, take one, drop other
if (a == HIGH) break main;
buffer[k++] = a; a = list[i++]; polarity ^= 1;
b = other[j++]; polarity ^= 2;
}
break;
case 1: // a second, b first;
if (a < b) { // NO OVERLAP, drop a
a = list[i++]; polarity ^= 1;
} else if (b < a) { // OVERLAP, take b
buffer[k++] = b; b = other[j++]; polarity ^= 2;
} else { // a == b, drop both!
if (a == HIGH) break main;
a = list[i++]; polarity ^= 1;
b = other[j++]; polarity ^= 2;
}
break;
case 2: // a first, b second; if a < b, overlap
if (b < a) { // no overlap, drop b
b = other[j++]; polarity ^= 2;
} else if (a < b) { // OVERLAP, take a
buffer[k++] = a; a = list[i++]; polarity ^= 1;
} else { // a == b, drop both!
if (a == HIGH) break main;
a = list[i++]; polarity ^= 1;
b = other[j++]; polarity ^= 2;
}
break;
}
}
buffer[k++] = HIGH; // terminate
len = k;
// swap list and buffer
int[] temp = list;
list = buffer;
buffer = temp;
return this;
}
private static final int max(int a, int b) {
return (a > b) ? a : b;
}
//----------------------------------------------------------------
// Generic filter-based scanning code
//----------------------------------------------------------------
private static interface Filter {
boolean contains(int codePoint);
}
private static final VersionInfo NO_VERSION = VersionInfo.getInstance(0, 0, 0, 0);
private static class VersionFilter implements Filter {
VersionInfo version;
VersionFilter(VersionInfo version) { this.version = version; }
public boolean contains(int ch) {
VersionInfo v = UCharacter.getAge(ch);
// Reference comparison ok; VersionInfo caches and reuses
// unique objects.
return v != NO_VERSION &&
v.compareTo(version) <= 0;
}
}
private static synchronized UnicodeSet getInclusions(int src) {
if (src != UCharacterProperty.SRC_PROPSVEC) {
throw new IllegalStateException("UnicodeSet.getInclusions(unknown src "+src+")");
}
if (INCLUSION == null) {
UnicodeSet incl = new UnicodeSet();
UCharacterProperty.INSTANCE.upropsvec_addPropertyStarts(incl);
INCLUSION = incl;
}
return INCLUSION;
}
/**
* Generic filter-based scanning code for UCD property UnicodeSets.
*/
private UnicodeSet applyFilter(Filter filter, int src) {
// Logically, walk through all Unicode characters, noting the start
// and end of each range for which filter.contain(c) is
// true. Add each range to a set.
//
// To improve performance, use an inclusions set which
// encodes information about character ranges that are known
// to have identical properties.
// getInclusions(src) contains exactly the first characters of
// same-value ranges for the given properties "source".
clear();
int startHasProperty = -1;
UnicodeSet inclusions = getInclusions(src);
int limitRange = inclusions.getRangeCount();
for (int j=0; j
*
* When a set contains multi-code point strings, then these statements may not be true, depending on the strings in
* the set (for example, whether they overlap with each other) and the string that is processed. For a set with
* strings:
*
*
* Note: If it is important to get the same boundaries whether iterating forward or backward through a string, then
* either only span() should be used and the boundaries cached for backward operation, or an ICU BreakIterator could
* be used.
* (OR of each set element)*.
* (Java/ICU/Perl regex stops at the first match of an OR.)
*
* @stable ICU 4.4
*/
CONTAINED,
/**
* Continues a span() while there is a set element at the current position.
* Increments by the longest matching element at each position.
* (For characters only, this is like while contains(current)==true).
*