RED-5562 improved pattern compile
This commit is contained in:
parent
37edffc75f
commit
5d68955be6
@ -8,13 +8,14 @@ import java.util.regex.Pattern;
|
|||||||
@Slf4j
|
@Slf4j
|
||||||
public class SeparatorUtils {
|
public class SeparatorUtils {
|
||||||
|
|
||||||
|
private final static Pattern punctuationPattern = Pattern.compile("\\p{Punct}");
|
||||||
private final static Set<Character> quotes = Set.of('\'', '\u0022', '\u00AB', '\u00BB', '\u2018', '\u2019', '\u201A', '\u201C', '\u201D', '\u201E'
|
private final static Set<Character> quotes = Set.of('\'', '\u0022', '\u00AB', '\u00BB', '\u2018', '\u2019', '\u201A', '\u201C', '\u201D', '\u201E'
|
||||||
, '\u2039', '\u203A');
|
, '\u2039', '\u203A');
|
||||||
|
|
||||||
private final static Set<Integer> japaneseAltPunctuationMarks = Set.of(65288, 65289, 65294, 65339, 65341, 65371, 65373, 65375, 65376, 12443, 12444, 65309);
|
private final static Set<Integer> japaneseAltPunctuationMarks = Set.of(65288, 65289, 65294, 65339, 65341, 65371, 65373, 65375, 65376, 12443, 12444, 65309);
|
||||||
|
|
||||||
public static boolean isSeparator(char c) {
|
public static boolean isSeparator(char c) {
|
||||||
return Character.isWhitespace(c) || Pattern.matches("\\p{Punct}", String.valueOf(c)) || quotes.contains(c) || isJapaneseSeparator(c);
|
return Character.isWhitespace(c) || punctuationPattern.matcher(String.valueOf(c)).matches() || quotes.contains(c) || isJapaneseSeparator(c);
|
||||||
}
|
}
|
||||||
|
|
||||||
public static boolean isJapaneseSeparator(char c) {
|
public static boolean isJapaneseSeparator(char c) {
|
||||||
|
|||||||
Loading…
x
Reference in New Issue
Block a user