Marmelator/redaction-service
1
0
Fork 0
Code Issues Pull Requests 14 Actions Packages Projects Releases 622 Wiki Activity

Pull request #500: RED-5562 34x

Browse Source 
Merge in RED/redaction-service from RED-5562-34x to release/3.163.x

* commit 'ba17fa77b40df8acb350fd53facfb3413f5af3b7':
  RED-5562 more symbols
  RED-5562 more symbols
This commit is contained in:
Timo Bejan 2022-11-15 10:40:36 +01:00 committed by Kresnadi Budisantoso
commit 884648f386
1 changed files with 1 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/redaction/utils/SeparatorUtils.java
View File

@ -12,7 +12,7 @@ public class SeparatorUtils {
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, 65306);
public static boolean isSeparator(char c) { public static boolean isSeparator(char c) {
return Character.isWhitespace(c) || punctuationPattern.matcher(String.valueOf(c)).matches() || quotes.contains(c) || isJapaneseSeparator(c); return Character.isWhitespace(c) || punctuationPattern.matcher(String.valueOf(c)).matches() || quotes.contains(c) || isJapaneseSeparator(c);