john-wagster commented on code in PR #14192:
URL: https://github.com/apache/lucene/pull/14192#discussion_r1941960613
##########
lucene/core/src/java/org/apache/lucene/util/automaton/RegExp.java:
##########
@@ -436,6 +478,160 @@ public enum Kind {
*/
@Deprecated public static final int DEPRECATED_COMPLEMENT = 0x10000;
+ /**
+ * See {@link #UNICODE_CASE_INSENSITIVE} for more details on the set of
known unstable alternative
+ * casings
+ */
+ static final Map<Integer, int[]> unstableUnicodeCharacters =
+ Map.ofEntries(
+ // these are the set of characters whose casing matches across
multiple characters
+ entry(181, new int[] {924, 956}),
+ entry(924, new int[] {181, 956}),
+ entry(956, new int[] {181, 924}),
+ entry(304, new int[] {105, 73, 305}),
+ entry(105, new int[] {304, 73, 305}),
+ entry(73, new int[] {304, 105, 305}),
+ entry(305, new int[] {304, 105, 73}),
+ entry(383, new int[] {83, 115}),
+ entry(83, new int[] {383, 115}),
+ entry(115, new int[] {383, 83}),
+ entry(837, new int[] {921, 953, 8126}),
+ entry(921, new int[] {837, 953, 8126}),
+ entry(953, new int[] {837, 921, 8126}),
+ entry(8126, new int[] {837, 921, 953}),
+ entry(962, new int[] {931, 963}),
+ entry(931, new int[] {962, 963}),
+ entry(963, new int[] {962, 931}),
+ entry(976, new int[] {914, 946}),
+ entry(914, new int[] {976, 946}),
+ entry(946, new int[] {976, 914}),
+ entry(977, new int[] {920, 952, 1012}),
+ entry(920, new int[] {977, 952, 1012}),
+ entry(952, new int[] {977, 920, 1012}),
+ entry(1012, new int[] {977, 920, 952}),
+ entry(981, new int[] {934, 966}),
+ entry(934, new int[] {981, 966}),
+ entry(966, new int[] {981, 934}),
+ entry(982, new int[] {928, 960}),
+ entry(928, new int[] {982, 960}),
+ entry(960, new int[] {982, 928}),
+ entry(1008, new int[] {922, 954}),
+ entry(922, new int[] {1008, 954}),
+ entry(954, new int[] {1008, 922}),
+ entry(1009, new int[] {929, 961}),
+ entry(929, new int[] {1009, 961}),
+ entry(961, new int[] {1009, 929}),
+ entry(1013, new int[] {917, 949}),
+ entry(917, new int[] {1013, 949}),
+ entry(949, new int[] {1013, 917}),
+ entry(7296, new int[] {1042, 1074}),
+ entry(1042, new int[] {7296, 1074}),
+ entry(1074, new int[] {7296, 1042}),
+ entry(7297, new int[] {1044, 1076}),
+ entry(1044, new int[] {7297, 1076}),
+ entry(1076, new int[] {7297, 1044}),
+ entry(7298, new int[] {1054, 1086}),
+ entry(1054, new int[] {7298, 1086}),
+ entry(1086, new int[] {7298, 1054}),
+ entry(7299, new int[] {1057, 1089}),
+ entry(1057, new int[] {7299, 1089}),
+ entry(1089, new int[] {7299, 1057}),
+ entry(7300, new int[] {1058, 1090, 7301}),
+ entry(1058, new int[] {7300, 1090, 7301}),
+ entry(1090, new int[] {7300, 1058, 7301}),
+ entry(7301, new int[] {7300, 1058, 1090}),
+ entry(7302, new int[] {1066, 1098}),
+ entry(1066, new int[] {7302, 1098}),
+ entry(1098, new int[] {7302, 1066}),
+ entry(7303, new int[] {1122, 1123}),
+ entry(1122, new int[] {7303, 1123}),
+ entry(1123, new int[] {7303, 1122}),
+ entry(7304, new int[] {42570, 42571}),
+ entry(42570, new int[] {7304, 42571}),
+ entry(42571, new int[] {7304, 42570}),
+ entry(7835, new int[] {7776, 7777}),
+ entry(7776, new int[] {7835, 7777}),
+ entry(7777, new int[] {7835, 7776}),
+ entry(8486, new int[] {969, 937}),
+ entry(969, new int[] {8486, 937}),
+ entry(937, new int[] {8486, 969}),
+ entry(8490, new int[] {107, 75}),
+ entry(107, new int[] {8490, 75}),
+ entry(75, new int[] {8490, 107}),
+ entry(8491, new int[] {229, 197}),
+ entry(229, new int[] {8491, 197}),
+ entry(197, new int[] {8491, 229}),
Review Comment:
I'm sold. I was comparing the CaseFolding spec with the Pattern and String
classes today and found a (albeit small) number of characters that neither
correctly supports from the unicode spec which seemed to me that should pretty
clearly be supported such as:
https://util.unicode.org/UnicodeJsps/character.jsp?a=1C8A&B1=Show
https://util.unicode.org/UnicodeJsps/character.jsp?a=0x1C89&B1=Show
I have some changes related to your other comments but I'm going to iterate
a bit further and I'll likely need some additional feedback on an approach that
incorporates that spec / we may need to talk about how to maintain the case
folding as the unicode spec iterates. I was hoping to rely heavily on the
Pattern class being updated so we didn't have to do as much maintenance on our
end to make sure case insensitivity continued to work appropriately.
Also just to be clear the intention of the PR is just handling character
casing not normalization (just forgot to address that yesterday).
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
