aboutsummaryrefslogtreecommitdiffstats
path: root/native/jni/src/utils/char_utils.cpp
diff options
context:
space:
mode:
Diffstat (limited to 'native/jni/src/utils/char_utils.cpp')
-rw-r--r--native/jni/src/utils/char_utils.cpp7
1 files changed, 6 insertions, 1 deletions
diff --git a/native/jni/src/utils/char_utils.cpp b/native/jni/src/utils/char_utils.cpp
index 3bb9055b2..a43e6dd62 100644
--- a/native/jni/src/utils/char_utils.cpp
+++ b/native/jni/src/utils/char_utils.cpp
@@ -1117,7 +1117,9 @@ static int compare_pair_capital(const void *a, const void *b) {
// TODO: Check if it's really acceptable to consider ΓΈ a diacritical variant of o
/* U+0100 */ 0x0041, 0x0061, 0x0041, 0x0061, 0x0041, 0x0061, 0x0043, 0x0063,
/* U+0108 */ 0x0043, 0x0063, 0x0043, 0x0063, 0x0043, 0x0063, 0x0044, 0x0064,
- /* U+0110 */ 0x0110, 0x0111, 0x0045, 0x0065, 0x0045, 0x0065, 0x0045, 0x0065,
+ /* U+0110 */ 0x0046, 0x0064, 0x0045, 0x0065, 0x0045, 0x0065, 0x0045, 0x0065,
+ // U+0110: Manually changed from 0110 to 0046
+ // U+0111: Manually changed from 0111 to 0064
/* U+0118 */ 0x0045, 0x0065, 0x0045, 0x0065, 0x0047, 0x0067, 0x0047, 0x0067,
/* U+0120 */ 0x0047, 0x0067, 0x0047, 0x0067, 0x0048, 0x0068, 0x0126, 0x0127,
/* U+0128 */ 0x0049, 0x0069, 0x0049, 0x0069, 0x0049, 0x0069, 0x0049, 0x0069,
@@ -1135,6 +1137,9 @@ static int compare_pair_capital(const void *a, const void *b) {
/* U+0170 */ 0x0055, 0x0075, 0x0055, 0x0075, 0x0057, 0x0077, 0x0059, 0x0079,
/* U+0178 */ 0x0059, 0x005A, 0x007A, 0x005A, 0x007A, 0x005A, 0x007A, 0x0073,
/* U+0180 */ 0x0180, 0x0181, 0x0182, 0x0183, 0x0184, 0x0185, 0x0186, 0x0187,
+ // TODO: A lot of letters are their own base code points, but for
+ // some (e.g. U+0180) it doesn't seem right. Ideally each code point should
+ // be checked individually with all languages it's used in.
/* U+0188 */ 0x0188, 0x0189, 0x018A, 0x018B, 0x018C, 0x018D, 0x018E, 0x018F,
/* U+0190 */ 0x0190, 0x0191, 0x0192, 0x0193, 0x0194, 0x0195, 0x0196, 0x0197,
/* U+0198 */ 0x0198, 0x0199, 0x019A, 0x019B, 0x019C, 0x019D, 0x019E, 0x019F,