From bc187f06cbe119957320f7ade4a6692fd278289f Mon Sep 17 00:00:00 2001 From: afresh1 Date: Sun, 16 May 2021 22:38:31 +0000 Subject: [PATCH] Update en_US.UTF-8.src to Unicode 13.0 This is what ships with the perl 5.32.1 OK schwarze@ --- share/locale/ctype/en_US.UTF-8.src | 391 +++++++++++++++++---------- share/locale/ctype/gen_ctype_utf8.pl | 4 +- 2 files changed, 255 insertions(+), 140 deletions(-) diff --git a/share/locale/ctype/en_US.UTF-8.src b/share/locale/ctype/en_US.UTF-8.src index 29f518c846e..3fd2888ee94 100644 --- a/share/locale/ctype/en_US.UTF-8.src +++ b/share/locale/ctype/en_US.UTF-8.src @@ -1,9 +1,9 @@ -/* $OpenBSD: en_US.UTF-8.src,v 1.10 2020/02/20 02:53:53 afresh1 Exp $ */ +/* $OpenBSD: en_US.UTF-8.src,v 1.11 2021/05/16 22:38:31 afresh1 Exp $ */ /* * COPYRIGHT AND PERMISSION NOTICE * - * Copyright (c) 1991-2019 Unicode, Inc. All rights reserved. + * Copyright (c) 1991-2021 Unicode, Inc. All rights reserved. * Distributed under the Terms of Use in * https://www.unicode.org/copyright.html. * @@ -40,7 +40,7 @@ ENCODING "UTF8" VARIABLE CODESET=UTF-8 -/* Unicode Version 12.1.0 */ +/* Unicode Version 13.0.0 */ /* * U+0000 - U+007F : Basic Latin @@ -1013,13 +1013,13 @@ SWIDTH1 0x0860 - 0x086a * U+08A0 - U+08FF : Arabic Extended-A */ -ALPHA 0x08a0 - 0x08b4 0x08b6 - 0x08bd 0x08d4 - 0x08df 0x08e3 - 0x08e9 +ALPHA 0x08a0 - 0x08b4 0x08b6 - 0x08c7 0x08d4 - 0x08df 0x08e3 - 0x08e9 ALPHA 0x08f0 - 0x08ff -GRAPH 0x08a0 - 0x08b4 0x08b6 - 0x08bd 0x08d3 - 0x08ff -PRINT 0x08a0 - 0x08b4 0x08b6 - 0x08bd 0x08d3 - 0x08ff +GRAPH 0x08a0 - 0x08b4 0x08b6 - 0x08c7 0x08d3 - 0x08ff +PRINT 0x08a0 - 0x08b4 0x08b6 - 0x08c7 0x08d3 - 0x08ff SPECIAL 0x08d3 0x08e0 - 0x08e2 0x08ea - 0x08ef SWIDTH0 0x08d3 - 0x08ff -SWIDTH1 0x08a0 - 0x08b4 0x08b6 - 0x08bd +SWIDTH1 0x08a0 - 0x08b4 0x08b6 - 0x08c7 /* @@ -1135,15 +1135,15 @@ ALPHA 0x0b5f - 0x0b63 0x0b71 DIGIT 0x0b66 - 0x0b6f GRAPH 0x0b01 - 0x0b03 0x0b05 - 0x0b0c 0x0b0f - 0x0b10 0x0b13 - 0x0b28 GRAPH 0x0b2a - 0x0b30 0x0b32 - 0x0b33 0x0b35 - 0x0b39 0x0b3c - 0x0b44 -GRAPH 0x0b47 - 0x0b48 0x0b4b - 0x0b4d 0x0b56 - 0x0b57 0x0b5c - 0x0b5d +GRAPH 0x0b47 - 0x0b48 0x0b4b - 0x0b4d 0x0b55 - 0x0b57 0x0b5c - 0x0b5d GRAPH 0x0b5f - 0x0b63 0x0b66 - 0x0b77 PUNCT 0x0b70 PRINT 0x0b01 - 0x0b03 0x0b05 - 0x0b0c 0x0b0f - 0x0b10 0x0b13 - 0x0b28 PRINT 0x0b2a - 0x0b30 0x0b32 - 0x0b33 0x0b35 - 0x0b39 0x0b3c - 0x0b44 -PRINT 0x0b47 - 0x0b48 0x0b4b - 0x0b4d 0x0b56 - 0x0b57 0x0b5c - 0x0b5d +PRINT 0x0b47 - 0x0b48 0x0b4b - 0x0b4d 0x0b55 - 0x0b57 0x0b5c - 0x0b5d PRINT 0x0b5f - 0x0b63 0x0b66 - 0x0b77 -SPECIAL 0x0b3c 0x0b4d 0x0b72 - 0x0b77 -SWIDTH0 0x0b01 0x0b3c 0x0b3f 0x0b41 - 0x0b44 0x0b4d 0x0b56 +SPECIAL 0x0b3c 0x0b4d 0x0b55 0x0b72 - 0x0b77 +SWIDTH0 0x0b01 0x0b3c 0x0b3f 0x0b41 - 0x0b44 0x0b4d 0x0b55 - 0x0b56 SWIDTH0 0x0b62 - 0x0b63 SWIDTH1 0x0b02 - 0x0b03 0x0b05 - 0x0b0c 0x0b0f - 0x0b10 0x0b13 - 0x0b28 SWIDTH1 0x0b2a - 0x0b30 0x0b32 - 0x0b33 0x0b35 - 0x0b39 0x0b3d - 0x0b3e @@ -1238,21 +1238,21 @@ TODIGIT < 0x0ce6 - 0x0cef : 0x0000 > * U+0D00 - U+0D7F : Malayalam */ -ALPHA 0x0d00 - 0x0d03 0x0d05 - 0x0d0c 0x0d0e - 0x0d10 0x0d12 - 0x0d3a -ALPHA 0x0d3d - 0x0d44 0x0d46 - 0x0d48 0x0d4a - 0x0d4c 0x0d4e -ALPHA 0x0d54 - 0x0d57 0x0d5f - 0x0d63 0x0d7a - 0x0d7f +ALPHA 0x0d00 - 0x0d0c 0x0d0e - 0x0d10 0x0d12 - 0x0d3a 0x0d3d - 0x0d44 +ALPHA 0x0d46 - 0x0d48 0x0d4a - 0x0d4c 0x0d4e 0x0d54 - 0x0d57 +ALPHA 0x0d5f - 0x0d63 0x0d7a - 0x0d7f DIGIT 0x0d66 - 0x0d6f -GRAPH 0x0d00 - 0x0d03 0x0d05 - 0x0d0c 0x0d0e - 0x0d10 0x0d12 - 0x0d44 -GRAPH 0x0d46 - 0x0d48 0x0d4a - 0x0d4f 0x0d54 - 0x0d63 0x0d66 - 0x0d7f +GRAPH 0x0d00 - 0x0d0c 0x0d0e - 0x0d10 0x0d12 - 0x0d44 0x0d46 - 0x0d48 +GRAPH 0x0d4a - 0x0d4f 0x0d54 - 0x0d63 0x0d66 - 0x0d7f PUNCT 0x0d4f 0x0d79 -PRINT 0x0d00 - 0x0d03 0x0d05 - 0x0d0c 0x0d0e - 0x0d10 0x0d12 - 0x0d44 -PRINT 0x0d46 - 0x0d48 0x0d4a - 0x0d4f 0x0d54 - 0x0d63 0x0d66 - 0x0d7f +PRINT 0x0d00 - 0x0d0c 0x0d0e - 0x0d10 0x0d12 - 0x0d44 0x0d46 - 0x0d48 +PRINT 0x0d4a - 0x0d4f 0x0d54 - 0x0d63 0x0d66 - 0x0d7f SPECIAL 0x0d3b - 0x0d3c 0x0d4d 0x0d58 - 0x0d5e 0x0d70 - 0x0d78 SWIDTH0 0x0d00 - 0x0d01 0x0d3b - 0x0d3c 0x0d41 - 0x0d44 0x0d4d SWIDTH0 0x0d62 - 0x0d63 -SWIDTH1 0x0d02 - 0x0d03 0x0d05 - 0x0d0c 0x0d0e - 0x0d10 0x0d12 - 0x0d3a -SWIDTH1 0x0d3d - 0x0d40 0x0d46 - 0x0d48 0x0d4a - 0x0d4c 0x0d4e - 0x0d4f -SWIDTH1 0x0d54 - 0x0d61 0x0d66 - 0x0d7f +SWIDTH1 0x0d02 - 0x0d0c 0x0d0e - 0x0d10 0x0d12 - 0x0d3a 0x0d3d - 0x0d40 +SWIDTH1 0x0d46 - 0x0d48 0x0d4a - 0x0d4c 0x0d4e - 0x0d4f 0x0d54 - 0x0d61 +SWIDTH1 0x0d66 - 0x0d7f TODIGIT < 0x0d66 - 0x0d70 : 0x0000 > TODIGIT < 0x0d71 100 > @@ -1262,19 +1262,19 @@ TODIGIT < 0x0d72 1000 > * U+0D80 - U+0DFF : Sinhala */ -ALPHA 0x0d82 - 0x0d83 0x0d85 - 0x0d96 0x0d9a - 0x0db1 0x0db3 - 0x0dbb +ALPHA 0x0d81 - 0x0d83 0x0d85 - 0x0d96 0x0d9a - 0x0db1 0x0db3 - 0x0dbb ALPHA 0x0dbd 0x0dc0 - 0x0dc6 0x0dcf - 0x0dd4 0x0dd6 0x0dd8 - 0x0ddf ALPHA 0x0df2 - 0x0df3 DIGIT 0x0de6 - 0x0def -GRAPH 0x0d82 - 0x0d83 0x0d85 - 0x0d96 0x0d9a - 0x0db1 0x0db3 - 0x0dbb +GRAPH 0x0d81 - 0x0d83 0x0d85 - 0x0d96 0x0d9a - 0x0db1 0x0db3 - 0x0dbb GRAPH 0x0dbd 0x0dc0 - 0x0dc6 0x0dca 0x0dcf - 0x0dd4 0x0dd6 GRAPH 0x0dd8 - 0x0ddf 0x0de6 - 0x0def 0x0df2 - 0x0df4 PUNCT 0x0df4 -PRINT 0x0d82 - 0x0d83 0x0d85 - 0x0d96 0x0d9a - 0x0db1 0x0db3 - 0x0dbb +PRINT 0x0d81 - 0x0d83 0x0d85 - 0x0d96 0x0d9a - 0x0db1 0x0db3 - 0x0dbb PRINT 0x0dbd 0x0dc0 - 0x0dc6 0x0dca 0x0dcf - 0x0dd4 0x0dd6 PRINT 0x0dd8 - 0x0ddf 0x0de6 - 0x0def 0x0df2 - 0x0df4 SPECIAL 0x0dca -SWIDTH0 0x0dca 0x0dd2 - 0x0dd4 0x0dd6 +SWIDTH0 0x0d81 0x0dca 0x0dd2 - 0x0dd4 0x0dd6 SWIDTH1 0x0d82 - 0x0d83 0x0d85 - 0x0d96 0x0d9a - 0x0db1 0x0db3 - 0x0dbb SWIDTH1 0x0dbd 0x0dc0 - 0x0dc6 0x0dcf - 0x0dd1 0x0dd8 - 0x0ddf SWIDTH1 0x0de6 - 0x0def 0x0df2 - 0x0df4 @@ -1679,10 +1679,11 @@ TODIGIT < 0x1a90 - 0x1a99 : 0x0000 > * U+1AB0 - U+1AFF : Combining Diacritical Marks Extended */ -GRAPH 0x1ab0 - 0x1abe -PRINT 0x1ab0 - 0x1abe +ALPHA 0x1abf - 0x1ac0 +GRAPH 0x1ab0 - 0x1ac0 +PRINT 0x1ab0 - 0x1ac0 SPECIAL 0x1ab0 - 0x1abe -SWIDTH0 0x1ab0 - 0x1abe +SWIDTH0 0x1ab0 - 0x1ac0 /* @@ -2529,11 +2530,11 @@ SWIDTH1 0x2a00 - 0x2aff * U+2B00 - U+2BFF : Miscellaneous Symbols and Arrows */ -GRAPH 0x2b00 - 0x2b73 0x2b76 - 0x2b95 0x2b98 - 0x2bff -PUNCT 0x2b00 - 0x2b73 0x2b76 - 0x2b95 0x2b98 - 0x2bff -PRINT 0x2b00 - 0x2b73 0x2b76 - 0x2b95 0x2b98 - 0x2bff +GRAPH 0x2b00 - 0x2b73 0x2b76 - 0x2b95 0x2b97 - 0x2bff +PUNCT 0x2b00 - 0x2b73 0x2b76 - 0x2b95 0x2b97 - 0x2bff +PRINT 0x2b00 - 0x2b73 0x2b76 - 0x2b95 0x2b97 - 0x2bff SWIDTH1 0x2b00 - 0x2b1a 0x2b1d - 0x2b4f 0x2b51 - 0x2b54 0x2b56 - 0x2b73 -SWIDTH1 0x2b76 - 0x2b95 0x2b98 - 0x2bff +SWIDTH1 0x2b76 - 0x2b95 0x2b97 - 0x2bff SWIDTH2 0x2b1b - 0x2b1c 0x2b50 0x2b55 @@ -2782,10 +2783,10 @@ SWIDTH0 0x2de0 - 0x2dff */ ALPHA 0x2e2f -GRAPH 0x2e00 - 0x2e4f -PUNCT 0x2e00 - 0x2e2e 0x2e30 - 0x2e4f -PRINT 0x2e00 - 0x2e4f -SWIDTH1 0x2e00 - 0x2e4f +GRAPH 0x2e00 - 0x2e52 +PUNCT 0x2e00 - 0x2e2e 0x2e30 - 0x2e52 +PRINT 0x2e00 - 0x2e52 +SWIDTH1 0x2e00 - 0x2e52 /* @@ -2905,10 +2906,10 @@ TODIGIT < 0x3192 - 0x3195 : 1 > * U+31A0 - U+31BF : Bopomofo Extended */ -ALPHA 0x31a0 - 0x31ba -GRAPH 0x31a0 - 0x31ba -PRINT 0x31a0 - 0x31ba -SWIDTH2 0x31a0 - 0x31ba +ALPHA 0x31a0 - 0x31bf +GRAPH 0x31a0 - 0x31bf +PRINT 0x31a0 - 0x31bf +SWIDTH2 0x31a0 - 0x31bf /* @@ -2972,10 +2973,10 @@ SWIDTH2 0x3300 - 0x33ff */ ALPHA 0x3400 - 0x3404 0x3406 - 0x3482 0x3484 - 0x3829 0x382b - 0x3b4c -ALPHA 0x3b4e - 0x4db5 -GRAPH 0x3400 - 0x4db5 -PRINT 0x3400 - 0x4db5 -SWIDTH2 0x3400 - 0x4db5 +ALPHA 0x3b4e - 0x4dbf +GRAPH 0x3400 - 0x4dbf +PRINT 0x3400 - 0x4dbf +SWIDTH2 0x3400 - 0x4dbf TODIGIT < 0x3405 5 > TODIGIT < 0x3483 2 > @@ -3006,10 +3007,10 @@ ALPHA 0x5e7b - 0x5efd 0x5f00 - 0x5f0b 0x5f0f 0x5f11 - 0x62fd ALPHA 0x62ff - 0x634b 0x634d - 0x67d1 0x67d3 - 0x6f05 0x6f07 - 0x7395 ALPHA 0x7397 - 0x767d 0x767f - 0x8085 0x8087 - 0x842b 0x842d - 0x8cad ALPHA 0x8caf - 0x8cb2 0x8cb4 - 0x8d2f 0x8d31 - 0x9620 0x9622 - 0x9645 -ALPHA 0x9647 - 0x964b 0x964d - 0x9677 0x9679 - 0x96f5 0x96f7 - 0x9fef -GRAPH 0x4e00 - 0x9fef -PRINT 0x4e00 - 0x9fef -SWIDTH2 0x4e00 - 0x9fef +ALPHA 0x9647 - 0x964b 0x964d - 0x9677 0x9679 - 0x96f5 0x96f7 - 0x9ffc +GRAPH 0x4e00 - 0x9ffc +PRINT 0x4e00 - 0x9ffc +SWIDTH2 0x4e00 - 0x9ffc TODIGIT < 0x4e00 1 > TODIGIT < 0x4e03 7 > @@ -3237,8 +3238,8 @@ SWIDTH1 0xa700 - 0xa71f * U+A720 - U+A7FF : Latin Extended-D */ -ALPHA 0xa722 - 0xa788 0xa78b - 0xa7bf 0xa7c2 - 0xa7c6 0xa7f7 - 0xa7ff -GRAPH 0xa720 - 0xa7bf 0xa7c2 - 0xa7c6 0xa7f7 - 0xa7ff +ALPHA 0xa722 - 0xa788 0xa78b - 0xa7bf 0xa7c2 - 0xa7ca 0xa7f5 - 0xa7ff +GRAPH 0xa720 - 0xa7bf 0xa7c2 - 0xa7ca 0xa7f5 - 0xa7ff LOWER 0xa723 0xa725 0xa727 0xa729 0xa72b 0xa72d 0xa72f - 0xa731 LOWER 0xa733 0xa735 0xa737 0xa739 0xa73b 0xa73d 0xa73f 0xa741 0xa743 LOWER 0xa745 0xa747 0xa749 0xa74b 0xa74d 0xa74f 0xa751 0xa753 0xa755 @@ -3247,7 +3248,7 @@ LOWER 0xa769 0xa76b 0xa76d 0xa76f 0xa771 - 0xa778 0xa77a 0xa77c LOWER 0xa77f 0xa781 0xa783 0xa785 0xa787 0xa78c 0xa78e 0xa791 LOWER 0xa793 - 0xa795 0xa797 0xa799 0xa79b 0xa79d 0xa79f 0xa7a1 LOWER 0xa7a3 0xa7a5 0xa7a7 0xa7a9 0xa7af 0xa7b5 0xa7b7 0xa7b9 0xa7bb -LOWER 0xa7bd 0xa7bf 0xa7c3 0xa7fa +LOWER 0xa7bd 0xa7bf 0xa7c3 0xa7c8 0xa7ca 0xa7f6 0xa7fa PUNCT 0xa720 - 0xa721 0xa789 - 0xa78a UPPER 0xa722 0xa724 0xa726 0xa728 0xa72a 0xa72c 0xa72e 0xa732 0xa734 UPPER 0xa736 0xa738 0xa73a 0xa73c 0xa73e 0xa740 0xa742 0xa744 0xa746 @@ -3257,9 +3258,9 @@ UPPER 0xa76c 0xa76e 0xa779 0xa77b 0xa77d - 0xa77e 0xa780 0xa782 UPPER 0xa784 0xa786 0xa78b 0xa78d 0xa790 0xa792 0xa796 0xa798 0xa79a UPPER 0xa79c 0xa79e 0xa7a0 0xa7a2 0xa7a4 0xa7a6 0xa7a8 UPPER 0xa7aa - 0xa7ae 0xa7b0 - 0xa7b4 0xa7b6 0xa7b8 0xa7ba 0xa7bc -UPPER 0xa7be 0xa7c2 0xa7c4 - 0xa7c6 -PRINT 0xa720 - 0xa7bf 0xa7c2 - 0xa7c6 0xa7f7 - 0xa7ff -SWIDTH1 0xa720 - 0xa7bf 0xa7c2 - 0xa7c6 0xa7f7 - 0xa7ff +UPPER 0xa7be 0xa7c2 0xa7c4 - 0xa7c7 0xa7c9 0xa7f5 +PRINT 0xa720 - 0xa7bf 0xa7c2 - 0xa7ca 0xa7f5 - 0xa7ff +SWIDTH1 0xa720 - 0xa7bf 0xa7c2 - 0xa7ca 0xa7f5 - 0xa7ff MAPUPPER < 0xa723 0xa722 > MAPUPPER < 0xa725 0xa724 > @@ -3327,6 +3328,9 @@ MAPUPPER < 0xa7bb 0xa7ba > MAPUPPER < 0xa7bd 0xa7bc > MAPUPPER < 0xa7bf 0xa7be > MAPUPPER < 0xa7c3 0xa7c2 > +MAPUPPER < 0xa7c8 0xa7c7 > +MAPUPPER < 0xa7ca 0xa7c9 > +MAPUPPER < 0xa7f6 0xa7f5 > MAPLOWER < 0xa722 0xa723 > MAPLOWER < 0xa724 0xa725 > MAPLOWER < 0xa726 0xa727 > @@ -3406,17 +3410,20 @@ MAPLOWER < 0xa7c2 0xa7c3 > MAPLOWER < 0xa7c4 0xa794 > MAPLOWER < 0xa7c5 0x0282 > MAPLOWER < 0xa7c6 0x1d8e > +MAPLOWER < 0xa7c7 0xa7c8 > +MAPLOWER < 0xa7c9 0xa7ca > +MAPLOWER < 0xa7f5 0xa7f6 > /* * U+A800 - U+A82F : Syloti Nagri */ ALPHA 0xa800 - 0xa805 0xa807 - 0xa827 -GRAPH 0xa800 - 0xa82b +GRAPH 0xa800 - 0xa82c PUNCT 0xa828 - 0xa82b -PRINT 0xa800 - 0xa82b -SPECIAL 0xa806 -SWIDTH0 0xa802 0xa806 0xa80b 0xa825 - 0xa826 +PRINT 0xa800 - 0xa82c +SPECIAL 0xa806 0xa82c +SWIDTH0 0xa802 0xa806 0xa80b 0xa825 - 0xa826 0xa82c SWIDTH1 0xa800 - 0xa801 0xa803 - 0xa805 0xa807 - 0xa80a 0xa80c - 0xa824 SWIDTH1 0xa827 - 0xa82b @@ -3613,12 +3620,12 @@ SWIDTH1 0xab28 - 0xab2e * U+AB30 - U+AB6F : Latin Extended-E */ -ALPHA 0xab30 - 0xab5a 0xab5c - 0xab67 -GRAPH 0xab30 - 0xab67 -LOWER 0xab30 - 0xab5a 0xab60 - 0xab67 -PUNCT 0xab5b -PRINT 0xab30 - 0xab67 -SWIDTH1 0xab30 - 0xab67 +ALPHA 0xab30 - 0xab5a 0xab5c - 0xab69 +GRAPH 0xab30 - 0xab6b +LOWER 0xab30 - 0xab5a 0xab60 - 0xab68 +PUNCT 0xab5b 0xab6a - 0xab6b +PRINT 0xab30 - 0xab6b +SWIDTH1 0xab30 - 0xab6b MAPUPPER < 0xab53 0xa7b3 > @@ -4000,10 +4007,10 @@ TODIGIT < 0x1018a 0x0000 > * U+10190 - U+101CF : Ancient Symbols */ -GRAPH 0x10190 - 0x1019b 0x101a0 -PUNCT 0x10190 - 0x1019b 0x101a0 -PRINT 0x10190 - 0x1019b 0x101a0 -SWIDTH1 0x10190 - 0x1019b 0x101a0 +GRAPH 0x10190 - 0x1019c 0x101a0 +PUNCT 0x10190 - 0x1019c 0x101a0 +PRINT 0x10190 - 0x1019c 0x101a0 +SWIDTH1 0x10190 - 0x1019c 0x101a0 /* @@ -4638,7 +4645,19 @@ TODIGIT < 0x10e79 800 > TODIGIT < 0x10e7a 900 > /* - * U+10E80 - U+10EFF : No_Block + * U+10E80 - U+10EBF : Yezidi + */ + +ALPHA 0x10e80 - 0x10ea9 0x10eab - 0x10eac 0x10eb0 - 0x10eb1 +GRAPH 0x10e80 - 0x10ea9 0x10eab - 0x10ead 0x10eb0 - 0x10eb1 +PUNCT 0x10ead +PRINT 0x10e80 - 0x10ea9 0x10eab - 0x10ead 0x10eb0 - 0x10eb1 +SWIDTH0 0x10eab - 0x10eac +SWIDTH1 0x10e80 - 0x10ea9 0x10ead 0x10eb0 - 0x10eb1 + + +/* + * U+10EC0 - U+10EFF : No_Block */ @@ -4676,10 +4695,25 @@ TODIGIT < 0x10f53 20 > TODIGIT < 0x10f54 100 > /* - * U+10F70 - U+10FDF : No_Block + * U+10F70 - U+10FAF : No_Block */ +/* + * U+10FB0 - U+10FDF : Chorasmian + */ + +ALPHA 0x10fb0 - 0x10fc4 +GRAPH 0x10fb0 - 0x10fcb +PRINT 0x10fb0 - 0x10fcb +SPECIAL 0x10fc5 - 0x10fcb +SWIDTH1 0x10fb0 - 0x10fcb + +TODIGIT < 0x10fc5 - 0x10fc8 : 1 > +TODIGIT < 0x10fc9 10 > +TODIGIT < 0x10fca 20 > +TODIGIT < 0x10fcb 100 > + /* * U+10FE0 - U+10FFF : Elymaic */ @@ -4747,14 +4781,14 @@ TODIGIT < 0x110f0 - 0x110f9 : 0x0000 > * U+11100 - U+1114F : Chakma */ -ALPHA 0x11100 - 0x11132 0x11144 - 0x11146 +ALPHA 0x11100 - 0x11132 0x11144 - 0x11147 DIGIT 0x11136 - 0x1113f -GRAPH 0x11100 - 0x11134 0x11136 - 0x11146 +GRAPH 0x11100 - 0x11134 0x11136 - 0x11147 PUNCT 0x11140 - 0x11143 -PRINT 0x11100 - 0x11134 0x11136 - 0x11146 +PRINT 0x11100 - 0x11134 0x11136 - 0x11147 SPECIAL 0x11133 - 0x11134 SWIDTH0 0x11100 - 0x11102 0x11127 - 0x1112b 0x1112d - 0x11134 -SWIDTH1 0x11103 - 0x11126 0x1112c 0x11136 - 0x11146 +SWIDTH1 0x11103 - 0x11126 0x1112c 0x11136 - 0x11147 TODIGIT < 0x11136 - 0x1113f : 0x0000 > @@ -4775,14 +4809,16 @@ SWIDTH1 0x11150 - 0x11172 0x11174 - 0x11176 * U+11180 - U+111DF : Sharada */ -ALPHA 0x11180 - 0x111bf 0x111c1 - 0x111c4 0x111da 0x111dc +ALPHA 0x11180 - 0x111bf 0x111c1 - 0x111c4 0x111ce - 0x111cf 0x111da +ALPHA 0x111dc DIGIT 0x111d0 - 0x111d9 -GRAPH 0x11180 - 0x111cd 0x111d0 - 0x111df +GRAPH 0x11180 - 0x111df PUNCT 0x111c5 - 0x111c8 0x111cd 0x111db 0x111dd - 0x111df -PRINT 0x11180 - 0x111cd 0x111d0 - 0x111df +PRINT 0x11180 - 0x111df SPECIAL 0x111c0 0x111c9 - 0x111cc -SWIDTH0 0x11180 - 0x11181 0x111b6 - 0x111be 0x111c9 - 0x111cc -SWIDTH1 0x11182 - 0x111b5 0x111bf - 0x111c8 0x111cd 0x111d0 - 0x111df +SWIDTH0 0x11180 - 0x11181 0x111b6 - 0x111be 0x111c9 - 0x111cc 0x111cf +SWIDTH1 0x11182 - 0x111b5 0x111bf - 0x111c8 0x111cd - 0x111ce +SWIDTH1 0x111d0 - 0x111df TODIGIT < 0x111d0 - 0x111d9 : 0x0000 > @@ -4892,15 +4928,16 @@ SWIDTH1 0x1135d - 0x11363 * U+11400 - U+1147F : Newa */ -ALPHA 0x11400 - 0x11441 0x11443 - 0x11445 0x11447 - 0x1144a 0x1145f +ALPHA 0x11400 - 0x11441 0x11443 - 0x11445 0x11447 - 0x1144a +ALPHA 0x1145f - 0x11461 DIGIT 0x11450 - 0x11459 -GRAPH 0x11400 - 0x11459 0x1145b 0x1145d - 0x1145f -PUNCT 0x1144b - 0x1144f 0x1145b 0x1145d -PRINT 0x11400 - 0x11459 0x1145b 0x1145d - 0x1145f +GRAPH 0x11400 - 0x1145b 0x1145d - 0x11461 +PUNCT 0x1144b - 0x1144f 0x1145a - 0x1145b 0x1145d +PRINT 0x11400 - 0x1145b 0x1145d - 0x11461 SPECIAL 0x11442 0x11446 0x1145e SWIDTH0 0x11438 - 0x1143f 0x11442 - 0x11444 0x11446 0x1145e -SWIDTH1 0x11400 - 0x11437 0x11440 - 0x11441 0x11445 0x11447 - 0x11459 -SWIDTH1 0x1145b 0x1145d 0x1145f +SWIDTH1 0x11400 - 0x11437 0x11440 - 0x11441 0x11445 0x11447 - 0x1145b +SWIDTH1 0x1145d 0x1145f - 0x11461 TODIGIT < 0x11450 - 0x11459 : 0x0000 > @@ -5050,7 +5087,30 @@ TODIGIT < 0x118f1 80 > TODIGIT < 0x118f2 90 > /* - * U+11900 - U+1199F : No_Block + * U+11900 - U+1195F : Dives Akuru + */ + +ALPHA 0x11900 - 0x11906 0x11909 0x1190c - 0x11913 0x11915 - 0x11916 +ALPHA 0x11918 - 0x11935 0x11937 - 0x11938 0x1193b - 0x1193c +ALPHA 0x1193f - 0x11942 +DIGIT 0x11950 - 0x11959 +GRAPH 0x11900 - 0x11906 0x11909 0x1190c - 0x11913 0x11915 - 0x11916 +GRAPH 0x11918 - 0x11935 0x11937 - 0x11938 0x1193b - 0x11946 +GRAPH 0x11950 - 0x11959 +PUNCT 0x11944 - 0x11946 +PRINT 0x11900 - 0x11906 0x11909 0x1190c - 0x11913 0x11915 - 0x11916 +PRINT 0x11918 - 0x11935 0x11937 - 0x11938 0x1193b - 0x11946 +PRINT 0x11950 - 0x11959 +SPECIAL 0x1193d - 0x1193e 0x11943 +SWIDTH0 0x1193b - 0x1193c 0x1193e 0x11943 +SWIDTH1 0x11900 - 0x11906 0x11909 0x1190c - 0x11913 0x11915 - 0x11916 +SWIDTH1 0x11918 - 0x11935 0x11937 - 0x11938 0x1193d 0x1193f - 0x11942 +SWIDTH1 0x11944 - 0x11946 0x11950 - 0x11959 + +TODIGIT < 0x11950 - 0x11959 : 0x0000 > + +/* + * U+11960 - U+1199F : No_Block */ @@ -5218,10 +5278,20 @@ SWIDTH1 0x11ee0 - 0x11ef2 0x11ef5 - 0x11ef8 /* - * U+11F00 - U+11FBF : No_Block + * U+11F00 - U+11FAF : No_Block */ +/* + * U+11FB0 - U+11FBF : Lisu Supplement + */ + +ALPHA 0x11fb0 +GRAPH 0x11fb0 +PRINT 0x11fb0 +SWIDTH1 0x11fb0 + + /* * U+11FC0 - U+11FFF : Tamil Supplement */ @@ -5457,11 +5527,13 @@ SWIDTH1 0x16f00 - 0x16f4a 0x16f50 - 0x16f87 0x16f93 - 0x16f9f * U+16FE0 - U+16FFF : Ideographic Symbols and Punctuation */ -ALPHA 0x16fe0 - 0x16fe1 0x16fe3 -GRAPH 0x16fe0 - 0x16fe3 +ALPHA 0x16fe0 - 0x16fe1 0x16fe3 0x16ff0 - 0x16ff1 +GRAPH 0x16fe0 - 0x16fe4 0x16ff0 - 0x16ff1 PUNCT 0x16fe2 -PRINT 0x16fe0 - 0x16fe3 -SWIDTH2 0x16fe0 - 0x16fe3 +PRINT 0x16fe0 - 0x16fe4 0x16ff0 - 0x16ff1 +SPECIAL 0x16fe4 +SWIDTH0 0x16fe4 +SWIDTH2 0x16fe0 - 0x16fe3 0x16ff0 - 0x16ff1 /* @@ -5478,14 +5550,34 @@ SWIDTH2 0x17000 - 0x187f7 * U+18800 - U+18AFF : Tangut Components */ -ALPHA 0x18800 - 0x18af2 -GRAPH 0x18800 - 0x18af2 -PRINT 0x18800 - 0x18af2 -SWIDTH2 0x18800 - 0x18af2 +ALPHA 0x18800 - 0x18aff +GRAPH 0x18800 - 0x18aff +PRINT 0x18800 - 0x18aff +SWIDTH2 0x18800 - 0x18aff + + +/* + * U+18B00 - U+18CFF : Khitan Small Script + */ + +ALPHA 0x18b00 - 0x18cd5 +GRAPH 0x18b00 - 0x18cd5 +PRINT 0x18b00 - 0x18cd5 +SWIDTH2 0x18b00 - 0x18cd5 /* - * U+18B00 - U+1AFFF : No_Block + * U+18D00 - U+18D8F : Tangut Supplement + */ + +ALPHA 0x18d00 - 0x18d08 +GRAPH 0x18d00 - 0x18d08 +PRINT 0x18d00 - 0x18d08 +SWIDTH2 0x18d00 - 0x18d08 + + +/* + * U+18D90 - U+1AFFF : No_Block */ @@ -6046,14 +6138,12 @@ SWIDTH2 0x1f0cf * U+1F100 - U+1F1FF : Enclosed Alphanumeric Supplement */ -GRAPH 0x1f100 - 0x1f10c 0x1f110 - 0x1f16c 0x1f170 - 0x1f1ac -GRAPH 0x1f1e6 - 0x1f1ff -PUNCT 0x1f110 - 0x1f16c 0x1f170 - 0x1f1ac 0x1f1e6 - 0x1f1ff -PRINT 0x1f100 - 0x1f10c 0x1f110 - 0x1f16c 0x1f170 - 0x1f1ac -PRINT 0x1f1e6 - 0x1f1ff +GRAPH 0x1f100 - 0x1f1ad 0x1f1e6 - 0x1f1ff +PUNCT 0x1f10d - 0x1f1ad 0x1f1e6 - 0x1f1ff +PRINT 0x1f100 - 0x1f1ad 0x1f1e6 - 0x1f1ff SPECIAL 0x1f100 - 0x1f10c -SWIDTH1 0x1f100 - 0x1f10c 0x1f110 - 0x1f16c 0x1f170 - 0x1f18d -SWIDTH1 0x1f18f - 0x1f190 0x1f19b - 0x1f1ac 0x1f1e6 - 0x1f1ff +SWIDTH1 0x1f100 - 0x1f18d 0x1f18f - 0x1f190 0x1f19b - 0x1f1ad +SWIDTH1 0x1f1e6 - 0x1f1ff SWIDTH2 0x1f18e 0x1f191 - 0x1f19a TODIGIT < 0x1f100 0x0000 > @@ -6119,13 +6209,13 @@ SWIDTH1 0x1f650 - 0x1f67f * U+1F680 - U+1F6FF : Transport and Map Symbols */ -GRAPH 0x1f680 - 0x1f6d5 0x1f6e0 - 0x1f6ec 0x1f6f0 - 0x1f6fa -PUNCT 0x1f680 - 0x1f6d5 0x1f6e0 - 0x1f6ec 0x1f6f0 - 0x1f6fa -PRINT 0x1f680 - 0x1f6d5 0x1f6e0 - 0x1f6ec 0x1f6f0 - 0x1f6fa +GRAPH 0x1f680 - 0x1f6d7 0x1f6e0 - 0x1f6ec 0x1f6f0 - 0x1f6fc +PUNCT 0x1f680 - 0x1f6d7 0x1f6e0 - 0x1f6ec 0x1f6f0 - 0x1f6fc +PRINT 0x1f680 - 0x1f6d7 0x1f6e0 - 0x1f6ec 0x1f6f0 - 0x1f6fc SWIDTH1 0x1f6c6 - 0x1f6cb 0x1f6cd - 0x1f6cf 0x1f6d3 - 0x1f6d4 SWIDTH1 0x1f6e0 - 0x1f6ea 0x1f6f0 - 0x1f6f3 -SWIDTH2 0x1f680 - 0x1f6c5 0x1f6cc 0x1f6d0 - 0x1f6d2 0x1f6d5 -SWIDTH2 0x1f6eb - 0x1f6ec 0x1f6f4 - 0x1f6fa +SWIDTH2 0x1f680 - 0x1f6c5 0x1f6cc 0x1f6d0 - 0x1f6d2 0x1f6d5 - 0x1f6d7 +SWIDTH2 0x1f6eb - 0x1f6ec 0x1f6f4 - 0x1f6fc /* @@ -6154,31 +6244,25 @@ SWIDTH2 0x1f7e0 - 0x1f7eb */ GRAPH 0x1f800 - 0x1f80b 0x1f810 - 0x1f847 0x1f850 - 0x1f859 -GRAPH 0x1f860 - 0x1f887 0x1f890 - 0x1f8ad +GRAPH 0x1f860 - 0x1f887 0x1f890 - 0x1f8ad 0x1f8b0 - 0x1f8b1 PUNCT 0x1f800 - 0x1f80b 0x1f810 - 0x1f847 0x1f850 - 0x1f859 -PUNCT 0x1f860 - 0x1f887 0x1f890 - 0x1f8ad +PUNCT 0x1f860 - 0x1f887 0x1f890 - 0x1f8ad 0x1f8b0 - 0x1f8b1 PRINT 0x1f800 - 0x1f80b 0x1f810 - 0x1f847 0x1f850 - 0x1f859 -PRINT 0x1f860 - 0x1f887 0x1f890 - 0x1f8ad +PRINT 0x1f860 - 0x1f887 0x1f890 - 0x1f8ad 0x1f8b0 - 0x1f8b1 SWIDTH1 0x1f800 - 0x1f80b 0x1f810 - 0x1f847 0x1f850 - 0x1f859 -SWIDTH1 0x1f860 - 0x1f887 0x1f890 - 0x1f8ad +SWIDTH1 0x1f860 - 0x1f887 0x1f890 - 0x1f8ad 0x1f8b0 - 0x1f8b1 /* * U+1F900 - U+1F9FF : Supplemental Symbols and Pictographs */ -GRAPH 0x1f900 - 0x1f90b 0x1f90d - 0x1f971 0x1f973 - 0x1f976 -GRAPH 0x1f97a - 0x1f9a2 0x1f9a5 - 0x1f9aa 0x1f9ae - 0x1f9ca -GRAPH 0x1f9cd - 0x1f9ff -PUNCT 0x1f900 - 0x1f90b 0x1f90d - 0x1f971 0x1f973 - 0x1f976 -PUNCT 0x1f97a - 0x1f9a2 0x1f9a5 - 0x1f9aa 0x1f9ae - 0x1f9ca -PUNCT 0x1f9cd - 0x1f9ff -PRINT 0x1f900 - 0x1f90b 0x1f90d - 0x1f971 0x1f973 - 0x1f976 -PRINT 0x1f97a - 0x1f9a2 0x1f9a5 - 0x1f9aa 0x1f9ae - 0x1f9ca -PRINT 0x1f9cd - 0x1f9ff -SWIDTH1 0x1f900 - 0x1f90b -SWIDTH2 0x1f90d - 0x1f971 0x1f973 - 0x1f976 0x1f97a - 0x1f9a2 -SWIDTH2 0x1f9a5 - 0x1f9aa 0x1f9ae - 0x1f9ca 0x1f9cd - 0x1f9ff +GRAPH 0x1f900 - 0x1f978 0x1f97a - 0x1f9cb 0x1f9cd - 0x1f9ff +PUNCT 0x1f900 - 0x1f978 0x1f97a - 0x1f9cb 0x1f9cd - 0x1f9ff +PRINT 0x1f900 - 0x1f978 0x1f97a - 0x1f9cb 0x1f9cd - 0x1f9ff +SWIDTH1 0x1f900 - 0x1f90b 0x1f93b 0x1f946 +SWIDTH2 0x1f90c - 0x1f93a 0x1f93c - 0x1f945 0x1f947 - 0x1f978 +SWIDTH2 0x1f97a - 0x1f9cb 0x1f9cd - 0x1f9ff /* @@ -6195,18 +6279,34 @@ SWIDTH1 0x1fa00 - 0x1fa53 0x1fa60 - 0x1fa6d * U+1FA70 - U+1FAFF : Symbols and Pictographs Extended-A */ -GRAPH 0x1fa70 - 0x1fa73 0x1fa78 - 0x1fa7a 0x1fa80 - 0x1fa82 -GRAPH 0x1fa90 - 0x1fa95 -PUNCT 0x1fa70 - 0x1fa73 0x1fa78 - 0x1fa7a 0x1fa80 - 0x1fa82 -PUNCT 0x1fa90 - 0x1fa95 -PRINT 0x1fa70 - 0x1fa73 0x1fa78 - 0x1fa7a 0x1fa80 - 0x1fa82 -PRINT 0x1fa90 - 0x1fa95 -SWIDTH2 0x1fa70 - 0x1fa73 0x1fa78 - 0x1fa7a 0x1fa80 - 0x1fa82 -SWIDTH2 0x1fa90 - 0x1fa95 +GRAPH 0x1fa70 - 0x1fa74 0x1fa78 - 0x1fa7a 0x1fa80 - 0x1fa86 +GRAPH 0x1fa90 - 0x1faa8 0x1fab0 - 0x1fab6 0x1fac0 - 0x1fac2 +GRAPH 0x1fad0 - 0x1fad6 +PUNCT 0x1fa70 - 0x1fa74 0x1fa78 - 0x1fa7a 0x1fa80 - 0x1fa86 +PUNCT 0x1fa90 - 0x1faa8 0x1fab0 - 0x1fab6 0x1fac0 - 0x1fac2 +PUNCT 0x1fad0 - 0x1fad6 +PRINT 0x1fa70 - 0x1fa74 0x1fa78 - 0x1fa7a 0x1fa80 - 0x1fa86 +PRINT 0x1fa90 - 0x1faa8 0x1fab0 - 0x1fab6 0x1fac0 - 0x1fac2 +PRINT 0x1fad0 - 0x1fad6 +SWIDTH2 0x1fa70 - 0x1fa74 0x1fa78 - 0x1fa7a 0x1fa80 - 0x1fa86 +SWIDTH2 0x1fa90 - 0x1faa8 0x1fab0 - 0x1fab6 0x1fac0 - 0x1fac2 +SWIDTH2 0x1fad0 - 0x1fad6 + + +/* + * U+1FB00 - U+1FBFF : Symbols for Legacy Computing + */ + +DIGIT 0x1fbf0 - 0x1fbf9 +GRAPH 0x1fb00 - 0x1fb92 0x1fb94 - 0x1fbca 0x1fbf0 - 0x1fbf9 +PUNCT 0x1fb00 - 0x1fb92 0x1fb94 - 0x1fbca +PRINT 0x1fb00 - 0x1fb92 0x1fb94 - 0x1fbca 0x1fbf0 - 0x1fbf9 +SWIDTH1 0x1fb00 - 0x1fb92 0x1fb94 - 0x1fbca 0x1fbf0 - 0x1fbf9 +TODIGIT < 0x1fbf0 - 0x1fbf9 : 0x0000 > /* - * U+1FB00 - U+1FFFF : No_Block + * U+1FC00 - U+1FFFF : No_Block */ @@ -6218,10 +6318,10 @@ ALPHA 0x20000 0x20002 - 0x20063 0x20065 - 0x200e1 0x200e3 - 0x20120 ALPHA 0x20122 - 0x20929 0x2092b - 0x20982 0x20984 - 0x2098b ALPHA 0x2098d - 0x2099b 0x2099d - 0x20ae9 0x20aeb - 0x20afc ALPHA 0x20afe - 0x20b18 0x20b1a - 0x2238f 0x22391 - 0x22997 -ALPHA 0x22999 - 0x23b1a 0x23b1c - 0x2626c 0x2626e - 0x2a6d6 -GRAPH 0x20000 - 0x2a6d6 -PRINT 0x20000 - 0x2a6d6 -SWIDTH2 0x20000 - 0x2a6d6 +ALPHA 0x22999 - 0x23b1a 0x23b1c - 0x2626c 0x2626e - 0x2a6dd +GRAPH 0x20000 - 0x2a6dd +PRINT 0x20000 - 0x2a6dd +SWIDTH2 0x20000 - 0x2a6dd TODIGIT < 0x20001 7 > TODIGIT < 0x20064 4 > @@ -6301,7 +6401,22 @@ SWIDTH2 0x2f800 - 0x2fa1d TODIGIT < 0x2f890 9 > /* - * U+2FA20 - U+DFFFF : No_Block + * U+2FA20 - U+2FFFF : No_Block + */ + + +/* + * U+30000 - U+3134F : CJK Unified Ideographs Extension G + */ + +ALPHA 0x30000 - 0x3134a +GRAPH 0x30000 - 0x3134a +PRINT 0x30000 - 0x3134a +SWIDTH2 0x30000 - 0x3134a + + +/* + * U+31350 - U+DFFFF : No_Block */ diff --git a/share/locale/ctype/gen_ctype_utf8.pl b/share/locale/ctype/gen_ctype_utf8.pl index e23472efb2c..559a1b349dc 100755 --- a/share/locale/ctype/gen_ctype_utf8.pl +++ b/share/locale/ctype/gen_ctype_utf8.pl @@ -1,5 +1,5 @@ #!/usr/bin/env perl -# $OpenBSD: gen_ctype_utf8.pl,v 1.5 2020/02/20 02:51:56 afresh1 Exp $ # +# $OpenBSD: gen_ctype_utf8.pl,v 1.6 2021/05/16 22:38:32 afresh1 Exp $ # use 5.022; use warnings; @@ -53,7 +53,7 @@ print <<'EOL'; /* * COPYRIGHT AND PERMISSION NOTICE * - * Copyright (c) 1991-2019 Unicode, Inc. All rights reserved. + * Copyright (c) 1991-2021 Unicode, Inc. All rights reserved. * Distributed under the Terms of Use in * https://www.unicode.org/copyright.html. * -- 2.20.1