Update Unicode data to 13.0.0.
Change-Id: Iff57514e09d6e4e141384dd0cf138314eb1435f1
Reviewed-on: https://code-review.googlesource.com/c/re2/+/53330
Reviewed-by: Paul Wankadia <junyer@google.com>
diff --git a/doc/syntax.html b/doc/syntax.html
index 8a268a1..47541e5 100644
--- a/doc/syntax.html
+++ b/doc/syntax.html
@@ -264,6 +264,7 @@
<tr><td colspan=2>Chakma</td></tr>
<tr><td colspan=2>Cham</td></tr>
<tr><td colspan=2>Cherokee</td></tr>
+<tr><td colspan=2>Chorasmian</td></tr>
<tr><td colspan=2>Common</td></tr>
<tr><td colspan=2>Coptic</td></tr>
<tr><td colspan=2>Cuneiform</td></tr>
@@ -271,6 +272,7 @@
<tr><td colspan=2>Cyrillic</td></tr>
<tr><td colspan=2>Deseret</td></tr>
<tr><td colspan=2>Devanagari</td></tr>
+<tr><td colspan=2>Dives_Akuru</td></tr>
<tr><td colspan=2>Dogra</td></tr>
<tr><td colspan=2>Duployan</td></tr>
<tr><td colspan=2>Egyptian_Hieroglyphs</td></tr>
@@ -302,6 +304,7 @@
<tr><td colspan=2>Katakana</td></tr>
<tr><td colspan=2>Kayah_Li</td></tr>
<tr><td colspan=2>Kharoshthi</td></tr>
+<tr><td colspan=2>Khitan_Small_Script</td></tr>
<tr><td colspan=2>Khmer</td></tr>
<tr><td colspan=2>Khojki</td></tr>
<tr><td colspan=2>Khudawadi</td></tr>
@@ -391,6 +394,7 @@
<tr><td colspan=2>Vai</td></tr>
<tr><td colspan=2>Wancho</td></tr>
<tr><td colspan=2>Warang_Citi</td></tr>
+<tr><td colspan=2>Yezidi</td></tr>
<tr><td colspan=2>Yi</td></tr>
<tr><td colspan=2>Zanabazar_Square</td></tr>
<tr><td></td></tr>
diff --git a/doc/syntax.txt b/doc/syntax.txt
index cb04bbf..ce87866 100644
--- a/doc/syntax.txt
+++ b/doc/syntax.txt
@@ -253,6 +253,7 @@
Chakma
Cham
Cherokee
+Chorasmian
Common
Coptic
Cuneiform
@@ -260,6 +261,7 @@
Cyrillic
Deseret
Devanagari
+Dives_Akuru
Dogra
Duployan
Egyptian_Hieroglyphs
@@ -291,6 +293,7 @@
Katakana
Kayah_Li
Kharoshthi
+Khitan_Small_Script
Khmer
Khojki
Khudawadi
@@ -380,6 +383,7 @@
Vai
Wancho
Warang_Citi
+Yezidi
Yi
Zanabazar_Square
diff --git a/re2/unicode.py b/re2/unicode.py
index 56ca811..e0f33ef 100644
--- a/re2/unicode.py
+++ b/re2/unicode.py
@@ -13,7 +13,7 @@
from six.moves import urllib
# Directory or URL where Unicode tables reside.
-_UNICODE_DIR = "https://www.unicode.org/Public/12.1.0/ucd"
+_UNICODE_DIR = "https://www.unicode.org/Public/13.0.0/ucd"
# Largest valid Unicode code value.
_RUNE_MAX = 0x10FFFF
diff --git a/re2/unicode_casefold.cc b/re2/unicode_casefold.cc
index 4ea2533..8424107 100644
--- a/re2/unicode_casefold.cc
+++ b/re2/unicode_casefold.cc
@@ -7,7 +7,7 @@
namespace re2 {
-// 1381 groups, 2792 pairs, 356 ranges
+// 1384 groups, 2798 pairs, 358 ranges
const CaseFold unicode_casefold[] = {
{ 65, 90, 32 },
{ 97, 106, -32 },
@@ -349,6 +349,8 @@
{ 42948, 42948, -48 },
{ 42949, 42949, -42307 },
{ 42950, 42950, -35384 },
+ { 42951, 42954, OddEven },
+ { 42997, 42998, OddEven },
{ 43859, 43859, -928 },
{ 43888, 43967, -38864 },
{ 65313, 65338, 32 },
@@ -366,9 +368,9 @@
{ 125184, 125217, 34 },
{ 125218, 125251, -34 },
};
-const int num_unicode_casefold = 356;
+const int num_unicode_casefold = 358;
-// 1381 groups, 1411 pairs, 198 ranges
+// 1384 groups, 1414 pairs, 200 ranges
const CaseFold unicode_tolower[] = {
{ 65, 90, 32 },
{ 181, 181, 775 },
@@ -560,6 +562,8 @@
{ 42948, 42948, -48 },
{ 42949, 42949, -42307 },
{ 42950, 42950, -35384 },
+ { 42951, 42953, OddEvenSkip },
+ { 42997, 42997, OddEven },
{ 43888, 43967, -38864 },
{ 65313, 65338, 32 },
{ 66560, 66599, 40 },
@@ -569,7 +573,7 @@
{ 93760, 93791, 32 },
{ 125184, 125217, 34 },
};
-const int num_unicode_tolower = 198;
+const int num_unicode_tolower = 200;
diff --git a/re2/unicode_groups.cc b/re2/unicode_groups.cc
index 63e6116..7b7a3c6 100644
--- a/re2/unicode_groups.cc
+++ b/re2/unicode_groups.cc
@@ -125,7 +125,7 @@
{ 2112, 2136 },
{ 2144, 2154 },
{ 2208, 2228 },
- { 2230, 2237 },
+ { 2230, 2247 },
{ 2308, 2361 },
{ 2365, 2365 },
{ 2384, 2384 },
@@ -201,7 +201,7 @@
{ 3294, 3294 },
{ 3296, 3297 },
{ 3313, 3314 },
- { 3333, 3340 },
+ { 3332, 3340 },
{ 3342, 3344 },
{ 3346, 3386 },
{ 3389, 3389 },
@@ -372,10 +372,10 @@
{ 12540, 12543 },
{ 12549, 12591 },
{ 12593, 12686 },
- { 12704, 12730 },
+ { 12704, 12735 },
{ 12784, 12799 },
- { 13312, 19893 },
- { 19968, 40943 },
+ { 13312, 19903 },
+ { 19968, 40956 },
{ 40960, 42124 },
{ 42192, 42237 },
{ 42240, 42508 },
@@ -387,8 +387,8 @@
{ 42775, 42783 },
{ 42786, 42888 },
{ 42891, 42943 },
- { 42946, 42950 },
- { 42999, 43009 },
+ { 42946, 42954 },
+ { 42997, 43009 },
{ 43011, 43013 },
{ 43015, 43018 },
{ 43020, 43042 },
@@ -425,7 +425,7 @@
{ 43808, 43814 },
{ 43816, 43822 },
{ 43824, 43866 },
- { 43868, 43879 },
+ { 43868, 43881 },
{ 43888, 44002 },
{ 44032, 55203 },
{ 55216, 55238 },
@@ -511,15 +511,19 @@
{ 68736, 68786 },
{ 68800, 68850 },
{ 68864, 68899 },
+ { 69248, 69289 },
+ { 69296, 69297 },
{ 69376, 69404 },
{ 69415, 69415 },
{ 69424, 69445 },
+ { 69552, 69572 },
{ 69600, 69622 },
{ 69635, 69687 },
{ 69763, 69807 },
{ 69840, 69864 },
{ 69891, 69926 },
{ 69956, 69956 },
+ { 69959, 69959 },
{ 69968, 70002 },
{ 70006, 70006 },
{ 70019, 70066 },
@@ -545,7 +549,7 @@
{ 70493, 70497 },
{ 70656, 70708 },
{ 70727, 70730 },
- { 70751, 70751 },
+ { 70751, 70753 },
{ 70784, 70831 },
{ 70852, 70853 },
{ 70855, 70855 },
@@ -558,7 +562,13 @@
{ 71424, 71450 },
{ 71680, 71723 },
{ 71840, 71903 },
- { 71935, 71935 },
+ { 71935, 71942 },
+ { 71945, 71945 },
+ { 71948, 71955 },
+ { 71957, 71958 },
+ { 71960, 71983 },
+ { 71999, 71999 },
+ { 72001, 72001 },
{ 72096, 72103 },
{ 72106, 72144 },
{ 72161, 72161 },
@@ -583,6 +593,7 @@
{ 73066, 73097 },
{ 73112, 73112 },
{ 73440, 73458 },
+ { 73648, 73648 },
{ 73728, 74649 },
{ 74880, 75075 },
{ 77824, 78894 },
@@ -601,7 +612,8 @@
{ 94176, 94177 },
{ 94179, 94179 },
{ 94208, 100343 },
- { 100352, 101106 },
+ { 100352, 101589 },
+ { 101632, 101640 },
{ 110592, 110878 },
{ 110928, 110930 },
{ 110948, 110951 },
@@ -680,12 +692,13 @@
{ 126625, 126627 },
{ 126629, 126633 },
{ 126635, 126651 },
- { 131072, 173782 },
+ { 131072, 173789 },
{ 173824, 177972 },
{ 177984, 178205 },
{ 178208, 183969 },
{ 183984, 191456 },
{ 194560, 195101 },
+ { 196608, 201546 },
};
static const URange16 Ll_range16[] = {
{ 97, 122 },
@@ -1289,9 +1302,12 @@
{ 42941, 42941 },
{ 42943, 42943 },
{ 42947, 42947 },
+ { 42952, 42952 },
+ { 42954, 42954 },
+ { 42998, 42998 },
{ 43002, 43002 },
{ 43824, 43866 },
- { 43872, 43879 },
+ { 43872, 43880 },
{ 43888, 43967 },
{ 64256, 64262 },
{ 64275, 64279 },
@@ -1386,6 +1402,7 @@
{ 43741, 43741 },
{ 43763, 43764 },
{ 43868, 43871 },
+ { 43881, 43881 },
{ 65392, 65392 },
{ 65438, 65439 },
};
@@ -1422,7 +1439,7 @@
{ 2112, 2136 },
{ 2144, 2154 },
{ 2208, 2228 },
- { 2230, 2237 },
+ { 2230, 2247 },
{ 2308, 2361 },
{ 2365, 2365 },
{ 2384, 2384 },
@@ -1498,7 +1515,7 @@
{ 3294, 3294 },
{ 3296, 3297 },
{ 3313, 3314 },
- { 3333, 3340 },
+ { 3332, 3340 },
{ 3342, 3344 },
{ 3346, 3386 },
{ 3389, 3389 },
@@ -1611,10 +1628,10 @@
{ 12543, 12543 },
{ 12549, 12591 },
{ 12593, 12686 },
- { 12704, 12730 },
+ { 12704, 12735 },
{ 12784, 12799 },
- { 13312, 19893 },
- { 19968, 40943 },
+ { 13312, 19903 },
+ { 19968, 40956 },
{ 40960, 40980 },
{ 40982, 42124 },
{ 42192, 42231 },
@@ -1740,15 +1757,19 @@
{ 68480, 68497 },
{ 68608, 68680 },
{ 68864, 68899 },
+ { 69248, 69289 },
+ { 69296, 69297 },
{ 69376, 69404 },
{ 69415, 69415 },
{ 69424, 69445 },
+ { 69552, 69572 },
{ 69600, 69622 },
{ 69635, 69687 },
{ 69763, 69807 },
{ 69840, 69864 },
{ 69891, 69926 },
{ 69956, 69956 },
+ { 69959, 69959 },
{ 69968, 70002 },
{ 70006, 70006 },
{ 70019, 70066 },
@@ -1774,7 +1795,7 @@
{ 70493, 70497 },
{ 70656, 70708 },
{ 70727, 70730 },
- { 70751, 70751 },
+ { 70751, 70753 },
{ 70784, 70831 },
{ 70852, 70853 },
{ 70855, 70855 },
@@ -1786,7 +1807,13 @@
{ 71352, 71352 },
{ 71424, 71450 },
{ 71680, 71723 },
- { 71935, 71935 },
+ { 71935, 71942 },
+ { 71945, 71945 },
+ { 71948, 71955 },
+ { 71957, 71958 },
+ { 71960, 71983 },
+ { 71999, 71999 },
+ { 72001, 72001 },
{ 72096, 72103 },
{ 72106, 72144 },
{ 72161, 72161 },
@@ -1811,6 +1838,7 @@
{ 73066, 73097 },
{ 73112, 73112 },
{ 73440, 73458 },
+ { 73648, 73648 },
{ 73728, 74649 },
{ 74880, 75075 },
{ 77824, 78894 },
@@ -1824,7 +1852,8 @@
{ 93952, 94026 },
{ 94032, 94032 },
{ 94208, 100343 },
- { 100352, 101106 },
+ { 100352, 101589 },
+ { 101632, 101640 },
{ 110592, 110878 },
{ 110928, 110930 },
{ 110948, 110951 },
@@ -1870,12 +1899,13 @@
{ 126625, 126627 },
{ 126629, 126633 },
{ 126635, 126651 },
- { 131072, 173782 },
+ { 131072, 173789 },
{ 173824, 177972 },
{ 177984, 178205 },
{ 178208, 183969 },
{ 183984, 191456 },
{ 194560, 195101 },
+ { 196608, 201546 },
};
static const URange16 Lt_range16[] = {
{ 453, 453 },
@@ -2487,7 +2517,9 @@
{ 42940, 42940 },
{ 42942, 42942 },
{ 42946, 42946 },
- { 42948, 42950 },
+ { 42948, 42951 },
+ { 42953, 42953 },
+ { 42997, 42997 },
{ 65313, 65338 },
};
static const URange32 Lu_range32[] = {
@@ -2588,7 +2620,7 @@
{ 2878, 2884 },
{ 2887, 2888 },
{ 2891, 2893 },
- { 2902, 2903 },
+ { 2901, 2903 },
{ 2914, 2915 },
{ 2946, 2946 },
{ 3006, 3010 },
@@ -2615,7 +2647,7 @@
{ 3402, 3405 },
{ 3415, 3415 },
{ 3426, 3427 },
- { 3458, 3459 },
+ { 3457, 3459 },
{ 3530, 3530 },
{ 3535, 3540 },
{ 3542, 3542 },
@@ -2662,7 +2694,7 @@
{ 6741, 6750 },
{ 6752, 6780 },
{ 6783, 6783 },
- { 6832, 6846 },
+ { 6832, 6848 },
{ 6912, 6916 },
{ 6964, 6980 },
{ 7019, 7027 },
@@ -2691,6 +2723,7 @@
{ 43014, 43014 },
{ 43019, 43019 },
{ 43043, 43047 },
+ { 43052, 43052 },
{ 43136, 43137 },
{ 43188, 43205 },
{ 43232, 43249 },
@@ -2728,6 +2761,7 @@
{ 68159, 68159 },
{ 68325, 68326 },
{ 68900, 68903 },
+ { 69291, 69292 },
{ 69446, 69456 },
{ 69632, 69634 },
{ 69688, 69702 },
@@ -2740,6 +2774,7 @@
{ 70016, 70018 },
{ 70067, 70080 },
{ 70089, 70092 },
+ { 70094, 70095 },
{ 70188, 70199 },
{ 70206, 70206 },
{ 70367, 70378 },
@@ -2762,6 +2797,11 @@
{ 71339, 71351 },
{ 71453, 71467 },
{ 71724, 71738 },
+ { 71984, 71989 },
+ { 71991, 71992 },
+ { 71995, 71998 },
+ { 72000, 72000 },
+ { 72002, 72003 },
{ 72145, 72151 },
{ 72154, 72160 },
{ 72164, 72164 },
@@ -2789,6 +2829,8 @@
{ 94031, 94031 },
{ 94033, 94087 },
{ 94095, 94098 },
+ { 94180, 94180 },
+ { 94192, 94193 },
{ 113821, 113822 },
{ 119141, 119145 },
{ 119149, 119154 },
@@ -2935,6 +2977,7 @@
{ 70018, 70018 },
{ 70067, 70069 },
{ 70079, 70080 },
+ { 70094, 70094 },
{ 70188, 70190 },
{ 70194, 70195 },
{ 70197, 70197 },
@@ -2966,6 +3009,11 @@
{ 71462, 71462 },
{ 71724, 71726 },
{ 71736, 71736 },
+ { 71984, 71989 },
+ { 71991, 71992 },
+ { 71997, 71997 },
+ { 72000, 72000 },
+ { 72002, 72002 },
{ 72145, 72147 },
{ 72156, 72159 },
{ 72164, 72164 },
@@ -2982,6 +3030,7 @@
{ 73110, 73110 },
{ 73461, 73462 },
{ 94033, 94087 },
+ { 94192, 94193 },
{ 119141, 119142 },
{ 119149, 119154 },
};
@@ -3051,7 +3100,7 @@
{ 2879, 2879 },
{ 2881, 2884 },
{ 2893, 2893 },
- { 2902, 2902 },
+ { 2901, 2902 },
{ 2914, 2915 },
{ 2946, 2946 },
{ 3008, 3008 },
@@ -3074,6 +3123,7 @@
{ 3393, 3396 },
{ 3405, 3405 },
{ 3426, 3427 },
+ { 3457, 3457 },
{ 3530, 3530 },
{ 3538, 3540 },
{ 3542, 3542 },
@@ -3131,6 +3181,7 @@
{ 6771, 6780 },
{ 6783, 6783 },
{ 6832, 6845 },
+ { 6847, 6848 },
{ 6912, 6915 },
{ 6964, 6964 },
{ 6966, 6970 },
@@ -3171,6 +3222,7 @@
{ 43014, 43014 },
{ 43019, 43019 },
{ 43045, 43046 },
+ { 43052, 43052 },
{ 43204, 43205 },
{ 43232, 43249 },
{ 43263, 43263 },
@@ -3212,6 +3264,7 @@
{ 68159, 68159 },
{ 68325, 68326 },
{ 68900, 68903 },
+ { 69291, 69292 },
{ 69446, 69456 },
{ 69633, 69633 },
{ 69688, 69702 },
@@ -3225,6 +3278,7 @@
{ 70016, 70017 },
{ 70070, 70078 },
{ 70089, 70092 },
+ { 70095, 70095 },
{ 70191, 70193 },
{ 70196, 70196 },
{ 70198, 70199 },
@@ -3260,6 +3314,9 @@
{ 71463, 71467 },
{ 71727, 71735 },
{ 71737, 71738 },
+ { 71995, 71996 },
+ { 71998, 71998 },
+ { 72003, 72003 },
{ 72148, 72151 },
{ 72154, 72155 },
{ 72160, 72160 },
@@ -3291,6 +3348,7 @@
{ 92976, 92982 },
{ 94031, 94031 },
{ 94095, 94098 },
+ { 94180, 94180 },
{ 113821, 113822 },
{ 119143, 119145 },
{ 119163, 119170 },
@@ -3413,6 +3471,7 @@
{ 69216, 69246 },
{ 69405, 69414 },
{ 69457, 69460 },
+ { 69573, 69579 },
{ 69714, 69743 },
{ 69872, 69881 },
{ 69942, 69951 },
@@ -3425,6 +3484,7 @@
{ 71360, 71369 },
{ 71472, 71483 },
{ 71904, 71922 },
+ { 72016, 72025 },
{ 72784, 72812 },
{ 73040, 73049 },
{ 73120, 73129 },
@@ -3447,6 +3507,7 @@
{ 126209, 126253 },
{ 126255, 126269 },
{ 127232, 127244 },
+ { 130032, 130041 },
};
static const URange16 Nd_range16[] = {
{ 48, 57 },
@@ -3501,6 +3562,7 @@
{ 71360, 71369 },
{ 71472, 71481 },
{ 71904, 71913 },
+ { 72016, 72025 },
{ 72784, 72793 },
{ 73040, 73049 },
{ 73120, 73129 },
@@ -3510,6 +3572,7 @@
{ 123200, 123209 },
{ 123632, 123641 },
{ 125264, 125273 },
+ { 130032, 130041 },
};
static const URange16 Nl_range16[] = {
{ 5870, 5872 },
@@ -3583,6 +3646,7 @@
{ 69216, 69246 },
{ 69405, 69414 },
{ 69457, 69460 },
+ { 69573, 69579 },
{ 69714, 69733 },
{ 70113, 70132 },
{ 71482, 71483 },
@@ -3692,6 +3756,7 @@
{ 11632, 11632 },
{ 11776, 11822 },
{ 11824, 11855 },
+ { 11858, 11858 },
{ 12289, 12291 },
{ 12296, 12305 },
{ 12308, 12319 },
@@ -3747,6 +3812,7 @@
{ 68336, 68342 },
{ 68409, 68415 },
{ 68505, 68508 },
+ { 69293, 69293 },
{ 69461, 69465 },
{ 69703, 69709 },
{ 69819, 69820 },
@@ -3760,7 +3826,7 @@
{ 70200, 70205 },
{ 70313, 70313 },
{ 70731, 70735 },
- { 70747, 70747 },
+ { 70746, 70747 },
{ 70749, 70749 },
{ 70854, 70854 },
{ 71105, 71127 },
@@ -3768,6 +3834,7 @@
{ 71264, 71276 },
{ 71484, 71486 },
{ 71739, 71739 },
+ { 72004, 72006 },
{ 72162, 72162 },
{ 72255, 72262 },
{ 72346, 72348 },
@@ -3814,6 +3881,9 @@
{ 65123, 65123 },
{ 65293, 65293 },
};
+static const URange32 Pd_range32[] = {
+ { 69293, 69293 },
+};
static const URange16 Pe_range16[] = {
{ 41, 41 },
{ 93, 93 },
@@ -4002,6 +4072,7 @@
{ 11836, 11839 },
{ 11841, 11841 },
{ 11843, 11855 },
+ { 11858, 11858 },
{ 12289, 12291 },
{ 12349, 12349 },
{ 12539, 12539 },
@@ -4069,7 +4140,7 @@
{ 70200, 70205 },
{ 70313, 70313 },
{ 70731, 70735 },
- { 70747, 70747 },
+ { 70746, 70747 },
{ 70749, 70749 },
{ 70854, 70854 },
{ 71105, 71127 },
@@ -4077,6 +4148,7 @@
{ 71264, 71276 },
{ 71484, 71486 },
{ 71739, 71739 },
+ { 72004, 72006 },
{ 72162, 72162 },
{ 72255, 72262 },
{ 72346, 72348 },
@@ -4274,8 +4346,9 @@
{ 10716, 10747 },
{ 10750, 11123 },
{ 11126, 11157 },
- { 11160, 11263 },
+ { 11159, 11263 },
{ 11493, 11498 },
+ { 11856, 11857 },
{ 11904, 11929 },
{ 11931, 12019 },
{ 12032, 12245 },
@@ -4304,6 +4377,7 @@
{ 43062, 43065 },
{ 43639, 43641 },
{ 43867, 43867 },
+ { 43882, 43883 },
{ 64297, 64297 },
{ 64434, 64449 },
{ 65020, 65021 },
@@ -4325,7 +4399,7 @@
{ 65847, 65855 },
{ 65913, 65929 },
{ 65932, 65934 },
- { 65936, 65947 },
+ { 65936, 65948 },
{ 65952, 65952 },
{ 66000, 66044 },
{ 67703, 67704 },
@@ -4372,16 +4446,15 @@
{ 127153, 127167 },
{ 127169, 127183 },
{ 127185, 127221 },
- { 127248, 127340 },
- { 127344, 127404 },
+ { 127245, 127405 },
{ 127462, 127490 },
{ 127504, 127547 },
{ 127552, 127560 },
{ 127568, 127569 },
{ 127584, 127589 },
- { 127744, 128725 },
+ { 127744, 128727 },
{ 128736, 128748 },
- { 128752, 128762 },
+ { 128752, 128764 },
{ 128768, 128883 },
{ 128896, 128984 },
{ 128992, 129003 },
@@ -4390,18 +4463,20 @@
{ 129104, 129113 },
{ 129120, 129159 },
{ 129168, 129197 },
- { 129280, 129291 },
- { 129293, 129393 },
- { 129395, 129398 },
- { 129402, 129442 },
- { 129445, 129450 },
- { 129454, 129482 },
+ { 129200, 129201 },
+ { 129280, 129400 },
+ { 129402, 129483 },
{ 129485, 129619 },
{ 129632, 129645 },
- { 129648, 129651 },
+ { 129648, 129652 },
{ 129656, 129658 },
- { 129664, 129666 },
- { 129680, 129685 },
+ { 129664, 129670 },
+ { 129680, 129704 },
+ { 129712, 129718 },
+ { 129728, 129730 },
+ { 129744, 129750 },
+ { 129792, 129938 },
+ { 129940, 129994 },
};
static const URange16 Sc_range16[] = {
{ 36, 36 },
@@ -4453,6 +4528,7 @@
{ 42784, 42785 },
{ 42889, 42890 },
{ 43867, 43867 },
+ { 43882, 43883 },
{ 64434, 64449 },
{ 65342, 65342 },
{ 65344, 65344 },
@@ -4610,8 +4686,9 @@
{ 11077, 11078 },
{ 11085, 11123 },
{ 11126, 11157 },
- { 11160, 11263 },
+ { 11159, 11263 },
{ 11493, 11498 },
+ { 11856, 11857 },
{ 11904, 11929 },
{ 11931, 12019 },
{ 12032, 12245 },
@@ -4646,7 +4723,7 @@
{ 65847, 65855 },
{ 65913, 65929 },
{ 65932, 65934 },
- { 65936, 65947 },
+ { 65936, 65948 },
{ 65952, 65952 },
{ 66000, 66044 },
{ 67703, 67704 },
@@ -4681,17 +4758,16 @@
{ 127153, 127167 },
{ 127169, 127183 },
{ 127185, 127221 },
- { 127248, 127340 },
- { 127344, 127404 },
+ { 127245, 127405 },
{ 127462, 127490 },
{ 127504, 127547 },
{ 127552, 127560 },
{ 127568, 127569 },
{ 127584, 127589 },
{ 127744, 127994 },
- { 128000, 128725 },
+ { 128000, 128727 },
{ 128736, 128748 },
- { 128752, 128762 },
+ { 128752, 128764 },
{ 128768, 128883 },
{ 128896, 128984 },
{ 128992, 129003 },
@@ -4700,18 +4776,20 @@
{ 129104, 129113 },
{ 129120, 129159 },
{ 129168, 129197 },
- { 129280, 129291 },
- { 129293, 129393 },
- { 129395, 129398 },
- { 129402, 129442 },
- { 129445, 129450 },
- { 129454, 129482 },
+ { 129200, 129201 },
+ { 129280, 129400 },
+ { 129402, 129483 },
{ 129485, 129619 },
{ 129632, 129645 },
- { 129648, 129651 },
+ { 129648, 129652 },
{ 129656, 129658 },
- { 129664, 129666 },
- { 129680, 129685 },
+ { 129664, 129670 },
+ { 129680, 129704 },
+ { 129712, 129718 },
+ { 129728, 129730 },
+ { 129744, 129750 },
+ { 129792, 129938 },
+ { 129940, 129994 },
};
static const URange16 Z_range16[] = {
{ 32, 32 },
@@ -4764,7 +4842,7 @@
{ 1758, 1791 },
{ 1872, 1919 },
{ 2208, 2228 },
- { 2230, 2237 },
+ { 2230, 2247 },
{ 2259, 2273 },
{ 2275, 2303 },
{ 64336, 64449 },
@@ -4814,8 +4892,7 @@
};
static const URange16 Armenian_range16[] = {
{ 1329, 1366 },
- { 1369, 1416 },
- { 1418, 1418 },
+ { 1369, 1418 },
{ 1421, 1423 },
{ 64275, 64279 },
};
@@ -4866,7 +4943,7 @@
static const URange16 Bopomofo_range16[] = {
{ 746, 747 },
{ 12549, 12591 },
- { 12704, 12730 },
+ { 12704, 12735 },
};
static const URange32 Brahmi_range32[] = {
{ 69632, 69709 },
@@ -4896,7 +4973,7 @@
};
static const URange32 Chakma_range32[] = {
{ 69888, 69940 },
- { 69942, 69958 },
+ { 69942, 69959 },
};
static const URange16 Cham_range16[] = {
{ 43520, 43574 },
@@ -4909,6 +4986,9 @@
{ 5112, 5117 },
{ 43888, 43967 },
};
+static const URange32 Chorasmian_range32[] = {
+ { 69552, 69579 },
+};
static const URange16 Common_range16[] = {
{ 0, 64 },
{ 91, 96 },
@@ -4924,7 +5004,6 @@
{ 894, 894 },
{ 901, 901 },
{ 903, 903 },
- { 1417, 1417 },
{ 1541, 1541 },
{ 1548, 1548 },
{ 1563, 1563 },
@@ -4963,8 +5042,8 @@
{ 9312, 10239 },
{ 10496, 11123 },
{ 11126, 11157 },
- { 11160, 11263 },
- { 11776, 11855 },
+ { 11159, 11263 },
+ { 11776, 11858 },
{ 12272, 12283 },
{ 12288, 12292 },
{ 12294, 12294 },
@@ -4987,6 +5066,7 @@
{ 43310, 43310 },
{ 43471, 43471 },
{ 43867, 43867 },
+ { 43882, 43883 },
{ 64830, 64831 },
{ 65040, 65049 },
{ 65072, 65106 },
@@ -5006,7 +5086,7 @@
{ 65792, 65794 },
{ 65799, 65843 },
{ 65847, 65855 },
- { 65936, 65947 },
+ { 65936, 65948 },
{ 66000, 66044 },
{ 66273, 66299 },
{ 94178, 94179 },
@@ -5050,18 +5130,16 @@
{ 127153, 127167 },
{ 127169, 127183 },
{ 127185, 127221 },
- { 127232, 127244 },
- { 127248, 127340 },
- { 127344, 127404 },
+ { 127232, 127405 },
{ 127462, 127487 },
{ 127489, 127490 },
{ 127504, 127547 },
{ 127552, 127560 },
{ 127568, 127569 },
{ 127584, 127589 },
- { 127744, 128725 },
+ { 127744, 128727 },
{ 128736, 128748 },
- { 128752, 128762 },
+ { 128752, 128764 },
{ 128768, 128883 },
{ 128896, 128984 },
{ 128992, 129003 },
@@ -5070,18 +5148,21 @@
{ 129104, 129113 },
{ 129120, 129159 },
{ 129168, 129197 },
- { 129280, 129291 },
- { 129293, 129393 },
- { 129395, 129398 },
- { 129402, 129442 },
- { 129445, 129450 },
- { 129454, 129482 },
+ { 129200, 129201 },
+ { 129280, 129400 },
+ { 129402, 129483 },
{ 129485, 129619 },
{ 129632, 129645 },
- { 129648, 129651 },
+ { 129648, 129652 },
{ 129656, 129658 },
- { 129664, 129666 },
- { 129680, 129685 },
+ { 129664, 129670 },
+ { 129680, 129704 },
+ { 129712, 129718 },
+ { 129728, 129730 },
+ { 129744, 129750 },
+ { 129792, 129938 },
+ { 129940, 129994 },
+ { 130032, 130041 },
{ 917505, 917505 },
{ 917536, 917631 },
};
@@ -5123,6 +5204,16 @@
{ 2406, 2431 },
{ 43232, 43263 },
};
+static const URange32 Dives_Akuru_range32[] = {
+ { 71936, 71942 },
+ { 71945, 71945 },
+ { 71948, 71955 },
+ { 71957, 71958 },
+ { 71960, 71989 },
+ { 71991, 71992 },
+ { 71995, 72006 },
+ { 72016, 72025 },
+};
static const URange32 Dogra_range32[] = {
{ 71680, 71739 },
};
@@ -5310,18 +5401,20 @@
{ 12295, 12295 },
{ 12321, 12329 },
{ 12344, 12347 },
- { 13312, 19893 },
- { 19968, 40943 },
+ { 13312, 19903 },
+ { 19968, 40956 },
{ 63744, 64109 },
{ 64112, 64217 },
};
static const URange32 Han_range32[] = {
- { 131072, 173782 },
+ { 94192, 94193 },
+ { 131072, 173789 },
{ 173824, 177972 },
{ 177984, 178205 },
{ 178208, 183969 },
{ 183984, 191456 },
{ 194560, 195101 },
+ { 196608, 201546 },
};
static const URange16 Hangul_range16[] = {
{ 4352, 4607 },
@@ -5381,7 +5474,7 @@
{ 1611, 1621 },
{ 1648, 1648 },
{ 2385, 2388 },
- { 6832, 6846 },
+ { 6832, 6848 },
{ 7376, 7378 },
{ 7380, 7392 },
{ 7394, 7400 },
@@ -5466,6 +5559,10 @@
{ 68159, 68168 },
{ 68176, 68184 },
};
+static const URange32 Khitan_Small_Script_range32[] = {
+ { 94180, 94180 },
+ { 101120, 101589 },
+};
static const URange16 Khmer_range16[] = {
{ 6016, 6109 },
{ 6112, 6121 },
@@ -5518,11 +5615,11 @@
{ 11360, 11391 },
{ 42786, 42887 },
{ 42891, 42943 },
- { 42946, 42950 },
- { 42999, 43007 },
+ { 42946, 42954 },
+ { 42997, 43007 },
{ 43824, 43866 },
{ 43868, 43876 },
- { 43878, 43879 },
+ { 43878, 43881 },
{ 64256, 64262 },
{ 65313, 65338 },
{ 65345, 65370 },
@@ -5556,6 +5653,9 @@
static const URange16 Lisu_range16[] = {
{ 42192, 42239 },
};
+static const URange32 Lisu_range32[] = {
+ { 73648, 73648 },
+};
static const URange32 Lycian_range32[] = {
{ 66176, 66204 },
};
@@ -5570,8 +5670,7 @@
{ 73440, 73464 },
};
static const URange16 Malayalam_range16[] = {
- { 3328, 3331 },
- { 3333, 3340 },
+ { 3328, 3340 },
{ 3342, 3344 },
{ 3346, 3396 },
{ 3398, 3400 },
@@ -5674,9 +5773,8 @@
{ 6622, 6623 },
};
static const URange32 Newa_range32[] = {
- { 70656, 70745 },
- { 70747, 70747 },
- { 70749, 70751 },
+ { 70656, 70747 },
+ { 70749, 70753 },
};
static const URange16 Nko_range16[] = {
{ 1984, 2042 },
@@ -5737,7 +5835,7 @@
{ 2876, 2884 },
{ 2887, 2888 },
{ 2891, 2893 },
- { 2902, 2903 },
+ { 2901, 2903 },
{ 2908, 2909 },
{ 2911, 2915 },
{ 2918, 2935 },
@@ -5792,8 +5890,7 @@
{ 43214, 43225 },
};
static const URange32 Sharada_range32[] = {
- { 70016, 70093 },
- { 70096, 70111 },
+ { 70016, 70111 },
};
static const URange32 Shavian_range32[] = {
{ 66640, 66687 },
@@ -5808,7 +5905,7 @@
{ 121505, 121519 },
};
static const URange16 Sinhala_range16[] = {
- { 3458, 3459 },
+ { 3457, 3459 },
{ 3461, 3478 },
{ 3482, 3505 },
{ 3507, 3515 },
@@ -5839,7 +5936,7 @@
{ 7360, 7367 },
};
static const URange16 Syloti_Nagri_range16[] = {
- { 43008, 43051 },
+ { 43008, 43052 },
};
static const URange16 Syriac_range16[] = {
{ 1792, 1805 },
@@ -5900,7 +5997,8 @@
static const URange32 Tangut_range32[] = {
{ 94176, 94176 },
{ 94208, 100343 },
- { 100352, 101106 },
+ { 100352, 101119 },
+ { 101632, 101640 },
};
static const URange16 Telugu_range16[] = {
{ 3072, 3084 },
@@ -5956,6 +6054,11 @@
{ 71840, 71922 },
{ 71935, 71935 },
};
+static const URange32 Yezidi_range32[] = {
+ { 69248, 69289 },
+ { 69291, 69293 },
+ { 69296, 69297 },
+};
static const URange16 Yi_range16[] = {
{ 40960, 42124 },
{ 42128, 42182 },
@@ -5963,13 +6066,13 @@
static const URange32 Zanabazar_Square_range32[] = {
{ 72192, 72263 },
};
-// 3987 16-bit ranges, 1525 32-bit ranges
+// 4001 16-bit ranges, 1602 32-bit ranges
const UGroup unicode_groups[] = {
{ "Adlam", +1, 0, 0, Adlam_range32, 3 },
{ "Ahom", +1, 0, 0, Ahom_range32, 3 },
{ "Anatolian_Hieroglyphs", +1, 0, 0, Anatolian_Hieroglyphs_range32, 1 },
{ "Arabic", +1, Arabic_range16, 22, Arabic_range32, 35 },
- { "Armenian", +1, Armenian_range16, 5, 0, 0 },
+ { "Armenian", +1, Armenian_range16, 4, 0, 0 },
{ "Avestan", +1, 0, 0, Avestan_range32, 2 },
{ "Balinese", +1, Balinese_range16, 2, 0, 0 },
{ "Bamum", +1, Bamum_range16, 1, Bamum_range32, 1 },
@@ -5991,8 +6094,9 @@
{ "Chakma", +1, 0, 0, Chakma_range32, 2 },
{ "Cham", +1, Cham_range16, 4, 0, 0 },
{ "Cherokee", +1, Cherokee_range16, 3, 0, 0 },
+ { "Chorasmian", +1, 0, 0, Chorasmian_range32, 1 },
{ "Co", +1, Co_range16, 1, Co_range32, 2 },
- { "Common", +1, Common_range16, 91, Common_range32, 81 },
+ { "Common", +1, Common_range16, 91, Common_range32, 82 },
{ "Coptic", +1, Coptic_range16, 3, 0, 0 },
{ "Cs", +1, Cs_range16, 1, 0, 0 },
{ "Cuneiform", +1, 0, 0, Cuneiform_range32, 4 },
@@ -6000,6 +6104,7 @@
{ "Cyrillic", +1, Cyrillic_range16, 8, 0, 0 },
{ "Deseret", +1, 0, 0, Deseret_range32, 1 },
{ "Devanagari", +1, Devanagari_range16, 4, 0, 0 },
+ { "Dives_Akuru", +1, 0, 0, Dives_Akuru_range32, 8 },
{ "Dogra", +1, 0, 0, Dogra_range32, 1 },
{ "Duployan", +1, 0, 0, Duployan_range32, 5 },
{ "Egyptian_Hieroglyphs", +1, 0, 0, Egyptian_Hieroglyphs_range32, 2 },
@@ -6014,7 +6119,7 @@
{ "Gujarati", +1, Gujarati_range16, 14, 0, 0 },
{ "Gunjala_Gondi", +1, 0, 0, Gunjala_Gondi_range32, 6 },
{ "Gurmukhi", +1, Gurmukhi_range16, 16, 0, 0 },
- { "Han", +1, Han_range16, 11, Han_range32, 6 },
+ { "Han", +1, Han_range16, 11, Han_range32, 8 },
{ "Hangul", +1, Hangul_range16, 14, 0, 0 },
{ "Hanifi_Rohingya", +1, 0, 0, Hanifi_Rohingya_range32, 2 },
{ "Hanunoo", +1, Hanunoo_range16, 1, 0, 0 },
@@ -6031,33 +6136,34 @@
{ "Katakana", +1, Katakana_range16, 7, Katakana_range32, 2 },
{ "Kayah_Li", +1, Kayah_Li_range16, 2, 0, 0 },
{ "Kharoshthi", +1, 0, 0, Kharoshthi_range32, 8 },
+ { "Khitan_Small_Script", +1, 0, 0, Khitan_Small_Script_range32, 2 },
{ "Khmer", +1, Khmer_range16, 4, 0, 0 },
{ "Khojki", +1, 0, 0, Khojki_range32, 2 },
{ "Khudawadi", +1, 0, 0, Khudawadi_range32, 2 },
- { "L", +1, L_range16, 380, L_range32, 229 },
+ { "L", +1, L_range16, 380, L_range32, 242 },
{ "Lao", +1, Lao_range16, 11, 0, 0 },
{ "Latin", +1, Latin_range16, 32, 0, 0 },
{ "Lepcha", +1, Lepcha_range16, 3, 0, 0 },
{ "Limbu", +1, Limbu_range16, 5, 0, 0 },
{ "Linear_A", +1, 0, 0, Linear_A_range32, 3 },
{ "Linear_B", +1, 0, 0, Linear_B_range32, 7 },
- { "Lisu", +1, Lisu_range16, 1, 0, 0 },
- { "Ll", +1, Ll_range16, 608, Ll_range32, 34 },
- { "Lm", +1, Lm_range16, 54, Lm_range32, 6 },
- { "Lo", +1, Lo_range16, 290, Lo_range32, 186 },
+ { "Lisu", +1, Lisu_range16, 1, Lisu_range32, 1 },
+ { "Ll", +1, Ll_range16, 611, Ll_range32, 34 },
+ { "Lm", +1, Lm_range16, 55, Lm_range32, 6 },
+ { "Lo", +1, Lo_range16, 290, Lo_range32, 199 },
{ "Lt", +1, Lt_range16, 10, 0, 0 },
- { "Lu", +1, Lu_range16, 599, Lu_range32, 37 },
+ { "Lu", +1, Lu_range16, 601, Lu_range32, 37 },
{ "Lycian", +1, 0, 0, Lycian_range32, 1 },
{ "Lydian", +1, 0, 0, Lydian_range32, 2 },
- { "M", +1, M_range16, 186, M_range32, 94 },
+ { "M", +1, M_range16, 187, M_range32, 103 },
{ "Mahajani", +1, 0, 0, Mahajani_range32, 1 },
{ "Makasar", +1, 0, 0, Makasar_range32, 1 },
- { "Malayalam", +1, Malayalam_range16, 8, 0, 0 },
+ { "Malayalam", +1, Malayalam_range16, 7, 0, 0 },
{ "Mandaic", +1, Mandaic_range16, 2, 0, 0 },
{ "Manichaean", +1, 0, 0, Manichaean_range32, 2 },
{ "Marchen", +1, 0, 0, Marchen_range32, 3 },
{ "Masaram_Gondi", +1, 0, 0, Masaram_Gondi_range32, 7 },
- { "Mc", +1, Mc_range16, 109, Mc_range32, 59 },
+ { "Mc", +1, Mc_range16, 109, Mc_range32, 66 },
{ "Me", +1, Me_range16, 5, 0, 0 },
{ "Medefaidrin", +1, 0, 0, Medefaidrin_range32, 1 },
{ "Meetei_Mayek", +1, Meetei_Mayek_range16, 3, 0, 0 },
@@ -6065,21 +6171,21 @@
{ "Meroitic_Cursive", +1, 0, 0, Meroitic_Cursive_range32, 3 },
{ "Meroitic_Hieroglyphs", +1, 0, 0, Meroitic_Hieroglyphs_range32, 1 },
{ "Miao", +1, 0, 0, Miao_range32, 3 },
- { "Mn", +1, Mn_range16, 207, Mn_range32, 111 },
+ { "Mn", +1, Mn_range16, 210, Mn_range32, 117 },
{ "Modi", +1, 0, 0, Modi_range32, 2 },
{ "Mongolian", +1, Mongolian_range16, 6, Mongolian_range32, 1 },
{ "Mro", +1, 0, 0, Mro_range32, 3 },
{ "Multani", +1, 0, 0, Multani_range32, 5 },
{ "Myanmar", +1, Myanmar_range16, 3, 0, 0 },
- { "N", +1, N_range16, 67, N_range32, 63 },
+ { "N", +1, N_range16, 67, N_range32, 66 },
{ "Nabataean", +1, 0, 0, Nabataean_range32, 2 },
{ "Nandinagari", +1, 0, 0, Nandinagari_range32, 3 },
- { "Nd", +1, Nd_range16, 37, Nd_range32, 22 },
+ { "Nd", +1, Nd_range16, 37, Nd_range32, 24 },
{ "New_Tai_Lue", +1, New_Tai_Lue_range16, 4, 0, 0 },
- { "Newa", +1, 0, 0, Newa_range32, 3 },
+ { "Newa", +1, 0, 0, Newa_range32, 2 },
{ "Nko", +1, Nko_range16, 2, 0, 0 },
{ "Nl", +1, Nl_range16, 7, Nl_range32, 5 },
- { "No", +1, No_range16, 29, No_range32, 41 },
+ { "No", +1, No_range16, 29, No_range32, 42 },
{ "Nushu", +1, 0, 0, Nushu_range32, 2 },
{ "Nyiakeng_Puachue_Hmong", +1, 0, 0, Nyiakeng_Puachue_Hmong_range32, 4 },
{ "Ogham", +1, Ogham_range16, 1, 0, 0 },
@@ -6095,34 +6201,34 @@
{ "Oriya", +1, Oriya_range16, 14, 0, 0 },
{ "Osage", +1, 0, 0, Osage_range32, 2 },
{ "Osmanya", +1, 0, 0, Osmanya_range32, 2 },
- { "P", +1, P_range16, 131, P_range32, 51 },
+ { "P", +1, P_range16, 132, P_range32, 53 },
{ "Pahawh_Hmong", +1, 0, 0, Pahawh_Hmong_range32, 5 },
{ "Palmyrene", +1, 0, 0, Palmyrene_range32, 1 },
{ "Pau_Cin_Hau", +1, 0, 0, Pau_Cin_Hau_range32, 1 },
{ "Pc", +1, Pc_range16, 6, 0, 0 },
- { "Pd", +1, Pd_range16, 17, 0, 0 },
+ { "Pd", +1, Pd_range16, 17, Pd_range32, 1 },
{ "Pe", +1, Pe_range16, 72, 0, 0 },
{ "Pf", +1, Pf_range16, 10, 0, 0 },
{ "Phags_Pa", +1, Phags_Pa_range16, 1, 0, 0 },
{ "Phoenician", +1, 0, 0, Phoenician_range32, 2 },
{ "Pi", +1, Pi_range16, 11, 0, 0 },
- { "Po", +1, Po_range16, 128, Po_range32, 51 },
+ { "Po", +1, Po_range16, 129, Po_range32, 52 },
{ "Ps", +1, Ps_range16, 75, 0, 0 },
{ "Psalter_Pahlavi", +1, 0, 0, Psalter_Pahlavi_range32, 3 },
{ "Rejang", +1, Rejang_range16, 2, 0, 0 },
{ "Runic", +1, Runic_range16, 2, 0, 0 },
- { "S", +1, S_range16, 146, S_range32, 80 },
+ { "S", +1, S_range16, 148, S_range32, 81 },
{ "Samaritan", +1, Samaritan_range16, 2, 0, 0 },
{ "Saurashtra", +1, Saurashtra_range16, 2, 0, 0 },
{ "Sc", +1, Sc_range16, 18, Sc_range32, 3 },
- { "Sharada", +1, 0, 0, Sharada_range32, 2 },
+ { "Sharada", +1, 0, 0, Sharada_range32, 1 },
{ "Shavian", +1, 0, 0, Shavian_range32, 1 },
{ "Siddham", +1, 0, 0, Siddham_range32, 2 },
{ "SignWriting", +1, 0, 0, SignWriting_range32, 3 },
{ "Sinhala", +1, Sinhala_range16, 12, Sinhala_range32, 1 },
- { "Sk", +1, Sk_range16, 28, Sk_range32, 1 },
+ { "Sk", +1, Sk_range16, 29, Sk_range32, 1 },
{ "Sm", +1, Sm_range16, 53, Sm_range32, 11 },
- { "So", +1, So_range16, 111, So_range32, 69 },
+ { "So", +1, So_range16, 112, So_range32, 70 },
{ "Sogdian", +1, 0, 0, Sogdian_range32, 1 },
{ "Sora_Sompeng", +1, 0, 0, Sora_Sompeng_range32, 2 },
{ "Soyombo", +1, 0, 0, Soyombo_range32, 1 },
@@ -6136,7 +6242,7 @@
{ "Tai_Viet", +1, Tai_Viet_range16, 2, 0, 0 },
{ "Takri", +1, 0, 0, Takri_range32, 2 },
{ "Tamil", +1, Tamil_range16, 16, Tamil_range32, 2 },
- { "Tangut", +1, 0, 0, Tangut_range32, 3 },
+ { "Tangut", +1, 0, 0, Tangut_range32, 4 },
{ "Telugu", +1, Telugu_range16, 12, 0, 0 },
{ "Thaana", +1, Thaana_range16, 1, 0, 0 },
{ "Thai", +1, Thai_range16, 2, 0, 0 },
@@ -6147,6 +6253,7 @@
{ "Vai", +1, Vai_range16, 1, 0, 0 },
{ "Wancho", +1, 0, 0, Wancho_range32, 2 },
{ "Warang_Citi", +1, 0, 0, Warang_Citi_range32, 2 },
+ { "Yezidi", +1, 0, 0, Yezidi_range32, 3 },
{ "Yi", +1, Yi_range16, 2, 0, 0 },
{ "Z", +1, Z_range16, 8, 0, 0 },
{ "Zanabazar_Square", +1, 0, 0, Zanabazar_Square_range32, 1 },
@@ -6154,7 +6261,7 @@
{ "Zp", +1, Zp_range16, 1, 0, 0 },
{ "Zs", +1, Zs_range16, 7, 0, 0 },
};
-const int num_unicode_groups = 188;
+const int num_unicode_groups = 192;
} // namespace re2
