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>
google · Mar 15, 2020 · ca93436 · ca93436
1 parent 2695ecf
commit ca93436
Show file tree

Hide file tree

Showing 5 changed files with 251 additions and 132 deletions.
diff --git a/doc/syntax.html b/doc/syntax.html
@@ -264,13 +264,15 @@ <h1>RE2 regular expression syntax reference</h1>
 <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>
 <tr><td colspan=2>Cypriot</td></tr>
 <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 @@ <h1>RE2 regular expression syntax reference</h1>
 <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 @@ <h1>RE2 regular expression syntax reference</h1>
 <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
@@ -253,13 +253,15 @@ Caucasian_Albanian
 Chakma
 Cham
 Cherokee
+Chorasmian
 Common
 Coptic
 Cuneiform
 Cypriot
 Cyrillic
 Deseret
 Devanagari
+Dives_Akuru
 Dogra
 Duployan
 Egyptian_Hieroglyphs
@@ -291,6 +293,7 @@ Kannada
 Katakana
 Kayah_Li
 Kharoshthi
+Khitan_Small_Script
 Khmer
 Khojki
 Khudawadi
@@ -380,6 +383,7 @@ Ugaritic
 Vai
 Wancho
 Warang_Citi
+Yezidi
 Yi
 Zanabazar_Square
 

diff --git 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
@@ -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 @@ const CaseFold unicode_casefold[] = {
 	{ 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 @@ const CaseFold unicode_casefold[] = {
 	{ 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 @@ const CaseFold unicode_tolower[] = {
 	{ 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 @@ const CaseFold unicode_tolower[] = {
 	{ 93760, 93791, 32 },
 	{ 125184, 125217, 34 },
 };
-const int num_unicode_tolower = 198;
+const int num_unicode_tolower = 200;