Skip to content

Commit cd177d9

Browse files
BurntSushiBurntSushi
committed
syntax: update to Unicode 13
1 parent 934e6e7 commit cd177d9

15 files changed

+1577
-961
lines changed

‎regex-syntax/src/unicode.rs

+1
Original file line numberDiff line numberDiff line change
@@ -595,6 +595,7 @@ fn ages(canonical_age: &str) -> Result<impl Iterator<Item = Range>> {
595595
("V11_0", age::V11_0),
596596
("V12_0", age::V12_0),
597597
("V12_1", age::V12_1),
598+
("V13_0", age::V13_0),
598599
];
599600
assert_eq!(AGES.len(), age::BY_NAME.len(), "ages are out of sync");
600601

‎regex-syntax/src/unicode_tables/age.rs

+123-60
Original file line numberDiff line numberDiff line change
@@ -1,14 +1,15 @@
11
// DO NOT EDIT THIS FILE. IT WAS AUTOMATICALLY GENERATED BY:
22
//
3-
// ucd-generate age /tmp/ucd/12.1.0/ --chars
3+
// ucd-generate age ucd-13.0.0 --chars
44
//
5-
// ucd-generate is available on crates.io.
5+
// ucd-generate 0.2.7 is available on crates.io.
66

77
pub const BY_NAME: &'static [(&'static str, &'static [(char, char)])] = &[
88
("V10_0", V10_0),
99
("V11_0", V11_0),
1010
("V12_0", V12_0),
1111
("V12_1", V12_1),
12+
("V13_0", V13_0),
1213
("V1_1", V1_1),
1314
("V2_0", V2_0),
1415
("V2_1", V2_1),
@@ -135,69 +136,131 @@ pub const V11_0: &'static [(char, char)] = &[
135136
];
136137

137138
pub const V12_0: &'static [(char, char)] = &[
138-
('\u{c77}', '\u{c77}'),
139-
('\u{e86}', '\u{e86}'),
140-
('\u{e89}', '\u{e89}'),
141-
('\u{e8c}', '\u{e8c}'),
142-
('\u{e8e}', '\u{e93}'),
143-
('\u{e98}', '\u{e98}'),
144-
('\u{ea0}', '\u{ea0}'),
145-
('\u{ea8}', '\u{ea9}'),
146-
('\u{eac}', '\u{eac}'),
139+
('౷', '౷'),
140+
('ຆ', 'ຆ'),
141+
('ຉ', 'ຉ'),
142+
('ຌ', 'ຌ'),
143+
('ຎ', 'ຓ'),
144+
('ຘ', 'ຘ'),
145+
('ຠ', 'ຠ'),
146+
('ຨ', 'ຩ'),
147+
('ຬ', 'ຬ'),
147148
('\u{eba}', '\u{eba}'),
148-
('\u{1cfa}', '\u{1cfa}'),
149-
('\u{2bc9}', '\u{2bc9}'),
150-
('\u{2bff}', '\u{2bff}'),
151-
('\u{2e4f}', '\u{2e4f}'),
152-
('\u{a7ba}', '\u{a7bf}'),
153-
('\u{a7c2}', '\u{a7c6}'),
154-
('\u{ab66}', '\u{ab67}'),
155-
('\u{10fe0}', '\u{10ff6}'),
156-
('\u{1145f}', '\u{1145f}'),
157-
('\u{116b8}', '\u{116b8}'),
158-
('\u{119a0}', '\u{119a7}'),
159-
('\u{119aa}', '\u{119d7}'),
160-
('\u{119da}', '\u{119e4}'),
161-
('\u{11a84}', '\u{11a85}'),
162-
('\u{11fc0}', '\u{11ff1}'),
163-
('\u{11fff}', '\u{11fff}'),
149+
('ᳺ', 'ᳺ'),
150+
('⯉', '⯉'),
151+
('⯿', '⯿'),
152+
('⹏', '⹏'),
153+
('Ꞻ', 'ꞿ'),
154+
('Ꟃ', 'Ᶎ'),
155+
('ꭦ', 'ꭧ'),
156+
('𐿠', '𐿶'),
157+
('𑑟', '𑑟'),
158+
('𑚸', '𑚸'),
159+
('𑦠', '𑦧'),
160+
('𑦪', '\u{119d7}'),
161+
('\u{119da}', '𑧤'),
162+
('𑪄', '𑪅'),
163+
('𑿀', '𑿱'),
164+
('𑿿', '𑿿'),
164165
('\u{13430}', '\u{13438}'),
165-
('\u{16f45}', '\u{16f4a}'),
166+
('𖽅', '𖽊'),
166167
('\u{16f4f}', '\u{16f4f}'),
167-
('\u{16f7f}', '\u{16f87}'),
168-
('\u{16fe2}', '\u{16fe3}'),
169-
('\u{187f2}', '\u{187f7}'),
170-
('\u{1b150}', '\u{1b152}'),
171-
('\u{1b164}', '\u{1b167}'),
172-
('\u{1e100}', '\u{1e12c}'),
173-
('\u{1e130}', '\u{1e13d}'),
174-
('\u{1e140}', '\u{1e149}'),
175-
('\u{1e14e}', '\u{1e14f}'),
176-
('\u{1e2c0}', '\u{1e2f9}'),
177-
('\u{1e2ff}', '\u{1e2ff}'),
178-
('\u{1e94b}', '\u{1e94b}'),
179-
('\u{1ed01}', '\u{1ed3d}'),
180-
('\u{1f16c}', '\u{1f16c}'),
181-
('\u{1f6d5}', '\u{1f6d5}'),
182-
('\u{1f6fa}', '\u{1f6fa}'),
183-
('\u{1f7e0}', '\u{1f7eb}'),
184-
('\u{1f90d}', '\u{1f90f}'),
185-
('\u{1f93f}', '\u{1f93f}'),
186-
('\u{1f971}', '\u{1f971}'),
187-
('\u{1f97b}', '\u{1f97b}'),
188-
('\u{1f9a5}', '\u{1f9aa}'),
189-
('\u{1f9ae}', '\u{1f9af}'),
190-
('\u{1f9ba}', '\u{1f9bf}'),
191-
('\u{1f9c3}', '\u{1f9ca}'),
192-
('\u{1f9cd}', '\u{1f9cf}'),
193-
('\u{1fa00}', '\u{1fa53}'),
194-
('\u{1fa70}', '\u{1fa73}'),
195-
('\u{1fa78}', '\u{1fa7a}'),
196-
('\u{1fa80}', '\u{1fa82}'),
197-
('\u{1fa90}', '\u{1fa95}'),
168+
('𖽿', '𖾇'),
169+
('𖿢', '𖿣'),
170+
('𘟲', '𘟷'),
171+
('𛅐', '𛅒'),
172+
('𛅤', '𛅧'),
173+
('𞄀', '𞄬'),
174+
('\u{1e130}', '𞄽'),
175+
('𞅀', '𞅉'),
176+
('𞅎', '𞅏'),
177+
('𞋀', '𞋹'),
178+
('𞋿', '𞋿'),
179+
('𞥋', '𞥋'),
180+
('𞴁', '𞴽'),
181+
('🅬', '🅬'),
182+
('🛕', '🛕'),
183+
('🛺', '🛺'),
184+
('🟠', '🟫'),
185+
('🤍', '🤏'),
186+
('🤿', '🤿'),
187+
('🥱', '🥱'),
188+
('🥻', '🥻'),
189+
('🦥', '🦪'),
190+
('🦮', '🦯'),
191+
('🦺', '🦿'),
192+
('🧃', '🧊'),
193+
('🧍', '🧏'),
194+
('🨀', '🩓'),
195+
('🩰', '🩳'),
196+
('🩸', '🩺'),
197+
('🪀', '🪂'),
198+
('🪐', '🪕'),
198199
];
199200

200-
pub const V12_1: &'static [(char, char)] = &[('\u{32ff}', '\u{32ff}')];
201+
pub const V12_1: &'static [(char, char)] = &[('㋿', '㋿')];
202+
203+
pub const V13_0: &'static [(char, char)] = &[
204+
('\u{8be}', '\u{8c7}'),
205+
('\u{b55}', '\u{b55}'),
206+
('\u{d04}', '\u{d04}'),
207+
('\u{d81}', '\u{d81}'),
208+
('\u{1abf}', '\u{1ac0}'),
209+
('\u{2b97}', '\u{2b97}'),
210+
('\u{2e50}', '\u{2e52}'),
211+
('\u{31bb}', '\u{31bf}'),
212+
('\u{4db6}', '\u{4dbf}'),
213+
('\u{9ff0}', '\u{9ffc}'),
214+
('\u{a7c7}', '\u{a7ca}'),
215+
('\u{a7f5}', '\u{a7f6}'),
216+
('\u{a82c}', '\u{a82c}'),
217+
('\u{ab68}', '\u{ab6b}'),
218+
('\u{1019c}', '\u{1019c}'),
219+
('\u{10e80}', '\u{10ea9}'),
220+
('\u{10eab}', '\u{10ead}'),
221+
('\u{10eb0}', '\u{10eb1}'),
222+
('\u{10fb0}', '\u{10fcb}'),
223+
('\u{11147}', '\u{11147}'),
224+
('\u{111ce}', '\u{111cf}'),
225+
('\u{1145a}', '\u{1145a}'),
226+
('\u{11460}', '\u{11461}'),
227+
('\u{11900}', '\u{11906}'),
228+
('\u{11909}', '\u{11909}'),
229+
('\u{1190c}', '\u{11913}'),
230+
('\u{11915}', '\u{11916}'),
231+
('\u{11918}', '\u{11935}'),
232+
('\u{11937}', '\u{11938}'),
233+
('\u{1193b}', '\u{11946}'),
234+
('\u{11950}', '\u{11959}'),
235+
('\u{11fb0}', '\u{11fb0}'),
236+
('\u{16fe4}', '\u{16fe4}'),
237+
('\u{16ff0}', '\u{16ff1}'),
238+
('\u{18af3}', '\u{18cd5}'),
239+
('\u{18d00}', '\u{18d08}'),
240+
('\u{1f10d}', '\u{1f10f}'),
241+
('\u{1f16d}', '\u{1f16f}'),
242+
('\u{1f1ad}', '\u{1f1ad}'),
243+
('\u{1f6d6}', '\u{1f6d7}'),
244+
('\u{1f6fb}', '\u{1f6fc}'),
245+
('\u{1f8b0}', '\u{1f8b1}'),
246+
('\u{1f90c}', '\u{1f90c}'),
247+
('\u{1f972}', '\u{1f972}'),
248+
('\u{1f977}', '\u{1f978}'),
249+
('\u{1f9a3}', '\u{1f9a4}'),
250+
('\u{1f9ab}', '\u{1f9ad}'),
251+
('\u{1f9cb}', '\u{1f9cb}'),
252+
('\u{1fa74}', '\u{1fa74}'),
253+
('\u{1fa83}', '\u{1fa86}'),
254+
('\u{1fa96}', '\u{1faa8}'),
255+
('\u{1fab0}', '\u{1fab6}'),
256+
('\u{1fac0}', '\u{1fac2}'),
257+
('\u{1fad0}', '\u{1fad6}'),
258+
('\u{1fb00}', '\u{1fb92}'),
259+
('\u{1fb94}', '\u{1fbca}'),
260+
('\u{1fbf0}', '\u{1fbf9}'),
261+
('\u{2a6d7}', '\u{2a6dd}'),
262+
('\u{30000}', '\u{3134a}'),
263+
];
201264

202265
pub const V1_1: &'static [(char, char)] = &[
203266
('\u{0}', 'ǵ'),

‎regex-syntax/src/unicode_tables/case_folding_simple.rs

+22-16
Original file line numberDiff line numberDiff line change
@@ -1,8 +1,8 @@
11
// DO NOT EDIT THIS FILE. IT WAS AUTOMATICALLY GENERATED BY:
22
//
3-
// ucd-generate case-folding-simple /tmp/ucd/12.1.0/ --chars --all-pairs
3+
// ucd-generate case-folding-simple ucd-13.0.0 --chars --all-pairs
44
//
5-
// ucd-generate is available on crates.io.
5+
// ucd-generate 0.2.7 is available on crates.io.
66

77
pub const CASE_FOLDING_SIMPLE: &'static [(char, &'static [char])] = &[
88
('A', &['a']),
@@ -459,7 +459,7 @@ pub const CASE_FOLDING_SIMPLE: &'static [(char, &'static [char])] = &[
459459
('ɵ', &['Ɵ']),
460460
('ɽ', &['Ɽ']),
461461
('ʀ', &['Ʀ']),
462-
('ʂ', &['\u{a7c5}']),
462+
('ʂ', &['Ʂ']),
463463
('ʃ', &['Ʃ']),
464464
('ʇ', &['Ʇ']),
465465
('ʈ', &['Ʈ']),
@@ -1199,7 +1199,7 @@ pub const CASE_FOLDING_SIMPLE: &'static [(char, &'static [char])] = &[
11991199
('Ჿ', &['ჿ']),
12001200
('ᵹ', &['Ᵹ']),
12011201
('ᵽ', &['Ᵽ']),
1202-
('ᶎ', &['\u{a7c6}']),
1202+
('ᶎ', &['Ᶎ']),
12031203
('Ḁ', &['ḁ']),
12041204
('ḁ', &['Ḁ']),
12051205
('Ḃ', &['ḃ']),
@@ -2167,7 +2167,7 @@ pub const CASE_FOLDING_SIMPLE: &'static [(char, &'static [char])] = &[
21672167
('ꞑ', &['Ꞑ']),
21682168
('Ꞓ', &['ꞓ']),
21692169
('ꞓ', &['Ꞓ']),
2170-
('ꞔ', &['\u{a7c4}']),
2170+
('ꞔ', &['Ꞔ']),
21712171
('Ꞗ', &['ꞗ']),
21722172
('ꞗ', &['Ꞗ']),
21732173
('Ꞙ', &['ꞙ']),
@@ -2203,17 +2203,23 @@ pub const CASE_FOLDING_SIMPLE: &'static [(char, &'static [char])] = &[
22032203
('ꞷ', &['Ꞷ']),
22042204
('Ꞹ', &['ꞹ']),
22052205
('ꞹ', &['Ꞹ']),
2206-
('\u{a7ba}', &['\u{a7bb}']),
2207-
('\u{a7bb}', &['\u{a7ba}']),
2208-
('\u{a7bc}', &['\u{a7bd}']),
2209-
('\u{a7bd}', &['\u{a7bc}']),
2210-
('\u{a7be}', &['\u{a7bf}']),
2211-
('\u{a7bf}', &['\u{a7be}']),
2212-
('\u{a7c2}', &['\u{a7c3}']),
2213-
('\u{a7c3}', &['\u{a7c2}']),
2214-
('\u{a7c4}', &['ꞔ']),
2215-
('\u{a7c5}', &['ʂ']),
2216-
('\u{a7c6}', &['ᶎ']),
2206+
('Ꞻ', &['ꞻ']),
2207+
('ꞻ', &['Ꞻ']),
2208+
('Ꞽ', &['ꞽ']),
2209+
('ꞽ', &['Ꞽ']),
2210+
('Ꞿ', &['ꞿ']),
2211+
('ꞿ', &['Ꞿ']),
2212+
('Ꟃ', &['ꟃ']),
2213+
('ꟃ', &['Ꟃ']),
2214+
('Ꞔ', &['ꞔ']),
2215+
('Ʂ', &['ʂ']),
2216+
('Ᶎ', &['ᶎ']),
2217+
('\u{a7c7}', &['\u{a7c8}']),
2218+
('\u{a7c8}', &['\u{a7c7}']),
2219+
('\u{a7c9}', &['\u{a7ca}']),
2220+
('\u{a7ca}', &['\u{a7c9}']),
2221+
('\u{a7f5}', &['\u{a7f6}']),
2222+
('\u{a7f6}', &['\u{a7f5}']),
22172223
('ꭓ', &['Ꭓ']),
22182224
('ꭰ', &['Ꭰ']),
22192225
('ꭱ', &['Ꭱ']),

0 commit comments

Comments
 (0)