Skip to content

Commit 9430972

Browse files
ChALkeRChALkeR
committed
fix unicode tests in accordance to pattern/patternProperties spec
1 parent 76b529f commit 9430972

File tree

6 files changed

+720
-24
lines changed

6 files changed

+720
-24
lines changed

‎tests/draft-future/optional/unicode.json‎

Lines changed: 120 additions & 4 deletions
Original file line numberDiff line numberDiff line change
@@ -1,8 +1,13 @@
11
[
22
{
33
"description": "unicode semantics should be used for all pattern matching",
4-
"schema": { "pattern": "\\wcole" },
4+
"schema": { "pattern": "\\p{Letter}cole" },
55
"tests": [
6+
{
7+
"description": "ascii character in json string",
8+
"data": "Les hivers de mon enfance etaient des saisons longues, longues. Nous vivions en trois lieux: l'ecole, l'eglise et la patinoire; mais la vraie vie etait sur la patinoire.",
9+
"valid": true
10+
},
611
{
712
"description": "literal unicode character in json string",
813
"data": "Les hivers de mon enfance étaient des saisons longues, longues. Nous vivions en trois lieux: l'école, l'église et la patinoire; mais la vraie vie était sur la patinoire.",
@@ -20,6 +25,32 @@
2025
}
2126
]
2227
},
28+
{
29+
"description": "\\w in patterns matches [A-Za-z0-9_], not unicode letters",
30+
"schema": { "pattern": "\\wcole" },
31+
"tests": [
32+
{
33+
"description": "ascii character in json string",
34+
"data": "Les hivers de mon enfance etaient des saisons longues, longues. Nous vivions en trois lieux: l'ecole, l'eglise et la patinoire; mais la vraie vie etait sur la patinoire.",
35+
"valid": true
36+
},
37+
{
38+
"description": "literal unicode character in json string",
39+
"data": "Les hivers de mon enfance étaient des saisons longues, longues. Nous vivions en trois lieux: l'école, l'église et la patinoire; mais la vraie vie était sur la patinoire.",
40+
"valid": false
41+
},
42+
{
43+
"description": "unicode character in hex format in string",
44+
"data": "Les hivers de mon enfance étaient des saisons longues, longues. Nous vivions en trois lieux: l'\u00e9cole, l'église et la patinoire; mais la vraie vie était sur la patinoire.",
45+
"valid": false
46+
},
47+
{
48+
"description": "unicode matching is case-sensitive",
49+
"data": "LES HIVERS DE MON ENFANCE ÉTAIENT DES SAISONS LONGUES, LONGUES. NOUS VIVIONS EN TROIS LIEUX: L'ÉCOLE, L'ÉGLISE ET LA PATINOIRE; MAIS LA VRAIE VIE ÉTAIT SUR LA PATINOIRE.",
50+
"valid": false
51+
}
52+
]
53+
},
2354
{
2455
"description": "unicode characters do not match ascii ranges",
2556
"schema": { "pattern": "[a-z]cole" },
@@ -42,8 +73,29 @@
4273
]
4374
},
4475
{
45-
"description": "unicode digits are more than 0 through 9",
76+
"description": "\\d in pattern matches [0-9], not unicode digits",
4677
"schema": { "pattern": "^\\d+$" },
78+
"tests": [
79+
{
80+
"description": "ascii digits",
81+
"data": "42",
82+
"valid": true
83+
},
84+
{
85+
"description": "ascii non-digits",
86+
"data": "-%#",
87+
"valid": false
88+
},
89+
{
90+
"description": "non-ascii digits (BENGALI DIGIT FOUR, BENGALI DIGIT TWO)",
91+
"data": "৪২",
92+
"valid": false
93+
}
94+
]
95+
},
96+
{
97+
"description": "unicode digits are more than 0 through 9",
98+
"schema": { "pattern": "^\\p{digit}+$" },
4799
"tests": [
48100
{
49101
"description": "ascii digits",
@@ -67,11 +119,16 @@
67119
"schema": {
68120
"type": "object",
69121
"patternProperties": {
70-
"\\wcole": true
122+
"\\p{Letter}cole": true
71123
},
72124
"additionalProperties": false
73125
},
74126
"tests": [
127+
{
128+
"description": "ascii character in json string",
129+
"data": { "l'ecole": "pas de vraie vie" },
130+
"valid": true
131+
},
75132
{
76133
"description": "literal unicode character in json string",
77134
"data": { "l'école": "pas de vraie vie" },
@@ -89,6 +146,38 @@
89146
}
90147
]
91148
},
149+
{
150+
"description": "\\w in patternProperties matches [A-Za-z0-9_], not unicode letters",
151+
"schema": {
152+
"type": "object",
153+
"patternProperties": {
154+
"\\wcole": true
155+
},
156+
"additionalProperties": false
157+
},
158+
"tests": [
159+
{
160+
"description": "ascii character in json string",
161+
"data": { "l'ecole": "pas de vraie vie" },
162+
"valid": true
163+
},
164+
{
165+
"description": "literal unicode character in json string",
166+
"data": { "l'école": "pas de vraie vie" },
167+
"valid": false
168+
},
169+
{
170+
"description": "unicode character in hex format in string",
171+
"data": { "l'\u00e9cole": "pas de vraie vie" },
172+
"valid": false
173+
},
174+
{
175+
"description": "unicode matching is case-sensitive",
176+
"data": { "L'ÉCOLE": "PAS DE VRAIE VIE" },
177+
"valid": false
178+
}
179+
]
180+
},
92181
{
93182
"description": "unicode characters do not match ascii ranges",
94183
"schema": {
@@ -117,14 +206,41 @@
117206
]
118207
},
119208
{
120-
"description": "unicode digits are more than 0 through 9",
209+
"description": "\\d in patternProperties matches [0-9], not unicode digits",
121210
"schema": {
122211
"type": "object",
123212
"patternProperties": {
124213
"^\\d+$": true
125214
},
126215
"additionalProperties": false
127216
},
217+
"tests": [
218+
{
219+
"description": "ascii digits",
220+
"data": { "42": "life, the universe, and everything" },
221+
"valid": true
222+
},
223+
{
224+
"description": "ascii non-digits",
225+
"data": { "-%#": "spending the year dead for tax reasons" },
226+
"valid": false
227+
},
228+
{
229+
"description": "non-ascii digits (BENGALI DIGIT FOUR, BENGALI DIGIT TWO)",
230+
"data": { "৪২": "khajit has wares if you have coin" },
231+
"valid": false
232+
}
233+
]
234+
},
235+
{
236+
"description": "unicode digits are more than 0 through 9",
237+
"schema": {
238+
"type": "object",
239+
"patternProperties": {
240+
"^\\p{digit}+$": true
241+
},
242+
"additionalProperties": false
243+
},
128244
"tests": [
129245
{
130246
"description": "ascii digits",

‎tests/draft2019-09/optional/unicode.json‎

Lines changed: 120 additions & 4 deletions
Original file line numberDiff line numberDiff line change
@@ -1,8 +1,13 @@
11
[
22
{
33
"description": "unicode semantics should be used for all pattern matching",
4-
"schema": { "pattern": "\\wcole" },
4+
"schema": { "pattern": "\\p{Letter}cole" },
55
"tests": [
6+
{
7+
"description": "ascii character in json string",
8+
"data": "Les hivers de mon enfance etaient des saisons longues, longues. Nous vivions en trois lieux: l'ecole, l'eglise et la patinoire; mais la vraie vie etait sur la patinoire.",
9+
"valid": true
10+
},
611
{
712
"description": "literal unicode character in json string",
813
"data": "Les hivers de mon enfance étaient des saisons longues, longues. Nous vivions en trois lieux: l'école, l'église et la patinoire; mais la vraie vie était sur la patinoire.",
@@ -20,6 +25,32 @@
2025
}
2126
]
2227
},
28+
{
29+
"description": "\\w in patterns matches [A-Za-z0-9_], not unicode letters",
30+
"schema": { "pattern": "\\wcole" },
31+
"tests": [
32+
{
33+
"description": "ascii character in json string",
34+
"data": "Les hivers de mon enfance etaient des saisons longues, longues. Nous vivions en trois lieux: l'ecole, l'eglise et la patinoire; mais la vraie vie etait sur la patinoire.",
35+
"valid": true
36+
},
37+
{
38+
"description": "literal unicode character in json string",
39+
"data": "Les hivers de mon enfance étaient des saisons longues, longues. Nous vivions en trois lieux: l'école, l'église et la patinoire; mais la vraie vie était sur la patinoire.",
40+
"valid": false
41+
},
42+
{
43+
"description": "unicode character in hex format in string",
44+
"data": "Les hivers de mon enfance étaient des saisons longues, longues. Nous vivions en trois lieux: l'\u00e9cole, l'église et la patinoire; mais la vraie vie était sur la patinoire.",
45+
"valid": false
46+
},
47+
{
48+
"description": "unicode matching is case-sensitive",
49+
"data": "LES HIVERS DE MON ENFANCE ÉTAIENT DES SAISONS LONGUES, LONGUES. NOUS VIVIONS EN TROIS LIEUX: L'ÉCOLE, L'ÉGLISE ET LA PATINOIRE; MAIS LA VRAIE VIE ÉTAIT SUR LA PATINOIRE.",
50+
"valid": false
51+
}
52+
]
53+
},
2354
{
2455
"description": "unicode characters do not match ascii ranges",
2556
"schema": { "pattern": "[a-z]cole" },
@@ -42,8 +73,29 @@
4273
]
4374
},
4475
{
45-
"description": "unicode digits are more than 0 through 9",
76+
"description": "\\d in pattern matches [0-9], not unicode digits",
4677
"schema": { "pattern": "^\\d+$" },
78+
"tests": [
79+
{
80+
"description": "ascii digits",
81+
"data": "42",
82+
"valid": true
83+
},
84+
{
85+
"description": "ascii non-digits",
86+
"data": "-%#",
87+
"valid": false
88+
},
89+
{
90+
"description": "non-ascii digits (BENGALI DIGIT FOUR, BENGALI DIGIT TWO)",
91+
"data": "৪২",
92+
"valid": false
93+
}
94+
]
95+
},
96+
{
97+
"description": "unicode digits are more than 0 through 9",
98+
"schema": { "pattern": "^\\p{digit}+$" },
4799
"tests": [
48100
{
49101
"description": "ascii digits",
@@ -67,11 +119,16 @@
67119
"schema": {
68120
"type": "object",
69121
"patternProperties": {
70-
"\\wcole": true
122+
"\\p{Letter}cole": true
71123
},
72124
"additionalProperties": false
73125
},
74126
"tests": [
127+
{
128+
"description": "ascii character in json string",
129+
"data": { "l'ecole": "pas de vraie vie" },
130+
"valid": true
131+
},
75132
{
76133
"description": "literal unicode character in json string",
77134
"data": { "l'école": "pas de vraie vie" },
@@ -89,6 +146,38 @@
89146
}
90147
]
91148
},
149+
{
150+
"description": "\\w in patternProperties matches [A-Za-z0-9_], not unicode letters",
151+
"schema": {
152+
"type": "object",
153+
"patternProperties": {
154+
"\\wcole": true
155+
},
156+
"additionalProperties": false
157+
},
158+
"tests": [
159+
{
160+
"description": "ascii character in json string",
161+
"data": { "l'ecole": "pas de vraie vie" },
162+
"valid": true
163+
},
164+
{
165+
"description": "literal unicode character in json string",
166+
"data": { "l'école": "pas de vraie vie" },
167+
"valid": false
168+
},
169+
{
170+
"description": "unicode character in hex format in string",
171+
"data": { "l'\u00e9cole": "pas de vraie vie" },
172+
"valid": false
173+
},
174+
{
175+
"description": "unicode matching is case-sensitive",
176+
"data": { "L'ÉCOLE": "PAS DE VRAIE VIE" },
177+
"valid": false
178+
}
179+
]
180+
},
92181
{
93182
"description": "unicode characters do not match ascii ranges",
94183
"schema": {
@@ -117,14 +206,41 @@
117206
]
118207
},
119208
{
120-
"description": "unicode digits are more than 0 through 9",
209+
"description": "\\d in patternProperties matches [0-9], not unicode digits",
121210
"schema": {
122211
"type": "object",
123212
"patternProperties": {
124213
"^\\d+$": true
125214
},
126215
"additionalProperties": false
127216
},
217+
"tests": [
218+
{
219+
"description": "ascii digits",
220+
"data": { "42": "life, the universe, and everything" },
221+
"valid": true
222+
},
223+
{
224+
"description": "ascii non-digits",
225+
"data": { "-%#": "spending the year dead for tax reasons" },
226+
"valid": false
227+
},
228+
{
229+
"description": "non-ascii digits (BENGALI DIGIT FOUR, BENGALI DIGIT TWO)",
230+
"data": { "৪২": "khajit has wares if you have coin" },
231+
"valid": false
232+
}
233+
]
234+
},
235+
{
236+
"description": "unicode digits are more than 0 through 9",
237+
"schema": {
238+
"type": "object",
239+
"patternProperties": {
240+
"^\\p{digit}+$": true
241+
},
242+
"additionalProperties": false
243+
},
128244
"tests": [
129245
{
130246
"description": "ascii digits",

0 commit comments

Comments
 (0)