this repo has no description
encoding/jsonschema: detect Perl regexps
We want to treat Perl syntax as a missing feature
and make genuinely invalid regular expressions
an error.
Signed-off-by: Roger Peppe <rogpeppe@gmail.com>
Change-Id: I187be5f8846e02c9af514ec808fa19a8598e41ce
Reviewed-on: https://review.gerrithub.io/c/cue-lang/cue/+/1201127
TryBot-Result: CUEcueckoo <cueckoo@cuelang.org>
Reviewed-by: Daniel Martí <mvdan@mvdan.cc>
+93
-50
+3
-10
encoding/jsonschema/constraints_string.go
+3
-10
encoding/jsonschema/constraints_string.go
···
15
15
package jsonschema
16
16
17
17
import (
18
-
"regexp"
19
18
"sync"
20
19
21
20
"cuelang.org/go/cue"
···
50
49
}
51
50
52
51
func constraintPattern(key string, n cue.Value, s *state) {
53
-
str, _ := s.strValue(n)
54
-
if _, err := regexp.Compile(str); err != nil {
55
-
if s.cfg.StrictFeatures {
56
-
// TODO check if the error is only because of an unsupported
57
-
// regexp feature (e.g. perl regexp) or because the regexp is just
58
-
// bad. If the latter, this should be an error even if Strict is false.
59
-
s.errf(n, "unsupported regexp: %v", err)
60
-
}
52
+
str, ok := s.regexpValue(n)
53
+
if !ok {
61
54
return
62
55
}
63
-
s.add(n, stringType, &ast.UnaryExpr{Op: token.MAT, X: s.string(n)})
56
+
s.add(n, stringType, &ast.UnaryExpr{Op: token.MAT, X: str})
64
57
}
65
58
66
59
type formatFuncInfo struct {
+26
encoding/jsonschema/decode.go
+26
encoding/jsonschema/decode.go
···
22
22
"fmt"
23
23
"math"
24
24
"net/url"
25
+
"regexp/syntax"
25
26
"sort"
26
27
"strconv"
27
28
"strings"
···
254
255
return "", false
255
256
}
256
257
return s, true
258
+
}
259
+
260
+
func (d *decoder) regexpValue(n cue.Value) (ast.Expr, bool) {
261
+
s, ok := d.strValue(n)
262
+
if !ok {
263
+
return nil, false
264
+
}
265
+
_, err := syntax.Parse(s, syntax.Perl)
266
+
if err == nil {
267
+
return d.string(n), true
268
+
}
269
+
var regErr *syntax.Error
270
+
if errors.As(err, ®Err) && regErr.Code == syntax.ErrInvalidPerlOp {
271
+
// It's Perl syntax that we'll never support because the CUE evaluation
272
+
// engine uses Go's regexp implementation and because the missing
273
+
// features are usually not there for good reason (e.g. exponential
274
+
// runtime). In other words, this is a missing feature but not an invalid
275
+
// regular expression as such.
276
+
if d.cfg.StrictFeatures {
277
+
d.errf(n, "unsupported Perl regexp syntax in %q: %v", s, err)
278
+
}
279
+
return nil, false
280
+
}
281
+
d.errf(n, "invalid regexp %q: %v", s, err)
282
+
return nil, false
257
283
}
258
284
259
285
// const draftCutoff = 5
+8
-8
encoding/jsonschema/testdata/external/tests/draft2019-09/optional/ecmascript-regex.json
+8
-8
encoding/jsonschema/testdata/external/tests/draft2019-09/optional/ecmascript-regex.json
···
47
47
"pattern": "^\\cC$"
48
48
},
49
49
"skip": {
50
-
"v2": "extract error: unsupported regexp: error parsing regexp: invalid escape sequence: `\\c`",
51
-
"v3": "extract error: unsupported regexp: error parsing regexp: invalid escape sequence: `\\c`"
50
+
"v2": "extract error: invalid regexp \"^\\\\cC$\": error parsing regexp: invalid escape sequence: `\\c`",
51
+
"v3": "extract error: invalid regexp \"^\\\\cC$\": error parsing regexp: invalid escape sequence: `\\c`"
52
52
},
53
53
"tests": [
54
54
{
···
79
79
"pattern": "^\\cc$"
80
80
},
81
81
"skip": {
82
-
"v2": "extract error: unsupported regexp: error parsing regexp: invalid escape sequence: `\\c`",
83
-
"v3": "extract error: unsupported regexp: error parsing regexp: invalid escape sequence: `\\c`"
82
+
"v2": "extract error: invalid regexp \"^\\\\cc$\": error parsing regexp: invalid escape sequence: `\\c`",
83
+
"v3": "extract error: invalid regexp \"^\\\\cc$\": error parsing regexp: invalid escape sequence: `\\c`"
84
84
},
85
85
"tests": [
86
86
{
···
370
370
"pattern": "\\p{Letter}cole"
371
371
},
372
372
"skip": {
373
-
"v2": "extract error: unsupported regexp: error parsing regexp: invalid character class range: `\\p{Letter}`",
374
-
"v3": "extract error: unsupported regexp: error parsing regexp: invalid character class range: `\\p{Letter}`"
373
+
"v2": "extract error: invalid regexp \"\\\\p{Letter}cole\": error parsing regexp: invalid character class range: `\\p{Letter}`",
374
+
"v3": "extract error: invalid regexp \"\\\\p{Letter}cole\": error parsing regexp: invalid character class range: `\\p{Letter}`"
375
375
},
376
376
"tests": [
377
377
{
···
496
496
"pattern": "^\\p{digit}+$"
497
497
},
498
498
"skip": {
499
-
"v2": "extract error: unsupported regexp: error parsing regexp: invalid character class range: `\\p{digit}`",
500
-
"v3": "extract error: unsupported regexp: error parsing regexp: invalid character class range: `\\p{digit}`"
499
+
"v2": "extract error: invalid regexp \"^\\\\p{digit}+$\": error parsing regexp: invalid character class range: `\\p{digit}`",
500
+
"v3": "extract error: invalid regexp \"^\\\\p{digit}+$\": error parsing regexp: invalid character class range: `\\p{digit}`"
501
501
},
502
502
"tests": [
503
503
{
+8
-8
encoding/jsonschema/testdata/external/tests/draft2020-12/optional/ecmascript-regex.json
+8
-8
encoding/jsonschema/testdata/external/tests/draft2020-12/optional/ecmascript-regex.json
···
47
47
"pattern": "^\\cC$"
48
48
},
49
49
"skip": {
50
-
"v2": "extract error: unsupported regexp: error parsing regexp: invalid escape sequence: `\\c`",
51
-
"v3": "extract error: unsupported regexp: error parsing regexp: invalid escape sequence: `\\c`"
50
+
"v2": "extract error: invalid regexp \"^\\\\cC$\": error parsing regexp: invalid escape sequence: `\\c`",
51
+
"v3": "extract error: invalid regexp \"^\\\\cC$\": error parsing regexp: invalid escape sequence: `\\c`"
52
52
},
53
53
"tests": [
54
54
{
···
79
79
"pattern": "^\\cc$"
80
80
},
81
81
"skip": {
82
-
"v2": "extract error: unsupported regexp: error parsing regexp: invalid escape sequence: `\\c`",
83
-
"v3": "extract error: unsupported regexp: error parsing regexp: invalid escape sequence: `\\c`"
82
+
"v2": "extract error: invalid regexp \"^\\\\cc$\": error parsing regexp: invalid escape sequence: `\\c`",
83
+
"v3": "extract error: invalid regexp \"^\\\\cc$\": error parsing regexp: invalid escape sequence: `\\c`"
84
84
},
85
85
"tests": [
86
86
{
···
370
370
"pattern": "\\p{Letter}cole"
371
371
},
372
372
"skip": {
373
-
"v2": "extract error: unsupported regexp: error parsing regexp: invalid character class range: `\\p{Letter}`",
374
-
"v3": "extract error: unsupported regexp: error parsing regexp: invalid character class range: `\\p{Letter}`"
373
+
"v2": "extract error: invalid regexp \"\\\\p{Letter}cole\": error parsing regexp: invalid character class range: `\\p{Letter}`",
374
+
"v3": "extract error: invalid regexp \"\\\\p{Letter}cole\": error parsing regexp: invalid character class range: `\\p{Letter}`"
375
375
},
376
376
"tests": [
377
377
{
···
514
514
"pattern": "^\\p{digit}+$"
515
515
},
516
516
"skip": {
517
-
"v2": "extract error: unsupported regexp: error parsing regexp: invalid character class range: `\\p{digit}`",
518
-
"v3": "extract error: unsupported regexp: error parsing regexp: invalid character class range: `\\p{digit}`"
517
+
"v2": "extract error: invalid regexp \"^\\\\p{digit}+$\": error parsing regexp: invalid character class range: `\\p{digit}`",
518
+
"v3": "extract error: invalid regexp \"^\\\\p{digit}+$\": error parsing regexp: invalid character class range: `\\p{digit}`"
519
519
},
520
520
"tests": [
521
521
{
+8
-8
encoding/jsonschema/testdata/external/tests/draft4/optional/ecmascript-regex.json
+8
-8
encoding/jsonschema/testdata/external/tests/draft4/optional/ecmascript-regex.json
···
44
44
"pattern": "^\\cC$"
45
45
},
46
46
"skip": {
47
-
"v2": "extract error: unsupported regexp: error parsing regexp: invalid escape sequence: `\\c`",
48
-
"v3": "extract error: unsupported regexp: error parsing regexp: invalid escape sequence: `\\c`"
47
+
"v2": "extract error: invalid regexp \"^\\\\cC$\": error parsing regexp: invalid escape sequence: `\\c`",
48
+
"v3": "extract error: invalid regexp \"^\\\\cC$\": error parsing regexp: invalid escape sequence: `\\c`"
49
49
},
50
50
"tests": [
51
51
{
···
75
75
"pattern": "^\\cc$"
76
76
},
77
77
"skip": {
78
-
"v2": "extract error: unsupported regexp: error parsing regexp: invalid escape sequence: `\\c`",
79
-
"v3": "extract error: unsupported regexp: error parsing regexp: invalid escape sequence: `\\c`"
78
+
"v2": "extract error: invalid regexp \"^\\\\cc$\": error parsing regexp: invalid escape sequence: `\\c`",
79
+
"v3": "extract error: invalid regexp \"^\\\\cc$\": error parsing regexp: invalid escape sequence: `\\c`"
80
80
},
81
81
"tests": [
82
82
{
···
359
359
"pattern": "\\p{Letter}cole"
360
360
},
361
361
"skip": {
362
-
"v2": "extract error: unsupported regexp: error parsing regexp: invalid character class range: `\\p{Letter}`",
363
-
"v3": "extract error: unsupported regexp: error parsing regexp: invalid character class range: `\\p{Letter}`"
362
+
"v2": "extract error: invalid regexp \"\\\\p{Letter}cole\": error parsing regexp: invalid character class range: `\\p{Letter}`",
363
+
"v3": "extract error: invalid regexp \"\\\\p{Letter}cole\": error parsing regexp: invalid character class range: `\\p{Letter}`"
364
364
},
365
365
"tests": [
366
366
{
···
481
481
"pattern": "^\\p{digit}+$"
482
482
},
483
483
"skip": {
484
-
"v2": "extract error: unsupported regexp: error parsing regexp: invalid character class range: `\\p{digit}`",
485
-
"v3": "extract error: unsupported regexp: error parsing regexp: invalid character class range: `\\p{digit}`"
484
+
"v2": "extract error: invalid regexp \"^\\\\p{digit}+$\": error parsing regexp: invalid character class range: `\\p{digit}`",
485
+
"v3": "extract error: invalid regexp \"^\\\\p{digit}+$\": error parsing regexp: invalid character class range: `\\p{digit}`"
486
486
},
487
487
"tests": [
488
488
{
+8
-8
encoding/jsonschema/testdata/external/tests/draft6/optional/ecmascript-regex.json
+8
-8
encoding/jsonschema/testdata/external/tests/draft6/optional/ecmascript-regex.json
···
44
44
"pattern": "^\\cC$"
45
45
},
46
46
"skip": {
47
-
"v2": "extract error: unsupported regexp: error parsing regexp: invalid escape sequence: `\\c`",
48
-
"v3": "extract error: unsupported regexp: error parsing regexp: invalid escape sequence: `\\c`"
47
+
"v2": "extract error: invalid regexp \"^\\\\cC$\": error parsing regexp: invalid escape sequence: `\\c`",
48
+
"v3": "extract error: invalid regexp \"^\\\\cC$\": error parsing regexp: invalid escape sequence: `\\c`"
49
49
},
50
50
"tests": [
51
51
{
···
75
75
"pattern": "^\\cc$"
76
76
},
77
77
"skip": {
78
-
"v2": "extract error: unsupported regexp: error parsing regexp: invalid escape sequence: `\\c`",
79
-
"v3": "extract error: unsupported regexp: error parsing regexp: invalid escape sequence: `\\c`"
78
+
"v2": "extract error: invalid regexp \"^\\\\cc$\": error parsing regexp: invalid escape sequence: `\\c`",
79
+
"v3": "extract error: invalid regexp \"^\\\\cc$\": error parsing regexp: invalid escape sequence: `\\c`"
80
80
},
81
81
"tests": [
82
82
{
···
359
359
"pattern": "\\p{Letter}cole"
360
360
},
361
361
"skip": {
362
-
"v2": "extract error: unsupported regexp: error parsing regexp: invalid character class range: `\\p{Letter}`",
363
-
"v3": "extract error: unsupported regexp: error parsing regexp: invalid character class range: `\\p{Letter}`"
362
+
"v2": "extract error: invalid regexp \"\\\\p{Letter}cole\": error parsing regexp: invalid character class range: `\\p{Letter}`",
363
+
"v3": "extract error: invalid regexp \"\\\\p{Letter}cole\": error parsing regexp: invalid character class range: `\\p{Letter}`"
364
364
},
365
365
"tests": [
366
366
{
···
481
481
"pattern": "^\\p{digit}+$"
482
482
},
483
483
"skip": {
484
-
"v2": "extract error: unsupported regexp: error parsing regexp: invalid character class range: `\\p{digit}`",
485
-
"v3": "extract error: unsupported regexp: error parsing regexp: invalid character class range: `\\p{digit}`"
484
+
"v2": "extract error: invalid regexp \"^\\\\p{digit}+$\": error parsing regexp: invalid character class range: `\\p{digit}`",
485
+
"v3": "extract error: invalid regexp \"^\\\\p{digit}+$\": error parsing regexp: invalid character class range: `\\p{digit}`"
486
486
},
487
487
"tests": [
488
488
{
+8
-8
encoding/jsonschema/testdata/external/tests/draft7/optional/ecmascript-regex.json
+8
-8
encoding/jsonschema/testdata/external/tests/draft7/optional/ecmascript-regex.json
···
44
44
"pattern": "^\\cC$"
45
45
},
46
46
"skip": {
47
-
"v2": "extract error: unsupported regexp: error parsing regexp: invalid escape sequence: `\\c`",
48
-
"v3": "extract error: unsupported regexp: error parsing regexp: invalid escape sequence: `\\c`"
47
+
"v2": "extract error: invalid regexp \"^\\\\cC$\": error parsing regexp: invalid escape sequence: `\\c`",
48
+
"v3": "extract error: invalid regexp \"^\\\\cC$\": error parsing regexp: invalid escape sequence: `\\c`"
49
49
},
50
50
"tests": [
51
51
{
···
75
75
"pattern": "^\\cc$"
76
76
},
77
77
"skip": {
78
-
"v2": "extract error: unsupported regexp: error parsing regexp: invalid escape sequence: `\\c`",
79
-
"v3": "extract error: unsupported regexp: error parsing regexp: invalid escape sequence: `\\c`"
78
+
"v2": "extract error: invalid regexp \"^\\\\cc$\": error parsing regexp: invalid escape sequence: `\\c`",
79
+
"v3": "extract error: invalid regexp \"^\\\\cc$\": error parsing regexp: invalid escape sequence: `\\c`"
80
80
},
81
81
"tests": [
82
82
{
···
359
359
"pattern": "\\p{Letter}cole"
360
360
},
361
361
"skip": {
362
-
"v2": "extract error: unsupported regexp: error parsing regexp: invalid character class range: `\\p{Letter}`",
363
-
"v3": "extract error: unsupported regexp: error parsing regexp: invalid character class range: `\\p{Letter}`"
362
+
"v2": "extract error: invalid regexp \"\\\\p{Letter}cole\": error parsing regexp: invalid character class range: `\\p{Letter}`",
363
+
"v3": "extract error: invalid regexp \"\\\\p{Letter}cole\": error parsing regexp: invalid character class range: `\\p{Letter}`"
364
364
},
365
365
"tests": [
366
366
{
···
481
481
"pattern": "^\\p{digit}+$"
482
482
},
483
483
"skip": {
484
-
"v2": "extract error: unsupported regexp: error parsing regexp: invalid character class range: `\\p{digit}`",
485
-
"v3": "extract error: unsupported regexp: error parsing regexp: invalid character class range: `\\p{digit}`"
484
+
"v2": "extract error: invalid regexp \"^\\\\p{digit}+$\": error parsing regexp: invalid character class range: `\\p{digit}`",
485
+
"v3": "extract error: invalid regexp \"^\\\\p{digit}+$\": error parsing regexp: invalid character class range: `\\p{digit}`"
486
486
},
487
487
"tests": [
488
488
{
+11
encoding/jsonschema/testdata/txtar/perl_pattern.txtar
+11
encoding/jsonschema/testdata/txtar/perl_pattern.txtar
···
1
+
Invalid perl syntax should not result in failure when #strictFeatures
2
+
isn't enabled.
3
+
4
+
-- schema.json --
5
+
{
6
+
"$schema": "https://json-schema.org/draft/2020-12/schema",
7
+
"pattern": "^(?![ \\t\\n]*\\(default(.*)\\))[\\s\\S]*"
8
+
}
9
+
-- out/decode/extract --
10
+
@jsonschema(schema="https://json-schema.org/draft/2020-12/schema")
11
+
_
+13
encoding/jsonschema/testdata/txtar/perl_pattern_strict.txtar
+13
encoding/jsonschema/testdata/txtar/perl_pattern_strict.txtar
···
1
+
Invalid perl syntax should result in failure when #strictFeatures
2
+
is enabled.
3
+
#strictFeatures
4
+
5
+
-- schema.json --
6
+
{
7
+
"$schema": "https://json-schema.org/draft/2020-12/schema",
8
+
"pattern": "^(?![ \\t\\n]*\\(default(.*)\\))[\\s\\S]*"
9
+
}
10
+
-- out/decode/extract --
11
+
ERROR:
12
+
unsupported Perl regexp syntax in "^(?![ \\t\\n]*\\(default(.*)\\))[\\s\\S]*": error parsing regexp: invalid or unsupported Perl syntax: `(?!`:
13
+
schema.json:3:5