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 <[email protected]> Change-Id: I187be5f8846e02c9af514ec808fa19a8598e41ce Reviewed-on: https://review.gerrithub.io/c/cue-lang/cue/+/1201127 TryBot-Result: CUEcueckoo <[email protected]> Reviewed-by: Daniel Martí <[email protected]>
cue-lang · Sep 12, 2024 · 67f9dfa · 67f9dfa
1 parent 9000dc3
commit 67f9dfa
Show file tree

Hide file tree

Showing 9 changed files with 93 additions and 50 deletions.
diff --git a/encoding/jsonschema/constraints_string.go b/encoding/jsonschema/constraints_string.go
@@ -15,7 +15,6 @@
 package jsonschema
 
 import (
-	"regexp"
 	"sync"
 
 	"cuelang.org/go/cue"
@@ -50,17 +49,11 @@ func constraintMinLength(key string, n cue.Value, s *state) {
 }
 
 func constraintPattern(key string, n cue.Value, s *state) {
-	str, _ := s.strValue(n)
-	if _, err := regexp.Compile(str); err != nil {
-		if s.cfg.StrictFeatures {
-			// TODO check if the error is only because of an unsupported
-			// regexp feature (e.g. perl regexp) or because the regexp is just
-			// bad. If the latter, this should be an error even if Strict is false.
-			s.errf(n, "unsupported regexp: %v", err)
-		}
+	str, ok := s.regexpValue(n)
+	if !ok {
 		return
 	}
-	s.add(n, stringType, &ast.UnaryExpr{Op: token.MAT, X: s.string(n)})
+	s.add(n, stringType, &ast.UnaryExpr{Op: token.MAT, X: str})
 }
 
 type formatFuncInfo struct {

diff --git a/encoding/jsonschema/decode.go b/encoding/jsonschema/decode.go
@@ -22,6 +22,7 @@ import (
 	"fmt"
 	"math"
 	"net/url"
+	"regexp/syntax"
 	"sort"
 	"strconv"
 	"strings"
@@ -256,6 +257,31 @@ func (d *decoder) strValue(n cue.Value) (s string, ok bool) {
 	return s, true
 }
 
+func (d *decoder) regexpValue(n cue.Value) (ast.Expr, bool) {
+	s, ok := d.strValue(n)
+	if !ok {
+		return nil, false
+	}
+	_, err := syntax.Parse(s, syntax.Perl)
+	if err == nil {
+		return d.string(n), true
+	}
+	var regErr *syntax.Error
+	if errors.As(err, &regErr) && regErr.Code == syntax.ErrInvalidPerlOp {
+		// It's Perl syntax that we'll never support because the CUE evaluation
+		// engine uses Go's regexp implementation and because the missing
+		// features are usually not there for good reason (e.g. exponential
+		// runtime). In other words, this is a missing feature but not an invalid
+		// regular expression as such.
+		if d.cfg.StrictFeatures {
+			d.errf(n, "unsupported Perl regexp syntax in %q: %v", s, err)
+		}
+		return nil, false
+	}
+	d.errf(n, "invalid regexp %q: %v", s, err)
+	return nil, false
+}
+
 // const draftCutoff = 5
 
 type coreType int

diff --git a/encoding/jsonschema/testdata/external/tests/draft2019-09/optional/ecmascript-regex.json b/encoding/jsonschema/testdata/external/tests/draft2019-09/optional/ecmascript-regex.json
@@ -47,8 +47,8 @@
 			"pattern": "^\\cC$"
 		},
 		"skip": {
-			"v2": "extract error: unsupported regexp: error parsing regexp: invalid escape sequence: `\\c`",
-			"v3": "extract error: unsupported regexp: error parsing regexp: invalid escape sequence: `\\c`"
+			"v2": "extract error: invalid regexp \"^\\\\cC$\": error parsing regexp: invalid escape sequence: `\\c`",
+			"v3": "extract error: invalid regexp \"^\\\\cC$\": error parsing regexp: invalid escape sequence: `\\c`"
 		},
 		"tests": [
 			{
@@ -79,8 +79,8 @@
 			"pattern": "^\\cc$"
 		},
 		"skip": {
-			"v2": "extract error: unsupported regexp: error parsing regexp: invalid escape sequence: `\\c`",
-			"v3": "extract error: unsupported regexp: error parsing regexp: invalid escape sequence: `\\c`"
+			"v2": "extract error: invalid regexp \"^\\\\cc$\": error parsing regexp: invalid escape sequence: `\\c`",
+			"v3": "extract error: invalid regexp \"^\\\\cc$\": error parsing regexp: invalid escape sequence: `\\c`"
 		},
 		"tests": [
 			{
@@ -370,8 +370,8 @@
 			"pattern": "\\p{Letter}cole"
 		},
 		"skip": {
-			"v2": "extract error: unsupported regexp: error parsing regexp: invalid character class range: `\\p{Letter}`",
-			"v3": "extract error: unsupported regexp: error parsing regexp: invalid character class range: `\\p{Letter}`"
+			"v2": "extract error: invalid regexp \"\\\\p{Letter}cole\": error parsing regexp: invalid character class range: `\\p{Letter}`",
+			"v3": "extract error: invalid regexp \"\\\\p{Letter}cole\": error parsing regexp: invalid character class range: `\\p{Letter}`"
 		},
 		"tests": [
 			{
@@ -496,8 +496,8 @@
 			"pattern": "^\\p{digit}+$"
 		},
 		"skip": {
-			"v2": "extract error: unsupported regexp: error parsing regexp: invalid character class range: `\\p{digit}`",
-			"v3": "extract error: unsupported regexp: error parsing regexp: invalid character class range: `\\p{digit}`"
+			"v2": "extract error: invalid regexp \"^\\\\p{digit}+$\": error parsing regexp: invalid character class range: `\\p{digit}`",
+			"v3": "extract error: invalid regexp \"^\\\\p{digit}+$\": error parsing regexp: invalid character class range: `\\p{digit}`"
 		},
 		"tests": [
 			{

diff --git a/encoding/jsonschema/testdata/external/tests/draft2020-12/optional/ecmascript-regex.json b/encoding/jsonschema/testdata/external/tests/draft2020-12/optional/ecmascript-regex.json
@@ -47,8 +47,8 @@
 			"pattern": "^\\cC$"
 		},
 		"skip": {
-			"v2": "extract error: unsupported regexp: error parsing regexp: invalid escape sequence: `\\c`",
-			"v3": "extract error: unsupported regexp: error parsing regexp: invalid escape sequence: `\\c`"
+			"v2": "extract error: invalid regexp \"^\\\\cC$\": error parsing regexp: invalid escape sequence: `\\c`",
+			"v3": "extract error: invalid regexp \"^\\\\cC$\": error parsing regexp: invalid escape sequence: `\\c`"
 		},
 		"tests": [
 			{
@@ -79,8 +79,8 @@
 			"pattern": "^\\cc$"
 		},
 		"skip": {
-			"v2": "extract error: unsupported regexp: error parsing regexp: invalid escape sequence: `\\c`",
-			"v3": "extract error: unsupported regexp: error parsing regexp: invalid escape sequence: `\\c`"
+			"v2": "extract error: invalid regexp \"^\\\\cc$\": error parsing regexp: invalid escape sequence: `\\c`",
+			"v3": "extract error: invalid regexp \"^\\\\cc$\": error parsing regexp: invalid escape sequence: `\\c`"
 		},
 		"tests": [
 			{
@@ -370,8 +370,8 @@
 			"pattern": "\\p{Letter}cole"
 		},
 		"skip": {
-			"v2": "extract error: unsupported regexp: error parsing regexp: invalid character class range: `\\p{Letter}`",
-			"v3": "extract error: unsupported regexp: error parsing regexp: invalid character class range: `\\p{Letter}`"
+			"v2": "extract error: invalid regexp \"\\\\p{Letter}cole\": error parsing regexp: invalid character class range: `\\p{Letter}`",
+			"v3": "extract error: invalid regexp \"\\\\p{Letter}cole\": error parsing regexp: invalid character class range: `\\p{Letter}`"
 		},
 		"tests": [
 			{
@@ -514,8 +514,8 @@
 			"pattern": "^\\p{digit}+$"
 		},
 		"skip": {
-			"v2": "extract error: unsupported regexp: error parsing regexp: invalid character class range: `\\p{digit}`",
-			"v3": "extract error: unsupported regexp: error parsing regexp: invalid character class range: `\\p{digit}`"
+			"v2": "extract error: invalid regexp \"^\\\\p{digit}+$\": error parsing regexp: invalid character class range: `\\p{digit}`",
+			"v3": "extract error: invalid regexp \"^\\\\p{digit}+$\": error parsing regexp: invalid character class range: `\\p{digit}`"
 		},
 		"tests": [
 			{

diff --git a/encoding/jsonschema/testdata/external/tests/draft4/optional/ecmascript-regex.json b/encoding/jsonschema/testdata/external/tests/draft4/optional/ecmascript-regex.json
@@ -44,8 +44,8 @@
 			"pattern": "^\\cC$"
 		},
 		"skip": {
-			"v2": "extract error: unsupported regexp: error parsing regexp: invalid escape sequence: `\\c`",
-			"v3": "extract error: unsupported regexp: error parsing regexp: invalid escape sequence: `\\c`"
+			"v2": "extract error: invalid regexp \"^\\\\cC$\": error parsing regexp: invalid escape sequence: `\\c`",
+			"v3": "extract error: invalid regexp \"^\\\\cC$\": error parsing regexp: invalid escape sequence: `\\c`"
 		},
 		"tests": [
 			{
@@ -75,8 +75,8 @@
 			"pattern": "^\\cc$"
 		},
 		"skip": {
-			"v2": "extract error: unsupported regexp: error parsing regexp: invalid escape sequence: `\\c`",
-			"v3": "extract error: unsupported regexp: error parsing regexp: invalid escape sequence: `\\c`"
+			"v2": "extract error: invalid regexp \"^\\\\cc$\": error parsing regexp: invalid escape sequence: `\\c`",
+			"v3": "extract error: invalid regexp \"^\\\\cc$\": error parsing regexp: invalid escape sequence: `\\c`"
 		},
 		"tests": [
 			{
@@ -359,8 +359,8 @@
 			"pattern": "\\p{Letter}cole"
 		},
 		"skip": {
-			"v2": "extract error: unsupported regexp: error parsing regexp: invalid character class range: `\\p{Letter}`",
-			"v3": "extract error: unsupported regexp: error parsing regexp: invalid character class range: `\\p{Letter}`"
+			"v2": "extract error: invalid regexp \"\\\\p{Letter}cole\": error parsing regexp: invalid character class range: `\\p{Letter}`",
+			"v3": "extract error: invalid regexp \"\\\\p{Letter}cole\": error parsing regexp: invalid character class range: `\\p{Letter}`"
 		},
 		"tests": [
 			{
@@ -481,8 +481,8 @@
 			"pattern": "^\\p{digit}+$"
 		},
 		"skip": {
-			"v2": "extract error: unsupported regexp: error parsing regexp: invalid character class range: `\\p{digit}`",
-			"v3": "extract error: unsupported regexp: error parsing regexp: invalid character class range: `\\p{digit}`"
+			"v2": "extract error: invalid regexp \"^\\\\p{digit}+$\": error parsing regexp: invalid character class range: `\\p{digit}`",
+			"v3": "extract error: invalid regexp \"^\\\\p{digit}+$\": error parsing regexp: invalid character class range: `\\p{digit}`"
 		},
 		"tests": [
 			{

diff --git a/encoding/jsonschema/testdata/external/tests/draft6/optional/ecmascript-regex.json b/encoding/jsonschema/testdata/external/tests/draft6/optional/ecmascript-regex.json
@@ -44,8 +44,8 @@
 			"pattern": "^\\cC$"
 		},
 		"skip": {
-			"v2": "extract error: unsupported regexp: error parsing regexp: invalid escape sequence: `\\c`",
-			"v3": "extract error: unsupported regexp: error parsing regexp: invalid escape sequence: `\\c`"
+			"v2": "extract error: invalid regexp \"^\\\\cC$\": error parsing regexp: invalid escape sequence: `\\c`",
+			"v3": "extract error: invalid regexp \"^\\\\cC$\": error parsing regexp: invalid escape sequence: `\\c`"
 		},
 		"tests": [
 			{
@@ -75,8 +75,8 @@
 			"pattern": "^\\cc$"
 		},
 		"skip": {
-			"v2": "extract error: unsupported regexp: error parsing regexp: invalid escape sequence: `\\c`",
-			"v3": "extract error: unsupported regexp: error parsing regexp: invalid escape sequence: `\\c`"
+			"v2": "extract error: invalid regexp \"^\\\\cc$\": error parsing regexp: invalid escape sequence: `\\c`",
+			"v3": "extract error: invalid regexp \"^\\\\cc$\": error parsing regexp: invalid escape sequence: `\\c`"
 		},
 		"tests": [
 			{
@@ -359,8 +359,8 @@
 			"pattern": "\\p{Letter}cole"
 		},
 		"skip": {
-			"v2": "extract error: unsupported regexp: error parsing regexp: invalid character class range: `\\p{Letter}`",
-			"v3": "extract error: unsupported regexp: error parsing regexp: invalid character class range: `\\p{Letter}`"
+			"v2": "extract error: invalid regexp \"\\\\p{Letter}cole\": error parsing regexp: invalid character class range: `\\p{Letter}`",
+			"v3": "extract error: invalid regexp \"\\\\p{Letter}cole\": error parsing regexp: invalid character class range: `\\p{Letter}`"
 		},
 		"tests": [
 			{
@@ -481,8 +481,8 @@
 			"pattern": "^\\p{digit}+$"
 		},
 		"skip": {
-			"v2": "extract error: unsupported regexp: error parsing regexp: invalid character class range: `\\p{digit}`",
-			"v3": "extract error: unsupported regexp: error parsing regexp: invalid character class range: `\\p{digit}`"
+			"v2": "extract error: invalid regexp \"^\\\\p{digit}+$\": error parsing regexp: invalid character class range: `\\p{digit}`",
+			"v3": "extract error: invalid regexp \"^\\\\p{digit}+$\": error parsing regexp: invalid character class range: `\\p{digit}`"
 		},
 		"tests": [
 			{

diff --git a/encoding/jsonschema/testdata/external/tests/draft7/optional/ecmascript-regex.json b/encoding/jsonschema/testdata/external/tests/draft7/optional/ecmascript-regex.json
@@ -44,8 +44,8 @@
 			"pattern": "^\\cC$"
 		},
 		"skip": {
-			"v2": "extract error: unsupported regexp: error parsing regexp: invalid escape sequence: `\\c`",
-			"v3": "extract error: unsupported regexp: error parsing regexp: invalid escape sequence: `\\c`"
+			"v2": "extract error: invalid regexp \"^\\\\cC$\": error parsing regexp: invalid escape sequence: `\\c`",
+			"v3": "extract error: invalid regexp \"^\\\\cC$\": error parsing regexp: invalid escape sequence: `\\c`"
 		},
 		"tests": [
 			{
@@ -75,8 +75,8 @@
 			"pattern": "^\\cc$"
 		},
 		"skip": {
-			"v2": "extract error: unsupported regexp: error parsing regexp: invalid escape sequence: `\\c`",
-			"v3": "extract error: unsupported regexp: error parsing regexp: invalid escape sequence: `\\c`"
+			"v2": "extract error: invalid regexp \"^\\\\cc$\": error parsing regexp: invalid escape sequence: `\\c`",
+			"v3": "extract error: invalid regexp \"^\\\\cc$\": error parsing regexp: invalid escape sequence: `\\c`"
 		},
 		"tests": [
 			{
@@ -359,8 +359,8 @@
 			"pattern": "\\p{Letter}cole"
 		},
 		"skip": {
-			"v2": "extract error: unsupported regexp: error parsing regexp: invalid character class range: `\\p{Letter}`",
-			"v3": "extract error: unsupported regexp: error parsing regexp: invalid character class range: `\\p{Letter}`"
+			"v2": "extract error: invalid regexp \"\\\\p{Letter}cole\": error parsing regexp: invalid character class range: `\\p{Letter}`",
+			"v3": "extract error: invalid regexp \"\\\\p{Letter}cole\": error parsing regexp: invalid character class range: `\\p{Letter}`"
 		},
 		"tests": [
 			{
@@ -481,8 +481,8 @@
 			"pattern": "^\\p{digit}+$"
 		},
 		"skip": {
-			"v2": "extract error: unsupported regexp: error parsing regexp: invalid character class range: `\\p{digit}`",
-			"v3": "extract error: unsupported regexp: error parsing regexp: invalid character class range: `\\p{digit}`"
+			"v2": "extract error: invalid regexp \"^\\\\p{digit}+$\": error parsing regexp: invalid character class range: `\\p{digit}`",
+			"v3": "extract error: invalid regexp \"^\\\\p{digit}+$\": error parsing regexp: invalid character class range: `\\p{digit}`"
 		},
 		"tests": [
 			{

diff --git a/encoding/jsonschema/testdata/txtar/perl_pattern.txtar b/encoding/jsonschema/testdata/txtar/perl_pattern.txtar
@@ -0,0 +1,11 @@
+Invalid perl syntax should not result in failure when #strictFeatures
+isn't enabled.
+
+-- schema.json --
+{
+    "$schema": "https://json-schema.org/draft/2020-12/schema",
+    "pattern": "^(?![ \\t\\n]*\\(default(.*)\\))[\\s\\S]*"
+}
+-- out/decode/extract --
+@jsonschema(schema="https://json-schema.org/draft/2020-12/schema")
+_
diff --git a/encoding/jsonschema/testdata/txtar/perl_pattern_strict.txtar b/encoding/jsonschema/testdata/txtar/perl_pattern_strict.txtar
@@ -0,0 +1,13 @@
+Invalid perl syntax should result in failure when #strictFeatures
+is enabled.
+#strictFeatures
+
+-- schema.json --
+{
+    "$schema": "https://json-schema.org/draft/2020-12/schema",
+    "pattern": "^(?![ \\t\\n]*\\(default(.*)\\))[\\s\\S]*"
+}
+-- out/decode/extract --
+ERROR:
+unsupported Perl regexp syntax in "^(?![ \\t\\n]*\\(default(.*)\\))[\\s\\S]*": error parsing regexp: invalid or unsupported Perl syntax: `(?!`:
+    schema.json:3:5