use sync.OnceValue for various regular expressions, require go1.21

Using regex.MustCompile consumes a significant amount of memory when
importing the package, even if those regular expressions are not used.

This changes compiling the regular expressions to use a sync.OnceValue
so that they're only compiled the first time they're used.

There are various regular expressions remaining that are still compiled
on import, but these are exported, so changing them to a sync.OnceValue
would be a breaking change; we can still decide to do so, but leaving
that for a follow-up.

It's worth noting that sync.OnceValue requires go1.21 or up, so raising
the minimum version accordingly.

Signed-off-by: Sebastiaan van Stijn <[email protected]>

Author: thaJeztah

go.mod

@@ -1,5 +1,5 @@
 module github.com/distribution/reference
 
-go 1.20
+go 1.21
 
 require github.com/opencontainers/go-digest v1.0.0

normalize.go

@@ -64,7 +64,7 @@ type normalizedNamed interface {
 //		_ "crypto/sha256"
 //	)
 func ParseNormalizedNamed(s string) (Named, error) {
-	if ok := anchoredIdentifierRegexp.MatchString(s); ok {
+	if ok := anchoredIdentifierRegexp().MatchString(s); ok {
 		return nil, fmt.Errorf("invalid repository name (%s), cannot specify 64-byte hexadecimal strings", s)
 	}
 	domain, remainder := splitDockerDomain(s)
@@ -274,7 +274,7 @@ func TagNameOnly(ref Named) Named {
 //		_ "crypto/sha256"
 //	)
 func ParseAnyReference(ref string) (Reference, error) {
-	if ok := anchoredIdentifierRegexp.MatchString(ref); ok {
+	if ok := anchoredIdentifierRegexp().MatchString(ref); ok {
 		return digestReference("sha256:" + ref), nil
 	}
 	if dgst, err := digest.Parse(ref); err == nil {

reference.go

@@ -207,7 +207,7 @@ func Path(named Named) (name string) {
 // If no valid hostname is found, the hostname is empty and the full value
 // is returned as name
 func splitDomain(name string) (string, string) {
-	match := anchoredNameRegexp.FindStringSubmatch(name)
+	match := anchoredNameRegexp().FindStringSubmatch(name)
 	if len(match) != 3 {
 		return "", name
 	}
@@ -241,7 +241,7 @@ func Parse(s string) (Reference, error) {
 
 	var repo repository
 
-	nameMatch := anchoredNameRegexp.FindStringSubmatch(matches[1])
+	nameMatch := anchoredNameRegexp().FindStringSubmatch(matches[1])
 	if len(nameMatch) == 3 {
 		repo.domain = nameMatch[1]
 		repo.path = nameMatch[2]
@@ -292,7 +292,7 @@ func ParseNamed(s string) (Named, error) {
 // WithName returns a named object representing the given string. If the input
 // is invalid ErrReferenceInvalidFormat will be returned.
 func WithName(name string) (Named, error) {
-	match := anchoredNameRegexp.FindStringSubmatch(name)
+	match := anchoredNameRegexp().FindStringSubmatch(name)
 	if match == nil || len(match) != 3 {
 		return nil, ErrReferenceInvalidFormat
 	}
@@ -310,7 +310,7 @@ func WithName(name string) (Named, error) {
 // WithTag combines the name from "name" and the tag from "tag" to form a
 // reference incorporating both the name and the tag.
 func WithTag(name Named, tag string) (NamedTagged, error) {
-	if !anchoredTagRegexp.MatchString(tag) {
+	if !anchoredTagRegexp().MatchString(tag) {
 		return nil, ErrTagInvalidFormat
 	}
 	var repo repository
@@ -336,7 +336,7 @@ func WithTag(name Named, tag string) (NamedTagged, error) {
 // WithDigest combines the name from "name" and the digest from "digest" to form
 // a reference incorporating both the name and the digest.
 func WithDigest(name Named, digest digest.Digest) (Canonical, error) {
-	if !anchoredDigestRegexp.MatchString(digest.String()) {
+	if !anchoredDigestRegexp().MatchString(digest.String()) {
 		return nil, ErrDigestInvalidFormat
 	}
 	var repo repository

regexp.go

@@ -3,6 +3,7 @@ package reference
 import (
 	"regexp"
 	"strings"
+	"sync"
 )
 
 // DigestRegexp matches well-formed digests, including algorithm (e.g. "sha256:<encoded>").
@@ -111,11 +112,15 @@ var (
 
 	// anchoredTagRegexp matches valid tag names, anchored at the start and
 	// end of the matched string.
-	anchoredTagRegexp = regexp.MustCompile(anchored(tag))
+	anchoredTagRegexp = sync.OnceValue(func() *regexp.Regexp {
+		return regexp.MustCompile(anchored(tag))
+	})
 
 	// anchoredDigestRegexp matches valid digests, anchored at the start and
 	// end of the matched string.
-	anchoredDigestRegexp = regexp.MustCompile(anchored(digestPat))
+	anchoredDigestRegexp = sync.OnceValue(func() *regexp.Regexp {
+		return regexp.MustCompile(anchored(digestPat))
+	})
 
 	// pathComponent restricts path-components to start with an alphanumeric
 	// character, with following parts able to be separated by a separator
@@ -131,13 +136,17 @@ var (
 
 	// anchoredNameRegexp is used to parse a name value, capturing the
 	// domain and trailing components.
-	anchoredNameRegexp = regexp.MustCompile(anchored(optional(capture(domainAndPort), `/`), capture(remoteName)))
+	anchoredNameRegexp = sync.OnceValue(func() *regexp.Regexp {
+		return regexp.MustCompile(anchored(optional(capture(domainAndPort), `/`), capture(remoteName)))
+	})
 
 	referencePat = anchored(capture(namePat), optional(`:`, capture(tag)), optional(`@`, capture(digestPat)))
 
 	// anchoredIdentifierRegexp is used to check or match an
 	// identifier value, anchored at start and end of string.
-	anchoredIdentifierRegexp = regexp.MustCompile(anchored(identifier))
+	anchoredIdentifierRegexp = sync.OnceValue(func() *regexp.Regexp {
+		return regexp.MustCompile(anchored(identifier))
+	})
 )
 
 // optional wraps the expression in a non-capturing group and makes the

regexp_test.go

@@ -176,9 +176,9 @@ func TestDomainRegexp(t *testing.T) {
 
 func TestFullNameRegexp(t *testing.T) {
 	t.Parallel()
-	if anchoredNameRegexp.NumSubexp() != 2 {
+	if anchoredNameRegexp().NumSubexp() != 2 {
 		t.Fatalf("anchored name regexp should have two submatches: %v, %v != 2",
-			anchoredNameRegexp, anchoredNameRegexp.NumSubexp())
+			anchoredNameRegexp(), anchoredNameRegexp().NumSubexp())
 	}
 
 	tests := []regexpMatch{
@@ -469,7 +469,7 @@ func TestFullNameRegexp(t *testing.T) {
 		tc := tc
 		t.Run(tc.input, func(t *testing.T) {
 			t.Parallel()
-			checkRegexp(t, anchoredNameRegexp, tc)
+			checkRegexp(t, anchoredNameRegexp(), tc)
 		})
 	}
 }
@@ -580,7 +580,7 @@ func TestIdentifierRegexp(t *testing.T) {
 		tc := tc
 		t.Run(tc.input, func(t *testing.T) {
 			t.Parallel()
-			match := anchoredIdentifierRegexp.MatchString(tc.input)
+			match := anchoredIdentifierRegexp().MatchString(tc.input)
 			if match != tc.match {
 				t.Errorf("Expected match=%t, got %t", tc.match, match)
 			}