Skip to content

policy: normalize HTTP schemes, host casing, and default ports #4068

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
crazy-max wants to merge 4 commits into
base: master
Choose a base branch
from
+352 −1
Open

policy: normalize HTTP schemes, host casing, and default ports #4068

Show file tree
Hide file tree
Changes from 3 commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
e2d156f
policy: strip default ports from input.http.host
locker95 Sep 9, 2026
65cbef9
policy: simplify HTTP host normalization and strengthen coverage
crazy-max Sep 9, 2026
f4567ef
policy: normalize HTTP scheme and host casing
crazy-max Sep 10, 2026
3c4f6c6
policy: canonicalize HTTP hosts and preserve Unicode destinations
crazy-max Sep 10, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
16 changes: 15 additions & 1 deletion policy/validate.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -532,6 +532,19 @@ func (p *Policy) Print(ctx print.Context, msg string) error {
return nil
}

func normalizeHTTPHost(u *url.URL) string {
host := u.Host
port := strings.TrimLeft(u.Port(), "0")
if u.Scheme == "http" && port == "80" || u.Scheme == "https" && port == "443" {
host = strings.TrimSuffix(host, ":"+u.Port())
Comment thread
crazy-max marked this conversation as resolved.
Outdated
}
// Preserve IPv6 zone identifiers, which may be case-sensitive interface names.
if i := strings.IndexByte(host, '%'); strings.HasPrefix(host, "[") && i >= 0 {
return strings.ToLower(host[:i]) + host[i:]
}
Comment thread
crazy-max marked this conversation as resolved.
Outdated
return strings.ToLower(host)
}

func sourceToInput(ctx context.Context, getVerifier PolicyVerifierProvider, src *gwpb.ResolveSourceMetaResponse, platform *ocispecs.Platform, logf func(logrus.Level, string)) (Input, []string, error) {
var inp Input
var unknowns []string
Expand All @@ -544,6 +557,7 @@ func sourceToInput(ctx context.Context, getVerifier PolicyVerifierProvider, src
if !ok {
return inp, nil, errors.Errorf("invalid source identifier: %s", src.Source.Identifier)
}
scheme = strings.ToLower(scheme)

switch scheme {
case "http", "https":
Expand All @@ -554,7 +568,7 @@ func sourceToInput(ctx context.Context, getVerifier PolicyVerifierProvider, src
inp.HTTP = &HTTP{
URL: src.Source.Identifier,
Schema: scheme,
Host: u.Host,
Host: normalizeHTTPHost(u),
Path: u.Path,
Query: u.Query(),
}
Expand Down
275 changes: 275 additions & 0 deletions policy/validate_test.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -7,13 +7,16 @@ import (
"encoding/json"
"errors"
"fmt"
"net/url"
"testing"
"time"

slsa02 "github.com/in-toto/in-toto-golang/in_toto/slsa_provenance/v0.2"
slsa1 "github.com/in-toto/in-toto-golang/in_toto/slsa_provenance/v1"
gwpb "github.com/moby/buildkit/frontend/gateway/pb"
"github.com/moby/buildkit/solver/pb"
policyaction "github.com/moby/buildkit/sourcepolicy/pb"
"github.com/moby/buildkit/sourcepolicy/policysession"
policyverifier "github.com/moby/policy-helpers"
policyimage "github.com/moby/policy-helpers/image"
policytypes "github.com/moby/policy-helpers/types"
Expand All @@ -23,6 +26,44 @@ import (
"github.com/stretchr/testify/require"
)

func TestNormalizeHTTPHost(t *testing.T) {
tests := []struct {
scheme, host, want string
}{
{"https", "example.com", "example.com"},
{"http", "Example.com", "example.com"},
{"https", "EXAMPLE.COM:000443", "example.com"},
{"https", "eXaMpLe.CoM:8443", "example.com:8443"},
{"https", "example.com:443", "example.com"},
{"http", "example.com:80", "example.com"},
Comment thread
crazy-max marked this conversation as resolved.
{"https", ":443", ""},
{"http", ":80", ""},
{"https", "example.com:000443", "example.com"},
{"http", "example.com:00080", "example.com"},
{"https", "example.com:00080", "example.com:00080"},
{"http", "example.com:000443", "example.com:000443"},
{"https", "example.com:008443", "example.com:008443"},
{"http", "example.com:000", "example.com:000"},
{"https", "example.com:80", "example.com:80"},
{"http", "example.com:443", "example.com:443"},
{"https", "example.com:8443", "example.com:8443"},
{"http", "example.com:8080", "example.com:8080"},
Comment thread
crazy-max marked this conversation as resolved.
{"https", "[2001:db8::1]", "[2001:db8::1]"},
Comment thread
crazy-max marked this conversation as resolved.
{"https", "[2001:DB8::ABCD]:443", "[2001:db8::abcd]"},
{"https", "[FE80::ABCD%Eth0]:000443", "[fe80::abcd%Eth0]"},
{"https", "[FE80::ABCD%Eth0]:8443", "[fe80::abcd%Eth0]:8443"},

@thaJeztah thaJeztah Sep 10, 2026

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

See my other comment; this is an invalid address, and would be rejected.

@crazy-max crazy-max Sep 10, 2026

Copy link
Copy Markdown
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This test constructs url.URL{Host: ...} directly, so the decoded %Eth0 is intentional. The https-ipv6-zone-case-preserved source-conversion test separately parses a URL containing %25Eth0 and verifies the decoded host.

{"https", "[2001:db8::1]:443", "[2001:db8::1]"},
{"https", "[2001:db8::1]:000443", "[2001:db8::1]"},
{"http", "[2001:db8::1]:00080", "[2001:db8::1]"},
{"https", "[2001:db8::1]:8443", "[2001:db8::1]:8443"},
}
for _, tt := range tests {
t.Run(tt.scheme+" "+tt.host, func(t *testing.T) {
require.Equal(t, tt.want, normalizeHTTPHost(&url.URL{Scheme: tt.scheme, Host: tt.host}))
})
}
}

func TestSourceToInputSingleSource(t *testing.T) {
tm := time.Date(2024, 1, 2, 3, 4, 5, 0, time.UTC)

Expand All @@ -48,6 +89,68 @@ func TestSourceToInputSingleSource(t *testing.T) {
},
expErrMsg: "invalid source identifier: not-a-source",
},
{
name: "http-ipv6-without-brackets",
src: &gwpb.ResolveSourceMetaResponse{
Source: &pb.SourceOp{Identifier: "http://::1:443"},
},
expErrMsg: "failed to parse http source url",
},
{
name: "https-ipv6-without-brackets",
src: &gwpb.ResolveSourceMetaResponse{
Source: &pb.SourceOp{Identifier: "https://::1:443"},
},
expErrMsg: "failed to parse http source url",
},
{
name: "http-mixed-case-scheme-and-host",
src: &gwpb.ResolveSourceMetaResponse{
Source: &pb.SourceOp{Identifier: "hTtP://User:PaSs@eXaMpLe.CoM:00080/Case/Path?Key=VaLuE#Frag"},
},
expInput: Input{
HTTP: &HTTP{
URL: "hTtP://User:PaSs@eXaMpLe.CoM:00080/Case/Path?Key=VaLuE#Frag",
Schema: "http",
Host: "example.com",
Path: "/Case/Path",
Query: map[string][]string{"Key": {"VaLuE"}},
},
},
expUnk: []string{"input.http.checksum"},
},
{
name: "https-mixed-case-host-with-non-default-port",
src: &gwpb.ResolveSourceMetaResponse{
Source: &pb.SourceOp{Identifier: "HTTPS://EXAMPLE.COM:8443/Case?Key=Value"},
},
expInput: Input{
HTTP: &HTTP{
URL: "HTTPS://EXAMPLE.COM:8443/Case?Key=Value",
Schema: "https",
Host: "example.com:8443",
Path: "/Case",
Query: map[string][]string{"Key": {"Value"}},
},
},
expUnk: []string{"input.http.checksum"},
},
{
name: "https-ipv6-zone-case-preserved",
src: &gwpb.ResolveSourceMetaResponse{
Source: &pb.SourceOp{Identifier: "HTTPS://[FE80::ABCD%25Eth0]:000443/Case"},
},
expInput: Input{
HTTP: &HTTP{
URL: "HTTPS://[FE80::ABCD%25Eth0]:000443/Case",
Schema: "https",
Host: "[fe80::abcd%Eth0]",
Path: "/Case",
Query: map[string][]string{},
},
},
expUnk: []string{"input.http.checksum"},
},
{
name: "http-source-with-checksum-and-auth",
src: &gwpb.ResolveSourceMetaResponse{
Expand Down Expand Up @@ -133,6 +236,120 @@ func TestSourceToInputSingleSource(t *testing.T) {
},
},
},
{
name: "https-default-port-stripped-from-host",
src: &gwpb.ResolveSourceMetaResponse{
Source: &pb.SourceOp{
Identifier: "https://example.com:443/foo.tar.gz",
},
HTTP: &gwpb.ResolveSourceHTTPResponse{
Checksum: "sha256:aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa",
},
},
expInput: Input{
HTTP: &HTTP{
URL: "https://example.com:443/foo.tar.gz",
Schema: "https",
Host: "example.com",
Path: "/foo.tar.gz",
Query: map[string][]string{},
Checksum: "sha256:aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa",
},
},
},
{
name: "http-default-port-stripped-from-host",
src: &gwpb.ResolveSourceMetaResponse{
Source: &pb.SourceOp{
Identifier: "http://example.com:80/foo.tar.gz",
},
HTTP: &gwpb.ResolveSourceHTTPResponse{
Checksum: "sha256:bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb",
},
},
expInput: Input{
HTTP: &HTTP{
URL: "http://example.com:80/foo.tar.gz",
Schema: "http",
Host: "example.com",
Path: "/foo.tar.gz",
Query: map[string][]string{},
Checksum: "sha256:bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb",
},
},
},
{
name: "http-empty-host-with-default-port",
src: &gwpb.ResolveSourceMetaResponse{
Source: &pb.SourceOp{Identifier: "http://:80"},
},
expInput: Input{
HTTP: &HTTP{
URL: "http://:80",
Schema: "http",
Host: "",
Query: map[string][]string{},
},
},
expUnk: []string{"input.http.checksum"},
},
{
name: "https-empty-host-with-default-port",
src: &gwpb.ResolveSourceMetaResponse{
Source: &pb.SourceOp{Identifier: "https://:443"},
},
expInput: Input{
HTTP: &HTTP{
URL: "https://:443",
Schema: "https",
Host: "",
Query: map[string][]string{},
},
},
expUnk: []string{"input.http.checksum"},
},
{
name: "https-non-default-port-kept-on-host",
src: &gwpb.ResolveSourceMetaResponse{
Source: &pb.SourceOp{
Identifier: "https://example.com:8443/foo.tar.gz",
},
HTTP: &gwpb.ResolveSourceHTTPResponse{
Checksum: "sha256:cccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccc",
},
},
expInput: Input{
HTTP: &HTTP{
URL: "https://example.com:8443/foo.tar.gz",
Schema: "https",
Host: "example.com:8443",
Path: "/foo.tar.gz",
Query: map[string][]string{},
Checksum: "sha256:cccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccc",
},
},
},
{
name: "https-ipv6-default-port-stripped-from-host",
src: &gwpb.ResolveSourceMetaResponse{
Source: &pb.SourceOp{
Identifier: "https://[2001:db8::1]:443/foo.tar.gz",
},
HTTP: &gwpb.ResolveSourceHTTPResponse{
Checksum: "sha256:dddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddd",
},
},
expInput: Input{
HTTP: &HTTP{
URL: "https://[2001:db8::1]:443/foo.tar.gz",
Schema: "https",
Host: "[2001:db8::1]",
Path: "/foo.tar.gz",
Query: map[string][]string{},
Checksum: "sha256:dddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddddd",
},
},
},
{
name: "local-source",
src: &gwpb.ResolveSourceMetaResponse{
Expand Down Expand Up @@ -883,6 +1100,64 @@ func TestSourceToInputSingleSource(t *testing.T) {
}
}

func TestCheckPolicyHTTPHost(t *testing.T) {
for _, tc := range []struct {
url string
allow bool
}{
{"https://example.com/", true},
{"Http://Example.com/", true},
{"hTtP://eXaMpLe.CoM/", true},
{"HTTP://EXAMPLE.COM/", true},
{"HTTPS://EXAMPLE.COM:000443/", true},
{"HTTPS://EXAMPLE.COM:8443/", false},
{"https://example.com:443/", true},
{"http://example.com:80/", true},
{"https://example.com:000443/", true},
{"http://example.com:00080/", true},
{"https://example.com:00080/", false},
{"http://example.com:000443/", false},
{"https://example.com:008443/", false},
{"https://example.com:80/", false},
{"http://example.com:443/", false},
{"https://example.com:8443/", false},
{"https://[2001:db8::1]:443/", true},
{"https://[2001:db8::1]:000443/", true},
{"http://[2001:db8::1]:00080/", true},
{"https://[2001:db8::1]:8443/", false},
} {
t.Run(tc.url, func(t *testing.T) {
p := NewPolicy(Opt{
Files: []File{{
Filename: "policy.rego",
Data: []byte(`
package docker

default allow := false
allow if input.http.host in ["example.com", "[2001:db8::1]"]
decision := {"allow": allow}
`),
}},
})

// Exec proxy requests do not include HTTP checksum metadata.
decision, next, err := p.CheckPolicy(t.Context(), &policysession.CheckPolicyRequest{
Source: &gwpb.ResolveSourceMetaResponse{
Source: &pb.SourceOp{Identifier: tc.url},
},
})
require.NoError(t, err)
require.Nil(t, next)
require.NotNil(t, decision)
want := policyaction.PolicyAction_DENY
if tc.allow {
want = policyaction.PolicyAction_ALLOW
}
require.Equal(t, want, decision.Action)
})
}
}

func TestCheckCaps(t *testing.T) {
p := NewPolicy(Opt{
Files: []File{{
Expand Down
Loading