[bugfix] add stricter checks during all stages of dereferencing remote AS objects (#2639)

* add stricter checks during all stages of dereferencing remote AS objects

* a comment

cmd/gotosocial/action/testrig/testrig.go

@@ -104,6 +104,9 @@
 		return &http.Response{
 			StatusCode: 200,
 			Body:       r,
+			Header: http.Header{
+				"Content-Type": req.Header.Values("Accept"),
+			},
 		}, nil
 	}, ""))
 	mediaManager := testrig.NewTestMediaManager(&state)

internal/api/util/mime.go

@@ -17,6 +17,8 @@
 
 package util
 
+import "strings"
+
 const (
 	// Possible GoToSocial mimetypes.
 	AppJSON           = `application/json`
@@ -24,7 +26,8 @@
 	AppXMLXRD         = `application/xrd+xml`
 	AppRSSXML         = `application/rss+xml`
 	AppActivityJSON   = `application/activity+json`
-	AppActivityLDJSON = `application/ld+json; profile="https://www.w3.org/ns/activitystreams"`
+	appActivityLDJSON = `application/ld+json` // without profile
+	AppActivityLDJSON = appActivityLDJSON + `; profile="https://www.w3.org/ns/activitystreams"`
 	AppJRDJSON        = `application/jrd+json` // https://www.rfc-editor.org/rfc/rfc7033#section-10.2
 	AppForm           = `application/x-www-form-urlencoded`
 	MultipartForm     = `multipart/form-data`
@@ -32,3 +35,112 @@
 	TextHTML          = `text/html`
 	TextCSS           = `text/css`
 )
+
+// JSONContentType returns whether is application/json(;charset=utf-8)? content-type.
+func JSONContentType(ct string) bool {
+	p := splitContentType(ct)
+	p, ok := isUTF8ContentType(p)
+	return ok && len(p) == 1 &&
+		p[0] == AppJSON
+}
+
+// JSONJRDContentType returns whether is application/(jrd+)?json(;charset=utf-8)? content-type.
+func JSONJRDContentType(ct string) bool {
+	p := splitContentType(ct)
+	p, ok := isUTF8ContentType(p)
+	return ok && len(p) == 1 &&
+		p[0] == AppJSON ||
+		p[0] == AppJRDJSON
+}
+
+// XMLContentType returns whether is application/xml(;charset=utf-8)? content-type.
+func XMLContentType(ct string) bool {
+	p := splitContentType(ct)
+	p, ok := isUTF8ContentType(p)
+	return ok && len(p) == 1 &&
+		p[0] == AppXML
+}
+
+// XMLXRDContentType returns whether is application/(xrd+)?xml(;charset=utf-8)? content-type.
+func XMLXRDContentType(ct string) bool {
+	p := splitContentType(ct)
+	p, ok := isUTF8ContentType(p)
+	return ok && len(p) == 1 &&
+		p[0] == AppXML ||
+		p[0] == AppXMLXRD
+}
+
+// ASContentType returns whether is valid ActivityStreams content-types:
+// - application/activity+json
+// - application/ld+json;profile=https://w3.org/ns/activitystreams
+func ASContentType(ct string) bool {
+	p := splitContentType(ct)
+	p, ok := isUTF8ContentType(p)
+	if !ok {
+		return false
+	}
+	switch len(p) {
+	case 1:
+		return p[0] == AppActivityJSON
+	case 2:
+		return p[0] == appActivityLDJSON &&
+			p[1] == "profile=https://www.w3.org/ns/activitystreams" ||
+			p[1] == "profile=\"https://www.w3.org/ns/activitystreams\""
+	default:
+		return false
+	}
+}
+
+// NodeInfo2ContentType returns whether is nodeinfo schema 2.0 content-type.
+func NodeInfo2ContentType(ct string) bool {
+	p := splitContentType(ct)
+	p, ok := isUTF8ContentType(p)
+	if !ok {
+		return false
+	}
+	switch len(p) {
+	case 1:
+		return p[0] == AppJSON
+	case 2:
+		return p[0] == AppJSON &&
+			p[1] == "profile=\"http://nodeinfo.diaspora.software/ns/schema/2.0#\"" ||
+			p[1] == "profile=http://nodeinfo.diaspora.software/ns/schema/2.0#"
+	default:
+		return false
+	}
+}
+
+// isUTF8ContentType checks for a provided charset in given
+// type parts list, removes it and returns whether is utf-8.
+func isUTF8ContentType(p []string) ([]string, bool) {
+	const charset = "charset="
+	const charsetUTF8 = charset + "utf-8"
+	for i, part := range p {
+
+		// Only handle charset slice parts.
+		if part[:len(charset)] == charset {
+
+			// Check if is UTF-8 charset.
+			ok := (part == charsetUTF8)
+
+			// Drop this slice part.
+			_ = copy(p[i:], p[i+1:])
+			p = p[:len(p)-1]
+
+			return p, ok
+		}
+	}
+	return p, true
+}
+
+// splitContentType splits content-type into semi-colon
+// separated parts. useful when a charset is provided.
+// note this also maps all chars to their lowercase form.
+func splitContentType(ct string) []string {
+	s := strings.Split(ct, ";")
+	for i := range s {
+		s[i] = strings.TrimSpace(s[i])
+		s[i] = strings.ToLower(s[i])
+	}
+	return s
+}

internal/api/util/mime_test.go

@@ -0,0 +1,75 @@
+package util_test
+
+import (
+	"testing"
+
+	"github.com/superseriousbusiness/gotosocial/internal/api/util"
+)
+
+func TestIsASContentType(t *testing.T) {
+	for _, test := range []struct {
+		Input  string
+		Expect bool
+	}{
+		{
+			Input:  "application/activity+json",
+			Expect: true,
+		},
+		{
+			Input:  "application/activity+json; charset=utf-8",
+			Expect: true,
+		},
+		{
+			Input:  "application/activity+json;charset=utf-8",
+			Expect: true,
+		},
+		{
+			Input:  "application/activity+json ;charset=utf-8",
+			Expect: true,
+		},
+		{
+			Input:  "application/activity+json ; charset=utf-8",
+			Expect: true,
+		},
+		{
+			Input:  "application/ld+json;profile=https://www.w3.org/ns/activitystreams",
+			Expect: true,
+		},
+		{
+			Input:  "application/ld+json;profile=\"https://www.w3.org/ns/activitystreams\"",
+			Expect: true,
+		},
+		{
+			Input:  "application/ld+json ;profile=https://www.w3.org/ns/activitystreams",
+			Expect: true,
+		},
+		{
+			Input:  "application/ld+json ;profile=\"https://www.w3.org/ns/activitystreams\"",
+			Expect: true,
+		},
+		{
+			Input:  "application/ld+json ; profile=https://www.w3.org/ns/activitystreams",
+			Expect: true,
+		},
+		{
+			Input:  "application/ld+json ; profile=\"https://www.w3.org/ns/activitystreams\"",
+			Expect: true,
+		},
+		{
+			Input:  "application/ld+json; profile=https://www.w3.org/ns/activitystreams",
+			Expect: true,
+		},
+		{
+			Input:  "application/ld+json; profile=\"https://www.w3.org/ns/activitystreams\"",
+			Expect: true,
+		},
+		{
+			Input:  "application/ld+json",
+			Expect: false,
+		},
+	} {
+		if util.ASContentType(test.Input) != test.Expect {
+			t.Errorf("did not get expected result %v for input: %s", test.Expect, test.Input)
+		}
+	}
+}

internal/federation/dereferencing/account.go

@@ -586,6 +586,16 @@ func (d *Dereferencer) enrichAccount(
 	}
 
 	if account.Username == "" {
+		// Assume the host from the
+		// ActivityPub representation.
+		id := ap.GetJSONLDId(apubAcc)
+		if id == nil {
+			return nil, nil, gtserror.New("no id property found on person, or id was not an iri")
+		}
+
+		// Get IRI host value.
+		accHost := id.Host
+
 		// No username was provided, so no webfinger was attempted earlier.
 		//
 		// Now we have a username we can attempt again, to ensure up-to-date
@@ -596,42 +606,37 @@ func (d *Dereferencer) enrichAccount(
 		// https://example.org/@someone@somewhere.else and we've been redirected
 		// from example.org to somewhere.else: we want to take somewhere.else
 		// as the accountDomain then, not the example.org we were redirected from.
-
-		// Assume the host from the returned
-		// ActivityPub representation.
-		id := ap.GetJSONLDId(apubAcc)
-		if id == nil {
-			return nil, nil, gtserror.New("no id property found on person, or id was not an iri")
-		}
-
-		// Get IRI host value.
-		accHost := id.Host
-
 		latestAcc.Domain, _, err = d.fingerRemoteAccount(ctx,
 			tsport,
 			latestAcc.Username,
 			accHost,
 		)
 		if err != nil {
-			// We still couldn't webfinger the account, so we're not certain
-			// what the accountDomain actually is. Still, we can make a solid
-			// guess that it's the Host of the ActivityPub URI of the account.
-			// If we're wrong, we can just try again in a couple days.
-			log.Errorf(ctx, "error webfingering[2] remote account %s@%s: %v", latestAcc.Username, accHost, err)
-			latestAcc.Domain = accHost
+			// Webfingering account still failed, so we're not certain
+			// what the accountDomain actually is. Exit here for safety.
+			return nil, nil, gtserror.Newf(
+				"error webfingering remote account %s@%s: %w",
+				latestAcc.Username, accHost, err,
+			)
 		}
 	}
 
 	if latestAcc.Domain == "" {
 		// Ensure we have a domain set by this point,
 		// otherwise it gets stored as a local user!
-		//
-		// TODO: there is probably a more granular way
-		// way of checking this in each of the above parts,
-		// and honestly it could do with a smol refactor.
 		return nil, nil, gtserror.Newf("empty domain for %s", uri)
 	}
 
+	// Ensure the final parsed account URI / URL matches
+	// the input URI we fetched (or received) it as.
+	if expect := uri.String(); latestAcc.URI != expect &&
+		latestAcc.URL != expect {
+		return nil, nil, gtserror.Newf(
+			"dereferenced account uri %s does not match %s",
+			latestAcc.URI, expect,
+		)
+	}
+
 	/*
 		BY THIS POINT we have more or less a fullly-formed
 		representation of the target account, derived from

internal/federation/dereferencing/account_test.go

@@ -19,6 +19,7 @@
 
 import (
 	"context"
+	"fmt"
 	"testing"
 	"time"
 
@@ -207,6 +208,28 @@ func (suite *AccountTestSuite) TestDereferenceLocalAccountWithUnknownUserURI() {
 	suite.Nil(fetchedAccount)
 }
 
+func (suite *AccountTestSuite) TestDereferenceRemoteAccountWithNonMatchingURI() {
+	fetchingAccount := suite.testAccounts["local_account_1"]
+
+	const (
+		remoteURI    = "https://turnip.farm/users/turniplover6969"
+		remoteAltURI = "https://turnip.farm/users/turniphater420"
+	)
+
+	// Create a copy of this remote account at alternative URI.
+	remotePerson := suite.client.TestRemotePeople[remoteURI]
+	suite.client.TestRemotePeople[remoteAltURI] = remotePerson
+
+	// Attempt to fetch account at alternative URI, it should fail!
+	fetchedAccount, _, err := suite.dereferencer.GetAccountByURI(
+		context.Background(),
+		fetchingAccount.Username,
+		testrig.URLMustParse(remoteAltURI),
+	)
+	suite.Equal(err.Error(), fmt.Sprintf("enrichAccount: dereferenced account uri %s does not match %s", remoteURI, remoteAltURI))
+	suite.Nil(fetchedAccount)
+}
+
 func TestAccountTestSuite(t *testing.T) {
 	suite.Run(t, new(AccountTestSuite))
 }

internal/federation/dereferencing/dereferencer_test.go

@@ -35,6 +35,7 @@ type DereferencerStandardTestSuite struct {
 	db      db.DB
 	storage *storage.Driver
 	state   state.State
+	client  *testrig.MockHTTPClient
 
 	testRemoteStatuses    map[string]vocab.ActivityStreamsNote
 	testRemotePeople      map[string]vocab.ActivityStreamsPerson
@@ -72,11 +73,12 @@ func (suite *DereferencerStandardTestSuite) SetupTest() {
 		converter,
 	)
 
+	suite.client = testrig.NewMockHTTPClient(nil, "../../../testrig/media")
 	suite.storage = testrig.NewInMemoryStorage()
 	suite.state.DB = suite.db
 	suite.state.Storage = suite.storage
 	media := testrig.NewTestMediaManager(&suite.state)
-	suite.dereferencer = dereferencing.NewDereferencer(&suite.state, converter, testrig.NewTestTransportController(&suite.state, testrig.NewMockHTTPClient(nil, "../../../testrig/media")), media)
+	suite.dereferencer = dereferencing.NewDereferencer(&suite.state, converter, testrig.NewTestTransportController(&suite.state, suite.client), media)
 	testrig.StandardDBSetup(suite.db, nil)
 }
 

internal/federation/dereferencing/finger.go

@@ -21,9 +21,9 @@
 	"context"
 	"encoding/json"
 	"net/url"
-	"strings"
 
 	apimodel "github.com/superseriousbusiness/gotosocial/internal/api/model"
+	apiutil "github.com/superseriousbusiness/gotosocial/internal/api/util"
 	"github.com/superseriousbusiness/gotosocial/internal/gtserror"
 	"github.com/superseriousbusiness/gotosocial/internal/log"
 	"github.com/superseriousbusiness/gotosocial/internal/transport"
@@ -74,10 +74,12 @@ func (d *Dereferencer) fingerRemoteAccount(
 		return "", nil, err
 	}
 
-	_, accountDomain, err := util.ExtractWebfingerParts(resp.Subject)
+	accUsername, accDomain, err := util.ExtractWebfingerParts(resp.Subject)
 	if err != nil {
 		err = gtserror.Newf("error extracting subject parts for %s: %w", target, err)
 		return "", nil, err
+	} else if accUsername != username {
+		return "", nil, gtserror.Newf("response username does not match input for %s: %w", target, err)
 	}
 
 	// Look through links for the first
@@ -92,8 +94,7 @@ func (d *Dereferencer) fingerRemoteAccount(
 			continue
 		}
 
-		if !strings.EqualFold(link.Type, "application/activity+json") &&
-			!strings.EqualFold(link.Type, "application/ld+json; profile=\"https://www.w3.org/ns/activitystreams\"") {
+		if !apiutil.ASContentType(link.Type) {
 			// Not an AP type, ignore.
 			continue
 		}
@@ -121,7 +122,7 @@ func (d *Dereferencer) fingerRemoteAccount(
 		}
 
 		// All looks good, return happily!
-		return accountDomain, uri, nil
+		return accDomain, uri, nil
 	}
 
 	return "", nil, gtserror.Newf("no suitable self, AP-type link found in webfinger response for %s", target)

internal/federation/dereferencing/status.go

@@ -396,7 +396,7 @@ func (d *Dereferencer) enrichStatus(
 	}
 
 	// Ensure we have the author account of the status dereferenced (+ up-to-date). If this is a new status
-	// (i.e. status.AccountID == "") then any error here is irrecoverable. AccountID must ALWAYS be set.
+	// (i.e. status.AccountID == "") then any error here is irrecoverable. status.AccountID must ALWAYS be set.
 	if _, _, err := d.getAccountByURI(ctx, requestUser, attributedTo); err != nil && status.AccountID == "" {
 		return nil, nil, gtserror.Newf("failed to dereference status author %s: %w", uri, err)
 	}
@@ -408,13 +408,33 @@ func (d *Dereferencer) enrichStatus(
 		return nil, nil, gtserror.Newf("error converting statusable to gts model for status %s: %w", uri, err)
 	}
 
-	// Check if we've previously
-	// stored this status in the DB.
-	// If we have, it'll be ID'd.
-	var isNew = (status.ID == "")
-	if isNew {
-		// No ID, we haven't stored this status before.
-		// Generate new status ID from the status publication time.
+	// Ensure final status isn't attempting
+	// to claim being authored by local user.
+	if latestStatus.Account.IsLocal() {
+		return nil, nil, gtserror.Newf(
+			"dereferenced status %s claiming to be local",
+			latestStatus.URI,
+		)
+	}
+
+	// Ensure the final parsed status URI / URL matches
+	// the input URI we fetched (or received) it as.
+	if expect := uri.String(); latestStatus.URI != expect &&
+		latestStatus.URL != expect {
+		return nil, nil, gtserror.Newf(
+			"dereferenced status uri %s does not match %s",
+			latestStatus.URI, expect,
+		)
+	}
+
+	var isNew bool
+
+	// Based on the original provided
+	// status model, determine whether
+	// this is a new insert / update.
+	if isNew = (status.ID == ""); isNew {
+
+		// Generate new status ID from the provided creation date.
 		latestStatus.ID, err = id.NewULIDFromTime(latestStatus.CreatedAt)
 		if err != nil {
 			log.Errorf(ctx, "invalid created at date (falling back to 'now'): %v", err)

internal/federation/dereferencing/status_test.go

@@ -19,6 +19,7 @@
 
 import (
 	"context"
+	"fmt"
 	"testing"
 
 	"github.com/stretchr/testify/suite"
@@ -218,6 +219,28 @@ func (suite *StatusTestSuite) TestDereferenceStatusWithImageAndNoContent() {
 	suite.NoError(err)
 }
 
+func (suite *StatusTestSuite) TestDereferenceStatusWithNonMatchingURI() {
+	fetchingAccount := suite.testAccounts["local_account_1"]
+
+	const (
+		remoteURI    = "https://turnip.farm/users/turniplover6969/statuses/70c53e54-3146-42d5-a630-83c8b6c7c042"
+		remoteAltURI = "https://turnip.farm/users/turniphater420/statuses/70c53e54-3146-42d5-a630-83c8b6c7c042"
+	)
+
+	// Create a copy of this remote account at alternative URI.
+	remoteStatus := suite.client.TestRemoteStatuses[remoteURI]
+	suite.client.TestRemoteStatuses[remoteAltURI] = remoteStatus
+
+	// Attempt to fetch account at alternative URI, it should fail!
+	fetchedStatus, _, err := suite.dereferencer.GetStatusByURI(
+		context.Background(),
+		fetchingAccount.Username,
+		testrig.URLMustParse(remoteAltURI),
+	)
+	suite.Equal(err.Error(), fmt.Sprintf("enrichStatus: dereferenced status uri %s does not match %s", remoteURI, remoteAltURI))
+	suite.Nil(fetchedStatus)
+}
+
 func TestStatusTestSuite(t *testing.T) {
 	suite.Run(t, new(StatusTestSuite))
 }

internal/federation/federatingactor.go

@@ -23,70 +23,18 @@
 	"fmt"
 	"net/http"
 	"net/url"
-	"strings"
 
 	errorsv2 "codeberg.org/gruf/go-errors/v2"
 	"codeberg.org/gruf/go-kv"
 	"github.com/superseriousbusiness/activity/pub"
 	"github.com/superseriousbusiness/activity/streams/vocab"
 	"github.com/superseriousbusiness/gotosocial/internal/ap"
+	apiutil "github.com/superseriousbusiness/gotosocial/internal/api/util"
 	"github.com/superseriousbusiness/gotosocial/internal/db"
 	"github.com/superseriousbusiness/gotosocial/internal/gtserror"
 	"github.com/superseriousbusiness/gotosocial/internal/log"
 )
 
-// IsASMediaType will return whether the given content-type string
-// matches one of the 2 possible ActivityStreams incoming content types:
-// - application/activity+json
-// - application/ld+json;profile=https://w3.org/ns/activitystreams
-//
-// Where for the above we are leniant with whitespace, quotes, and charset.
-func IsASMediaType(ct string) bool {
-	var (
-		// First content-type part,
-		// contains the application/...
-		p1 string = ct //nolint:revive
-
-		// Second content-type part,
-		// contains AS IRI or charset
-		// if provided.
-		p2 string
-	)
-
-	// Split content-type by semi-colon.
-	sep := strings.IndexByte(ct, ';')
-	if sep >= 0 {
-		p1 = ct[:sep]
-
-		// Trim all start/end
-		// space of second part.
-		p2 = ct[sep+1:]
-		p2 = strings.Trim(p2, " ")
-	}
-
-	// Trim any ending space from the
-	// main content-type part of string.
-	p1 = strings.TrimRight(p1, " ")
-
-	switch p1 {
-	case "application/activity+json":
-		// Accept with or without charset.
-		// This should be case insensitive.
-		// https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Content-Type#charset
-		return p2 == "" || strings.EqualFold(p2, "charset=utf-8")
-
-	case "application/ld+json":
-		// Drop any quotes around the URI str.
-		p2 = strings.ReplaceAll(p2, "\"", "")
-
-		// End part must be a ref to the main AS namespace IRI.
-		return p2 == "profile=https://www.w3.org/ns/activitystreams"
-
-	default:
-		return false
-	}
-}
-
 // federatingActor wraps the pub.FederatingActor
 // with some custom GoToSocial-specific logic.
 type federatingActor struct {
@@ -124,7 +72,7 @@ func (f *federatingActor) PostInboxScheme(ctx context.Context, w http.ResponseWr
 
 	// Ensure valid ActivityPub Content-Type.
 	// https://www.w3.org/TR/activitypub/#server-to-server-interactions
-	if ct := r.Header.Get("Content-Type"); !IsASMediaType(ct) {
+	if ct := r.Header.Get("Content-Type"); !apiutil.ASContentType(ct) {
 		const ct1 = "application/activity+json"
 		const ct2 = "application/ld+json;profile=https://w3.org/ns/activitystreams"
 		err := fmt.Errorf("Content-Type %s not acceptable, this endpoint accepts: [%q %q]", ct, ct1, ct2)

internal/federation/federatingactor_test.go

@@ -154,71 +154,3 @@ func (suite *FederatingActorTestSuite) TestSendRemoteFollower() {
 func TestFederatingActorTestSuite(t *testing.T) {
 	suite.Run(t, new(FederatingActorTestSuite))
 }
-
-func TestIsASMediaType(t *testing.T) {
-	for _, test := range []struct {
-		Input  string
-		Expect bool
-	}{
-		{
-			Input:  "application/activity+json",
-			Expect: true,
-		},
-		{
-			Input:  "application/activity+json; charset=utf-8",
-			Expect: true,
-		},
-		{
-			Input:  "application/activity+json;charset=utf-8",
-			Expect: true,
-		},
-		{
-			Input:  "application/activity+json ;charset=utf-8",
-			Expect: true,
-		},
-		{
-			Input:  "application/activity+json ; charset=utf-8",
-			Expect: true,
-		},
-		{
-			Input:  "application/ld+json;profile=https://www.w3.org/ns/activitystreams",
-			Expect: true,
-		},
-		{
-			Input:  "application/ld+json;profile=\"https://www.w3.org/ns/activitystreams\"",
-			Expect: true,
-		},
-		{
-			Input:  "application/ld+json ;profile=https://www.w3.org/ns/activitystreams",
-			Expect: true,
-		},
-		{
-			Input:  "application/ld+json ;profile=\"https://www.w3.org/ns/activitystreams\"",
-			Expect: true,
-		},
-		{
-			Input:  "application/ld+json ; profile=https://www.w3.org/ns/activitystreams",
-			Expect: true,
-		},
-		{
-			Input:  "application/ld+json ; profile=\"https://www.w3.org/ns/activitystreams\"",
-			Expect: true,
-		},
-		{
-			Input:  "application/ld+json; profile=https://www.w3.org/ns/activitystreams",
-			Expect: true,
-		},
-		{
-			Input:  "application/ld+json; profile=\"https://www.w3.org/ns/activitystreams\"",
-			Expect: true,
-		},
-		{
-			Input:  "application/ld+json",
-			Expect: false,
-		},
-	} {
-		if federation.IsASMediaType(test.Input) != test.Expect {
-			t.Errorf("did not get expected result %v for input: %s", test.Expect, test.Input)
-		}
-	}
-}

internal/transport/dereference.go

@@ -64,9 +64,16 @@ func (t *transport) Dereference(ctx context.Context, iri *url.URL) ([]byte, erro
 	}
 	defer rsp.Body.Close()
 
+	// Ensure a non-error status response.
 	if rsp.StatusCode != http.StatusOK {
 		return nil, gtserror.NewFromResponse(rsp)
 	}
 
+	// Ensure that the incoming request content-type is expected.
+	if ct := rsp.Header.Get("Content-Type"); !apiutil.ASContentType(ct) {
+		err := gtserror.Newf("non activity streams response: %s", ct)
+		return nil, gtserror.SetMalformed(err)
+	}
+
 	return io.ReadAll(rsp.Body)
 }

internal/transport/derefinstance.go

@@ -101,10 +101,17 @@ func dereferenceByAPIV1Instance(ctx context.Context, t *transport, iri *url.URL)
 	}
 	defer resp.Body.Close()
 
+	// Ensure a non-error status response.
 	if resp.StatusCode != http.StatusOK {
 		return nil, gtserror.NewFromResponse(resp)
 	}
 
+	// Ensure that the incoming request content-type is expected.
+	if ct := resp.Header.Get("Content-Type"); !apiutil.JSONContentType(ct) {
+		err := gtserror.Newf("non json response type: %s", ct)
+		return nil, gtserror.SetMalformed(err)
+	}
+
 	b, err := io.ReadAll(resp.Body)
 	if err != nil {
 		return nil, err
@@ -251,20 +258,27 @@ func callNodeInfoWellKnown(ctx context.Context, t *transport, iri *url.URL) (*ur
 	}
 	defer resp.Body.Close()
 
+	// Ensure a non-error status response.
 	if resp.StatusCode != http.StatusOK {
 		return nil, gtserror.NewFromResponse(resp)
 	}
 
+	// Ensure that the incoming request content-type is expected.
+	if ct := resp.Header.Get("Content-Type"); !apiutil.JSONContentType(ct) {
+		err := gtserror.Newf("non json response type: %s", ct)
+		return nil, gtserror.SetMalformed(err)
+	}
+
 	b, err := io.ReadAll(resp.Body)
 	if err != nil {
 		return nil, err
 	} else if len(b) == 0 {
-		return nil, errors.New("callNodeInfoWellKnown: response bytes was len 0")
+		return nil, gtserror.New("response bytes was len 0")
 	}
 
 	wellKnownResp := &apimodel.WellKnownResponse{}
 	if err := json.Unmarshal(b, wellKnownResp); err != nil {
-		return nil, fmt.Errorf("callNodeInfoWellKnown: could not unmarshal server response as WellKnownResponse: %s", err)
+		return nil, gtserror.Newf("could not unmarshal server response as WellKnownResponse: %w", err)
 	}
 
 	// look through the links for the first one that matches the nodeinfo schema, this is what we need
@@ -275,11 +289,11 @@ func callNodeInfoWellKnown(ctx context.Context, t *transport, iri *url.URL) (*ur
 		}
 		nodeinfoHref, err = url.Parse(l.Href)
 		if err != nil {
-			return nil, fmt.Errorf("callNodeInfoWellKnown: couldn't parse url %s: %s", l.Href, err)
+			return nil, gtserror.Newf("couldn't parse url %s: %w", l.Href, err)
 		}
 	}
 	if nodeinfoHref == nil {
-		return nil, errors.New("callNodeInfoWellKnown: could not find nodeinfo rel in well known response")
+		return nil, gtserror.New("could not find nodeinfo rel in well known response")
 	}
 
 	return nodeinfoHref, nil
@@ -302,20 +316,27 @@ func callNodeInfo(ctx context.Context, t *transport, iri *url.URL) (*apimodel.No
 	}
 	defer resp.Body.Close()
 
+	// Ensure a non-error status response.
 	if resp.StatusCode != http.StatusOK {
 		return nil, gtserror.NewFromResponse(resp)
 	}
 
+	// Ensure that the incoming request content-type is expected.
+	if ct := resp.Header.Get("Content-Type"); !apiutil.NodeInfo2ContentType(ct) {
+		err := gtserror.Newf("non nodeinfo schema 2.0 response: %s", ct)
+		return nil, gtserror.SetMalformed(err)
+	}
+
 	b, err := io.ReadAll(resp.Body)
 	if err != nil {
 		return nil, err
 	} else if len(b) == 0 {
-		return nil, errors.New("callNodeInfo: response bytes was len 0")
+		return nil, gtserror.New("response bytes was len 0")
 	}
 
 	niResp := &apimodel.Nodeinfo{}
 	if err := json.Unmarshal(b, niResp); err != nil {
-		return nil, fmt.Errorf("callNodeInfo: could not unmarshal server response as Nodeinfo: %s", err)
+		return nil, gtserror.Newf("could not unmarshal server response as Nodeinfo: %w", err)
 	}
 
 	return niResp, nil

internal/transport/finger.go

@@ -97,9 +97,17 @@ func (t *transport) Finger(ctx context.Context, targetUsername string, targetDom
 			// again here to renew the TTL
 			t.controller.state.Caches.GTS.Webfinger().Set(targetDomain, url)
 		}
+
 		if rsp.StatusCode == http.StatusGone {
 			return nil, fmt.Errorf("account has been deleted/is gone")
 		}
+
+		// Ensure that the incoming request content-type is expected.
+		if ct := rsp.Header.Get("Content-Type"); !apiutil.JSONJRDContentType(ct) {
+			err := gtserror.Newf("non webfinger type response: %s", ct)
+			return nil, gtserror.SetMalformed(err)
+		}
+
 		return io.ReadAll(rsp.Body)
 	}
 
@@ -192,6 +200,12 @@ func (t *transport) webfingerFromHostMeta(ctx context.Context, targetDomain stri
 		return "", fmt.Errorf("GET request for %s failed: %s", req.URL.String(), rsp.Status)
 	}
 
+	// Ensure that the incoming request content-type is expected.
+	if ct := rsp.Header.Get("Content-Type"); !apiutil.XMLXRDContentType(ct) {
+		err := gtserror.Newf("non host-meta type response: %s", ct)
+		return "", gtserror.SetMalformed(err)
+	}
+
 	e := xml.NewDecoder(rsp.Body)
 	var hm apimodel.HostMeta
 	if err := e.Decode(&hm); err != nil {

testrig/transportcontroller.go

@@ -245,9 +245,7 @@ func NewMockHTTPClient(do func(req *http.Request) (*http.Response, error), relat
 			StatusCode:    responseCode,
 			Body:          readCloser,
 			ContentLength: int64(responseContentLength),
-			Header: http.Header{
-				"content-type": {responseContentType},
-			},
+			Header:        http.Header{"Content-Type": {responseContentType}},
 		}, nil
 	}