fix(gateway): only convert between dag-* variants

core/corehttp/gateway_handler_codec.go

@@ -32,15 +32,11 @@ var codecToContentType = map[uint64]string{
 	uint64(mc.DagCbor): "application/vnd.ipld.dag-cbor",
 }
 
-// contentTypeToCodecs maps the HTTP Content Type to the respective
-// possible codecs. If the original data is in one of those codecs,
-// we stream the raw bytes. Otherwise, we encode in the last codec
-// of the list.
-var contentTypeToCodecs = map[string][]uint64{
-	"application/json":              {uint64(mc.Json), uint64(mc.DagJson)},
-	"application/vnd.ipld.dag-json": {uint64(mc.DagJson)},
-	"application/cbor":              {uint64(mc.Cbor), uint64(mc.DagCbor)},
-	"application/vnd.ipld.dag-cbor": {uint64(mc.DagCbor)},
+// convertibleContentTypeToCodec maps the HTTP Content Type to the respective
+// possible codec to convert.
+var convertibleContentTypeToCodec = map[string]uint64{
+	"application/vnd.ipld.dag-json": uint64(mc.DagJson),
+	"application/vnd.ipld.dag-cbor": uint64(mc.DagCbor),
 }
 
 // contentTypeToExtension maps the HTTP Content Type to the respective file
@@ -103,38 +99,17 @@ func (i *gatewayHandler) serveCodec(ctx context.Context, w http.ResponseWriter,
 		return
 	}
 
-	// Otherwise, the user has requested a specific content type. Let's first get
-	// the codecs that can be used with this content type.
-	codecs, ok := contentTypeToCodecs[requestedContentType]
-	if !ok {
-		// This is never supposed to happen unless function is called with wrong parameters.
-		err := fmt.Errorf("unsupported content type: %s", requestedContentType)
-		webError(w, err.Error(), err, http.StatusInternalServerError)
-		return
-	}
-
-	// If we need to convert, use the last codec (strict dag- variant)
-	toCodec := codecs[len(codecs)-1]
-
-	// If the requested content type has "dag-", ALWAYS go through the encoding
-	// process in order to validate the content.
-	if strings.Contains(requestedContentType, "dag-") {
+	// If the user has requested a specific content type that we can convert to
+	// (dag-*), we check the codecs and go through the encoding process in order
+	// to validate the content. Otherwise, we directly stream the raw data.
+	// serveRawBlock cannot be directly used here as it sets different headers.
+	// Note that we cannot convert between json and cbor. Only between dag-json
+	// and dag-cbor.
+	if toCodec, ok := convertibleContentTypeToCodec[requestedContentType]; ok {
 		i.serveCodecConverted(ctx, w, r, resolvedPath, contentPath, toCodec, modtime)
-		return
-	}
-
-	// Otherwise, check if the data is encoded with the requested content type.
-	// If so, we can directly stream the raw data. serveRawBlock cannot be directly
-	// used here as it sets different headers.
-	for _, codec := range codecs {
-		if resolvedPath.Cid().Prefix().Codec == codec {
-			i.serveCodecRaw(ctx, w, r, resolvedPath, contentPath, name, modtime)
-			return
-		}
+	} else {
+		i.serveCodecRaw(ctx, w, r, resolvedPath, contentPath, name, modtime)
 	}
-
-	// Finally, if nothing of the above is true, we have to actually convert the codec.
-	i.serveCodecConverted(ctx, w, r, resolvedPath, contentPath, toCodec, modtime)
 }
 
 func (i *gatewayHandler) serveCodecHTML(ctx context.Context, w http.ResponseWriter, r *http.Request, resolvedPath ipath.Resolved, contentPath ipath.Path) {

test/sharness/t0123-gateway-json-cbor.sh

@@ -81,12 +81,6 @@ test_dag_pb () {
     ipfs dag get --output-codec dag-$format $DIR_CID > ipfs_dag_get_output 2>&1 &&
     test_cmp ipfs_dag_get_output curl_output
   '
-
-  test_expect_success "GET UnixFS as $name with format=dag-$format and format=$format produce same output" '
-    curl -s "http://127.0.0.1:$GWAY_PORT/ipfs/$DIR_CID?format=dag-$format" > curl_output_1 2>&1 &&
-    curl -s "http://127.0.0.1:$GWAY_PORT/ipfs/$DIR_CID?format=$format" > curl_output_2 2>&1 &&
-    test_cmp curl_output_1 curl_output_2
-  '
 }
 
 test_dag_pb "DAG-JSON" "json"
@@ -146,23 +140,6 @@ test_cmp_dag_get "JSON" "json" "inline"
 test_cmp_dag_get "CBOR" "cbor" "attachment"
 
 
-## Lossless conversion between JSON and CBOR
-
-test_expect_success "GET JSON as CBOR produces DAG-CBOR output" '
-  CID=$(echo "{ \"test\": \"json\" }" | ipfs dag put --input-codec json --store-codec json) &&
-  curl -s "http://127.0.0.1:$GWAY_PORT/ipfs/$CID?format=cbor" > curl_output 2>&1 &&
-  ipfs dag get --output-codec dag-cbor $CID > ipfs_dag_get_output 2>&1 &&
-  test_cmp ipfs_dag_get_output curl_output
-'
-
-test_expect_success "GET CBOR as JSON produces DAG-JSON output" '
-  CID=$(echo "{ \"test\": \"json\" }" | ipfs dag put --input-codec json --store-codec cbor) &&
-  curl -s "http://127.0.0.1:$GWAY_PORT/ipfs/$CID?format=json" > curl_output 2>&1 &&
-  ipfs dag get --output-codec dag-json $CID > ipfs_dag_get_output 2>&1 &&
-  test_cmp ipfs_dag_get_output curl_output
-'
-
-
 ## Pathing, traversal
 
 DAG_CBOR_TRAVERSAL_CID="bafyreibs4utpgbn7uqegmd2goqz4bkyflre2ek2iwv743fhvylwi4zeeim"
@@ -204,12 +181,6 @@ test_expect_success "GET DAG-CBOR traverses multiple links" '
   test_cmp expected actual
 '
 
-# test_expect_success "GET DAG-PB has expected output" '
-#   curl -s "http://127.0.0.1:$GWAY_PORT/ipfs/$DAG_PB_CID?format=dag-json" > curl_output 2>&1 &&
-#   jq --sort-keys . curl_output > actual &&
-#   test_cmp ../t0123-gateway-json-cbor/dag-pb.json actual
-# '
-
 
 ## NATIVE TESTS:
 ## DAG- regression tests for core behaviors when native DAG-(CBOR|JSON) is requested
@@ -302,11 +273,11 @@ test_native_dag () {
     test_should_contain "Content-Type: application/vnd.ipld.dag-$format" output &&
     test_should_contain "Content-Length: " output
   '
-  test_expect_success "HEAD $name with an explicit JSON format returns HTTP 200" '
-    curl -I "http://127.0.0.1:$GWAY_PORT/ipfs/$CID?format=json" -o output &&
+  test_expect_success "HEAD $name with an explicit DAG-JSON format returns HTTP 200" '
+    curl -I "http://127.0.0.1:$GWAY_PORT/ipfs/$CID?format=dag-json" -o output &&
     test_should_contain "HTTP/1.1 200 OK" output &&
-    test_should_contain "Etag: \"$CID.json\"" output &&
-    test_should_contain "Content-Type: application/json" output &&
+    test_should_contain "Etag: \"$CID.dag-json\"" output &&
+    test_should_contain "Content-Type: application/vnd.ipld.dag-json" output &&
     test_should_contain "Content-Length: " output
   '
   test_expect_success "HEAD dag-pb with ?format=$format returns HTTP 200" '