Implement Binary Upload (Fixes #2126)

.github/scripts/write-binary-content-via-raw-data.sh

@@ -0,0 +1,26 @@
+#!/bin/bash -e
+
+# This script creates a large binary resource (8 MiB) and verifies that its binary content
+# can be read correctly via both JSON/XML wrappers and direct binary upload.
+
+BASE="http://localhost:8080/fhir"
+
+# 8 MiB of random data (8388608 bytes), kept in memory
+DATA="$(openssl rand 8388608 | tr -d '\n')"
+
+echo "Testing direct binary upload and download..."
+
+# Create Binary resource via direct binary upload
+ID=$(curl -s -H 'Content-Type: application/octet-stream' --data-binary "$DATA" "$BASE/Binary" | jq -r '.id')
+
+echo "Created Binary resource via direct binary upload with ID: $ID"
+
+# Retrieve and verify via direct binary
+BASE64_ENCODED_BINARY_RESOURCE_VIA_DIRECT=$(curl -s -H 'Accept: application/octet-stream' "$BASE/Binary/$ID")
+
+if [ "$DATA" = "$BASE64_ENCODED_BINARY_RESOURCE_VIA_DIRECT" ]; then
+    echo "✅ Direct Binary: Successfully verified 8 MiB binary content integrity"
+else
+    echo "🆘 Direct Binary: Content verification failed"
+    exit 1
+fi

.github/workflows/build.yml

@@ -1074,6 +1074,9 @@ jobs:
     - name: Binary Content Download - found (via XML)
       run: .github/scripts/read-binary-content-via-xml-found.sh
 
+    - name: Binary Content Upload (via raw data)
+      run: .github/scripts/write-binary-content-via-raw-data.sh
+
     - name: Conditional Delete - Check Referential Integrity Violated
       run: .github/scripts/conditional-delete-type/check-referential-integrity-violated.sh
 

modules/rest-api/src/blaze/rest_api/routes.clj

@@ -35,6 +35,10 @@
   {:name :auth-guard
    :wrap auth-guard/wrap-auth-guard})
 
+(def ^:private wrap-binary-data
+  {:name :binary-data
+   :wrap resource/wrap-binary-data})
+
 (def ^:private wrap-resource
   {:name :resource
    :wrap resource/wrap-resource})
@@ -111,7 +115,9 @@
                                    :blaze.rest-api.interaction/handler)})
          (contains? interactions :create)
          (assoc :post {:interaction "create"
-                       :middleware [wrap-resource]
+                       :middleware (if (= name "Binary")
+                                     [wrap-binary-data]
+                                     [wrap-resource])
                        :handler (-> interactions :create
                                     :blaze.rest-api.interaction/handler)})
          (contains? interactions :conditional-delete-type)
@@ -179,7 +185,9 @@
                                       :blaze.rest-api.interaction/handler)})
             (contains? interactions :update)
             (assoc :put {:interaction "update"
-                         :middleware [wrap-resource]
+                         :middleware (if (= name "Binary")
+                                       [wrap-binary-data]
+                                       [wrap-resource])
                          :handler (-> interactions :update
                                       :blaze.rest-api.interaction/handler)})
             (contains? interactions :delete)

modules/rest-util/src/blaze/middleware/fhir/resource.clj

@@ -7,6 +7,7 @@
    [blaze.anomaly :as ba :refer [if-ok when-ok]]
    [blaze.async.comp :as ac]
    [blaze.fhir.spec :as fhir-spec]
+   [blaze.fhir.spec.type :as type]
    [clojure.data.xml.jvm.parse :as xml-jvm]
    [clojure.data.xml.tree :as xml-tree]
    [clojure.java.io :as io]
@@ -16,7 +17,10 @@
    [ring.util.request :as request])
   (:import
    [com.ctc.wstx.api WstxInputProperties]
+   [java.io InputStream]
    [java.io Reader]
+   [java.util Base64$Encoder]
+   [java.util Base64]
    [javax.xml.stream XMLInputFactory]))
 
 (set! *warn-on-reflection* true)
@@ -95,6 +99,25 @@
       (assoc request :body resource))
     (ba/incorrect "Missing HTTP body.")))
 
+(def ^:private ^Base64$Encoder b64-encoder
+  (.withoutPadding (Base64/getUrlEncoder)))
+
+(defn- get-binary-data [body]
+  (with-open [_ (prom/timer parse-duration-seconds "binary")
+              ^InputStream input body]
+    (.encodeToString b64-encoder input)))
+
+(defn- resource-request-binary-data [{:keys [body headers] :as request}]
+  (if body
+    (when-ok [b64-encoded-data (get-binary-data body)]
+      (let [content-type (get headers "content-type")]
+        (assoc request :body
+               {:fhir/type :fhir/Binary
+                :resourceType "Binary"
+                :contentType (type/code content-type)
+                :data (type/base64Binary b64-encoded-data)})))
+    (ba/incorrect "Missing HTTP body.")))
+
 (defn- unsupported-media-type-msg [media-type]
   (format "Unsupported media type `%s` expect one of `application/fhir+json` or `application/fhir+xml`."
           media-type))
@@ -109,19 +132,43 @@
                     :http/status 415))
     (if (str/blank? (slurp (:body request)))
       (assoc request :body nil)
-      (ba/incorrect "Content-Type header expected, but is missing."))))
+      (ba/incorrect "Expected Content-Type header for FHIR resources."))))
+
+(defn- binary-resource-request [request]
+  (if-let [content-type (request/content-type request)]
+    (cond
+      (json-request? content-type) (resource-request-json request)
+      (xml-request? content-type) (resource-request-xml request)
+      :else
+      (resource-request-binary-data request))
+    (ba/incorrect "Expected Content-Type header for binary resources.")))
 
 (defn wrap-resource
   "Middleware to parse a resource from the body according the content-type
   header.
 
-  Updates the :body key in the request map on successful parsing and conforming
-  the resource to the internal format.
+  If the resource is successfully parsed and conformed to the internal format,
+  updates the :body key in the request map.
 
-  Returns an OperationOutcome in the internal format, skipping the handler, with
-  an appropriate error on parsing and conforming errors."
+  In case on errors, returns an OperationOutcome in the internal format with the
+  appropriate error and skips the handler."
   [handler]
   (fn [request]
     (if-ok [request (resource-request request)]
       (handler request)
       ac/completed-future)))
+
+(defn wrap-binary-data
+  "Middleware to parse binary data from the body according the content-type
+  header.
+
+  If the resource is successfully parsed and conformed to the internal format,
+  updates the :body key in the request map.
+
+  In case on errors, returns an OperationOutcome in the internal format with the
+  appropriate error and skips the handler."
+  [handler]
+  (fn [request]
+    (if-ok [request (binary-resource-request request)]
+      (handler request)
+      ac/completed-future)))

modules/rest-util/test/blaze/middleware/fhir/resource_test.clj

@@ -2,9 +2,10 @@
   (:require
    [blaze.async.comp :as ac]
    [blaze.fhir.spec :as fhir-spec]
+   [blaze.fhir.spec.type :as type]
    [blaze.fhir.test-util]
    [blaze.handler.util :as handler-util]
-   [blaze.middleware.fhir.resource :refer [wrap-resource]]
+   [blaze.middleware.fhir.resource :refer [wrap-binary-data wrap-resource]]
    [blaze.test-util :as tu :refer [satisfies-prop]]
    [clojure.spec.test.alpha :as st]
    [clojure.string :as str]
@@ -23,18 +24,24 @@
 
 (test/use-fixtures :each tu/fixture)
 
-(defn wrap-error [handler]
+(defn- wrap-error [handler]
   (fn [request]
     (-> (handler request)
         (ac/exceptionally handler-util/error-response))))
 
-(def resource-handler
-  "A handler which just returns the :body from the request."
+(def ^:private resource-handler
+  "A handler which just returns the `:body` from a non-binary resource request."
   (-> (comp ac/completed-future :body)
       wrap-resource
       wrap-error))
 
-(defn input-stream
+(def ^:private binary-resource-handler
+  "A handler which just returns the `:body` from a binary resource request."
+  (-> (comp ac/completed-future :body)
+      wrap-binary-data
+      wrap-error))
+
+(defn- input-stream
   ([^String s]
    (ByteArrayInputStream. (.getBytes s StandardCharsets/UTF_8)))
   ([^String s closed?]
@@ -193,6 +200,42 @@
               :body (input-stream (str "<Binary xmlns=\"http://hl7.org/fhir\"><data value=\"" (apply str (repeat (* 8 1024 1024) \a)) "\"/></Binary>"))})
       fhir-spec/fhir-type := :fhir/Binary)))
 
+(deftest binary-test
+  (testing "when sending FHIR-wrapped Binary data"
+    (testing "both handlers should return the same for both wrappers (JSON and XML)"
+      (let [b64-encoded-binary-data "MTA1NjE0Cg==" ;; raw data: 105614
+            binary-resource-data (type/base64Binary b64-encoded-binary-data)
+            binary-resource-content-type (type/code "text/plain")]
+        (doseq [handler [resource-handler binary-resource-handler]
+                [fhir-content-type resource-string-representation]
+                [["application/fhir+json;charset=utf-8" (str "{\"data\" : \"" b64-encoded-binary-data "\", \"resourceType\" : \"Binary\", \"contentType\" : \"" binary-resource-content-type "\"}")]
+                 ["application/fhir+xml;charset=utf-8" (str "<Binary xmlns=\"http://hl7.org/fhir\"><data value=\"" b64-encoded-binary-data "\"/><contentType value=\"" binary-resource-content-type "\"/></Binary>")]]]
+          (let [closed? (atom false)]
+            (given @(handler
+                     {:headers {"content-type" fhir-content-type}
+                      :body (input-stream resource-string-representation closed?)})
+              {:fhir/type :fhir/Binary
+               :contentType binary-resource-content-type
+               :data binary-resource-data})
+            (is closed?))))))
+
+  (testing "when sending raw binary resource handling"
+    (testing "small raw binary data"
+      (let [closed? (atom false)
+            raw-binary-data "test-data"]
+        (given @(binary-resource-handler
+                 {:headers {"content-type" "application/octet-stream"}
+                  :body (input-stream raw-binary-data closed?)})
+          raw-binary-data)
+        (is @closed?)))
+
+    (testing "large raw binary resource"
+      (let [large-data (apply str (repeat (* 8 1024 1024) \a))]  ; 8MB of data
+        (given @(binary-resource-handler
+                 {:headers {"content-type" "application/octet-stream"}
+                  :body (input-stream large-data)})
+          large-data)))))
+
 (def ^:private whitespace
   (gen/fmap str/join (gen/vector (gen/elements [" " "\n" "\r" "\t"]))))