Implement Binary Upload (Fixes #2126)

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

@@ -0,0 +1,54 @@
+#!/bin/bash -e
+
+# This script creates a large binary resource (8 MiB) and verifies that its binary content
+# * can be read correctly via direct binary upload, and that
+# * it's properly base64-encoded.
+
+BASE="http://localhost:8080/fhir"
+
+# Create temporary files for original and downloaded data
+TEMP_ORIGINAL=$(mktemp)
+TEMP_DOWNLOAD=$(mktemp)
+TEMP_JSON=$(mktemp)
+
+# Ensure cleanup of temporary files
+trap 'rm -f "$TEMP_ORIGINAL" "$TEMP_DOWNLOAD" "$TEMP_JSON"' EXIT
+
+echo "Testing direct binary upload and download..."
+
+# Generate 8 MiB of random binary data
+dd if=/dev/urandom bs=8388608 count=1 2>/dev/null > "$TEMP_ORIGINAL"
+
+# Create Binary resource via direct binary upload
+ID=$(curl -s \
+     -H 'Content-Type: application/octet-stream' \
+     --data-binary "@$TEMP_ORIGINAL" \
+     "$BASE/Binary" | \
+     jq -r '.id')
+
+echo "Created Binary resource via direct binary upload with ID: $ID"
+
+# Download as JSON format to verify base64 encoding
+curl -s \
+     -H 'Accept: application/fhir+json' \
+     "$BASE/Binary/$ID" > "$TEMP_JSON"
+
+# Extract the base64 content and decode it
+jq -r '.data' "$TEMP_JSON" | base64 -d > "$TEMP_DOWNLOAD"
+
+# Compare files directly
+if [ -n "$ID" ] && cmp -s "$TEMP_ORIGINAL" "$TEMP_DOWNLOAD"; then
+    echo "✅ Direct Binary: Successfully verified 8 MiB binary content integrity and base64 encoding"
+else
+    echo "🆘 Direct Binary: Content verification failed"
+    echo "Server response (JSON):"
+    cat "$TEMP_JSON"
+    echo "Original size: $(wc -c < "$TEMP_ORIGINAL") bytes"
+    echo "Downloaded size: $(wc -c < "$TEMP_DOWNLOAD") bytes"
+    # Show first few bytes of both files in hex for debugging
+    echo "First 32 bytes of original:"
+    hexdump -C "$TEMP_ORIGINAL" | head -n 2
+    echo "First 32 bytes of downloaded:"
+    hexdump -C "$TEMP_DOWNLOAD" | head -n 2
+    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 +99,23 @@
       (assoc request :body resource))
     (ba/incorrect "Missing HTTP body.")))
 
+(defn- get-binary-data [body]
+  (with-open [_ (prom/timer parse-duration-seconds "binary")]
+    (.encodeToString ^Base64$Encoder (Base64/getEncoder) (.readAllBytes ^InputStream body))))
+
+(defn- resource-request-binary-data [{:keys [body headers] :as request}]
+  (if body
+    ;; `when-ok` is not needed here because:
+    ;;  * Binary data is not parsed nor validated, and
+    ;;  * Base64-encoding should not fail under normal circumstances.
+    (let [b64-encoded-data (get-binary-data body)
+          content-type (get headers "content-type")]
+      (assoc request :body
+             {:fhir/type :fhir/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 +130,43 @@
                     :http/status 415))
     (if (str/blank? (slurp (:body request)))
       (assoc request :body nil)
-      (ba/incorrect "Content-Type header expected, but is missing."))))
+      (ba/incorrect "Missing Content-Type header for FHIR resources."))))
+
+(defn- binary-resource-request [request]
+  (if-let [content-type (request/content-type request)]
+    (cond
+      (str/starts-with? content-type "application/fhir+json") (resource-request-json request)
+      (str/starts-with? content-type "application/fhir+xml") (resource-request-xml request)
+      :else
+      (resource-request-binary-data request))
+    (ba/incorrect "Missing 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)))