From 61c49255301cf9ad5ca997836211db4a776d36a2 Mon Sep 17 00:00:00 2001
From: Leandro Doctors <leandro.doctors@dkfz-heidelberg.de>
Date: Sat, 14 Dec 2024 13:18:04 +0100
Subject: [PATCH] Fix Failing Binary Emit case

The body should contain a JSON payload with the OperationOutcome, not
the OperationOutcome itself.
We use JSON because it is our default.
---
 .../read-binary-content-via-json-invalid.sh   | 90 +++++++++++++++++++
 .github/workflows/build.yml                   |  3 +
 .../src/blaze/middleware/fhir/output.clj      |  2 +-
 .../blaze/middleware/fhir/output_test.clj     |  6 +-
 4 files changed, 97 insertions(+), 4 deletions(-)
 create mode 100755 .github/scripts/read-binary-content-via-json-invalid.sh

diff --git a/.github/scripts/read-binary-content-via-json-invalid.sh b/.github/scripts/read-binary-content-via-json-invalid.sh
new file mode 100755
index 000000000..71a86db50
--- /dev/null
+++ b/.github/scripts/read-binary-content-via-json-invalid.sh
@@ -0,0 +1,90 @@
+#!/bin/bash -e
+
+# This script creates a binary resource with invalid data and verifies that,
+# when trying to read its binary content, an error is generated.
+
+BASE="http://localhost:8080/fhir"
+
+# 10 KiB of random data, base64 encoded
+ORIGINAL_DATA="$(openssl rand -base64 10240 | tr -d '\n')"
+
+
+# Higher-order function to apply a transformation based on a condition
+apply_generic_transformation() {
+    local data="$1"
+    local transform_function="$2"
+
+    local result=""
+
+    # Iterate over the data and apply the transform function
+    for i in $(seq 0 ${#data}-1); do
+        result+=$(eval "$transform_function \"${data:$i:1}\" $i")
+    done
+
+    echo "$result"
+}
+
+remove_parts_transform() {
+    local char="$1"
+    local index="$2"
+
+    # Randomly decide whether to remove this character (skip it)
+    if (( RANDOM % 10 == 0 )); then
+        echo ""  # Skip (remove this character)
+    else
+        echo "$char"  # Keep the character
+    fi
+}
+
+insert_non_base64_transform() {
+    local char="$1"
+    local index="$2"
+
+    # Every 5th character, insert a random non-base64 character
+    if (( index % 5 == 0 )); then
+        echo "$((RANDOM % 10))"  # Insert a non-base64 digit
+    else
+        echo "$char"  # Keep the character
+    fi
+}
+
+# Apply Remove Parts of the Data
+FAULTY_DATA=$(apply_generic_transformation "$ORIGINAL_DATA" "remove_parts_transform")
+# echo "Faulty Data (Removed Parts): $FAULTY_DATA"
+
+# Apply Insert Non-Base64 Characters
+FAULTY_DATA=$(apply_generic_transformation "$FAULTY_DATA" "insert_non_base64_transform")
+# echo "Faulty Data (also with Non-Base64 Characters): $FAULTY_DATA"
+
+DATA=$FAULTY_DATA
+
+
+binary() {
+cat <<END
+{
+  "resourceType": "Binary",
+  "contentType": "application/pdf",
+  "data": "$DATA"
+}
+END
+}
+
+# Create a Binary resource that contains that faulty data, and get its ID (via JSON)
+ID_VIA_JSON=$(curl -s -H 'Content-Type: application/fhir+json' -d "$(binary)" "$BASE/Binary" | jq -r '.id')
+
+echo "Created Binary resource that contains the Random Data"
+echo "  - via JSON, with ID: $ID_VIA_JSON"
+
+
+# Retrieve the Binary resource, and Base64 encode it so it can be safely handled by Bash (JSON)
+BASE64_ENCODED_BINARY_RESOURCE_VIA_JSON=$(curl -s -H 'Accept: application/pdf' "$BASE/Binary/$ID_VIA_JSON" | base64 | tr -d '\n')
+
+
+echo "Binary data retrieved. Verifying content... (JSON version)"
+
+if [ "$DATA" = "$BASE64_ENCODED_BINARY_RESOURCE_VIA_JSON" ]; then
+    echo "✅ Base64 encoding of both the Original Data and the Retrieved Resource Data match (JSON)"
+else
+    echo "🆘 Base64 encoding of both the Original Data and the Retrieved Resource Data are different (JSON)"
+    exit 1
+fi
diff --git a/.github/workflows/build.yml b/.github/workflows/build.yml
index fbf2061c9..3fd8c47e3 100644
--- a/.github/workflows/build.yml
+++ b/.github/workflows/build.yml
@@ -1069,6 +1069,9 @@ jobs:
     - name: Binary Content Download - not found
       run: .github/scripts/read-binary-content-not-found.sh
 
+    - name: Binary Content Download - invalid (via JSON)
+      run: .github/scripts/read-binary-content-via-json-invalid.sh
+
     - name: Binary Content Download - found (via JSON)
       run: .github/scripts/read-binary-content-via-json-found.sh
 
diff --git a/modules/rest-util/src/blaze/middleware/fhir/output.clj b/modules/rest-util/src/blaze/middleware/fhir/output.clj
index 08fa16fbf..6ebd34e08 100644
--- a/modules/rest-util/src/blaze/middleware/fhir/output.clj
+++ b/modules/rest-util/src/blaze/middleware/fhir/output.clj
@@ -68,7 +68,7 @@
     (update response :body generate-binary**)
     (catch Throwable e
       (assoc response
-             :body (generate-error-payload identity e)
+             :body (generate-error-payload generate-json e)
              :status 500))))
 
 (defn- generate-binary [response]
diff --git a/modules/rest-util/test/blaze/middleware/fhir/output_test.clj b/modules/rest-util/test/blaze/middleware/fhir/output_test.clj
index 1d8fdcb8e..e3b176183 100644
--- a/modules/rest-util/test/blaze/middleware/fhir/output_test.clj
+++ b/modules/rest-util/test/blaze/middleware/fhir/output_test.clj
@@ -214,12 +214,12 @@
           [:headers "Content-Type"] := nil
           :body := nil)))
 
-    (testing "failing binary emit (with invalid data)"
+    (testing "failing binary emit (with invalid data, using the JSON default)"
       (given (call (binary-resource-handler-200 {:content-type "application/pdf" :data "MTANjECg=="}) {:headers {"accept" "text/plain"}})
         :status := 500
         [:headers "Content-Type"] := "application/pdf"
-        [:body :fhir/type] := :fhir/OperationOutcome
-        [:body :issue 0 :diagnostics] := "Input byte array has wrong 4-byte ending unit"))))
+        [:body parse-json :fhir/type] := :fhir/OperationOutcome
+        [:body parse-json :issue 0 :diagnostics] := "Input byte array has wrong 4-byte ending unit"))))
 
 (deftest not-acceptable-test
   (is (nil? (call resource-handler-200-with-patient {:headers {"accept" "text/plain"}}))))