Change content type back to application/json for H2 files

app/controllers/resources_controller.rb

@@ -150,7 +150,15 @@ def get_fileset_uuid(external_id)
   def metadata_for_blob(blob, file)
     file.delete(:externalIdentifier)
     file[:size] = blob.byte_size
-    file[:hasMimeType] = blob.content_type || 'application/octet-stream'
+    # Invalid JSON files uploaded for deposit with a JSON content type will trigger 400 errors in sdr-api since they are
+    # parsed as JSON and rejected.  The work around is to change the content_type in the request for uploads like this
+    # to something specific that will be changed back to application/json after upload is complete.
+    # There is a corresponding translation in sdr-client.  See https://github.com/sul-dlss/happy-heron/issues/3075
+    file[:hasMimeType] = if blob.content_type == 'application/x-stanford-json'
+                           'application/json'
+                         else
+                           blob.content_type || 'application/octet-stream'
+                         end
     declared_md5 = file[:hasMessageDigests].find { |digest| digest.fetch(:type) == 'md5' }.fetch(:digest)
     calculated_md5 = base64_to_hexdigest(blob.checksum)
     raise BlobError, "MD5 mismatch for #{file[:filename]}" if declared_md5 != calculated_md5

openapi.yml

@@ -40,7 +40,12 @@ paths:
     put:
       tags:
         - files
-      summary: Upload the binary file resource
+      summary: |
+       Upload the binary file resource
+       Note that the 'content_type' of the file posted will be set to 'application/octet-stream' if not supplied.
+       Additional values in 'content_type' after a semicolon will be removed (e.g. ';v15' is removed in 'application/text;v15')
+       You may post JSON files with 'content_type' of 'application/x-stanford-json' to avoid 400 errors if posted JSON file is invalid.
+       The system will automatically change the content type to 'application/json' after upload.
       responses:
         '200':
           description: OK

spec/requests/create_dro_spec.rb

@@ -49,18 +49,19 @@
   let(:checksum) { 'f5nXiniiM+u/gexbNkOA/A==' }
 
   let(:blob) do
-    ActiveStorage::Blob.create!(key: 'tozuehlw6e8du20vn1xfzmiifyok',
+    ActiveStorage::Blob.create!(key: 'tozuehlw6e8du20vn1xfzmiifyok', content_type: 'application/text',
                                 filename: 'file2.txt', byte_size: 10, checksum:)
   end
   let(:signed_id) do
     ActiveStorage.verifier.generate(blob.id, purpose: :blob_id)
   end
 
+  let(:expected_content_type) { 'application/text' }
   let(:expected_model_params) do
     model_params = dro.to_h
     file_params = model_params.dig(:structural, :contains, 0, :structural, :contains, 0)
     file_params.delete(:externalIdentifier)
-    file_params[:hasMimeType] = 'application/octet-stream'
+    file_params[:hasMimeType] = expected_content_type
     file_params[:size] = 10
     model_params.with_indifferent_access
   end
@@ -218,6 +219,56 @@
     end
   end
 
+  context 'when the file is application/x-stanford-json' do
+    let(:blob) do
+      ActiveStorage::Blob.create!(key: 'tozuehlw6e8du20vn1xfzmiifyok',
+                                  filename: 'file2.txt', content_type: 'application/x-stanford-json',
+                                  byte_size: 10, checksum:)
+    end
+    let(:expected_content_type) { 'application/json' }
+
+    it 'switches the content type to application/json' do
+      post '/v1/resources',
+           params: request,
+           headers: { 'Content-Type' => 'application/json', 'Authorization' => "Bearer #{jwt}" }
+      expect(response).to be_created
+      expect(response.location).to be_present
+      expect(JSON.parse(response.body)['jobId']).to be_present
+      expect(IngestJob).to have_received(:perform_later).with(model_params: expected_model_params,
+                                                              background_job_result: instance_of(BackgroundJobResult),
+                                                              signed_ids: { 'file2.txt' => signed_id },
+                                                              globus_ids: {},
+                                                              start_workflow: false,
+                                                              assign_doi: false,
+                                                              priority: 'default')
+    end
+  end
+
+  context 'when the file supplies a nil content_type' do
+    let(:blob) do
+      ActiveStorage::Blob.create!(key: 'tozuehlw6e8du20vn1xfzmiifyok',
+                                  filename: 'file2.txt', content_type: nil,
+                                  byte_size: 10, checksum:)
+    end
+    let(:expected_content_type) { 'application/octet-stream' }
+
+    it 'switches the content type to application/octet-stream' do
+      post '/v1/resources',
+           params: request,
+           headers: { 'Content-Type' => 'application/json', 'Authorization' => "Bearer #{jwt}" }
+      expect(response).to be_created
+      expect(response.location).to be_present
+      expect(JSON.parse(response.body)['jobId']).to be_present
+      expect(IngestJob).to have_received(:perform_later).with(model_params: expected_model_params,
+                                                              background_job_result: instance_of(BackgroundJobResult),
+                                                              signed_ids: { 'file2.txt' => signed_id },
+                                                              globus_ids: {},
+                                                              start_workflow: false,
+                                                              assign_doi: false,
+                                                              priority: 'default')
+    end
+  end
+
   context 'when limited user is authorized for the collection' do
     let(:limited_user) { create(:user, collections: ['druid:fg123hj4567'], full_access: false) }
 

spec/requests/direct_uploads_spec.rb

@@ -3,10 +3,12 @@
 require 'rails_helper'
 
 RSpec.describe 'Direct upload' do
-  let(:json) do
-    '{"blob":{"filename":"Gemfile.lock","byte_size":1751,"checksum":"vQ0xN+GwJBg9iEAcD4v73g==",' \
-      '"content_type":"text/html"}}'
-  end
+  let(:filename) { 'test.txt' }
+  let(:data) { 'text' }
+  let(:checksum) { OpenSSL::Digest::MD5.base64digest(data) }
+  let(:byte_size) { data.length }
+  let(:content_type) { 'text/plain' }
+  let(:json) { JSON.dump({ blob: { filename:, byte_size:, checksum:, content_type: } }) }
 
   context 'when unauthorized' do
     it 'returns 401' do
@@ -25,4 +27,80 @@
       expect(response).to be_successful
     end
   end
+
+  context 'when uploading a file' do
+    it 'returns 204' do
+      post '/v1/direct_uploads',
+           params: json,
+           headers: { 'Content-Type' => 'application/json', 'Authorization' => "Bearer #{jwt}" }
+
+      expect(response).to be_successful
+      expect(response.media_type).to eq 'application/json'
+      direct_upload = response.parsed_body
+
+      signed_id = direct_upload['signed_id']
+      expect(signed_id).to be_truthy
+
+      direct_upload_uri = URI.parse(direct_upload['direct_upload']['url'])
+      expect(direct_upload_uri.path).to start_with('/v1/disk/')
+
+      put direct_upload_uri.path,
+          params: data,
+          headers: { 'Content-Type' => content_type, 'Authorization' => "Bearer #{jwt}" }
+
+      expect(response).to have_http_status(:no_content) # Status: 204
+    end
+  end
+
+  context 'when uploading an invalid json file with application/json' do
+    let(:content_type) { 'application/json' }
+
+    it 'returns 400' do
+      post '/v1/direct_uploads',
+           params: json,
+           headers: { 'Content-Type' => 'application/json', 'Authorization' => "Bearer #{jwt}" }
+
+      expect(response).to be_successful
+      expect(response.media_type).to eq 'application/json'
+      direct_upload = response.parsed_body
+
+      signed_id = direct_upload['signed_id']
+      expect(signed_id).to be_truthy
+
+      direct_upload_uri = URI.parse(direct_upload['direct_upload']['url'])
+      expect(direct_upload_uri.path).to start_with('/v1/disk/')
+
+      put direct_upload_uri.path,
+          params: data,
+          headers: { 'Content-Type' => content_type, 'Authorization' => "Bearer #{jwt}" }
+
+      expect(response).to have_http_status(:bad_request) # Status: 400
+    end
+  end
+
+  context 'when uploading an invalid json file with application/x-stanford-json' do
+    let(:content_type) { 'application/x-stanford-json' }
+
+    it 'returns 204' do
+      post '/v1/direct_uploads',
+           params: json,
+           headers: { 'Content-Type' => 'application/json', 'Authorization' => "Bearer #{jwt}" }
+
+      expect(response).to be_successful
+      expect(response.media_type).to eq 'application/json'
+      direct_upload = response.parsed_body
+
+      signed_id = direct_upload['signed_id']
+      expect(signed_id).to be_truthy
+
+      direct_upload_uri = URI.parse(direct_upload['direct_upload']['url'])
+      expect(direct_upload_uri.path).to start_with('/v1/disk/')
+
+      put direct_upload_uri.path,
+          params: data,
+          headers: { 'Content-Type' => content_type, 'Authorization' => "Bearer #{jwt}" }
+
+      expect(response).to have_http_status(:no_content) # Status: 204
+    end
+  end
 end