OAProc: ensure binary outputs are not UTF-8 decoded (#2304)

Author: tomkralidis

pygeoapi/process/hello_world.py

@@ -3,7 +3,7 @@
 # Authors: Tom Kralidis <tomkralidis@gmail.com>
 #          Francesco Martinelli <francesco.martinelli@ingv.it>
 #
-# Copyright (c) 2022 Tom Kralidis
+# Copyright (c) 2026 Tom Kralidis
 # Copyright (c) 2024 Francesco Martinelli
 #
 # Permission is hereby granted, free of charge, to any person
@@ -29,6 +29,7 @@
 #
 # =================================================================
 
+import json
 import logging
 
 from pygeoapi.process.base import BaseProcessor, ProcessorExecuteError
@@ -82,6 +83,17 @@
             'minOccurs': 0,
             'maxOccurs': 1,
             'keywords': ['message']
+        },
+        'as_bytes': {
+            'title': 'As bytes',
+            'description': 'Whether to force return as bytes',
+            'schema': {
+                'type': 'bool',
+                'default': False
+            },
+            'minOccurs': 0,
+            'maxOccurs': 1,
+            'keywords': ['as_bytes']
         }
     },
     'outputs': {
@@ -136,6 +148,9 @@ def execute(self, data, outputs=None):
                 'value': value
             }
 
+        if data.get('as_bytes', False):
+            json.dumps(produced_outputs).encode('utf-8')
+
         return mimetype, produced_outputs
 
     def __repr__(self):

pygeoapi/process/manager/base.py

@@ -277,7 +277,8 @@ def _execute_handler_sync(self, p: BaseProcessor, job_id: str,
             current_status = JobStatus.running
             jfmt, outputs = p.execute(data_dict, **extra_execute_parameters)
 
-            if isinstance(outputs, bytes):
+            if isinstance(outputs, bytes) and outputs.isascii():
+                LOGGER.debug('output is ASCII; decoding utf-8')
                 outputs = outputs.decode('utf-8')
 
             if requested_response == RequestedResponse.document.value:

tests/api/test_processes.py

@@ -95,7 +95,7 @@ def test_describe_processes(config, api_):
     assert process['title'] == 'Hello World'
     assert len(process['keywords']) == 3
     assert len(process['links']) == 6
-    assert len(process['inputs']) == 2
+    assert len(process['inputs']) == 3
     assert len(process['outputs']) == 1
     assert len(process['outputTransmission']) == 1
     assert len(process['jobControlOptions']) == 2
@@ -242,6 +242,12 @@ def test_execute_process(config, api_):
             'name': 'Test document'
         }
     }
+    req_body_10 = {
+        'inputs': {
+            'name': 'Test document as bytes response',
+            'as_bytes': True
+        }
+    }
 
     cleanup_jobs = set()
 
@@ -410,6 +416,12 @@ def test_execute_process(config, api_):
     response2 = '{"id":"echo","value":"Hello Test document!"}'
     assert response == response2
 
+    req = mock_api_request(data=req_body_10)
+    rsp_headers, code, response = execute_process(api_, req, 'hello-world')
+
+    response2 = '{"id":"echo","value":"Hello Test document as bytes response!"}'  # noqa
+    assert response == response2
+
     # Cleanup
     time.sleep(2)  # Allow time for any outstanding async jobs
     for _, job_id in cleanup_jobs: