Fix bug when trying to get job result in binary format (geopython#1798)

Author: MartinPontius

pygeoapi/process/manager/mongodb_.py

@@ -32,6 +32,7 @@
 
 from pymongo import MongoClient
 
+from pygeoapi.api import FORMAT_TYPES, F_JSON, F_JSONLD
 from pygeoapi.process.base import (
     JobNotFoundError,
     JobResultNotFoundError,
@@ -151,8 +152,16 @@ def get_job_result(self, job_id):
             if entry["status"] != "successful":
                 LOGGER.info("JOBMANAGER - job not finished or failed")
                 return (None,)
-            with open(entry["location"], "r") as file:
-                data = json.load(file)
+            if not entry["location"]:
+                LOGGER.warning(f"job {job_id!r} -  unknown result location")
+                raise JobResultNotFoundError()
+            if entry["mimetype"] in (None, FORMAT_TYPES[F_JSON],
+                                     FORMAT_TYPES[F_JSONLD]):
+                with open(entry["location"], "r") as file:
+                    data = json.load(file)
+            else:
+                with open(entry["location"], "rb") as file:
+                    data = file.read()
             LOGGER.info("JOBMANAGER - MongoDB job result queried")
             return entry["mimetype"], data
         except Exception as err:

pygeoapi/process/manager/postgresql.py

@@ -49,6 +49,7 @@
 from sqlalchemy.engine import make_url
 from sqlalchemy.orm import Session
 
+from pygeoapi.api import FORMAT_TYPES, F_JSON, F_JSONLD
 from pygeoapi.process.base import (
     JobNotFoundError,
     JobResultNotFoundError,
@@ -292,8 +293,13 @@ def get_job_result(self, job_id: str) -> Tuple[str, Any]:
         else:
             try:
                 location = Path(location)
-                with location.open(encoding='utf-8') as fh:
-                    result = json.load(fh)
+                if mimetype in (None, FORMAT_TYPES[F_JSON],
+                                FORMAT_TYPES[F_JSONLD]):
+                    with location.open('r', encoding='utf-8') as fh:
+                        result = json.load(fh)
+                else:
+                    with location.open('rb') as fh:
+                        result = fh.read()
             except (TypeError, FileNotFoundError, json.JSONDecodeError):
                 raise JobResultNotFoundError()
             else:

pygeoapi/process/manager/tinydb_.py

@@ -37,6 +37,7 @@
 import tinydb
 from filelock import FileLock
 
+from pygeoapi.api import FORMAT_TYPES, F_JSON, F_JSONLD
 from pygeoapi.process.base import (
     JobNotFoundError,
     JobResultNotFoundError,
@@ -211,8 +212,13 @@ def get_job_result(self, job_id: str) -> Tuple[str, Any]:
         else:
             try:
                 location = Path(location)
-                with location.open('r', encoding='utf-8') as filehandler:
-                    result = json.load(filehandler)
+                if mimetype in (None, FORMAT_TYPES[F_JSON],
+                                FORMAT_TYPES[F_JSONLD]):
+                    with location.open('r', encoding='utf-8') as filehandler:
+                        result = json.load(filehandler)
+                else:
+                    with location.open('rb') as filehandler:
+                        result = filehandler.read()
             except (TypeError, FileNotFoundError, json.JSONDecodeError):
                 raise JobResultNotFoundError()
             else:

tests/test_manager.py

@@ -30,9 +30,11 @@
 
 import pytest
 
-from pygeoapi.process.base import UnknownProcessError
+from pygeoapi.process.base import UnknownProcessError, JobNotFoundError
 from pygeoapi.process.manager.base import get_manager
 
+from .util import get_test_file_path
+
 
 @pytest.fixture()
 def config() -> Dict:
@@ -41,6 +43,7 @@ def config() -> Dict:
             'manager': {
                 'name': 'TinyDB',
                 'output_dir': '/tmp',
+                'connection': '/tmp/pygeoapi-process-manager-test.db'
             }
         },
         'resources': {
@@ -71,3 +74,28 @@ def test_get_processor_raises_exception(config):
     manager = get_manager(config)
     with pytest.raises(expected_exception=UnknownProcessError):
         manager.get_processor('foo')
+
+
+def test_get_job_result_binary(config):
+    manager = get_manager(config)
+    nc_file = get_test_file_path("tests/data/coads_sst.nc")
+    job_id = "15eeae38-608c-11ef-81c8-0242ac130002"
+    job_metadata = {
+        "type": "process",
+        "identifier": job_id,
+        "process_id": "dummy",
+        "job_start_datetime": "2024-08-22T12:00:00.000000Z",
+        "job_end_datetime": "2024-08-22T12:00:01.000000Z",
+        "status": "successful",
+        "location": nc_file,
+        "mimetype": "application/x-netcdf",
+        "message": "Job complete",
+        "progress": 100
+    }
+    try:
+        manager.get_job(job_id)
+    except JobNotFoundError:
+        manager.add_job(job_metadata)
+    mimetype, result = manager.get_job_result(job_id)
+    assert mimetype == "application/x-netcdf"
+    assert isinstance(result, bytes)