Issue #571/#595 make JobDatabaseInterface an abstract base class

openeo/extra/job_management.py

@@ -1,3 +1,4 @@
+import abc
 import contextlib
 import datetime
 import json
@@ -32,19 +33,27 @@ class _Backend(NamedTuple):
 MAX_RETRIES = 5
 
 
-class JobDatabaseInterface:
+class JobDatabaseInterface(metaclass=abc.ABCMeta):
+    """
+    Interface for a database of job metadata to use with the :py:class:`MultiBackendJobManager`,
+    allowing to regularly persist the job metadata while polling the job statuses
+    and resume/restart the job tracking after it was interrupted.
+    """
 
+    @abc.abstractmethod
     def exists(self) -> bool:
-        """Is there existing data in the database?"""
-        return False
+        """Does the job database already exist, to read job data from?"""
+        ...
 
+    @abc.abstractmethod
     def read(self) -> pd.DataFrame:
-        """Read job data from the database ad pandas DataFrame."""
-        raise NotImplementedError()
+        """Read job data from the database as pandas DataFrame."""
+        ...
 
+    @abc.abstractmethod
     def persist(self, df: pd.DataFrame):
         """Store job data to the database."""
-        raise NotImplementedError()
+        ...
 
 
 class MultiBackendJobManager:
@@ -512,7 +521,6 @@ def ignore_connection_errors(context: Optional[str] = None):
         time.sleep(5)
 
 
-
 class CsvJobDatabase(JobDatabaseInterface):
     def __init__(self, path: Union[str, Path]):
         self.path = Path(path)

tests/extra/test_job_management.py

@@ -481,26 +481,34 @@ def test_read_non_wkt(self, tmp_path):
         df = CsvJobDatabase(path).read()
         assert isinstance(df.geometry[0], str)
 
-    def test_persist(self, tmp_path):
-        df = pd.DataFrame(
+    def test_persist_and_read(self, tmp_path):
+        orig = pd.DataFrame(
             {
-                "some_number": [3, 2, 1],
+                "numbers": [3, 2, 1],
+                "names": ["apple", "banana", "coconut"],
             }
         )
-
         path = tmp_path / "jobs.csv"
-        CsvJobDatabase(path).persist(df)
-        assert CsvJobDatabase(path).read().equals(df)
+        CsvJobDatabase(path).persist(orig)
+        assert path.exists()
+
+        loaded = CsvJobDatabase(path).read()
+        assert list(loaded.dtypes) == list(orig.dtypes)
+        assert loaded.equals(orig)
 
 
 class TestParquetJobDatabase:
-    def test_read_persist(self, tmp_path):
-        df = pd.DataFrame(
+    def test_persist_and_read(self, tmp_path):
+        orig = pd.DataFrame(
             {
-                "some_number": [3, 2, 1],
+                "numbers": [3, 2, 1],
+                "names": ["apple", "banana", "coconut"],
             }
         )
-
         path = tmp_path / "jobs.parquet"
-        ParquetJobDatabase(path).persist(df)
-        assert ParquetJobDatabase(path).read().equals(df)
+        ParquetJobDatabase(path).persist(orig)
+        assert path.exists()
+
+        loaded = ParquetJobDatabase(path).read()
+        assert list(loaded.dtypes) == list(orig.dtypes)
+        assert loaded.equals(orig)