fix(cdk): support start_date format without microseconds in file-based connectors

Co-Authored-By: Daryna Ishchenko <darina.ishchenko17@gmail.com>

Author: devin-ai-integration[bot]

airbyte_cdk/sources/file_based/file_based_stream_reader.py

@@ -98,14 +98,22 @@ def get_matching_files(
         """
         ...
 
+    @staticmethod
+    def _parse_start_date(start_date_str: str) -> datetime:
+        """Parse a start_date string, supporting both with and without microseconds."""
+        try:
+            return datetime.strptime(start_date_str, AbstractFileBasedStreamReader.DATE_TIME_FORMAT)
+        except ValueError:
+            return datetime.strptime(start_date_str, "%Y-%m-%dT%H:%M:%SZ")
+
     def filter_files_by_globs_and_start_date(
         self, files: List[RemoteFile], globs: List[str]
     ) -> Iterable[RemoteFile]:
         """
         Utility method for filtering files based on globs.
         """
         start_date = (
-            datetime.strptime(self.config.start_date, self.DATE_TIME_FORMAT)
+            self._parse_start_date(self.config.start_date)
             if self.config and self.config.start_date
             else None
         )

unit_tests/sources/file_based/test_file_based_stream_reader.py

@@ -401,6 +401,13 @@ def documentation_url(cls) -> AnyUrl:
             set(),
             id="all_csvs_modified_before_start_date",
         ),
+        pytest.param(
+            ["**/*.csv"],
+            {"start_date": "2023-06-01T03:54:07Z", "streams": []},
+            {"a.csv", "a/b.csv", "a/c.csv", "a/b/c.csv", "a/c/c.csv", "a/b/c/d.csv"},
+            set(),
+            id="all_csvs_start_date_without_microseconds",
+        ),
         pytest.param(
             ["**/*.csv"],
             {"start_date": "2023-06-05T03:54:07.000Z", "streams": []},
@@ -494,6 +501,42 @@ def test_preserve_sub_directories_scenarios(
     assert file_paths[AbstractFileBasedStreamReader.FILE_FOLDER] == path.dirname(source_file_path)
 
 
+@pytest.mark.parametrize(
+    "start_date_str, expected",
+    [
+        pytest.param(
+            "2025-01-01T00:00:00.000000Z",
+            datetime(2025, 1, 1, 0, 0, 0),
+            id="with_microseconds_zero",
+        ),
+        pytest.param(
+            "2025-06-15T12:30:45.123456Z",
+            datetime(2025, 6, 15, 12, 30, 45, 123456),
+            id="with_microseconds_nonzero",
+        ),
+        pytest.param(
+            "2025-01-01T00:00:00Z",
+            datetime(2025, 1, 1, 0, 0, 0),
+            id="without_microseconds",
+        ),
+        pytest.param(
+            "2025-12-31T23:59:59Z",
+            datetime(2025, 12, 31, 23, 59, 59),
+            id="without_microseconds_end_of_day",
+        ),
+    ],
+)
+def test_parse_start_date(start_date_str: str, expected: datetime) -> None:
+    reader = TestStreamReader()
+    assert reader._parse_start_date(start_date_str) == expected
+
+
+def test_parse_start_date_invalid_raises() -> None:
+    reader = TestStreamReader()
+    with pytest.raises(ValueError):
+        reader._parse_start_date("not-a-date")
+
+
 def test_upload_with_file_transfer_reader():
     stream_reader = TestStreamReaderWithDefaultUpload()