Support ZSAV export

Author: TonisOrmisson

src/openstatspec/api.py

@@ -12,7 +12,7 @@
 def capabilities() -> Mapping[str, Any]:
     return CapabilityDeclaration(
         specification="OpenStatSpec strict wide-table SPSS profile (initial)",
-        formats={"SAV": {"import": True, "export": True}, "ZSAV": {"import": True, "export": False}},
+        formats={"SAV": {"import": True, "export": True}, "ZSAV": {"import": True, "export": True}},
         database_profiles=declared_profiles(),
     ).as_dict()
 

src/openstatspec/spss/sav.py

@@ -86,8 +86,8 @@ def import_sav_dataset(*, source: str | Path, database_url: str, dataset_id: str
 
 def export_sav_dataset(*, database_url: str, dataset_id: str, destination: str | Path) -> dict[str, Any]:
     destination_path = Path(destination)
-    if destination_path.suffix.lower() != ".sav":
-        raise UnsupportedOperationError("This initial profile exports SAV only; ZSAV output is not yet supported.")
+    if destination_path.suffix.lower() not in {".sav", ".zsav"}:
+        raise UnsupportedOperationError("Export destinations must use the .sav or .zsav extension.")
     dataset, variables, rows = read_wide_dataset(database_url=database_url, dataset_id=dataset_id)
     import pandas as pd
     frame = pd.DataFrame(
@@ -108,6 +108,7 @@ def export_sav_dataset(*, database_url: str, dataset_id: str, destination: str |
     }
     pyreadstat.write_sav(
         frame, destination_path, file_label=dataset["file_label"], column_labels=labels,
+        compress=destination_path.suffix.lower() == ".zsav",
         variable_format=formats, variable_measure=measures,
         variable_display_width=displays, variable_value_labels=value_labels,
         missing_ranges=missing_ranges, note=json.loads(dataset["documents"]),

tests/test_sav_sqlite.py

@@ -49,6 +49,10 @@ def test_sav_round_trip_uses_one_wide_table_and_catalog(tmp_path) -> None:
     assert meta.variable_value_labels == {"age": {34.0: "thirty-four"}}
     assert meta.original_variable_types == {"age": "F8.0", "name": "A12"}
     assert meta.variable_measure == {"age": "scale", "name": "nominal"}
+    zsav = tmp_path / "roundtrip.zsav"
+    openstatspec.export_sav(database_url=database, dataset_id="tiny", destination=zsav)
+    compressed_frame, _ = pyreadstat.read_sav(zsav, user_missing=True)
+    assert compressed_frame["name"].tolist() == ["Ada", ""]
     assert meta.missing_ranges == {"age": [{"lo": 34.0, "hi": 34.0}]}
     assert meta.notes == ["Fixture note"]