Improve error handling, tests of "tdc check" CLI

Author: khaeru

transport_data/cli/__init__.py

@@ -69,8 +69,19 @@ def check(structure_urn: str, path: Path, sheets, verbose, **options):  # noqa:
     from transport_data import STORE
     from transport_data.util.sdmx import read_csv
 
+    # Pieces of any error message
+    message = []
+
     # Handle `structure_urn`: retrieve a data structure that describes the data
-    structure = STORE.get(structure_urn)
+    try:
+        structure = STORE.get(structure_urn)
+    except Exception:
+        message.append(f"Structure {structure_urn!r} could not be loaded")
+        structure = structure_cls = structure_id = None
+    else:
+        structure_cls = type(structure).__name__.lower().replace("definition", "")
+        structure_id = sdmx.urn.shorten(structure.urn).split("=")[-1]
+
     if isinstance(structure, common.BaseDataflow):
         # Also retrieve the data structure definition
         STORE.resolve(structure, "structure")
@@ -79,10 +90,8 @@ def check(structure_urn: str, path: Path, sheets, verbose, **options):  # noqa:
     # Construct keyword arguments for CSVAdapter
     # TODO Check if this works for full SDMX-CSV
     adapt = {
-        "structure": options.pop("structure")
-        or type(structure).__name__.lower().replace("definition", ""),
-        "structure_id": options.pop("structure_id")
-        or sdmx.urn.shorten(structure.urn).split("=")[-1],
+        "structure": options.pop("structure") or structure_cls,
+        "structure_id": options.pop("structure_id") or structure_id,
         "action": options.pop("action"),
     }
 
@@ -125,18 +134,16 @@ def check(structure_urn: str, path: Path, sheets, verbose, **options):  # noqa:
         try:
             dm = read_csv(p, structure, adapt)
         except Exception as e:
-            message = [f"read failed with\n{type(e).__name__}: {' '.join(e.args)}"]
+            message.append(f"read failed with\n{type(e).__name__}: {' '.join(e.args)}")
 
             if len(e.args) and "line 1" in e.args[0]:
                 message.append(
                     "Hint: try giving --structure= or --structure-id argument(s) to "
                     "adapt to SDMX-CSV."
                 )
-            elif isinstance(e, KeyError):
-                message.append(
-                    "Hint: try giving --action argument to adapt to SDMX-CSV."
-                )
-            else:
+            elif structure is None:
+                pass
+            else:  # pragma: no cover
                 message.append("\n".join(format_exception(e)))
 
             print("")

transport_data/tests/test_cli.py

@@ -30,26 +30,26 @@ def test_cli(tdc_cli, command):
     tdc_cli.invoke(command)
 
 
-def test_check(tdc_cli, test_data_path, tmp_store):
+CHECK_ARGS = [
+    "--structure=dataflow",
+    "--structure-id=FOO",
+    "--action=I",
+    "--sheets=sheet_2",
+    "Dataflow=TDCI:EMBER_001(1.0.0)",
+]
+
+
+def test_check0(tdc_cli, test_data_path, tmp_store):
+    """Check a successful read of a .xlsx file."""
     ember_dfd(tmp_store)
 
-    result = tdc_cli.invoke(
-        [
-            "check",
-            "--sheets=sheet_2",
-            "--structure=dataflow",
-            "--structure-id=FOO",
-            "--action=I",
-            "Dataflow=TDCI:EMBER_001(1.0.0)",
-            str(test_data_path.joinpath("read-csv-2.xlsx")),
-        ]
-    )
+    path = test_data_path.joinpath("read-csv-2.xlsx")
+    result = tdc_cli.invoke(["check"] + CHECK_ARGS + [str(path)])
 
     # Command runs without error
     assert 0 == result.exit_code, result.output
 
     # Data was located, converted to CSV, read, and summary displayed
-    print(result.output)
     assert re.fullmatch(
         r"""
 File: .*read-csv-2.xlsx
@@ -64,6 +64,40 @@ def test_check(tdc_cli, test_data_path, tmp_store):
     )
 
 
+@pytest.mark.parametrize(
+    "args, exit_code, text",
+    (
+        (["X"], 1, "line 1, field 1"),  # Omit all options and URN
+        (CHECK_ARGS[:1] + ["X"], 1, "line 1, field 2"),  # Only --structure=
+        (CHECK_ARGS[:2] + ["X"], 1, "'X' could not be loaded"),  # Invalid structure URN
+        (CHECK_ARGS + ["-v"], 0, "..."),  # Show pd.DataFrame abbreviated string repr
+        (CHECK_ARGS + ["-vv"], 0, ""),  # Show pd.DataFrame full string repr
+    ),
+)
+def test_check1(tdc_cli, test_data_path, tmp_store, args, exit_code, text):
+    """Check various other argument combinations."""
+    ember_dfd(tmp_store)
+
+    path = test_data_path.joinpath("read-csv-1.csv")
+    result = tdc_cli.invoke(["check"] + args + [str(path)])
+
+    # Command gives the expected exit code
+    assert exit_code == result.exit_code, result.output
+
+    # Expected text is contained in the output
+    if text:
+        assert text in result.output
+
+
+def test_check2(tdc_cli, tmp_path):
+    path = tmp_path.joinpath("foo.txt")
+    path.touch()
+    result = tdc_cli.invoke(["check", "X", str(path)])
+
+    assert 2 == result.exit_code, result.output
+    assert "Unsupported file extension" in result.output
+
+
 # ‘Script’ of CLI input to produce a data structure definition
 SCRIPT = [
     # Maintainer ID: accept the default (TDCI)