BUG: setting Data_version in write_cdf

imap_processing/cdf/utils.py

@@ -115,10 +115,11 @@ def write_cdf(
             "No Data_version attribute found in dataset. Using default v999.",
             stacklevel=2,
         )
-        version = "v999"
-    elif not re.match(r"v\d{3}", version):
+        version = "999"
+        dataset.attrs["Data_version"] = version
+    elif not re.match(r"\d{3}", version):
         raise ValueError(
-            f"The Data_version attribute {version} does not match expected format vXXX."
+            f"The Data_version attribute {version} does not match expected format XXX."
         )
 
     repointing = dataset.attrs.get("Repointing", None)
@@ -129,7 +130,7 @@ def write_cdf(
         data_level=data_level,
         descriptor=descriptor,
         start_time=start_date,
-        version=version,
+        version=f"v{version}",  # Ensure version is prefixed with 'v'
         repointing=repointing_int,
     )
     file_path = Path(science_file.construct_path())

imap_processing/cli.py

@@ -569,7 +569,7 @@ def post_processing(
         products = []
         for ds in processed_data:
             if isinstance(ds, xr.Dataset):
-                ds.attrs["Data_version"] = self.version
+                ds.attrs["Data_version"] = self.version[1:]  # Strip 'v' from version
                 if self.repointing is not None:
                     ds.attrs["Repointing"] = self.repointing
                 ds.attrs["Start_date"] = self.start_date

imap_processing/tests/cdf/test_utils.py

@@ -26,7 +26,7 @@ def test_dataset():
     # Load the CDF attrs
     swe_attrs = ImapCdfAttributes()
     swe_attrs.add_instrument_global_attrs("swe")
-    swe_attrs.add_global_attribute("Data_version", "v001")
+    swe_attrs.add_global_attribute("Data_version", "001")
 
     dataset = xr.Dataset(
         {

imap_processing/tests/idex/test_idex_l2a.py

@@ -46,7 +46,7 @@ def test_l2a_logical_source_and_cdf(l2a_dataset: xr.Dataset):
     expected_src = "imap_idex_l2a_sci-1week"
     assert l2a_dataset.attrs["Logical_source"] == expected_src
     # Verify the CDF file can be created with no errors.
-    l2a_dataset.attrs["Data_version"] = "v999"
+    l2a_dataset.attrs["Data_version"] = "999"
     file_name = write_cdf(l2a_dataset)
     assert file_name.exists()
     assert file_name.name == "imap_idex_l2a_sci-1week_20231218_v999.cdf"

imap_processing/tests/idex/test_idex_l2b.py

@@ -40,7 +40,7 @@ def test_l2b_logical_source_and_cdf(l2b_dataset: xr.Dataset):
     expected_src = "imap_idex_l2b_sci-1week"
     assert l2b_dataset.attrs["Logical_source"] == expected_src
     # Verify the CDF file can be created with no errors.
-    l2b_dataset.attrs["Data_version"] = "v999"
+    l2b_dataset.attrs["Data_version"] = "999"
     file_name = write_cdf(l2b_dataset)
 
     assert file_name.exists()

imap_processing/tests/idex/test_idex_l2c.py

@@ -61,8 +61,8 @@ def test_l2c_attrs_and_vars(l2c_datasets: list[xr.Dataset], l1b_dataset: xr.Data
         "rectangular_lat_pixel": int(180 / IDEX_SPACING_DEG),
         "epoch": 1,
     }
-    healpix_ds.attrs["Data_version"] = "v999"
-    rect_ds.attrs["Data_version"] = "v999"
+    healpix_ds.attrs["Data_version"] = "999"
+    rect_ds.attrs["Data_version"] = "999"
     # Check the attributes of the dataset by writing to a CDF file
     hp_file_name = write_cdf(healpix_ds)
     rect_file_name = write_cdf(rect_ds)

imap_processing/tests/mag/test_mag_decom.py

@@ -15,7 +15,7 @@ def cdf_attrs():
     test_attrs.add_instrument_global_attrs("mag")
     test_attrs.add_instrument_variable_attrs("mag", "l1a")
     # Default v001 expected when writing to file and re-loading
-    test_attrs.add_global_attribute("Data_version", "v001")
+    test_attrs.add_global_attribute("Data_version", "001")
     return test_attrs
 
 

imap_processing/tests/swe/test_swe_l1b.py

@@ -168,7 +168,7 @@ def test_swe_l1b(mock_get_file_paths, l1b_validation_df):
     l1b_input = l1a_datasets[0]
 
     # Set these two as cli.py -> post_processing would have.
-    l1b_input.attrs["Data_version"] = "v002"
+    l1b_input.attrs["Data_version"] = "002"
     # write data to CDF
     l1a_cdf_filepath = write_cdf(l1b_input)
     assert l1a_cdf_filepath.name == "imap_swe_l1a_sci_20240510_v002.cdf"
@@ -204,7 +204,7 @@ def get_file_paths_side_effect(descriptor):
     l1b_datasets = swe_l1b(dependencies)
 
     l1b_write_ds = l1b_datasets[0]
-    l1b_write_ds.attrs["Data_version"] = "v002"
+    l1b_write_ds.attrs["Data_version"] = "002"
 
     sci_l1b_filepath = write_cdf(l1b_write_ds)
 

imap_processing/tests/swe/test_swe_l2.py

@@ -262,7 +262,7 @@ def test_swe_l2(mock_get_file_paths, use_fake_spin_data_for_time):
     test_data_path = "tests/swe/l0_data/2024051010_SWE_SCIENCE_packet.bin"
     l1a_datasets = swe_l1a(imap_module_directory / test_data_path)
     l1a_ds = l1a_datasets[0]
-    l1a_ds.attrs["Data_version"] = "v000"
+    l1a_ds.attrs["Data_version"] = "000"
     l1a_cdf_filepath = write_cdf(l1a_ds)
     assert l1a_cdf_filepath.name == "imap_swe_l1a_sci_20240510_v000.cdf"
 
@@ -295,7 +295,7 @@ def get_file_paths_side_effect(descriptor):
     eu_anc = AncillaryInput("imap_swe_eu-conversion_20240510_v000.csv")
     dependencies = ProcessingInputCollection(science_input, inflight_anc, eu_anc)
     l1b_dataset = swe_l1b(dependencies)[0]
-    l1b_dataset.attrs["Data_version"] = "v000"
+    l1b_dataset.attrs["Data_version"] = "000"
     l2_dataset = swe_l2(l1b_dataset)
 
     assert isinstance(l2_dataset, xr.Dataset)
@@ -325,6 +325,6 @@ def get_file_paths_side_effect(descriptor):
     assert np.allclose(cal_factor, cal_factor[:, 0:1], rtol=1e-9, atol=0)
 
     # Write L2 to CDF
-    l2_dataset.attrs["Data_version"] = "v002"
+    l2_dataset.attrs["Data_version"] = "002"
     l2_cdf_filepath = write_cdf(l2_dataset)
     assert l2_cdf_filepath.name == "imap_swe_l2_sci_20240510_v002.cdf"

imap_processing/tests/ultra/mock_data.py

@@ -387,7 +387,7 @@ def mock_l1c_pset_product_healpix(
         attrs={
             "Logical_file_id": logical_file_id,
             "Logical_source": logical_source,
-            "Data_version": "v001",
+            "Data_version": "001",
         },
     )
 

imap_processing/tests/ultra/unit/test_ultra_l1a.py

@@ -71,7 +71,7 @@ def test_cdf_aux(ccsds_path_theta_0):
     """Tests that CDF file can be created."""
 
     test_data = ultra_l1a(ccsds_path_theta_0, apid_input=ULTRA_AUX.apid[0])
-    test_data[0].attrs["Data_version"] = "v999"
+    test_data[0].attrs["Data_version"] = "999"
     test_data[0].attrs["Repointing"] = "repoint99999"
     test_data_path = write_cdf(test_data[0], istp=True)
 
@@ -85,7 +85,7 @@ def test_cdf_aux(ccsds_path_theta_0):
 def test_cdf_rates(ccsds_path_theta_0):
     """Tests that CDF file can be created."""
     test_data = ultra_l1a(ccsds_path_theta_0, apid_input=ULTRA_RATES.apid[0])
-    test_data[0].attrs["Data_version"] = "v999"
+    test_data[0].attrs["Data_version"] = "999"
     test_data[0].attrs["Repointing"] = "repoint99999"
     test_data_path = write_cdf(test_data[0], istp=True)
 
@@ -99,7 +99,7 @@ def test_cdf_rates(ccsds_path_theta_0):
 def test_cdf_tof(ccsds_path_theta_0):
     """Tests that CDF file can be created."""
     test_data = ultra_l1a(ccsds_path_theta_0, apid_input=ULTRA_TOF.apid[0])
-    test_data[0].attrs["Data_version"] = "v999"
+    test_data[0].attrs["Data_version"] = "999"
     test_data[0].attrs["Repointing"] = "repoint99999"
     test_data_path = write_cdf(test_data[0], istp=True)
     assert test_data_path.exists()
@@ -112,7 +112,7 @@ def test_cdf_tof(ccsds_path_theta_0):
 def test_cdf_events(ccsds_path_theta_0):
     """Tests that CDF file can be created."""
     test_data = ultra_l1a(ccsds_path_theta_0, apid_input=ULTRA_EVENTS.apid[0])
-    test_data[0].attrs["Data_version"] = "v999"
+    test_data[0].attrs["Data_version"] = "999"
     test_data[0].attrs["Repointing"] = "repoint99999"
     test_data_path = write_cdf(test_data[0], istp=True)
 
@@ -127,7 +127,7 @@ def test_cdf_hk(ccsds_path_theta_0):
     """Tests that CDF file can be created."""
     test_data = ultra_l1a(ccsds_path_theta_0, apid_input=869)
     data = test_data[0]
-    data.attrs["Data_version"] = "v999"
+    data.attrs["Data_version"] = "999"
     data.attrs["Repointing"] = "repoint99999"
     test_data_path = write_cdf(data, istp=True)
 
@@ -143,7 +143,7 @@ def test_cdf_cmdtxt(ccsds_path_all_apids):
     """Tests that CDF file can be created."""
     test_data = ultra_l1a(ccsds_path_all_apids, apid_input=939)
     data = test_data[0]
-    data.attrs["Data_version"] = "v999"
+    data.attrs["Data_version"] = "999"
     test_data_path = write_cdf(data, istp=True)
 
     assert test_data_path.exists()
@@ -155,7 +155,7 @@ def test_cdf_monitorlimits(ccsds_path_all_apids):
     """Tests that CDF file can be created."""
     test_data = ultra_l1a(ccsds_path_all_apids, apid_input=937)
     data = test_data[0]
-    data.attrs["Data_version"] = "v999"
+    data.attrs["Data_version"] = "999"
     test_data_path = write_cdf(data, istp=True)
 
     assert test_data_path.exists()
@@ -169,7 +169,7 @@ def test_cdf_startup(ccsds_path_all_apids):
     """Tests that CDF file can be created."""
     test_data = ultra_l1a(ccsds_path_all_apids, apid_input=941)
     data = test_data[0]
-    data.attrs["Data_version"] = "v999"
+    data.attrs["Data_version"] = "999"
     test_data_path = write_cdf(data, istp=True)
 
     assert test_data_path.exists()

imap_processing/tests/ultra/unit/test_ultra_l1b.py

@@ -135,7 +135,7 @@ def side_effect_func(event_times, position, ultra_frame, dps_frame, sc_frame):
         == "IMAP-Ultra Instrument Level-1B Direct Event Data."
     )
 
-    l1b_de_dataset[0].attrs["Data_version"] = "v999"
+    l1b_de_dataset[0].attrs["Data_version"] = "999"
     l1b_de_dataset[0].attrs["Repointing"] = "repoint99999"
     test_data_path = write_cdf(l1b_de_dataset[0], istp=True)
     assert test_data_path.exists()
@@ -200,7 +200,7 @@ def test_cdf_extendedspin(use_fake_spin_data_for_time, faux_aux_dataset, rates_d
 
     l1b_extendedspin_dataset = ultra_l1b(data_dict)
     """Tests that CDF file is created and contains same attributes as xarray."""
-    l1b_extendedspin_dataset[0].attrs["Data_version"] = "v999"
+    l1b_extendedspin_dataset[0].attrs["Data_version"] = "999"
     l1b_extendedspin_dataset[0].attrs["Repointing"] = "repoint99999"
     test_data_path = write_cdf(l1b_extendedspin_dataset[0], istp=True)
     assert test_data_path.exists()
@@ -229,7 +229,7 @@ def test_cdf_cullingmask(use_fake_spin_data_for_time, faux_aux_dataset, rates_da
     data_dict["imap_ultra_l1a_45sensor-rates"] = rates_dataset
 
     l1b_extendedspin_dataset = ultra_l1b(data_dict)
-    l1b_extendedspin_dataset[1].attrs["Data_version"] = "v999"
+    l1b_extendedspin_dataset[1].attrs["Data_version"] = "999"
     l1b_extendedspin_dataset[1].attrs["Repointing"] = "repoint99999"
     test_data_path = write_cdf(l1b_extendedspin_dataset[1], istp=True)
     assert test_data_path.exists()
@@ -258,7 +258,7 @@ def test_cdf_badtimes(use_fake_spin_data_for_time, faux_aux_dataset, rates_datas
     data_dict["imap_ultra_l1a_45sensor-rates"] = rates_dataset
 
     l1b_extendedspin_dataset = ultra_l1b(data_dict)
-    l1b_extendedspin_dataset[2].attrs["Data_version"] = "v999"
+    l1b_extendedspin_dataset[2].attrs["Data_version"] = "999"
     l1b_extendedspin_dataset[2].attrs["Repointing"] = "repoint99999"
     test_data_path = write_cdf(l1b_extendedspin_dataset[2], istp=True)
     assert test_data_path.exists()

imap_processing/tests/ultra/unit/test_ultra_l1c.py

@@ -172,7 +172,7 @@ def test_calculate_spacecraft_pset_with_cdf():
     }
 
     output_datasets = ultra_l1c(data_dict)
-    output_datasets[0].attrs["Data_version"] = "v999"
+    output_datasets[0].attrs["Data_version"] = "999"
     output_datasets[0].attrs["Repointing"] = f"repoint{pointing + 1:05d}"
     test_data_path = write_cdf(output_datasets[0], istp=True)