disdrodb 0.1.2__py3-none-any.whl → 0.1.4__py3-none-any.whl

disdrodb/metadata/checks.py

@@ -137,8 +137,6 @@ def _check_metadata_measurement_interval(metadata):
     """Check metadata ``measurement_interval``."""
     from disdrodb.api.checks import check_measurement_intervals
 
-    if "measurement_interval" not in metadata:
-        raise ValueError("The metadata file does not contain the 'measurement_interval' key.")
     measurement_intervals = metadata["measurement_interval"]
     _ = check_measurement_intervals(measurement_intervals)
 
@@ -150,28 +148,8 @@ def _check_metadata_sensor_name(metadata):
     check_sensor_name(sensor_name)
 
 
-def check_station_metadata(data_source, campaign_name, station_name, metadata_archive_dir=None):
-    """Check DISDRODB metadata compliance."""
-    from disdrodb.l0.l0_reader import check_metadata_reader
-
-    metadata = read_station_metadata(
-        data_source=data_source,
-        campaign_name=campaign_name,
-        station_name=station_name,
-        metadata_archive_dir=metadata_archive_dir,
-    )
-    _check_metadata_keys(metadata)
-    _check_metadata_values(metadata)
-    _check_metadata_campaign_name(metadata, expected_name=campaign_name)
-    _check_metadata_data_source(metadata, expected_name=data_source)
-    _check_metadata_station_name(metadata, expected_name=station_name)
-    _check_metadata_sensor_name(metadata)
-    _check_metadata_measurement_interval(metadata)
-    check_metadata_reader(metadata)
-
-
 #### --------------------------------------------------------------------------.
-#### Metadata Archive Missing Information
+#### Geolocation Checks
 
 
 def _check_lonlat_type(longitude, latitude):
@@ -185,16 +163,18 @@ def _check_lonlat_type(longitude, latitude):
         raise ValueError("Unspecified longitude and latitude coordinates.")
 
 
-def _check_lonlat_validity(longitude, latitude):
+def _check_lonlat_validity(longitude, latitude, raise_error_if_unknown=True):
     if longitude == -9999 or latitude == -9999:
-        raise ValueError("Missing lat lon coordinates (-9999).")
+        if raise_error_if_unknown:
+            raise ValueError("Missing lat lon coordinates (-9999).")
+        return
     if longitude > 180 or longitude < -180:
         raise ValueError("Invalid longitude (outside [-180, 180])")
     if latitude > 90 or latitude < -90:
         raise ValueError("Invalid latitude (outside [-90, 90])")
 
 
-def check_station_metadata_geolocation(metadata) -> None:
+def check_station_metadata_geolocation(metadata, raise_error_if_unknown=True) -> None:
     """Identify metadata with missing or wrong geolocation."""
     # Get longitude, latitude and platform type
     longitude = metadata.get("longitude")
@@ -209,54 +189,11 @@ def check_station_metadata_geolocation(metadata) -> None:
             raise ValueError("For mobile platform_type, specify latitude and longitude -9999")
     # - If fixed platform
     else:
-        _check_lonlat_validity(longitude=longitude, latitude=latitude)
-
-
-def identify_missing_metadata_coords(metadata_filepaths: str) -> None:
-    """Identify missing coordinates.
-
-    Parameters
-    ----------
-    metadata_filepaths : str
-        Input YAML file path.
-
-    Raises
-    ------
-    TypeError
-        Error if ``latitude`` or ``longitude`` coordinates are not present or are wrongly formatted.
-
-    """
-    for filepath in metadata_filepaths:
-        metadata = read_yaml(filepath)
-        check_station_metadata_geolocation(metadata)
-
-
-def identify_empty_metadata_keys(metadata_filepaths: list, keys: Union[str, list]) -> None:
-    """Identify empty metadata keys.
-
-    Parameters
-    ----------
-    metadata_filepaths : str
-        Input YAML file path.
-    keys : Union[str,list]
-        Attributes to verify the presence.
-    """
-    if isinstance(keys, str):
-        keys = [keys]
-
-    for filepath in metadata_filepaths:
-        for key in keys:
-            metadata = read_yaml(filepath)
-            if len(str(metadata.get(key, ""))) == 0:  # ensure is string to avoid error
-                print(f"Empty {key} at: ", filepath)
-
-
-#### --------------------------------------------------------------------------.
-#### Check Metadata Archive
+        _check_lonlat_validity(longitude=longitude, latitude=latitude, raise_error_if_unknown=raise_error_if_unknown)
 
 
-def check_metadata_archive_keys(metadata_archive_dir: Optional[str] = None) -> bool:
-    """Check that all metadata files have valid keys.
+def check_metadata_archive_geolocation(metadata_archive_dir: Optional[str] = None, raise_error_if_unknown=True):
+    """Check the metadata files have missing or wrong geolocation..
 
     Parameters
     ----------
@@ -293,17 +230,39 @@ def check_metadata_archive_keys(metadata_archive_dir: Optional[str] = None) -> b
             station_name=station_name,
         )
         try:
-            _check_metadata_keys(metadata)
+            check_station_metadata_geolocation(metadata, raise_error_if_unknown=raise_error_if_unknown)
         except Exception as e:
-            print(f"Error for {data_source} {campaign_name} {station_name}.")
-            print(f"The error is: {e}.")
             is_valid = False
-
+            print(f"Missing information for {data_source} {campaign_name} {station_name}.")
+            print(f"The error is: {e}.")
     return is_valid
 
 
-def check_metadata_archive_campaign_name(metadata_archive_dir: Optional[str] = None) -> bool:
-    """Check metadata ``campaign_name``.
+####------------------------------------------------------------------------------------------------------------.
+#### Metadata Station and Archive Checking Routines
+def check_station_metadata(data_source, campaign_name, station_name, metadata_archive_dir=None):
+    """Check DISDRODB metadata compliance."""
+    from disdrodb.l0.l0_reader import check_metadata_reader
+
+    metadata = read_station_metadata(
+        data_source=data_source,
+        campaign_name=campaign_name,
+        station_name=station_name,
+        metadata_archive_dir=metadata_archive_dir,
+    )
+    _check_metadata_keys(metadata)
+    _check_metadata_values(metadata)
+    _check_metadata_campaign_name(metadata, expected_name=campaign_name)
+    _check_metadata_data_source(metadata, expected_name=data_source)
+    _check_metadata_station_name(metadata, expected_name=station_name)
+    _check_metadata_sensor_name(metadata)
+    _check_metadata_measurement_interval(metadata)
+    check_station_metadata_geolocation(metadata, raise_error_if_unknown=False)
+    check_metadata_reader(metadata)
+
+
+def check_metadata_archive(metadata_archive_dir: Optional[str] = None, raise_error=False):
+    """Check the archive metadata compliance.
 
     Parameters
     ----------
@@ -312,6 +271,9 @@ def check_metadata_archive_campaign_name(metadata_archive_dir: Optional[str] = N
         The directory path must end with ``<...>/DISDRODB``.
         If ``None``, it uses the ``metadata_archive_dir`` path specified
         in the DISDRODB active configuration.
+    raise_error: bool (optional)
+        Whether to raise an error and interrupt the archive check if a
+        metadata is not compliant. The default value is ``False``.
 
     Returns
     -------
@@ -332,24 +294,74 @@ def check_metadata_archive_campaign_name(metadata_archive_dir: Optional[str] = N
         data_source = infer_data_source_from_path(filepath)
         campaign_name = infer_campaign_name_from_path(filepath)
         station_name = os.path.basename(filepath).replace(".yml", "")
-
-        metadata = read_station_metadata(
-            metadata_archive_dir=metadata_archive_dir,
-            data_source=data_source,
-            campaign_name=campaign_name,
-            station_name=station_name,
-        )
+        # Check compliance
         try:
-            _check_metadata_campaign_name(metadata, expected_name=campaign_name)
+            check_station_metadata(
+                metadata_archive_dir=metadata_archive_dir,
+                data_source=data_source,
+                campaign_name=campaign_name,
+                station_name=station_name,
+            )
         except Exception as e:
             is_valid = False
-            print(f"Error for {data_source} {campaign_name} {station_name}.")
-            print(f"The error is: {e}.")
+            msg = f"Error for {data_source} {campaign_name} {station_name}."
+            msg = msg + f"The error is: {e}."
+            if raise_error:
+                raise ValueError(msg)
+            print(msg)
+
     return is_valid
 
 
-def check_metadata_archive_data_source(metadata_archive_dir: Optional[str] = None) -> bool:
-    """Check metadata ``data_source``.
+####-----------------------------------------------------------------------------------------------.
+#### Utilities
+
+
+def identify_missing_metadata_coords(metadata_filepaths: str) -> None:
+    """Identify missing coordinates.
+
+    Parameters
+    ----------
+    metadata_filepaths : str
+        Input YAML file path.
+
+    Raises
+    ------
+    TypeError
+        Error if ``latitude`` or ``longitude`` coordinates are not present or are wrongly formatted.
+
+    """
+    for filepath in metadata_filepaths:
+        metadata = read_yaml(filepath)
+        check_station_metadata_geolocation(metadata)
+
+
+def identify_empty_metadata_keys(metadata_filepaths: list, keys: Union[str, list]) -> None:
+    """Identify empty metadata keys.
+
+    Parameters
+    ----------
+    metadata_filepaths : str
+        Input YAML file path.
+    keys : Union[str,list]
+        Attributes to verify the presence.
+    """
+    if isinstance(keys, str):
+        keys = [keys]
+
+    for filepath in metadata_filepaths:
+        for key in keys:
+            metadata = read_yaml(filepath)
+            if len(str(metadata.get(key, ""))) == 0:  # ensure is string to avoid error
+                print(f"Empty {key} at: ", filepath)
+
+
+#### --------------------------------------------------------------------------.
+#### Metadata Archive Utilities
+
+
+def check_metadata_archive_keys(metadata_archive_dir: Optional[str] = None) -> bool:
+    """Check that all metadata files have valid keys.
 
     Parameters
     ----------
@@ -386,16 +398,17 @@ def check_metadata_archive_data_source(metadata_archive_dir: Optional[str] = Non
             station_name=station_name,
         )
         try:
-            _check_metadata_data_source(metadata, expected_name=data_source)
+            _check_metadata_keys(metadata)
         except Exception as e:
-            is_valid = False
             print(f"Error for {data_source} {campaign_name} {station_name}.")
             print(f"The error is: {e}.")
+            is_valid = False
+
     return is_valid
 
 
-def check_metadata_archive_sensor_name(metadata_archive_dir: Optional[str] = None) -> bool:
-    """Check metadata ``sensor_name``.
+def check_metadata_archive_campaign_name(metadata_archive_dir: Optional[str] = None) -> bool:
+    """Check metadata ``campaign_name``.
 
     Parameters
     ----------
@@ -432,7 +445,7 @@ def check_metadata_archive_sensor_name(metadata_archive_dir: Optional[str] = Non
             station_name=station_name,
         )
         try:
-            _check_metadata_sensor_name(metadata)
+            _check_metadata_campaign_name(metadata, expected_name=campaign_name)
         except Exception as e:
             is_valid = False
             print(f"Error for {data_source} {campaign_name} {station_name}.")
@@ -440,8 +453,8 @@ def check_metadata_archive_sensor_name(metadata_archive_dir: Optional[str] = Non
     return is_valid
 
 
-def check_metadata_archive_station_name(metadata_archive_dir: Optional[str] = None) -> bool:
-    """Check metadata ``station_name``.
+def check_metadata_archive_data_source(metadata_archive_dir: Optional[str] = None) -> bool:
+    """Check metadata ``data_source``.
 
     Parameters
     ----------
@@ -478,7 +491,7 @@ def check_metadata_archive_station_name(metadata_archive_dir: Optional[str] = No
             station_name=station_name,
         )
         try:
-            _check_metadata_station_name(metadata, expected_name=station_name)
+            _check_metadata_data_source(metadata, expected_name=data_source)
         except Exception as e:
             is_valid = False
             print(f"Error for {data_source} {campaign_name} {station_name}.")
@@ -486,8 +499,8 @@ def check_metadata_archive_station_name(metadata_archive_dir: Optional[str] = No
     return is_valid
 
 
-def check_metadata_archive_reader(metadata_archive_dir: Optional[str] = None) -> bool:
-    """Check if the ``reader`` key is available and there is the associated reader.
+def check_metadata_archive_sensor_name(metadata_archive_dir: Optional[str] = None) -> bool:
+    """Check metadata ``sensor_name``.
 
     Parameters
     ----------
@@ -502,8 +515,6 @@ def check_metadata_archive_reader(metadata_archive_dir: Optional[str] = None) ->
     bool
         If the check succeeds, the result is ``True``, otherwise ``False``.
     """
-    from disdrodb.l0.l0_reader import check_metadata_reader
-
     is_valid = True
     metadata_archive_dir = get_metadata_archive_dir(metadata_archive_dir)
     list_metadata_paths = get_list_metadata(
@@ -526,7 +537,7 @@ def check_metadata_archive_reader(metadata_archive_dir: Optional[str] = None) ->
             station_name=station_name,
         )
         try:
-            check_metadata_reader(metadata)
+            _check_metadata_sensor_name(metadata)
         except Exception as e:
             is_valid = False
             print(f"Error for {data_source} {campaign_name} {station_name}.")
@@ -534,8 +545,8 @@ def check_metadata_archive_reader(metadata_archive_dir: Optional[str] = None) ->
     return is_valid
 
 
-def check_metadata_archive(metadata_archive_dir: Optional[str] = None, raise_error=False):
-    """Check the archive metadata compliance.
+def check_metadata_archive_station_name(metadata_archive_dir: Optional[str] = None) -> bool:
+    """Check metadata ``station_name``.
 
     Parameters
     ----------
@@ -544,9 +555,6 @@ def check_metadata_archive(metadata_archive_dir: Optional[str] = None, raise_err
         The directory path must end with ``<...>/DISDRODB``.
         If ``None``, it uses the ``metadata_archive_dir`` path specified
         in the DISDRODB active configuration.
-    raise_error: bool (optional)
-        Whether to raise an error and interrupt the archive check if a
-        metadata is not compliant. The default value is ``False``.
 
     Returns
     -------
@@ -567,27 +575,24 @@ def check_metadata_archive(metadata_archive_dir: Optional[str] = None, raise_err
         data_source = infer_data_source_from_path(filepath)
         campaign_name = infer_campaign_name_from_path(filepath)
         station_name = os.path.basename(filepath).replace(".yml", "")
-        # Check compliance
+
+        metadata = read_station_metadata(
+            metadata_archive_dir=metadata_archive_dir,
+            data_source=data_source,
+            campaign_name=campaign_name,
+            station_name=station_name,
+        )
         try:
-            check_station_metadata(
-                metadata_archive_dir=metadata_archive_dir,
-                data_source=data_source,
-                campaign_name=campaign_name,
-                station_name=station_name,
-            )
+            _check_metadata_station_name(metadata, expected_name=station_name)
         except Exception as e:
             is_valid = False
-            msg = f"Error for {data_source} {campaign_name} {station_name}."
-            msg = msg + f"The error is: {e}."
-            if raise_error:
-                raise ValueError(msg)
-            print(msg)
-
+            print(f"Error for {data_source} {campaign_name} {station_name}.")
+            print(f"The error is: {e}.")
     return is_valid
 
 
-def check_metadata_archive_geolocation(metadata_archive_dir: Optional[str] = None):
-    """Check the metadata files have missing or wrong geolocation..
+def check_metadata_archive_reader(metadata_archive_dir: Optional[str] = None) -> bool:
+    """Check if the ``reader`` key is available and there is the associated reader.
 
     Parameters
     ----------
@@ -602,6 +607,8 @@ def check_metadata_archive_geolocation(metadata_archive_dir: Optional[str] = Non
     bool
         If the check succeeds, the result is ``True``, otherwise ``False``.
     """
+    from disdrodb.l0.l0_reader import check_metadata_reader
+
     is_valid = True
     metadata_archive_dir = get_metadata_archive_dir(metadata_archive_dir)
     list_metadata_paths = get_list_metadata(
@@ -624,9 +631,9 @@ def check_metadata_archive_geolocation(metadata_archive_dir: Optional[str] = Non
             station_name=station_name,
         )
         try:
-            check_station_metadata_geolocation(metadata)
+            check_metadata_reader(metadata)
         except Exception as e:
             is_valid = False
-            print(f"Missing information for {data_source} {campaign_name} {station_name}.")
+            print(f"Error for {data_source} {campaign_name} {station_name}.")
             print(f"The error is: {e}.")
     return is_valid

disdrodb/metadata/standards.py

@@ -24,9 +24,10 @@ the DISDRODB-METADATA repository
 
 # Define valid values for specific metadata keys
 METADATA_VALUES = {
+    "raw_data_format": ["txt", "netcdf"],
     "deployment_status": ["ongoing", "terminated"],
-    "platform_type": ["fixed", "mobile"],
     "deployment_mode": ["land", "ship", "truck", "cable"],
+    "platform_type": ["fixed", "mobile"],
     "platform_protection": ["shielded", "unshielded", ""],
 }
 
@@ -70,6 +71,7 @@ METADATA_KEYS = [
     "history",
     "comment",
     "station_id",
+    "koppen_geiger",
     "location",
     "country",
     "continent",