Proof of concept CLI

pyproject.toml

@@ -7,6 +7,7 @@ authors = [
     {name = "Jonatan Skogsfors", email = "jonatan.skogsfors@smhi.se"},
 ]
 dependencies = [
+    "click>=8.3.0",
     "nodc-codes",
     "pyyaml>=6.0.3",
     "sharkadm",
@@ -17,7 +18,7 @@ license = {file = "LICENSE"}
 
 
 [project.scripts]
-generate-text-from-yaml = "metadata.generate_text_file_from_yaml:main"
+metadata-from-package = "delivery_metadata.metadata_from_package:cli"
 
 [dependency-groups]
 dev = [

requirements.txt

@@ -63,7 +63,9 @@ charset-normalizer==3.4.3 \
 click==8.3.0 \
     --hash=sha256:9b9f285302c6e3064f4330c05f05b81945b2a39544279343e6e7c5f27a9baddc \
     --hash=sha256:e7b8232224eba16f4ebe410c25ced9f7875cb5f3263ffc93cc3e8da705e229c4
-    # via typer
+    # via
+    #   delivery-metadata
+    #   typer
 colorama==0.4.6 ; sys_platform == 'win32' \
     --hash=sha256:08695f5cb7ed6e0531a20572697297273c47b8cae5a63ffc6d6ed5c201be6e44 \
     --hash=sha256:4f1d9991f5acc0ca119f9d443620b77f9d6b33703e51011c16baf57afb285fc6
@@ -86,9 +88,9 @@ fastexcel==0.16.0 \
     --hash=sha256:e1059eac593f4b92843ac9d10901677cccc2a8152c67e315c9dfbd7ce7c722e7 \
     --hash=sha256:ec1c56b9b3b7b7ff2bde64dbe0e378a707287aff9deeb71ff6d0f8c3b7d24e34
     # via sharkadm
-filelock==3.19.1 \
-    --hash=sha256:66eda1888b0171c998b35be2bcc0f6d75c388a7ce20c3f3f37aa8e96c2dddf58 \
-    --hash=sha256:d38e30481def20772f5baf097c122c3babc4fcdb7e14e57049eb9d88c6dc017d
+filelock==3.20.0 \
+    --hash=sha256:339b4732ffda5cd79b13f4e2711a31b0365ce445d95d243bb996273d072546a2 \
+    --hash=sha256:711e943b4ec6be42e1d4e6690b48dc175c822967466bb31c0c293f34334c13f4
     # via virtualenv
 geopandas==1.1.1 \
     --hash=sha256:1745713f64d095c43e72e08e753dbd271678254b24f2e01db8cdb8debe1d293d \
@@ -289,9 +291,9 @@ pandas==2.3.3 \
     # via
     #   geopandas
     #   sharkadm
-platformdirs==4.4.0 \
-    --hash=sha256:abd01743f24e5287cd7a5db3752faf1a2d65353f38ec26d98e25a6db65958c85 \
-    --hash=sha256:ca753cf4d81dc309bc67b0ea38fd15dc97bc30ce419a7f58d13eb3bf14c4febf
+platformdirs==4.5.0 \
+    --hash=sha256:70ddccdd7c99fc5942e9fc25636a8b34d04c24b335100223152c2803e4063312 \
+    --hash=sha256:e578a81bb873cbb89a41fcc904c7ef523cc18284b7e3b3ccf06aca1403b7ebd3
     # via virtualenv
 pluggy==1.6.0 \
     --hash=sha256:7dcc130b76258d33b90f61b658791dede3486c3e6bfb003ee5c9bfb396dd22f3 \
@@ -503,9 +505,9 @@ requests==2.32.5 \
     # via
     #   nodc-codes
     #   sharkadm
-rich==14.1.0 \
-    --hash=sha256:536f5f1785986d6dbdea3c75205c473f970777b4a0d6c6dd1b696aa05a3fa04f \
-    --hash=sha256:e497a48b844b0320d45007cdebfeaeed8db2a4f4bcf49f15e455cfc4af11eaa8
+rich==14.2.0 \
+    --hash=sha256:73ff50c7c0c1c77c8243079283f4edb376f0f6442433aecb8ce7e6d0b92d1fe4 \
+    --hash=sha256:76bc51fe2e57d2b1be1f96c524b890b816e334ab4c1e45888799bfaab0021edd
     # via
     #   sharkadm
     #   typer
@@ -582,7 +584,7 @@ shapely==2.1.2 \
     # via
     #   geopandas
     #   sharkadm
-sharkadm @ git+https://github.com/nodc-sweden/SHARKadm@6f4c94ff7640f818e0b524223dd3a86a4db7efed
+sharkadm @ git+https://github.com/nodc-sweden/SHARKadm@7f9cc05006425e6a463e8ddc00df3d4ac9ab8606
     # via delivery-metadata
 shellingham==1.5.4 \
     --hash=sha256:7ecfff8f2fd72616f7481040475a65b2bf8af90a56c89140852d1120324e8686 \
@@ -608,9 +610,9 @@ urllib3==2.5.0 \
     --hash=sha256:3fc47733c7e419d4bc3f6b3dc2b4f890bb743906a30d56ba4a5bfa4bbff92760 \
     --hash=sha256:e6b01673c0fa6a13e374b50871808eb3bf7046c4b125b216f6bf1cc604cff0dc
     # via requests
-virtualenv==20.34.0 \
-    --hash=sha256:341f5afa7eee943e4984a9207c025feedd768baff6753cd660c857ceb3e36026 \
-    --hash=sha256:44815b2c9dee7ed86e387b842a84f20b93f7f417f95886ca1996a72a4138eb1a
+virtualenv==20.35.1 \
+    --hash=sha256:041dac43b6899858a91838b616599e80000e545dee01a21172a6a46746472cb2 \
+    --hash=sha256:1d9d93cd01d35b785476e2fa7af711a98d40d227a078941695bbae394f8737e2
     # via pre-commit
 xlsxwriter==3.2.9 \
     --hash=sha256:254b1c37a368c444eac6e2f867405cc9e461b0ed97a3233b2ac1e574efb4140c \

src/delivery_metadata/__init__.py

@@ -1 +1 @@
-from delivery_metadata.delivery_metadata import DeliveryMetadata
+from delivery_metadata.delivery_data import DeliveryData

src/delivery_metadata/delivery_data.py

@@ -0,0 +1,115 @@
+from pathlib import Path
+from typing import Callable, Self
+
+import polars as pl
+from sharkadm.data import get_polars_data_holder
+
+
+def _apply_on_column(function: Callable, column: str, dataframe: pl.DataFrame):
+    if column not in dataframe.columns:
+        return None
+
+    return function(dataframe[column])
+
+
+class DeliveryData:
+    _fields = (
+        "datatype",
+        "abstract",
+        "description",
+        "dataset_filename",
+        "discipline",
+        "measuring_area_type",
+        "coordinate_system",
+        "platform_class",
+        "access_constraints",
+        "min_year",
+        "max_year",
+        "min_date",
+        "max_date",
+        "min_longitude_dd",
+        "max_longitude_dd",
+        "min_latitude_dd",
+        "max_latitude_dd",
+        "taxonomic_coverage",
+        "originator",
+        "contact",
+        "orderer",
+        "data_holding_centre",
+        "data distributor",
+        "database_reference",
+        "internet_access",
+        "address",
+        "postal_code",
+        "city",
+        "phone",
+        "email",
+        "citation",
+    )
+
+    def __init__(
+        self,
+        data: pl.DataFrame | None = None,
+        delivery_note: dict | None = None,
+        source: str = "",
+    ):
+        self._data = pl.DataFrame() if data is None else data
+        self._delivery_note = delivery_note or {}
+        self._source = source
+
+    @property
+    def data(self):
+        return self._data
+
+    @property
+    def delivery_note(self):
+        return self._delivery_note
+
+    @property
+    def fields(self):
+        return self._fields
+
+    def generate_metadata(self):
+        print("\n".join(sorted(self._data.columns)))
+        return {
+            "datatype": self._delivery_note.get("DTYPE"),
+            "abstract": None,
+            "description": None,
+            "dataset_filename": self._source,
+            "discipline": None,
+            "measuring_area_type": None,
+            "coordinate_system": None,
+            "platform_class": None,
+            "access_constraints": None,
+            "min_year": _apply_on_column(min, "visit_year", self._data),
+            "max_year": _apply_on_column(max, "visit_year", self._data),
+            "min_date": _apply_on_column(min, "sample_date", self._data),
+            "max_date": _apply_on_column(max, "sample_date", self._data),
+            "min_longitude_dd": _apply_on_column(min, "sample_longitude_dd", self._data),
+            "max_longitude_dd": _apply_on_column(max, "sample_longitude_dd", self._data),
+            "min_latitude_dd": _apply_on_column(min, "sample_latitude_dd", self._data),
+            "max_latitude_dd": _apply_on_column(max, "sample_latitude_dd", self._data),
+            "taxonomic_coverage": None,
+            "originator": None,
+            "contact": None,
+            "orderer": None,
+            "data_holding_centre": None,
+            "data distributor": None,
+            "database_reference": None,
+            "internet_access": None,
+            "address": None,
+            "postal_code": None,
+            "city": None,
+            "phone": None,
+            "email": None,
+            "citation": None,
+        }
+
+    @classmethod
+    def from_shark_package(cls, package_path: Path) -> Self:
+        sharkadm_dataholder = get_polars_data_holder(package_path)
+        return cls(
+            data=sharkadm_dataholder.data,
+            delivery_note=sharkadm_dataholder.delivery_note.data,
+            source=package_path.name,
+        )

src/delivery_metadata/metadata_from_package.py

@@ -0,0 +1,18 @@
+import json
+from pathlib import Path
+
+import click
+
+from delivery_metadata import DeliveryData
+
+
+@click.command()
+@click.argument("package", type=click.Path(exists=True, path_type=Path))
+def cli(package: Path):
+    delivery_data = DeliveryData.from_shark_package(package)
+    metadata = delivery_data.generate_metadata()
+    print(json.dumps(metadata, indent=2))
+
+
+if __name__ == "__main__":
+    cli()

tests/test_parse_shark_data.py

@@ -3,7 +3,7 @@
 
 import polars as pl
 
-from delivery_metadata import DeliveryMetadata
+from delivery_metadata import DeliveryData
 
 
 def _write_data_to_package_folder(
@@ -48,7 +48,7 @@ def test_parse_unpacked_folder(
     )
 
     # When parsing data using path
-    metadata = DeliveryMetadata.from_shark_package(package_path)
+    metadata = DeliveryData.from_shark_package(package_path)
 
     # Then metadata holds data
     assert not metadata.data.is_empty()
@@ -92,7 +92,7 @@ def test_parse_zipped_folder(
     zipped_package_path = Path(shutil.make_archive(package_path, "zip", package_path))
 
     # When parsing data using path
-    metadata = DeliveryMetadata.from_shark_package(zipped_package_path)
+    metadata = DeliveryData.from_shark_package(zipped_package_path)
 
     # Then metadata holds data
     assert not metadata.data.is_empty()

tests/test_write_metadata.py

@@ -1,12 +1,14 @@
-from delivery_metadata import DeliveryMetadata
+from delivery_metadata import DeliveryData
 
 
 def test_metadata_has_expected_fields():
-    # Given a metadata object
-    metadata = DeliveryMetadata()
+    given_delivery_note_data = {"DTYPE": "Phytoplankton"}
+
+    # Given a delivery data object
+    delivery_data = DeliveryData(delivery_note=given_delivery_note_data)
 
     # When looking at all available fields
-    metadata_fields = metadata.fields
+    metadata = delivery_data.generate_metadata()
 
     # Then they correspond to the expected fields
     orderered_expected_fields = (
@@ -43,4 +45,4 @@ def test_metadata_has_expected_fields():
         "citation",
     )
 
-    assert metadata_fields == orderered_expected_fields
+    assert set(metadata.keys()) == set(orderered_expected_fields)