update output types of publishing assets, add cloud sensors

output_spec.md

@@ -30,6 +30,7 @@ The top level should contain:
     ├── metric_metadata.parquet
     ├── source_data_releases.parquet
     ├── data_publishers.parquet
+    ├── geometry_metadata.parquet
     ├── country_metadata.parquet
     ├── metrics/
     │   ├── {metric_filename_a}.parquet
@@ -51,6 +52,8 @@ with tabulated metadata:
   `SourceDataRelease`
 - A `data_publishers.parquet` file containing a serialised list of
   `DataPublisher`
+- A `geometry_metadata.parquet` file containing a serialised list of
+  `GeometryMetadata`
 - A `country_metadata.parquet` file containing a serialised list of
   `CountryMetadata` (most countries should only have one entry, but there may be
   more than one because of entities like the UK)
@@ -63,17 +66,20 @@ serialised as parquet files, and can be given any filename, as the
 `MetricMetadata` struct should contain the filepath to them.
 
 Likewise, geometries should be placed in the `geometries` subdirectory. Each set
-of geometries should consist of two files, with the same filename stem and
-different extensions:
-
-- `{filename}.fgb` - a FlatGeobuf file with the geoIDs stored in the `GEO_ID`
-  column
+of geometries should consist of four files, with the same filename stem and
+different extensions. The filename stem is specified inside the
+`GeometryMetadata` struct.
+
+- `{filename}.flatgeobuf` - a FlatGeobuf file with the geoIDs stored in the
+  `GEO_ID` column
+- `{filename}.geojsonseq` - corresponding GeoJSONSeq file
+- `{filename}.pmtiles` - corresponding PMTiles file
 - `{filename}.parquet` - a serialised dataframe storing the names of the
   corresponding areas. This dataframe must have:
 
   - a `GEO_ID` column which corresponds exactly to those in the FlatGeobuf file.
   - one or more other columns, whose names are
-    [lowercase ISO 639-3 chdes](https://iso639-3.sil.org/code_tables/639/data),
+    [lowercase ISO 639-3 codes](https://iso639-3.sil.org/code_tables/639/data),
     and contain the names of each region in those specified languages.
 
   For example, the parquet file corresponding to the Belgian regions (with

pyproject.toml

@@ -50,6 +50,8 @@ dependencies = [
   "jcs >=0.2.1", # For generating IDs from class attributes
 ]
 
+
+
 [project.optional-dependencies]
 test = [
   "pytest >=6",

python/popgetter/__init__.py

@@ -1,23 +1,43 @@
 from __future__ import annotations
 
+import os
+import warnings
 from collections.abc import Sequence
 from pathlib import Path
 
+from dagster import ExperimentalWarning
+
+from popgetter.io_managers.azure import (
+    AzureGeneralIOManager,
+    AzureGeoIOManager,
+    AzureMetadataIOManager,
+    AzureMetricsIOManager,
+)
+from popgetter.io_managers.local import (
+    LocalGeoIOManager,
+    LocalMetadataIOManager,
+    LocalMetricsIOManager,
+)
 from popgetter.utils import StagingDirResource
 
 __version__ = "0.1.0"
 
 __all__ = ["__version__"]
 
 
+if "IGNORE_EXPERIMENTAL_WARNINGS" in os.environ:
+    warnings.filterwarnings("ignore", category=ExperimentalWarning)
+
+
+import os
+
 from dagster import (
     AssetsDefinition,
     AssetSelection,
     Definitions,
     PipesSubprocessClient,
     SourceAsset,
     define_asset_job,
-    load_assets_from_modules,
     load_assets_from_package_module,
 )
 from dagster._core.definitions.cacheable_assets import (
@@ -33,7 +53,7 @@
     *load_assets_from_package_module(assets.us, group_name="us"),
     *load_assets_from_package_module(assets.be, group_name="be"),
     *load_assets_from_package_module(assets.uk, group_name="uk"),
-    *load_assets_from_modules([cloud_outputs], group_name="cloud_assets"),
+    *load_assets_from_package_module(cloud_outputs, group_name="cloud_outputs"),
 ]
 
 job_be: UnresolvedAssetJobDefinition = define_asset_job(
@@ -55,16 +75,37 @@
     description="Downloads UK data.",
 )
 
+resources_by_env = {
+    "prod": {
+        "metadata_io_manager": AzureMetadataIOManager(),
+        "geometry_io_manager": AzureGeoIOManager(),
+        "metrics_io_manager": AzureMetricsIOManager(),
+    },
+    "dev": {
+        "metadata_io_manager": LocalMetadataIOManager(),
+        "geometry_io_manager": LocalGeoIOManager(),
+        "metrics_io_manager": LocalMetricsIOManager(),
+    },
+}
+
+resources = {
+    "pipes_subprocess_client": PipesSubprocessClient(),
+    "staging_res": StagingDirResource(
+        staging_dir=str(Path(__file__).parent.joinpath("staging_dir").resolve())
+    ),
+    "azure_general_io_manager": AzureGeneralIOManager(".bin"),
+}
+
+resources.update(resources_by_env[os.getenv("ENV", "dev")])
 
 defs: Definitions = Definitions(
     assets=all_assets,
     schedules=[],
-    sensors=[cloud_outputs.country_outputs_sensor],
-    resources={
-        "pipes_subprocess_client": PipesSubprocessClient(),
-        "staging_res": StagingDirResource(
-            staging_dir=str(Path(__file__).parent.joinpath("staging_dir").resolve())
-        ),
-    },
+    sensors=[
+        cloud_outputs.metadata_sensor,
+        cloud_outputs.geometry_sensor,
+        cloud_outputs.metrics_sensor,
+    ],
+    resources=resources,
     jobs=[job_be, job_us, job_uk],
 )

python/popgetter/assets/be/__init__.py

@@ -1,28 +1,11 @@
 #!/usr/bin/python3
 from __future__ import annotations
 
-from dagster import (
-    asset,
-)
-
-from popgetter.metadata import (
-    CountryMetadata,
-)
-
 from . import (
     census_derived,  # noqa: F401
     census_geometry,  # noqa: F401
     census_tables,  # noqa: F401
 )
-from .belgium import asset_prefix, country
-
-
-@asset(key_prefix=asset_prefix)
-def get_country_metadata() -> CountryMetadata:
-    """
-    Returns a CountryMetadata of metadata about the country.
-    """
-    return country
 
 
 # @asset(key_prefix=asset_prefix)