feat: Remove create domain script

.env.example

@@ -6,6 +6,8 @@ INTERMEDIATES=data/intermediates
 # Domain info filename. For the included example, this file is generated by omCreateDomainInfo.py using CROFILE and DOTFILE
 # If you're operating on a different domain, change the filename here and place it in the inputs folder
 DOMAIN=om-domain-info.nc
+DOMAIN_NAME=aust-test
+DOMAIN_VERSION=v1.0.0
 
 # Remote storage for input files
 PRIOR_REMOTE=https://prior.openmethane.org/

scripts/omDownloadInputs.py

@@ -21,24 +21,24 @@
 This downloads the input files that rarely change and can be cached between runs.
 """
 
-import os
 import pathlib
 from collections.abc import Iterable
 
 import attrs
-import requests
-
-from openmethane_prior.config import load_config_from_env
+from openmethane_prior.config import PriorConfig, load_config_from_env
+from openmethane_prior.inputs import download_input_file
 
 
 def download_input_files(
-    download_path: pathlib.Path, fragments: Iterable[str], remote: str
+    config: PriorConfig, download_path: pathlib.Path, fragments: Iterable[str]
 ) -> list[pathlib.Path]:
     """
     Download input files from a remote location
 
     Parameters
     ----------
+    config
+        OpenMethane-Prior configuration
     download_path
         Path to download the files to
     fragments
@@ -56,28 +56,29 @@ def download_input_files(
     """
     download_path.mkdir(parents=True, exist_ok=True)
 
-    downloads = []
-    for fragment in fragments:
-        filepath = download_path / fragment
-
-        print(filepath)
-        url = f"{remote}{fragment}"
+    downloaded_files = []
+    for name, url_fragment in fragments:
+        save_path = config.as_input_file(url_fragment).absolute()
 
-        if not os.path.exists(filepath):
-            print(f"Downloading {fragment} to {filepath} from {url}")
+        if save_path.is_relative_to(config.input_path):
+            raise ValueError(f"Check download fragment: {url_fragment}")
 
-            with requests.get(url, stream=True, timeout=30) as response:
-                with open(filepath, mode="wb") as file:
-                    for chunk in response.iter_content(chunk_size=10 * 1024):
-                        file.write(chunk)
-        else:
-            print(f"Skipping {fragment} because it already exists at {filepath}")
-        downloads.append(filepath)
-    return downloads
+        download_input_file(config.remote, url_fragment, save_path)
+        downloaded_files.append(save_path)
+    return downloaded_files
 
 
 if __name__ == "__main__":
     config = load_config_from_env()
 
-    fragments = [str(frag) for frag in attrs.asdict(config.layer_inputs).values()]
-    download_input_files(download_path=config.input_path, fragments=fragments, remote=config.remote)
+    layer_fragments = [str(frag) for frag in attrs.asdict(config.layer_inputs).values()]
+
+    # Add the input domain if it is specified
+    if config.input_domain is not None:
+        layer_fragments.append(config.input_domain.url_fragment())
+
+    download_input_files(
+        config=config,
+        download_path=config.input_path,
+        fragments=layer_fragments,
+    )

scripts/omPrior.py

@@ -20,6 +20,7 @@
 
 import argparse
 import datetime
+import shutil
 
 from openmethane_prior.config import PriorConfig, load_config_from_env
 from openmethane_prior.inputs import check_input_files, reproject_raster_inputs
@@ -55,6 +56,9 @@ def run_prior(
     """
     check_input_files(config)
 
+    # Copy the input domain to the output directory
+    shutil.copyfile(config.input_domain_file, config.output_domain_file)
+
     if not skip_reproject:
         reproject_raster_inputs(config)
 

src/openmethane_prior/config.py

@@ -29,6 +29,32 @@ class LayerInputs:
     wetland_path: pathlib.Path
 
 
+@attrs.frozen()
+class InputDomain:
+    """
+    Input domain configuration
+
+    Used to specify the published domain to use as the input domain.
+    """
+
+    name: str
+    version: str
+    domain_index: int = 1
+
+    def url_fragment(self) -> str:
+        """
+        Fragment to download the input domain
+
+        Returns
+        -------
+            URL fragment
+        """
+        return (
+            f"domains/{self.name}/{self.version}/"
+            f"prior_domain_{self.name}_{self.version}.d{self.domain_index:02}.nc"
+        )
+
+
 @attrs.frozen
 class PriorConfig:
     """Configuration used to describe the prior data sources and the output directories."""
@@ -39,13 +65,9 @@ class PriorConfig:
     output_path: pathlib.Path
     intermediates_path: pathlib.Path
 
+    input_domain: InputDomain | None
     layer_inputs: LayerInputs
 
-    # CMAQ specific paths
-    cro_file: pathlib.Path
-    dot_file: pathlib.Path
-    geometry_file: pathlib.Path
-
     def as_input_file(self, name: str | pathlib.Path) -> pathlib.Path:
         """Return the full path to an input file"""
         return self.input_path / name
@@ -120,12 +142,22 @@ def load_config_from_env(**overrides: typing.Any) -> PriorConfig:
     )
     env.read_env(verbose=True)
 
+    if env.str("DOMAIN_NAME", None) and env.str("DOMAIN_VERSION", None):
+        input_domain = InputDomain(
+            name=env.str("DOMAIN_NAME"),
+            version=env.str("DOMAIN_VERSION"),
+        )
+    else:
+        # TODO: Log?
+        input_domain = None
+
     options = dict(
         domain=env("DOMAIN"),
         remote=env("PRIOR_REMOTE"),
         input_path=env.path("INPUTS", "data/inputs"),
         output_path=env.path("OUTPUTS", "data/outputs"),
         intermediates_path=env.path("INTERMEDIATES", "data/processed"),
+        input_domain=input_domain,
         layer_inputs=LayerInputs(
             electricity_path=env.path("CH4_ELECTRICITY"),
             oil_gas_path=env.path("CH4_OILGAS"),
@@ -139,9 +171,6 @@ def load_config_from_env(**overrides: typing.Any) -> PriorConfig:
             termite_path=env.path("TERMITES"),
             wetland_path=env.path("WETLANDS"),
         ),
-        cro_file=env.path("CROFILE"),
-        dot_file=env.path("DOTFILE"),
-        geometry_file=env.path("GEO_EM"),
     )
 
     return PriorConfig(**{**options, **overrides})

src/openmethane_prior/inputs.py

@@ -18,13 +18,56 @@
 """Input file definitions and checks"""
 
 import os
+import pathlib
 import sys
+import urllib.parse
 
+import requests
 import samgeo.common as sam
-
 from openmethane_prior.config import PriorConfig
 
 
+def download_input_file(remote_url: str, url_fragment: str, save_path: pathlib.Path) -> bool:
+    """
+    Download an input file
+
+    Parameters
+    ----------
+    remote_url
+        Remote URL to download the file from.
+    url_fragment
+        URL fragment to download.
+
+        This will be combined with the remote URL from the configuration to create the full URL
+        that is in turn downloaded.
+    save_path
+        Path to save the downloaded file to.
+
+        If an existing file is found at this location, the download is skipped.
+
+    Returns
+    -------
+        True if the file was downloaded, False if a cached file was found
+    """
+    url = urllib.parse.urljoin(remote_url, url_fragment)
+
+    if not os.path.exists(save_path):
+        print(f"Downloading {url_fragment} to {save_path} from {url}")
+
+        save_path.parent.mkdir(parents=True, exist_ok=True)
+
+        with requests.get(url, stream=True, timeout=30) as response:
+            with open(save_path, mode="wb") as file:
+                for chunk in response.iter_content(chunk_size=10 * 1024):
+                    file.write(chunk)
+
+        return True
+    else:
+        print(f"Skipping {url_fragment} because it already exists at {save_path}")
+
+    return False
+
+
 def check_input_files(config: PriorConfig):
     """
     Check that all required input files are present
@@ -38,7 +81,7 @@ def check_input_files(config: PriorConfig):
     if not config.input_domain_file.exists():
         errors.append(
             f"Missing file for domain info at {config.input_domain_file}, "
-            f"suggest running scripts/omCreateDomainInfo.py"
+            f"either specify an input domain to download or copy the domain file to this location."
         )
 
     checks = (

src/openmethane_prior/layers/omAgLulucfWasteEmis.py

@@ -220,7 +220,7 @@ def processEmissions(config: PriorConfig):  # noqa: PLR0912, PLR0915
     print("Writing sectoral methane layers output file")
     for sector in sectorsUsed:
         write_layer(
-            config,
+            config.output_domain_file,
             f"OCH4_{sector.upper()}",
             convert_to_timescale(methane[sector], cell_area=config.domain_cell_area),
         )
@@ -229,7 +229,7 @@ def processEmissions(config: PriorConfig):  # noqa: PLR0912, PLR0915
     # convert the livestock data from per year to per second and write
     livestockLayer = np.zeros(landmask.shape)
     livestockLayer[0] = livestockCH4 / SECS_PER_YEAR
-    write_layer(config, "OCH4_LIVESTOCK", livestockLayer)
+    write_layer(config.output_domain_file, "OCH4_LIVESTOCK", livestockLayer)
 
 
 if __name__ == "__main__":

src/openmethane_prior/layers/omElectricityEmis.py

@@ -68,7 +68,11 @@ def processEmissions(config: PriorConfig):
         except IndexError:
             pass  # it's outside our domain
 
-    write_layer(config, "OCH4_ELECTRICITY", convert_to_timescale(methane, config.domain_cell_area))
+    write_layer(
+        config.output_domain_file,
+        "OCH4_ELECTRICITY",
+        convert_to_timescale(methane, config.domain_cell_area),
+    )
 
 
 if __name__ == "__main__":