File/address validation

[barneydobson]

There's a lot of os.path.join - file addresses should be defined and validated in one place

[barneydobson]

• Use pathlib not OS.
• Should validate that a file exists at some point. I guess at two main points, after the downloaders are called in prepare_data, and after the preprocessing functions are called in pre_processing.
• Should graph functions define their file dependencies somewhere?
• Are addresses handled in the same way as the old repo? I.e., after validation, everything gets stored in a dict?

[barneydobson]

@dalonsoa I think the #10 has made pretty much cleared what i need to know to start on graph functions. We don't have to solve all of this now, but I wanted to check if I could still make the assumption that the graph functions are receiving paths as args in the same way as before.
e.g.,

@register_graphfcn
def double_directed(G, river_buffer_distance, address_to_some_raster: str | Path, **kwargs):
    print(river_buffer_distance)
    return G

[dalonsoa]

I would say so, given that it is impractical in some cases to pass fully loaded data objects like dataframes and the sort.

Ideally, pass Path objects, no strings, so you can start rolling out the use of pathlib.

[barneydobson]

Sounds good - don't need to decide this now but for consistency with params could also put addresses in pydantic classes too. Though would ideally make the address relative to the user's defined base directory in some way and also work with model instances... Don't need to work this out yet though

class DownloadAddresses(BaseModel):
    dir: Path = Field(default = Path('/base_dir/bounding_box_1/downloads', description = 'folder containing downloads for a given bbox')
    elevation: Path = Field(default = Path(dir / 'elevation.tif'), description="Raw elevation data.")
    
class NationalAddresses(BaseModel):
    dir: Path = Field(default = Path('/base_dir/national_downloads', description = 'folder containing large national scale downloads')
    buildings: Path = Field(default = Path(dir / 'GBR.parquet'), description="National building datasets.")