Toy script to scrape specfic collection of imagery

README

This toy script was put together to assist in trawling some imagery and building a small catalog.

trawl.py

"""
A basic script to trawl an imagery archie and build a custom catalog.
"""

from pathlib import Path
import click
import rasterio  # type: ignore
import pandas  # type: ignore
import geopandas  # type: ignore
from shapely.geometry import Polygon  # type: ignore

import structlog

_LOG = structlog.get_logger()


def image_info(pathname) -> geopandas.GeoDataFrame:
    """
    Open an image and retrieve the shape extents.
    Not getting the extents via the affine transformation as this
    data is not northup (has x and y rotations).
    """
    with rasterio.open(pathname) as src:
        # forming a polygon using UL, UR, LR, LL coordinates
        polygon = Polygon(
            [
                src.xy(0, 0),
                src.xy(0, src.shape[1]),
                src.xy(*src.shape),
                src.xy(src.shape[0], 0),
            ]
        )

        data = {
            "pathname": [str(pathname)],
            "vessel_side": [pathname.stem.strip("_")[-1]],  # ends in P or S
            "geometry": [polygon],
        }
        gdf = geopandas.GeoDataFrame(data, crs=src.crs)

    return gdf


@click.command()
@click.option(
    "--rootdir",
    type=click.Path(exists=True, readable=True),
    help="The root input directory",
)
@click.option(
    "--outdir",
    type=click.Path(exists=True, writable=True),
    help="The base output directory to contain the output file",
)
@click.option("--pattern", default="*.tif", help="The filename pattern to search for.")
def main(rootdir: str, outdir: str, pattern: str = "*.tif") -> None:
    """Find the imagery, and build a mini catalog as a vector file."""
    catalog = []
    for file in Path(rootdir).rglob(pattern):
        # we'll store the full pathname, not relative, nor basename
        pname = file.absolute()

        _LOG.info("processing datafile", pathname=str(pname))
        catalog.append(image_info(pname))

    _LOG.info("concatenating individual dataframes")
    gdf = pandas.concat(catalog)

    out_pathname = Path(outdir).joinpath("data-catalog.gpkg")
    _LOG.info("writing catalog as a GeoPackage", out_pathname=out_pathname)
    gdf.to_file(out_pathname, driver="GPKG")

    _LOG.info("finished")


if __name__ == "__main__":
    main()