Bounding Boxes cover the planet

[marklit]

$ mkdir -p meta

$ for FILENAME in `unzip -l 2023-10-31.zip | grep meta | grep -Eo '[a-z]{2}\/.*'`; do
      unzip 2023-10-31.zip $FILENAME -d meta/
  done

$ ipython

import json
from   pathlib import Path


def get_file_type(filename):
    if 'parcels' in filename:
        return 'parcels'

    if 'buildings' in filename:
        return 'buildings'

    if 'addresses' in filename:
        return 'addresses'

    return 'Unknown'


recs = []

for filename in Path('meta').rglob('*.geojson.meta'):
    rec = json.loads(open(filename).read())

    recs.append({
        'type': 'Feature',
        'properties': {**{key: val
                          for key, val in rec.items()
                          if key != 'bounds'},
                       **{'filename': str(filename),
                          'data_type': get_file_type(str(filename))}},
        'geometry': rec['bounds']})

with open('aggregate.geojson', 'w') as f:
    f.write(json.dumps({'type': 'FeatureCollection',
                        'features': recs}))

This is just the address bounding boxes:

I found a few examples that don't look to be using ESPG:4326

$ jq -S .bounds meta/us/ne/platte-addresses-county.geojson.meta

{
  "coordinates": [
    [
      [
        -1133.0434584,
        -1744.838695
      ],
      [
        88.5864698,
        -1744.838695
      ],
      [
        88.5864698,
        715.7039681
      ],
      [
        -1133.0434584,
        715.7039681
      ],
      [
        -1133.0434584,
        -1744.838695
      ]
    ]
  ],
  "type": "Polygon"
}

There are others that look to just use whole numbers for coordinates rather than the more granular values most records have.

$ jq -S .bounds meta/au/qld/brisbane_city_council-addresses-city.geojson.meta

{
  "coordinates": [
    [
      [
        -89.9710477,
        -86.9220497
      ],
      [
        0,
        -86.9220497
      ],
      [
        0,
        153.4656792
      ],
      [
        -89.9710477,
        153.4656792
      ],
      [
        -89.9710477,
        -86.9220497
      ]
    ]
  ],
  "type": "Polygon"
}

I filtered out some of the largest bounding boxes for the address datasets and it's revealing bounding boxes that stretch from the north to the south pole.