validata-geojson-stream

validata-geojson-stream.py

import collections
import json

import json_stream
import jsonschema

from jsonschema import Draft7Validator


with open("schema_amenagements_cyclables.json") as ifile:
    schema = json.load(ifile)
    # get schema for one item
    subschema = schema["properties"]["features"]["items"]["properties"]["properties"]

errors = collections.defaultdict(int)

with open("france-20220224.geojson") as f:
    data = json_stream.load(f)
    count = 0
    v = Draft7Validator(subschema)
    for feature in data["features"].persistent():
        count += 1
        print(count)
        props = dict(feature["properties"])
        for e in v.iter_errors(props):
            errors[e.message] += 1

print(errors)