Feed in some JSONL and get a report of the patterns present in the data.

json_shapes.py

#!/usr/bin/env python3

import csv
import json
from collections import OrderedDict
from collections import Counter

def trace(data, shape=None):
    if isinstance(data, dict):
        new_dict = OrderedDict()
        for k, v in data.items():
            new_v = trace(v)
            if new_v is not None:
                new_dict[k] = new_v
        return new_dict
    elif isinstance(data, list) and len(data) > 0:
        return list(filter(lambda e: e is not None, [trace(e) for e in data]))
    elif data is None:
        return None
    else:
        return ""

count = Counter()
examples = {}
for i, row in enumerate(csv.DictReader(open('orcids.csv'))):
    data = json.loads(row['identifier_json'])
    shape = json.dumps(trace(data), indent=2, sort_keys=True)
    examples[shape] = row['external_identifier'].strip('druid:')
    count[shape] += 1

for count, [shape, total] in enumerate(count.most_common()):
    print(f"Shape {count + 1} total={total} example=https://purl.stanford.edu/{examples[shape]}.json")
    print()
    print(shape)
    print()
    print()