gistfile1.txt

#!/usr/bin/env python3
"""Script to generate main summary test schema."""

import json


PROBES_FILE = "histograms_probes.json"

probe_lists = {"parent": [], "content": []}
with open(PROBES_FILE) as probes_file:
    probes = json.load(probes_file)
    for probe, processes in probes.items():
        for process in processes:
            probe_lists[process].append(
                {"type": "STRING", "name": probe, "mode": "NULLABLE"}
            )

json_schema_template = [
    {"type": "TIMESTAMP", "name": "submission_timestamp", "mode": "REQUIRED"},
    {"type": "INTEGER", "name": "sample_id", "mode": "REQUIRED"},
    {"type": "STRING", "name": "document_id", "mode": "NULLABLE"},
    {"type": "STRING", "name": "client_id", "mode": "NULLABLE"},
    {"type": "STRING", "name": "normalized_channel", "mode": "NULLABLE"},
    {"type": "STRING", "name": "normalized_os", "mode": "NULLABLE"},
    {
        "fields": [
            {"type": "STRING", "name": "version", "mode": "NULLABLE"},
            {"type": "STRING", "name": "build_id", "mode": "NULLABLE"},
        ],
        "type": "RECORD",
        "name": "application",
        "mode": "NULLABLE",
    },
    {
        "fields": [
            {
                "fields": [
                    {
                        "fields": [
                            {
                                "fields": probe_lists["content"],
                                "type": "RECORD",
                                "name": "histograms",
                                "mode": "NULLABLE",
                            }
                        ],
                        "type": "RECORD",
                        "name": "content",
                        "mode": "NULLABLE",
                    }
                ],
                "type": "RECORD",
                "name": "processes",
                "mode": "NULLABLE",
            },
            {
                "fields": probe_lists["parent"],
                "type": "RECORD",
                "name": "histograms",
                "mode": "NULLABLE",
            },
        ],
        "type": "RECORD",
        "name": "payload",
        "mode": "NULLABLE",
    },
]

print(json.dumps(json_schema_template, indent=4, sort_keys=True))