An example for fetching data from the EHRI GraphQL API and converting it to TSV

history.py

#!/usr/bin/env python3

# Fetch id, name, and scope-content data for documentary units
# and write as tab-separated values.

import sys, requests, csv, json

if len(sys.argv) < 1:
    sys.stderr.write("usage: history.py <initial-api-url>\n")
    sys.exit(1)

# Fetch 50 items per request
FETCH_NUM = 50
URL = sys.argv[1]

# The actual GraphQL query. The number of items per request
# is a mandatory parameter (which we could also hard-code).
# The cursor is an optional parameter which defaults to null.
QUERY = """
query getRepositoryHistory($num: Int!, $cursor: Cursor) {
    page: repositories(first: $num, from: $cursor) {
        items {
            id
            description {
                name
                history
            }
        }
        pageInfo {
            nextPage
        }
    }
}
"""

csvwriter = csv.writer(sys.stdout, delimiter="\t", quoting=csv.QUOTE_MINIMAL)
csvwriter.writerow(["id", "name", "history"]) # header

def get_history(page_num = 1, cursor=None):

    sys.stderr.write("Fetching page: %d\n" % page_num)

    args = dict(
        query = QUERY,
        variables = dict(
            num = FETCH_NUM,
            cursor = cursor
        )
    )

    r = requests.post(URL, 
            data = json.dumps(args), 
            headers = {"Content-type": "application/json"})
    data = r.json()

    page = data["data"]["page"]

    for item in page["items"]:
        # fetch the ID and first description...
        id = item["id"]
        description = item["description"]
        if description is not None:
            name = item["description"]["name"]
            history = item["description"]["history"]
        
            if not history is None:
                csvwriter.writerow([id, name, history])

    next = page["pageInfo"]["nextPage"]
    if next is not None:
        get_history(page_num + 1, next)

get_history()