Minimalist Hypothesis export script

hypothesis-export.py

import getpass
import json
import re
import sys

import requests

GROUP_URL_PATTERN = "https://hypothes.is/groups/([^/]+)/.*"

api_token = getpass.getpass(
    "Enter developer token (from https://hypothes.is/account/developer): "
)

group_url = input("Enter group URL: ")
group_id_match = re.match(GROUP_URL_PATTERN, group_url)
if not group_id_match:
    print(f"{group_url} is not a valid group URL", file=sys.stderr)
    sys.exit(1)
group_id = group_id_match.group(1)

search_api = "https://hypothes.is/api/search"
output_file = "annotations.json"

search_after = None
annotations = []
page_size = 200
total = None

while True:
    rsp = requests.get(
        search_api,
        params={
            "group": group_id,
            "limit": page_size,
            "search_after": search_after,
            "order": "asc",
            "sort": "created",
        },
        headers={"Authorization": f"Bearer {api_token}"},
    )
    rsp.raise_for_status()
    data = rsp.json()
    rows = data["rows"]
    total = data["total"]
    if not rows:
        break
    annotations += rows
    search_after = rows[-1]["created"]
    print(f"Fetched {len(annotations)} of {total} annotations...", file=sys.stderr)

with open(output_file, "w") as fp:
    json.dump(annotations, fp, indent=2, sort_keys=True)