Convert UUIDs from the annotations table in Hypothesis' database to annotation IDs as seen in the API

uuids-to-hexids.py

#!/usr/bin/env python

import argparse
import base64
import binascii
import sys
import uuid

ES_FLAKE_MAGIC_BYTE = ['e', '5']


def ann_id(u):
    hexstring = uuid.UUID(u).hex
    return _get_urlsafe_from_hex(hexstring)


def _get_urlsafe_from_hex(value):
    # Validate and normalise hex string
    hexstring = uuid.UUID(hex=value).hex

    is_flake_id = (hexstring[12] == ES_FLAKE_MAGIC_BYTE[0] and
                   hexstring[16] == ES_FLAKE_MAGIC_BYTE[1])

    if is_flake_id:
        # The hex representation of the flake ID is simply the UUID without the
        # two magic nibbles.
        data = binascii.unhexlify(hexstring[0:12] +
                                  hexstring[13:16] +
                                  hexstring[17:32])
        return base64.urlsafe_b64encode(data)

    # Encode UUID bytes and strip two bytes of padding
    data = binascii.unhexlify(hexstring)
    return base64.urlsafe_b64encode(data)[:-2]


if __name__ == '__main__':
    p = argparse.ArgumentParser()
    p.add_argument('--permalink', dest='permalink', action='store_true')
    p.add_argument('--direct-link', dest='direct_link', action='store_true')
    args = p.parse_args()

    for line in sys.stdin:
        id_ = ann_id(line.strip())

        if args.permalink:
            print('https://hypothes.is/a/{}'.format(id_))
        elif args.direct_link:
            print('https://hyp.is/{}'.format(id_))
        else:
            print(id_)