kinto-http is the Python library to interact with a Kinto server.

There is also a similar client in JavaScript.

Use pip:

$ pip install kinto-http

Here is an overview of what the API provides:

import kinto_http

client = kinto_http.Client(server_url="http://localhost:8888/v1",
                           auth=('alexis', 'p4ssw0rd'))

records = client.get_records(bucket='default', collection='todos')
for i, record in enumerate(records):
    record['title'] = 'Todo {}'.format(i)
    client.update_record(data=record)

The passed auth parameter is a requests authentication policy.

By default, a simple tuple will become a Basic Auth authorization request header, that can authenticate users with Kinto Accounts.

import kinto_http

auth = ('alexis', 'p4ssw0rd')

client = kinto_http.Client(server_url='http://localhost:8888/v1',
                           auth=auth)

It is also possible to pass a bucket ID and/or collection ID to set them as default values for the parameters of the client operations.

client = Client(bucket="payments", collection="receipts", auth=auth)

After creating a client, you can also replicate an existing one and overwrite some key arguments.

client2 = client.clone(collection="orders")

An asynchronous client is also available. It has all the same endpoints as the sync client except for the batch operations.

from kinto_http import AsyncClient

auth = ('alexis', 'p4ssw0rd')

client = AsyncClient(server_url='http://localhost:8888/v1', auth=auth)
info = await client.server_info()
assert 'schema' in info['capabilities'], "Server doesn't support schema validation."

The dry_mode parameter can be set to simulate requests without actually sending them over the network. When enabled, dry mode ensures that no external calls are made, making it useful for testing or debugging. Instead of performing real HTTP operations, the client logs the requests with DEBUG level.

import kinto_http

client = kinto_http.Client(auth=kinto_http.BearerTokenAuth("XYPJTNsFKV2"))

The authorization header is prefixed with Bearer by default. If the header_type is customized on the server, the client must specify the expected type: kinto_http.BearerTokenAuth("XYPJTNsFKV2", type="Bearer+OIDC")

import kinto_http

client = kinto_http.Client(server_url='http://localhost:8888/v1', auth=kinto_http.BrowserOAuth())

The client will open a browser page and will catch the Bearer token obtained after the OAuth dance.

Custom headers can be specified in the Client constructor, and will be sent in every request:

import kinto_http

client = kinto_http.Client(server_url="http://server/v1", headers={
    "Allow-Access": "CDN",
    "User-Agent": "blocklist-updater"
})

You can use the server_info() method to fetch the server information:

from kinto_http import Client

client = Client(server_url='http://localhost:8888/v1')
info = client.server_info()
assert 'schema' in info['capabilities'], "Server doesn't support schema validation."

• get_bucket(id=None, **kwargs): retrieve single bucket
• get_buckets(**kwargs): retrieve all readable buckets
• create_bucket(id=None, data=None, **kwargs): create a bucket
• update_bucket(id=None, data=None, **kwargs): create or replace an existing bucket
• patch_bucket(id=None, changes=None, **kwargs): modify some fields in an existing bucket
• delete_bucket(id=None, **kwargs): delete a bucket and everything under it
• delete_buckets(**kwargs): delete every writable buckets

• get_group(id=None, bucket=None, **kwargs): retrieve single group
• get_groups(bucket=None, **kwargs): retrieve all readable groups
• create_group(id=None, data=None, bucket=None, **kwargs): create a group
• update_group(id=None, data=None, bucket=None, **kwargs): create or replace an existing group
• patch_group(id=None, changes=None, bucket=None, **kwargs): modify some fields in an existing group
• delete_group(id=None, bucket=None, **kwargs): delete a group and everything under it
• delete_groups(bucket=None, **kwargs): delete every writable groups

• get_collection(id=None, bucket=None, **kwargs): retrieve single collection
• get_collections(bucket=None, **kwargs): retrieve all readable collections
• create_collection(id=None, data=None, bucket=None, **kwargs): create a collection
• update_collection(id=None, data=None, bucket=None, **kwargs): create or replace an existing collection
• patch_collection(id=None, changes=None, bucket=None, **kwargs): modify some fields in an existing collection
• delete_collection(id=None, bucket=None, **kwargs): delete a collection and everything under it
• delete_collections(bucket=None, **kwargs): delete every writable collections

• get_record(id=None, bucket=None, collection=None, **kwargs): retrieve single record
• get_records(bucket=None, collection=None, **kwargs): retrieve all readable records
• get_paginated_records(bucket=None, collection=None, **kwargs): paginated list of records
• get_records_timestamp(bucket=None, collection=None, **kwargs): return the records timestamp of this collection
• create_record(id=None, data=None, bucket=None, collection=None, **kwargs): create a record
• update_record(id=None, data=None, bucket=None, collection=None, **kwargs): create or replace an existing record
• patch_record(id=None, changes=None, bucket=None, collection=None, **kwargs): modify some fields in an existing record
• delete_record(id=None, bucket=None, collection=None, **kwargs): delete a record and everything under it
• delete_records(bucket=None, collection=None, **kwargs): delete every writable records

The objects permissions can be specified or modified by passing a permissions to create_*(), patch_*(), or update_*() methods:

client.create_record(data={'foo': 'bar'},
                     permissions={'read': ['group:groupid']})


record = client.get_record('123', collection='todos', bucket='alexis')
record['permissions']['write'].append('leplatrem')
client.update_record(data=record)

In order to obtain a list of all the permissions, on every object, use the get_permissions() method:

all_perms = client.get_permissions(exclude_resource_names=("record",))

has_collection_perms = any(
    p for p in all_perms
    if p["collection_id"] == "my-collection"
    and "write" in p["permissions"]
)

In some cases, you might want to create a bucket, collection, group or record only if it doesn't exist already. To do so, you can pass the if_not_exists=True to the create_*() methods:

client.create_bucket(id='blog', if_not_exists=True)
client.create_collection(id='articles', bucket='blog', if_not_exists=True)

In some cases, you might want to delete a bucket, collection, group or record only if it exists already. To do so, you can pass the if_exists=True to the delete_* methods:

client.delete_bucket(id='bucket', if_exists=True)

The .patch_*() operations receive a changes parameter.

from kinto_http.patch_type import BasicPatch, MergePatch, JSONPatch


client.patch_record(id='abc', changes=BasicPatch({'over': 'write'}))

client.patch_record(id='todo', changes=MergePatch({'assignee': 'bob'}))

client.patch_record(id='receipts', changes=JSONPatch([
    {'op': 'add', 'path': '/data/members/0', 'value': 'ldap:[email protected]'}
]))

The create_*(), patch_*(), and update_*() methods take a safe argument (default: True).

If True, the client will ensure that the object doesn't exist already for creations, or wasn't modified on the server side since we fetched it. The timestamp will be implicitly read from the last_modified field in the passed data object, or taken explicitly from the if_match parameter.

Rather than issuing a request for each and every operation, it is possible to batch several operations in one request (sync client only).

Using the batch() method as a Python context manager (with):

with client.batch() as batch:
    for idx in range(0, 100):
        batch.update_record(data={'id': idx})

Reading data from batch operations is achieved by using the results() method available after a batch context is closed.

with client.batch() as batch:
    batch.get_record('r1')
    batch.get_record('r2')
    batch.get_record('r3')

r1, r2, r3 = batch.results()

Failing operations will raise a KintoException, which has request and response attributes.

try:
    client.create_group(id="friends")
except kinto_http.KintoException as e:
    if e.response and e.response.status_code == 403:
        print("Not allowed!")

A timeout value in seconds can be specified in the client constructor:

client = KintoClient(server_url="...", timeout=5)

To distinguish the connect from the read timeout, use a tuple:

client = KintoClient(server_url="...", timeout=(3.05, 27))

For an infinit timeout, use None:

client = KintoClient(server_url="...", timeout=None)

See the timeout documentation of the underlying requests library.

When the server is throttled (under heavy load or maintenance) it can return error responses.

The client can hence retry to send the same request until it succeeds. To enable this, specify the number of retries on the client:

client = Client(server_url='http://localhost:8888/v1',
                auth=credentials,
                retry=10)

The Kinto protocol lets the server define the duration in seconds between retries. It is possible (but not recommended) to force this value in the clients:

client = Client(server_url='http://localhost:8888/v1',
                auth=credentials,
                retry=10,
                retry_after=5)

When the server responses are paginated, the client will download every page and merge them transparently.

The get_paginated_records() method returns a generator that will yield each page:

for page in client.get_paginated_records():
    records = page["data"]

It is possible to specify a limit for the number of items to be retrieved in one page:

records = client.get_records(_limit=10)

In order to retrieve every available pages with a limited number of items in each of them, you can specify the number of pages:

records = client.get_records(_limit=10, pages=float('inf'))  # Infinity

If the built-in history plugin is enabled, it is possible to retrieve the history of changes:

# Get the complete history of a bucket
changes = client.get_history(bucket='default')

# and optionally use filters
hist = client.get_history(bucket='default', _limit=2, _sort='-last_modified', _since='1533762576015')
hist = client.get_history(bucket='default', resource_name='collection')

The history of a bucket can also be purged with:

client.purge_history(bucket='default')

If the kinto-attachment plugin is enabled, it is possible to fetch, add, or remove attachments on records:

filepath = client.download_attachment(record_obj)

client.add_attachment(id="record-id", filepath="/path/to/image.png")

Or remove them:

client.remove_attachment(id="record-id")

The get_endpoint() method returns an endpoint URL on the server:

client = Client(server_url='http://localhost:8888/v1',
                auth=('token', 'your-token'),
                bucket="payments",
                collection="receipts")

print(client.get_endpoint("record",
                          id="c6894b2c-1856-11e6-9415-3c970ede22b0"))

# '/buckets/payments/collections/receipts/records/c6894b2c-1856-11e6-9415-3c970ede22b0'

In addition to the data types supported by JSON, kinto-http.py also supports native Python date and datetime objects.

In case a payload contain a date or a datetime object, kinto-http.py will encode it as an ISO formatted string.

Please note that this transformation is only one-way. While reading a record, if a string contains a ISO formated string, kinto-http.py will not convert it to a native Python date or datetime object.

If you know that a field will be a datetime, you might consider encoding it yourself to be more explicit about it being a string for Kinto.

In order to have common arguments and options for scripts, some utilities are provided to ease configuration and initialization of client from command-line arguments.

import argparse
import logging

from kinto_http import cli_utils

logger = logging.getLogger(__name__)

if __name__ == "__main__":
    parser = argparse.ArgumentParser(description="Download records")
    cli_utils.set_parser_server_options(parser)

    args = parser.parse_args()

    cli_utils.setup_logger(logger, args)

    logger.debug("Instantiate Kinto client.")
    client = cli_utils.create_client_from_args(args)

    logger.info("Fetch records.")
    records = client.get_records()
    logger.warn("{} records.".format(len(records)))

The script now accepts basic options:

$ python example.py --help

usage: example.py [-h] [-s SERVER] [-a AUTH] [-b BUCKET] [-c COLLECTION] [-v]
                  [-q] [-D]

Download records

optional arguments:
  -h, --help            show this help message and exit
  -s SERVER, --server SERVER
                        The location of the remote server (with prefix)
  -a AUTH, --auth AUTH  BasicAuth credentials: `token:my-secret` or
                        Authorization header: `Bearer token`
  -b BUCKET, --bucket BUCKET
                        Bucket name.
  -c COLLECTION, --collection COLLECTION
                        Collection name.
  --retry RETRY         Number of retries when a request fails
  --retry-after RETRY_AFTER
                        Delay in seconds between retries when requests fail
                        (default: provided by server)
  -v, --verbose         Show all messages.
  -q, --quiet           Show only critical errors.
  -D, --debug           Show all messages, including debug messages.

See contributing docs