kinto-http 10.3.0

Creating a client

The passed auth parameter is a requests authentication policy, allowing authenticating using whatever scheme fits you best.

By default, Kinto supports Firefox Accounts and Basic authentication policies.

from kinto_http import Client
credentials = ('alexis', 'p4ssw0rd')

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

It is also possible to pass the bucket and the collection to the client at creation time, so that this value will be used by default.

auth = ("token", "secret")
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")

Using a Bearer access token to authenticate (OpenID)

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")

Using FxA from a script with the email/password

from fxa.plugins.requests import FxABearerTokenAuth

auth = FxABearerTokenAuth(
    email, passwd,
    scopes=['kinto'],
    client_id="<FXA-CLIENT-ID>",
    account_server_url='https://api.accounts.firefox.com/v1',
    oauth_server_url='https://oauth.accounts.firefox.com/v1',
)
client = Client(bucket="payments", collection="receipts", auth=auth)

Getting server information

You can use the server_info method to get 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."

Handling buckets

All operations are rooted in a bucket. It makes little sense for one application to handle multiple buckets at once (but it is possible). If no specific bucket name is provided, the “default” bucket is used.

from kinto_http import Client
from kinto_http.patch_type import BasicPatch, MergePatch, JSONPatch
credentials = ('alexis', 'p4ssw0rd')

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

# To create a bucket.
client.create_bucket(id='payments')

# To get an existing bucket
bucket = client.get_bucket(id='payments')

# Or retrieve all readable buckets.
buckets = client.get_buckets()

# To create or replace an existing bucket.
client.update_bucket(id='payments', data={'description': 'My payments data.'})

# Or modify some fields in an existing bucket.
# The Kinto server supports different types of patches, which can be used from kinto_http.patch_type.
client.patch_bucket(id='payments', changes=BasicPatch({'status': 'updated'}))

# It is also possible to manipulate bucket permissions (see later)
client.patch_bucket(id='payments', changes=BasicPatch(permissions={}))

# Or delete a bucket and everything under.
client.delete_bucket(id='payment')

# Or even every writable buckets.
client.delete_buckets()

Groups

A group associates a name to a list of principals. It is useful in order to handle permissions.

# To create a group.
client.create_group(id='receipts', bucket='payments', data={'members': ['blah', 'foo']})

# Or get an existing one.
group = client.get_group(id='receipts', bucket='payments')

# Or retrieve all groups in the bucket.
groups = client.get_groups(bucket='payments')

# To create or replace an existing bucket.
client.update_group(id='receipts', bucket='payments', 'data'={'members':['foo']})

# Or modify some fields in an existing group.
# This uses the server's support for JSON patch, but any patch_type is accepted.
client.patch_group(id='receipts', bucket='payments',
    changes=JSONPatch([{'op': 'add', 'path': '/data/members/0', 'value': 'ldap:user@corp.com'}]))

# To delete an existing group.
client.delete_group(id='receipts', bucket='payments')

# Or all groups in a bucket.
client.delete_groups(bucket='payments')

Collections

A collection is where records are stored.

# To create a collection.
client.create_collection(id='receipts', bucket='payments')

# Or get an existing one.
collection = client.get_collection(id='receipts', bucket='payments')

# Or retrieve all of them inside a bucket.
collections = client.get_collections(bucket='payments')

# To create or replace an exiting collection.
client.update_collection(id='receipts', bucket='payments', data={'description':'bleeh'})

# Or modify some fields of an existing collection.
client.patch_collection(id='receipts', bucket='payments', changes=MergePatch({'status':'updated'}))

# To delete an existing collection.
client.delete_collection(id='receipts', bucket='payments')

# Or every collections in a bucket.
client.delete_collections(bucket='payments')

Records

Records can be retrieved from and saved to collections.

A record is a dict with the “permissions” and “data” keys.

 # You can pass a python dictionary to create the record.
 client.create_record(data={'status': 'done', title: 'Todo #1'},
                      collection='todos', bucket='default')

 # You can use id to specify the record id when creating it.
 client.create_record(id='todo2', data={'status': 'doing', 'title': 'Todo #2'},
                      collection='todos', bucket='default')

 # Or get an existing one by its id.
 record = client.get_record(id='todo2', collection='todos', bucket='default')

 # Or retrieve all records.
 records = client.get_records(collection='todos', bucket='default')

# Or retrieve records page by page.
for page in client.get_paginated_records(collection='todos', bucket='default'):
   # Do something with each page
   print(page)

 # Or retrieve records timestamp.
 records_timestamp = client.get_records_timestamp(collection='todos', bucket='default')

 # To replace a record using a previously fetched record
 client.update_record(data=record, collection='todos', bucket='default')

 # Or create or replace it by its id.
 client.update_record(data={'status': 'unknown'}, id='todo2', collection='todos', bucket='default')

 # Or modify some fields in an existing record.
 client.patch_record(changes=MergePatch({'assignee': 'bob'}), id='todo2', collection='todos', bucket='default')

 # To delete an existing record.
 client.delete_record(id='89881454-e4e9-4ef0-99a9-404d95900352',
                      collection='todos')

 # Or every records of a collection.
 client.delete_records(collection='todos')

Permissions

By default, authors will get read and write access to the manipulated objects. It is possible to change this behavior by passing a dict to the permissions parameter.

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

Buckets, collections and records have permissions which can be edited. For instance to give access to “leplatrem” to a specific record, you would do:

record = client.get_record(1234, collection='todos', bucket='alexis')
record['permissions']['write'].append('leplatrem')
client.update_record(record)

# During creation, it is possible to pass the permissions dict.
client.create_record(data={'foo': 'bar'}, permissions={})

Get or create

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='bucket', if_not_exists=True)

Delete

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)

Overwriting existing objects

Most of the methods take a safe argument, which defaults to True. If set to True and a last_modified field is present in the passed data, or if the if_match parameter is specified then a check will be added to the requests to ensure the record wasn’t modified on the server side in the meantime.

Batching operations

Rather than issuing a request for each and every operation, it is possible to batch the requests. The client will then issue as little requests as possible.

It is possible to do batch requests using 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()

Besides the results() method, a batch object shares all the same methods as another client.

Retry on error

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)

Pagination

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

However, 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

Generating endpoint paths

You may want to generate some endpoint paths, you can use the get_endpoint utility to do so:

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'

Handling datetime and date objects

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.

Command-line scripts

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 token:my-secret
  -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.

Moving records from one collection to another

Moving records is not directly supported by the CLI tools in this library. To do this you’ll need to install kinto-wizard.

pip install kinto-wizard

Once kinto-wizard is installed you can then run a dump command in order to get all the records from one collection in a file.

kinto-wizard dump \
   --full --server http://{kinto_url}:8888/v1 \
   --bucket {bucket_name} \
   --collection {collection_name} \
   --auth {user}:{password} \
   > {path_to_file.yaml}

This new local file has a description of the specified collection and its records. Now you can change the file in order to reflect the modifications you need. If you are moving items into a new collection you’ll have to search for the collection name in the YAML file and change it, do not forget to change the id field of that collection too.

After running the first dump command you might want to validate the dumped records, this will ensure that you can upload them successfully using the load command.

kinto-wizard validate {path_to_file.yaml}

With this local file you can now re-upload all records into a new collection by running the following load command.

kinto-wizard load \
   --server http://{kinto_url}:8888/v1 \
   --auth {user}:{password} \
   {path_to_file.yaml}

The old collection will not be deleted automatically, so you probably would want to delete it manually.