RelationalAI SDK for Python

This guide presents the API for the RelationalAI SDK for Python.

This guide presents the main features of the RelationalAI SDK for Python, used to interact with the Relational Knowledge Graph Management System (RKGMS).

The rai-sdk package is open source and is available in the Github repository RelationalAI/rai-sdk-python. It includes self-contained examples of the main API functionality. We welcome contributions and pull requests.

Note: This guide applies to rai-sdk, the latest iteration of the Python SDK. The relationalai-sdk package is deprecated.

The last section describes how the different Console notebook cell types can be implemented with this Python SDK API.


Python3 is required to interact with the RKGMS using the RelationalAI SDK for Python.


The RelationalAI SDK for Python is a standalone client library and can be installed using the poetry or pip Python package managers as shown below:

# using the poetry package manager
poetry add rai-sdk
# using the pip package manager
pip install rai-sdk

The Configuration File

To connect to the RAI Server using the Python SDK, you need a configuration file (~/.rai/config) with RAICredentials — an ID and secret key provided to you by an admin. The procedure to get these RAICredentials is described below:

  1. The Organization/Team/Person requires an account with RAI. Once the account is set up, an admin user is assigned to the account.
  2. This admin user can use the RAI Console to create OAuth Clients. Each client is identified by a Client ID and a secret keystring.
  3. The SDK users add the client_id and client_secret to their ~/.rai/config file, which should look like this:
region = us-east
host =
port = 443
client_id = ################ [from admin]
client_secret = ############## [from admin]

Note that in the place of ######### the actual credentials of the user should appear. This config file should be placed in the ~/.rai folder (~/.rai/config).

See the SDK README for more details.

If you get “local issuer certificate” errors you may need to install local certificates for your Python.

Multiple Profiles (Optional)

You can have other profiles besides [default] in separate sections of the config file. This is useful, for instance, if you have configurations for different regions.

The Python API function takes the configuration file and the profile name as optional arguments:

from railib import config
cfg ="~/.rai/config", profile = "default")


Each OAuth Client has its own set of permissions, which determine what operations it can execute. Depending on the permissions, some operations (such as listing other users, or creating or deleting compute engines) might fail.

Creating a Context

Most API operations use a context object, constructed with railib.api.Context. To use the default profile in your ~/.rai/config file, use:

from railib import api, config
cfg =
# to specify a non-default profile instead:
# cfg ='myprofile')
ctx = api.Context(**cfg)

As an alternative to loading a configuration and using **cfg, you can also specify the different fields directly using keyword arguments in api.Context. See help(api.Context) for details.

You can test your configuration and context by running:


This should return a list with database info, assuming your keys have the corresponding permissions (see Listing Databases below).

User Management

A client with the right permissions can list the users under the account:

Listing Users

rsp = api.list_users(ctx)
print(json.dumps(rsp, indent=2))

Retrieving User Details

import json
rsp = api.get_user(ctx, user)
print(json.dumps(rsp, indent=2))

Here, user is a string ID, e.g. "auth0|XXXXXXXXXXXXXXXXXX".

Managing Engines

To query and update RAICloud databases, you will need a running engine. The following API calls create and manage them:

Creating an Engine

You can create a new engine as follows:

from railib import api, config, show
import json
from urllib.request import HTTPError
cfg =
ctx = api.Context(**cfg)

    response = api.create_engine(ctx, engine_name, size)
    print(json.dumps(response, indent=2))
except HTTPError as e:

API requests return a JSON value, represented in Python as a Dict, which can be converted to a string with json.dumps and then printed, as above. Here is a sample result for api.create_engine:

{'engine': {
    'account_name': '#########',
    'created_by': '#########',
    'id': '#########',
    'name': 'python_sdk_engine',
    'region': '#########',
    'requested_on': '2021-08-17T15:32:47.371Z',
    'size': 'XS',
    'state': 'REQUESTED'

Valid sizes are given by the api.EngineSize enum:

  • XS (extra small)
  • S (small)
  • M (medium)
  • L (large)
  • XL (extra large)

Note: It may take some time before your engine is in the “PROVISIONED” state, where it is ready for queries. It will be in the “PROVISIONING” state before that.

To list all the engines that are currently provisioned, we can use api.list_engines:

from railib import api
import json
print(json.dumps(api.list_engines(ctx, "PROVISIONED"), indent=2))

If there is an error with the request, an urllib.request.HTTPError exception will be thrown.

Most of the API examples below assume we have a running engine (in “PROVISIONED” state) in the engine variable, and a test database in database:

# replace by your values for testing:
database = "mydatabase"
engine = "MYENGINE"

Deleting an Engine

An engine can be deleted with:

from railib import api, show
from urllib.request import HTTPError
    rsp = api.delete_engine(ctx, engine)
except HTTPError as e:

If successful, this will return:

    {'name': XXXXXX
    'state': 'DELETING',
    'message': 'engine XXXX deleted successfully'}

Note that since RAICloud decouples computation from storage, deleting an engine does not delete any cloud databases. See Managing Engines for more details.

Getting Info for an Engine

The details for a specific compute engine can be retrieved with api.get_engine:

api.get_engine(ctx, engine)

This will return an empty list if the engine does not exist.

Managing Databases

Creating a Database

You can create a database with api.create_database, as follows:

from railib import api, config, show
import json
# Create the database, but not overwrite it if it exists:
database = "mydatabase" # adjust as needed
engine = "MYENGINE"
cfg =
ctx = api.Context(**cfg)
    rsp = api.create_database(ctx, database, engine, overwrite=False)
    print(json.dumps(rsp, indent=2))
except HTTPError as e:

The result from a successful create_database call will look like this:

  "output": [],
  "version": 2,
  "problems": [],
  "actions": [],
  "debug_level": 0,
  "aborted": false,
  "type": "TransactionResult"

A failed call will have a non-empty problems list.

If you want to overwrite an existing database, you can use the overwrite = True option. This will erase all the data in the old database.

Cloning a Database

The call api.create_database can be used to clone a database by specifying a source argument:

from railib import api, config
import json
# Clone the database, but not overwrite the target if it exists:
cfg =
ctx = api.Context(**cfg)
from urllib.request import HTTPError
    rsp = api.create_database(ctx, "MYDBCLONE", engine, overwrite=False,
    print(json.dumps(rsp, indent=2))
except HTTPError as e:

With this API call, you can clone a database from database to "MYDBCLONE", creating an identical copy. Any subsequent changes to either database will not affect the other. Cloning a database only succeeds if the source database already exists.

If you want to clone into an existing database, overwriting it in the process, use the overwrite = True flag. This will erase all the data in the clone target database.

Retrieving Database Details

rsp = api.get_database(ctx, databasename)
print(json.dumps(rsp, indent=2))

The response is a JSON object. If the database does not exist, an [] will be returned.

Note that this call does not require a running engine.

Listing Databases

Using api.list_databases will list the databases available to the account:

rsp = api.list_databases(ctx)
# rsp = api.list_databases(ctx, state)
print(json.dumps(rsp, indent=2))

The optional variable state (default: None) can be used to filter databases by state. For example, “CREATED”, “CREATING”, or “CREATION_FAILED”.

Deleting Databases

A database can be deleted with api.delete_database, if the config has the right permissions.

The database is identified by its name (as used in create_database).

from railib import api, show
from urllib.request import HTTPError
    rsp = api.delete_database(ctx, database_name)
except HTTPError as e:

If successful, the response will be of the form

{'name': 'XXXXXXX', 'message': 'deleted successfully'}

Rel Model Sources

Rel model sources are collections of Rel code that can be added, updated, or deleted from a particular database. Since they update a database, a running engine is required to perform operations on sources.

Installing Rel Model Sources

The api.install_source function installs a Rel source in a given database. The last argument is a Python Dictionary, mapping names to sources, so that more than one named source can be installed.

For example, to add a Rel source code file to a database:

def source_string = """
def countries = {"United States of America"; "Germany"; "Japan"; "Greece"}
def oceans = {"Arctic"; "Atlantic"; "Indian"; "Pacific"; "Southern"}

api.install_source(ctx, database, engine, {"mysource" : sourcestring})

If the database already contains an installed source with the same given name, then it is replaced by the new source.

If you need to install from a file, read it into a string first. For example:

from os import path
sources = {}
with open(fname) as fp:
    sources[path.basename(fname)] =  # source name => source
api.install_source(ctx, database, engine, sources)

Deleting a Rel Model Source

You can delete a source from a database using the following code:

rsp = api.delete_source(ctx, database, engine, "sourcename")

Listing Installed Rel Model Sources

You can list the sources in a database using the following code:

api.list_sources(ctx, database, engine)

This returns a JSON array of names.

To see the contents of a named source, use:

api.get_source(ctx, database, engine, sourcename)

where sourcename is a string.

Querying a Database

A query is a high-level API call to run a single query/transaction against the database. It specifies a Rel program (which can be empty), and a set of output relations to be returned.

api.query(ctx: railib.api.Context, database: str, engine: str, command: str,
    inputs: dict = None,
    readonly: bool = True) -> dict

For example:

>>> rsp = api.query(ctx, database, engine, "def output[x in {1;2;3}] = x * 2")
>>> show.results(rsp)
// output (Int64)
1, 2;
2, 4;
3, 6

By convention, only the output relation is returned (as in Console notebooks), and readonly is True.

Queries meant to update EDB relations (with insert and delete) must use readonly=false.

For example, here is an API call to load some CSV-like data and store it in the EDB relation myedb:

data = """
api.query(ctx, database, engine, """
def config:schema:name="string"
def config:schema:lastname="string"
def config:schema:id="int"
def config:syntax:header_row=1
def config:data = mydata

def delete[:myedb] = myedb
def insert[:myedb] = load_csv[config]
""", inputs = {"mydata":data}, readonly=False)

Getting Multiple Relations Back

As in the notebooks, if you want to return multiple relations, you can define sub-relations of output. For example:

>>> resp = api.query(ctx, database, engine,
                     "def a = 1;2 def b = 3;4 def output:one = a def output:two = b")
>>> show.results(resp)
// output (:two*Int64)

// output (:one*Int64)

Result Structure

The response is a Python dictionary with one or more of the following keys:

outputA list of Python Dict’s, each one corresponding to an output relation — in this case, only one, output. (Future releases of the SDK may support multiple relations here.)
problemsinformation about any existing problems in the DB (which are not necessarily caused by the query)
actions[can ignore]
abortedIndicator whether an abort resulted — for example, if an Integrity Constraint was violated.
type[can ignore]

Each query is a complete transaction, executed in the context of the provided database.

Specifying Inputs

The api.query method takes an optional inputs dictionary that can be used to map relation names to string constants for the duration of the query, analogous to the file upload feature of Rel Console notebooks. Thus, for example,

api.query(ctx, database, engine, "def output = foo", inputs = {"foo" : "asdf"})

will return the string "asdf" back.

Functions that transform a file and write the results to an EDB relation can be easily written in this way. The calls api.load_csv and api.load_json are special cases of this.

Printing Responses

The function can be used to print API responses. For example:

>>> response = api.query(ctx, database, engine, "def output = 'a';'b';'c'")
>>> show.results(response)
// output (Char)

The function show.problems(response) prints problems.

See the examples folder for examples on how to handle HTTP exceptions.

Loading Data: load_csv and load_json

As a convenience, the Python API includes load_csv and load_json Python functions. (These are not strictly necessary, since the load utilities in Rel itself can be used in a non-read-only api.query that uses the inputs option.)

The call api.load_csv loads data and inserts the result into the EDB relation named by the relation argument.

load_csv(ctx: railib.api.Context, database: str, engine: str, relation: str, data, syntax: dict = {})
    # `syntax`:
    #   * header: a map from col number to name (base 1)
    #   * header_row: row number of header, 0 means no header (default: 1)
    #   * delim: default: ,
    #   * quotechar: default: "
    #   * escapechar: default: \
    # Schema: a map from col name to rel type name, eg:
    #   {'a': "int", 'b': "string"}

Similarly, load_json loads the data string as JSON, and inserts it into the EDB named by the relation argument:

load_json(ctx: railib.api.Context, database: str, engine: str, relation: str, data) -> dict


api.load_json(ctx, database, engine, "myjson",  """{"a" : "b"}""")

Note: in both cases, the relation EDB is not cleared, allowing for multi-part, incremental loads. To clear it, issue a non-read-only query of the form:

def delete[:relation] = relation

Listing EDB Relations

api.list_edb(ctx, database, engine)

will list the EDB relations in the given database. The result is a JSON list of objects.

Notebook Cell Types

As explained in Working with RAI Notebooks there are four notebook cell types: Query, Install, Update, and Markdown. In this section, we show how the three cell types (Query, Install, and Update) map to Python SDK API calls.

Query Cells

The Query cell in a Rel Console notebook corresponds to the relationalai.Connection.query() Python function call as shown below:

>>> rsp = api.query(ctx, database, engine, "def output = {(1,); (2,); (3,)}")
>>> show.results(rsp)
// output (Int64)

Note that no rules/definitions are persisted in this case.

Install Cells

If IDBs need to be persisted, we use the install_source API call. That is, the Install cell in a Rel Console notebook corresponds to the api.install_source Python function call, as shown below:

>>> rsp = api.install_source(ctx, database, engine,
    {"arity1_k_def" : "def k = {2;3;5}"} )
>>> show.problems(rsp)

Let’s check that the above relation was persisted using the query() API call:

>>> rsp = api.query(ctx, database, engine, "def output = k")
>>> show.results(rsp)
// output (Int64)

Update Cells

The Update cell in a Rel Console notebook corresponds to an api.query Python function call with the flag readonly=False. For example:

rsp = api.query(ctx, database, engine, """def insert:employee = {(1, \"Han Solo\");
         (2, \"Bart Simpson\")}""", readonly=False)

Now let’s modify the employee EDB:

rsp = api.query(ctx, database, engine, """def insert:employee = {(3, \"King\");
          (4, \"Queen\")}""", readonly=False)

Now let’s query the modified employee EDB:

>>> rsp = api.query(ctx, database, engine, "def output=employee")
>>> show.results(rsp)
// output (Int64*String)
1, "Han Solo";
2, "Bart Simpson";
3, "King";
4, "Queen"

To update EDB relations, always use the readonly=False option in the query() call.