relycomply_client 0.15.1

A python client for the RelyComply platform

RelyComply Python Client and CLI

This package contains the python client and CLI for the RelyComply platform:

RelyComply is an end-to-end Anti-Money Laundering (AML) Platform, managing detection, risk management and automation of your AML compliance requirements

The CLI makes configuration of the system substantially simpler and allows for a full configuration-as-devops experience.

The python client exposes both a lower level GraphQL client which makes it easy to interact with the GraphQL APi in a pythonic manner. As well as a higher-level integration client that provides useful routines for common integration tasks.

There is also a rely-sync tool that makes it easy to sync a local configuration to the application. This is somewhat similar to a tool like terraform.

Credentials

Credentials for the various clients can be loaded in multiple ways. In the following order of precedence:

• Constructor Arguments
• Environment Variables
• AWS Secrets
• Configuration Files (.rely.toml)
• Defaults

Credentials can stack, so you can define certain credentials in a config file for example, and override it with an environment variable.

The following credentials can be set:

• token: The API token of the user
• url: The url of the RelyComply application (default: https://app.relycomply.com)
• impersonate: If user impersonation is available for your user this allows you to make a command on behalf of another user. This should be in format <organisation_name>:<user_email>, for example relycomply:[e-mail].

Constructor Arguments

You can pass credentials when constructing a client as keyword arguments, e.g.

client = RelyComplyGQLClient(token="<token>")

Environment Variables

The client will look if there are matching environment variables of the format RELYCOMPLY_<CREDENTIAL>, for example: RELYCOMPLY_TOKEN=<token>.

AWS Secrets

The client can integrate with the AWS Secrets Manager. This is done by setting appropriate environment variables with the naming convention RELYCOMPLY_<CREDENTIAL>_AWS_SECRET and the value being the secrets path, for example:

RELYCOMPLY_TOKEN_AWS_SECRET=path/to/my/secret

For more information please consult the AWS Secrets Manager documentation.

Config Files

Credentials can be set in a .rely.toml file. The client will search the current working directory, and all parent directories in order, with the most local directory taking precedence. The .rely.toml file is a TOML file with the credentials as keys, e.g.

token="<token>"
url="https://relycomply.customer.com"

Default Credentials

The following credentials have default values:

url="https://app.relycomply.com"

RelyComplyGQLClient

A flexible and intelligent GraphQL client for RelyComply. This client will create methods that match the mutation sand queries of the RelyComply API, and expose them with familiar calling conventions. It also handles paging as well as simplifying the returned structures.

It can be constructed as below:

from relycomply_client import RelyComplyGQLClient

client = RelyComplyGQLClient()

# Or with specific credentials
client = RelyComplyGQLClient(token="<token>")

Queries can be called with their lowerCase field name and any filter arguments as kwargs, e.g.:

client.products(nameContain="ZA") # Will return a list of products
client.products(nameContain="ZA", _iter=True) # Will return a lazy generator
client.products(name="retailZA", _only=True) # Will return only the first object or None

The client will automatically collapse edge lists into plain lists of objects to make the output easier to work with.

Mutations can be called in a similar way, but arguments will be lifted into the $input variable

client.createProduct(name="retailZA", label="South African Retail") # Returns the created product

The interface is automatically generated from the GQL schema as well as the CLI support templates. Thus it should always be in sync with the latest features on the platform.

The client also exposes a raw GraphQl call when you need to make a more complex query. No post processing will be done on the results. This is mainly useful because it loads and manages the credentials. For example to query the first 10 products.

client.graphql(
    """
    products(first:$first) {
        edges {
            node {
                id
                name
            }
        }
    }
    """, 
    variables=dict(first=10)
)

RelyComplyClient

The RelyComplyCLient contains higher level methods that make common integration tasks simpler. It provides simple integration with various cloud services and common data tools like pandas.

from relycomply_client import RelyComplyGQLClient
rc = RelyComplyClient()

# Or for quick usage for a standard client
from relycomply_client.rc import rc

A quick overview of a common data integration with the transaction monitoring is shown below:

# Load a file with pandas
raw_df = pd.read_csv(file_path)

# Perform some cleaning
df = clean_raw_df(raw_df)

# Pull in a datafile from a known source. 
# This will automatically create a signed URL if an S3 path is passed
raw_data_file = rc.pull_to_datafile(
    file_path, "raw/" + file_name, wait_for_ready=True
)

# Upload a dataframe as parquet datafile
processed_data_file = rc.put_to_datafile(df, "processed/" + file_name)

# Ingest the given files

# Note that the responses from previous calls can be passed as is for the call
# arguments. Their id's will be automatically extracted.
data_source_version = rc.ingest_datasource(
    data_source, data_file=processed_data_file, raw_data_files=[raw_data_file]
)

# Run a monitor
rc.run_monitor(monitor_name, source_versions=[data_source_version])

The underlying RelyComplyGQLClient can be accessed with .gql property.

rc.gql.createProduct(name="bank_account", label="My Bank Account)

Command Line Interface (CLI)

The command line interface is an important part of our developer first mentality. It acts as a layer on top of the GraphQL API and makes it substantially easier to for power-users to explore and manipulate RelyComply.

GraphQL is excellent as an API for integration, but can be a lot extra overhead to quickly just see what is happening in the system. Primarily this is because the user has to define the output format they want. This greatly improves the flexibility but certainly is not as easy as just using curl on a rest endpoint. The rely CLI makes it easy to perform the standard queries and mutations on the GraphQL API, without extra effort by the user.

The CLI can be accessed with the rely command. The basic format is to call it with a type and an action. This will automatically be coerced into the appropriate GraphQL calls.

Arguments can be passed as keyword arguments of the form --key=value additionally a configuration file name can be passed as the final argument. The value can be a json string, which will be parsed correctly for complex arguments. Arguments are merged with the command line arguments taking precedence.

Queries can be performed with the list and retrieve actions, retrieve will return a single item, and list will display a table of items.

rely product list # Will list all the products
rely product list --nameContains="za" # Will list all the products with za in their name
rely product retrieve --id=123 # Will return just the specified product 

Mutations can called by their name, with the action being prepended to the type. The system is intelligent enough that the case of the action and type do not matter.

# Will call createProduct
rely product create --name="bank_account" --label="Bank Account" 

# Will update the given product (updateProduct) based on the given ID and the config file (pr_my_product.toml)
rely product update --id=10 pr_my_product.toml

Certain aliases are provided for convenience, e.g.

# This will call addCaseNote
rely case addNote --case=123 --note="This is my note"

The format of the output can be controlled with the --json, --yaml and --toml (default) flags.

rely-sync

The rely-sync command makes it easy to create and update large sets of configuration in a standard way without having to run the individual update or create commands.

This works by examining the *.toml files in directory (or recursively in directories), and checking an extra piece of metadata that explains the type of the object that it represents (e.g. "Product"), then using the name will check if it exists, if it does not it will offer to create it, otherwise it will update it.

The metadata is set as a magic comment at the top of the file, for example a Product definition would look like:

#% type = "Product"

name = "retail_account"
label = "Generic Retail Account"
description = "For people who need banks"

Note the magic comments start with a "#%" sequence which means they will be ignored by any other tools, but are differentiated from normal comments. The structure of the metadata itself is in TOML format. Currently only the type key is supported which should be the Type of the object. In the future this may be extended.

Assuming this file were placed in a directory, the following command would be called:

rely-sync <directory>

Found the following files with type information
  - tests/configuration/pr_retail.toml [type = Product]

The following items will be created:

Type     name            path
-------  --------------  ----------------------------------
Product  retail_account  tests/configuration/pr_retail.toml

Are you sure you would like to continue (yes/no)?: 

if you respond yes then you would further see:

rely Product create tests/configuration/pr_retail.toml

You would then be able to see the created Product on the application.

If you were to run it again and responded yes you would see:

Found the following files with type information
  - tests/configuration/pr_retail.toml [type = Product]

The following items may be updated:

Type     name            path
-------  --------------  ----------------------------------
Product  retail_account  tests/configuration/pr_retail.toml

Are you sure you would like to continue (yes/no)?: yes
rely Product update --id=retail_account tests/configuration/pr_retail.toml

Since the product would be updated instead of created.

The output will show you any errors in your metadata, or alerts about lack of metadata.

A slightly more complex example of configuration files can be found in the tests/configuration folder.

You can see additional help and arguments with rely-sync --help.

Caveats and known issues

The system is currently reasonably simple and there are known cases where things will not work:

• If you rename an object it will create a new object, leaving the old one intact. In the case use the CLI to delete the old object (BUT BE CAREFUL).
• If you change the type of an object (in the metadata) very strange things will happen. This will not be destructive though.

The system will be enhanced in the future to more gracefully deal with these cases.

rely-watch

The rely-watch command is similar to the rely-sync command but when given a directory it will watch for file changes in that directory (either additions or updates), and if applicable run rely-sync on that file. This allows for a live updating experience and is especially useful for editing complex rule sets.