Octo Bots
The bots infrastructure gives easy to define easy to work with interface for actions on events and background jobs
Each event that is triggered via github webhook, will be sent to the bots manager to execute different pre-defined bots
Each bot is defined within a yaml file and can define its operations and filters on which the bot can run on
Along with that, background jobs may also be defined to run every X time and perform operations, similar to the bots
Starting a new bots server
The first thing that needs to be done in order to create a bots server, is define its configuration
The yaml configuration of the server may look as follows:
settings:
parallel-background-jobs: 10
bots-endpoint: /events
client-validity-time-minutes: 10
parallel-bots: true
credentials:
github-app-credentials:
api-url: https://api.github.com
app-name: octo-bot-app
certificate-path: ../data/cacert.pem
app-creds-path: ../data/github/octo_bot_github_app_creds.yml
checkmarx-credentials:
api-url: https://checkmarx.api.com
creds-path: ../data/checkmarx/octo_bot_checkmarx_creds.yml
bots:
- ./bots/github_checkmarx_bot.yml
jobs:
- ./jobs/github_stale_job.yml
Where:
- settings - the global bots server settings
- parallel-background-jobs: Maximum amount of running parallel background jobs
- bots-endpoint: Listening endpoint which events will be sent to
- client-validity-time-minutes: For how long a client can be valid for connection (github/checkmarx)
- parallel-bots: Should the bots defined run in parallel or sequential
- credentials - credentials for each client, currently supports
- github-app-credentials
- checkmarx-credentials
- bots - List of files that define different bots
- jobs - List of files that define different background jobs
Once the configuration is done, the bots server can be started, either by the bots_executor.py or by a wrapping script for example:
flask_app = Flask(__name__)
bots_manager = BotsManager.create_bots_manager(args.config_path, flask_app)
bots_manager.start_bots_manager()
flask_app.run(ssl_context="adhoc", host="0.0.0.0", port=8443)
Defining a new bot
Each bot is defined with a list of operations and a list of filters
The list of filters will assert whether the bots needs to run its operations or not on any event
A set of bots are defined in a yaml configuration file, for example:
bots:
- name: Checkmarx scan
parallel: true
operations:
- pull-request-checkmarx:
risk-scheme:
min-high-vuls: 1
min-medium-vuls: 20
min-low-vuls: 100
min-info-vuls: 10000
scan-timeout: 30 minutes
filters:
- events-filter:
events:
- name: pull_request
actions:
- opened
- reopened
- ready_for_review
- synchronize
The above bot will define a name bot called Checkmarx scan
The bot will run its operations in parallel (currently only one operation)
The operation that will be ran is called pull-request-checkmarx
And it has a set of parameters, such as the risk-scheme
Along with that, a list of filters are also set, such as the events-filter, which u can filter out which webhooks will trigger this bot, such as the pull_request webhook
The operation itself is defined in a python script, which overrides a base Operation
operations/github/pull_request_checkmarx_operation.py
Once the bot is ready, it can be set within the main yaml configuration
Defining a new background job
The same idea as the bot is applied here for example:
jobs:
- name: Close stale issues and pull requests
parallel: true
operations:
- close-stale:
close-issues: true
close-prs: true
stale-expiration: 60 days
stale-comment: >
This issue has been automaticlly marked as stale because it has not had
recent activity. It will be closed and can be later re-opened.
Thanks for your contributions.
exempt-labels:
- hotfix
- release
every: "30 minutes"
For example here the job is ran every 30 minutes, and will clean up stale pull requests and issues, close them and write a comment (github will also send a mail to the subscribers of the repo accordingly)
Defining github app credentials
In order to connect to github, we also need to define which application we are working with
To do so, we also define a seperate creds yaml file, which can be locally (or in the future from conjur):
app-id: 1
client-id: some_client_id
client-secret: some_client_secret
private-key-path: /some/path/to/app/key
webhook-secret: some_webhook_secret
The above are the credentials which will be used in the main settings, to generate access tokens to github as the github app (bot) to act upon him and make actions
Adding a new operation
In order to add a new operation, we would need to create a new class which inherits the Operation class
The operation class is defined as follows:
operations/operation.py
class Operation:
@abc.abstractstaticmethod
def create_operation(config: dict) -> 'Operation':
pass
@abc.abstractstaticmethod
def operation_type() -> str:
pass
@abc.abstractmethod
def execute_operation(self, clients: Dict[str, BotsBaseClient], headers: dict, event: dict):
pass
Where:
- create_operation: Receives the config dict from the yaml where we defined the operation and its parameters, and returns the created operation
- operation_type: The operation type that we defined its name in the yaml
- execute_operation: The actual operation logic, with inputs of a set of clients which we can use (github, checkmarx) and the validated event itself, after all the filters we defined
Once the operation was implemented, we can use it on the yaml as seen above
Adding a new filter
Similarly, in order to add a new filter, we would implement the following interface
filters/filter.py
class Filter:
@abc.abstractstaticmethod
def create_filter(config: dict) -> "Filter":
pass
@abc.abstractstaticmethod
def filter_type() -> str:
pass
@abc.abstractmethod
def filter_event(self, clients: Dict[str, BotsBaseClient], headers: dict, event: dict) -> bool:
pass
Where:
- create_filter: Receives the config dict from the yaml where we defined the filter and its parameters, and returns the created filter
- filter_type: The filter type that we defined its name in the yaml
- execute_filter: The actual filter logic, with inputs of a set of clients which we can use (github, checkmarx) and the event itself, which we will filter (if it needs to be filtered, we will return True)
Adding a new client
To add a new client, such as github or checkmarx, we would need to implement two classes, one for the credentials, and one for the actual client itself
The credentials class is defined as follows
class BotsBaseCredentials:
@abstractmethod
def create_authenticated_client(self, validity_time) -> BotsBaseClient:
pass
@abstractstaticmethod
def creds_type() -> str:
pass
@abstractstaticmethod
def client_type() -> str:
pass
Where:
- create_authenticated_client: Will create the client itself per the credentials, alrdy authenticated to the resource
- creds_type: Name of the creds type
- client_type: Name of the client
And the client is defined as followsw:
class BotsBaseClient:
@abstractmethod
def is_valid_client(self) -> bool:
pass
@abstractmethod
def validate_request(self, request: Any) -> bool:
pass
@abstractstaticmethod
def client_type() -> str:
pass
Where:
- is_valid_client: Checks if the client itself is valid to be used, if not it will be recreated via the creds
- validate_request: Tries and validates a given request if it can be even used in the first place
- client_type: Name of the client
Once the above two are defined, the client will arrive to the operations as one of the clients in the dictionary, and can be casted and used accordingly