This Symphony bot client is written in an event handler architecture. The client keeps polling a datafeed and responds to different types of Real Time Events it receives.
To build a functional bot which responds to different types of incoming messages from datafeed (Connection, IM, Chat Room, etc....), the respective type of listener needs to be implemented by inheriting the interfaces in the listeners folder. Currently ConnectionListener, imListener, RoomListener interfaces are provided.
This client is compatible with Python 3.6 or above
Create a virtual environment by executing the following command (optional):
python3 -m venv ./venv
Activate the virtual environment (optional):
source ./venv/bin/activate
Install dependencies required for this client by executing the command below.
pip install -r requirements.txt
To develop this library run:
pip install -e .
This will give an editable install so the code can be changed while still being installed in the environment. Tests can then be run with:
pytest tests for the whole test suite
pytest tests/clients/test_datafeed_client.py for one test file
pytest tests -k test_get_text for a single test
An environment variable SYMPHONY_TEST_CONFIG can be set with the authentication type and configuration file separated by a colon.
SYMPHONY_TEST_CONFIG=RSA:~/.dev_config.json pytest
The Python client operates using a Symphony Service Account, which is a type of account that applications use to work with Symphony APIs. Please contact with Symphony Admin in your company to get the account.
The client currently supports two types of Service Account authentications, they are Client Certificates Authentication and RSA Public/Private Key Pair Authentication.
As an example, the roomListenerTestImp has been implemented to respond with "Hello World", to a chat room in which there is an incoming message. To respond to other types of events, respective Listeners need to be implemented.
RSA Public/Private Key Pair is the recommended authentication mechanism by Symphony, due to its robust security and simplicity.
To run the bot using the RSA Public/Private Key Pair, a rsa_config.json should be provided. In our example, the json file resides in the resources folder but it can be anywhere.
An example main_RSA.py has been provided to illustrate how all components work together.
To run the bot using the Client Certificates Authentication, a config.json should be provided. In our example, the json file resides in the resources folder but it can be anywhere.
An example main_certificate.py has been provided to illustrate how all components work together.
Notes: Most of the time, the port numbers do not need to be changed.
An example of json has been provided below. (The "botPrivateKeyPath" ends with a trailing "/")
{
"sessionAuthHost": "MY_ENVIRONMENT.symphony.com",
"sessionAuthPort": 443,
"keyAuthHost": "MY_ENVIRONMENT.symphony.com",
"keyAuthPort": 443,
"podHost": "MY_ENVIRONMENT.symphony.com",
"podPort": 443,
"agentHost": "MY_ENVIRONMENT.symphony.com",
"agentPort": 443,
// For bot RSA authentication
"botPrivateKeyPath":"./sym_api_client_python/resources/",
"botPrivateKeyName": "bot_private_key.pem",
// For bot cert authentication
"botCertPath": "/path/to/bot-cert/",
"botCertName": "/bot-cert.p12",
"botCertPassword": "bot-cert-password",
"botUsername": "YOUR_BOT_USERNAME",
"botEmailAddress": "YOUR_BOT_EMAIL_ADDRESS",
"appCertPath": "",
"appCertName": "",
"appCertPassword": "",
"authTokenRefreshPeriod": "30"
// Optional: If all the traffic goes through a single proxy, set this parameter. If using multiple proxies or only using a proxy for some of the components, set them below and don't useproxyURL
"proxyURL": "http://localhost:8888",
"proxyUsername": "proxy-username",
"proxyPassword": "proxy-password",
// Optional: set this if traffic to pod goes through a specific, unique proxy
"podProxyURL": "http://localhost:8888",
"podProxyUsername": "proxy-username",
"podProxyPassword": "proxy-password",
// Optional: set this if traffic to agent goes through a specific, unique proxy
"agentProxyURL": "http://localhost:8888",
"agentProxyUsername": "proxy-username",
"agentProxyPassword": "proxy-password",
// Optional: set this if traffic to KeyManager goes through a specific, unique proxy
"keyManagerProxyURL": "http://localhost:8888",
"keyManagerProxyUsername": "proxy-username",
"keyManagerProxyPassword": "proxy-password",
// Required: If a truststore is required to access on-prem components, provide a path to the python truststore. Needs to be .pem file. Instructions below for converting JKS to python pem truststore. If truststore is not needed, set value as empty string ("").
"truststorePath": "/path/to/truststore.pem",
// Optional: if set to true, the datafeed id will be stored on the filesystem and subsequently reused.
// Applies for DFv1 only. Default value is true.
"reuseDatafeedID": true,
// Optional: path to the folder where to store the datafeed id. Applies for DFv1 and if reuseDatafeedID set to true.
// Default value is os.getcwd().
"datafeedIdFilePath": "/some/folder/"
}
Adjust the following paths in the sample to match your configuration
- "sym_api_client_python/logs/example.log"
- "sym_api_client_python/resources/config.json"
Example Main Class:
import logging
from sym_api_client_python.configure.configure import SymConfig
from sym_api_client_python.auth.rsa_auth import SymBotRSAAuth
from sym_api_client_python.clients.sym_bot_client import SymBotClient
from sym_api_client_python.listeners.\
im_listener_test_imp import IMListenerTestImp
from sym_api_client_python.listeners.\
room_listener_test_imp import RoomListenerTestImp
def configure_logging():
logging.basicConfig(
filename='./logs/example.log',
format='%(asctime)s - %(name)s - %(levelname)s - %(message)s',
filemode='w', level=logging.DEBUG
)
logging.getLogger("urllib3").setLevel(logging.WARNING)
def main():
print('Python Client runs using RSA authentication')
# Configure log
configure_logging()
# RSA Auth flow: pass path to rsa config.json file
configure = SymConfig('../resources/config.json')
configure.load_config()
auth = SymBotRSAAuth(configure)
auth.authenticate()
# Initialize SymBotClient with auth and configure objects
bot_client = SymBotClient(auth, configure)
# Initialize datafeed service
datafeed_event_service = bot_client.get_datafeed_event_service()
# Initialize listener objects and append them to datafeed_event_service
# Datafeed_event_service polls the datafeed and the event listeners
# respond to the respective types of events
im_listener_test = IMListenerTestImp(bot_client)
datafeed_event_service.add_im_listener(im_listener_test)
room_listener_test = RoomListenerTestImp(bot_client)
datafeed_event_service.add_room_listener(room_listener_test)
# Create and read the datafeed
print('Starting datafeed')
datafeed_event_service.start_datafeed()
if __name__ == "__main__":
main()
Once the certificates are provided and example listeners are implemented, let's run the bot by executing following command:
python3 main_certificate.py.
Change the paths to both log, and config.json
- "sym_api_client_python/logs/example.log"
- "sym_api_client_python/resources/config.json"
Example Main Class:
import logging
from sym_api_client_python.configure.configure import SymConfig
from sym_api_client_python.auth.auth import Auth
from sym_api_client_python.clients.sym_bot_client import SymBotClient
from sym_api_client_python.listeners.im_listener_test_imp import \
IMListenerTestImp
from sym_api_client_python.listeners.room_listener_test_imp import \
RoomListenerTestImp
def configure_logging():
logging.basicConfig(
filename='./logs/example.log',
format='%(asctime)s - %(name)s - %(levelname)s - %(message)s',
filemode='w', level=logging.DEBUG
)
logging.getLogger("urllib3").setLevel(logging.WARNING)
def main():
print('Python Client runs using Cert authentication')
# Configure log
configure_logging()
# Cert Auth flow: pass path to certificate config.json file
configure = SymConfig('../resources/config.json')
configure.load_config()
auth = Auth(configure)
auth.authenticate()
# Initialize SymBotClient with auth and configure objects
bot_client = SymBotClient(auth, configure)
# Initialize datafeed service
datafeed_event_service = bot_client.get_datafeed_event_service()
# Initialize listener objects and append them to datafeed_event_service
# Datafeed_event_service polls the datafeed and the event listeners
# respond to the respective types of events
im_listener_test = IMListenerTestImp(bot_client)
datafeed_event_service.add_im_listener(im_listener_test)
room_listener_test = RoomListenerTestImp(bot_client)
datafeed_event_service.add_room_listener(room_listener_test)
# Create and read the datafeed
print('Starting datafeed')
datafeed_event_service.start_datafeed()
if __name__ == "__main__":
main()
The Python SDK truststore requires that your certificates be in a .pem file that is a collection of your certificates. You can convert JKSto a .p12 and then convert to .pem:
keytool -importkeystore -srckeystore myapp.jks -destkeystore myapp.p12 -srcalias myapp-dev -srcstoretype jks -deststoretype pkcs12
openssl pkcs12 -in myapp.p12 -out myapp.pem
The joke bot shows how a Symphony chat bot application works. In general, a chat bot application keeps polling the datafeed API for new messages, then it sends the messages to listeners to handle, depending on the message type.
To interact with the joke bot, try /bot joke
Symphony Elements allow developers to include native, interactive UI components as part of an inbound message. When a form Element is submitted, an event of type "SYMPHONYELEMENTSACTION" is sent across the datafeed.
In order to handle Elements events, implement an ElementsActionListener and parse the JSON payload using the SymElementsParser
from sym_api_client_python.listeners.elements_listener import ElementsActionListener
from sym_api_client_python.processors.sym_elements_parser import SymElementsParser
from .processors.action_processor import ActionProcessor
class ElementsListenerTestImp(ElementsActionListener):
def __init__(self, sym_bot_client):
self.bot_client = sym_bot_client
self.action_processor = ActionProcessor(self.bot_client)
def on_elements_action(self, action):
stream_type = self.bot_client.get_stream_client().stream_info_v2(SymElementsParser().get_stream_id(action))
if stream_type['streamType']['type'] == 'ROOM':
self.action_processor.process_room_action(action)
elif stream_type['streamType']['type'] == 'IM':
self.action_processor.process_im_action(action)
Use SymElementsParser class inside ActionProcessor:
from sym_api_client_python.processors.message_formatter import MessageFormatter
from sym_api_client_python.processors.sym_elements_parser import SymElementsParser
class ActionProcessor:
def __init__(self, bot_client):
self.bot_client = bot_client
def process_room_action(self, action):
try:
action_clicked = SymElementsParser().get_action(action)
if action_clicked == 'submit':
#do something
except:
raise
Note: to send Elements forms as messages, please refer to this documentation for valid messageML samples.
Symphony REST API offer a range of capabilities for application to integrate, visit the official documentation for more information.
Use pip install sym-api-client-python or pip install sym-api-client-python==0.2.0
A non-blocking version of the DataFeed has been created using asyncio. This supports asynchronous behaviour in two ways:
-
Listening on the event feed does not have to be blocking, this would allow for a bot that responds to Symphony events, but also external stimulus like timers or REST requests. Using asyncio means this can be done without threading, and therefore without careful threadsafe structures
-
Handlers can be asynchronous. This means that while responding to one client with an expensive request, other clients can still get timely responses
-
Successfully employing this strategy can make your bot handle far higher loads than would be possible in the synchronous case, without having to resort to threading or complex callbacks.
NOTE
There is an outstanding issue with aiohttp and Python 3.8+ on Windows. This affects our client when you specify a proxy on Windows with Python 3.8+.
If you are using a proxy, you may need to implement the following workaround before you start the async datafeed or make any async calls with aiohttp:
if sys.version_info[0] == 3 and sys.version_info[1] >= 8 and sys.platform.startswith('win'):
asyncio.set_event_loop_policy(asyncio.WindowsSelectorEventLoopPolicy())
Blocking listeners will however still block the datafeed. Listeners that need to make HTTP requests however can make them with aiohttp, instead of requests for example. aiohttp is already a dependency of this module. These can be awaited to release the thread.
# Blocking
requests.get(url)
# Not blocking
async def fetch(session, url):
async with session.get(url) as response:
return await response.text()
async def main():
async with aiohttp.ClientSession() as session:
html = await fetch(session, 'http://python.org')
print(html)
To see this functionality, run the main_async.py version in the examples folder. The key changes are:
main_async.py
datafeed_event_service = bot_client.get_async_datafeed_event_service()
...
async def timed_ringer(period_in_seconds, message):
while True:
await asyncio.sleep(period_in_seconds)
print(message)
# Create and read the datafeed
print('Starting datafeed')
loop = asyncio.get_event_loop()
awaitables = asyncio.gather(timed_ringer(2, "Ding"), timed_ringer(5, "Dong"), datafeed_event_service.start_datafeed())
loop.run_until_complete(awaitables)
im_listener_test_imp.py
class AsyncImListenerImp(IMListener):
"""Example implementation of RoomListener with asynchronous functionality"""
def __init__(self, sym_bot_client):
self.bot_client = sym_bot_client
self.message_parser = SymMessageParser()
async def on_im_msg(self, msg):
logging.debug('message received in IM', msg)
await asyncio.sleep(5)
msg_text = self.message_parser.get_text(msg)
msg_to_send = dict(
message='<messageML>Hello {}, sorry to keep you waiting!</messageML>'.format(
self.message_parser.get_im_first_name(msg))
)
if msg_text:
stream_id = self.message_parser.get_stream_id(msg)
self.bot_client.get_message_client(). \
send_msg(stream_id, msg_to_send)
See the release page.
- Added "cache-control" header value to pod and agent sessions
- Bug fixes in SymBotClient and StreamClient
- Updated requirements.txt to allow developers to pip install editable version
- fixed bug in test_MessagesClient.py
- Added AsyncElementsListener Class to elements_listener_test_imp.py
- Updated sub packages included in requirements.txt
- Updated SymElementsParser to read new JSON payload introduced in elements 1.1
- Introduced asynchronous datafeed listening service.
- Added Resiliency in read_datafeed() in DataFeedEventService
- Fixed Logging Issue in DataFeedEventService
- Handle JSONDecodeError
- fixed bug in search_users()
- Created Wall Post Listener
- Created Message Suppression Listener
- Fixed typos and comments
- Updated Error handling logic for Server Errors (500's) in datafeed_event_service.py
- Updated authentication retry logic for RSA based authentication
- User has a maximum of 5 times to reauthenticate. If limit is reached, MaxRetryException is raised.
- SDK Recovers and creates new datafeed if reauthenticate succeeds.
- Added Truststore support for RSA based authentication
- Updated examples
- Updated examples
- Added support for Elements actions coming across datafeed
- Added ElementsActionListener
- Added Processors, containing MessageFormatter, SymElementsParser, SymMessageParser. Provides helpers for accessing message and action payload information
- Added Templates folder, containing example templates for sending in Elements
- Added FormBuilder to programaticaly generate Elements forms
- Added examples for ExpenseBot using Elements
- Cleaned up typos and formatting
- Added AdminClient, SignalsClient, and ConnectionsClient
- Added docstrings to new methods
- Added clarification to README
- Re-released 0.1.15 due to merge issues
- Added podProxyURL, agentProxyURL, and keyManagerProxyURL as supported parameters in the config.json and config loader. If proxyURL is set, all of these proxies will be set to that URL. Otherwise, it will use the proxy address provided.
- merge to using the same method in configure.py to load RSA and Cert
- Rewrite clients to use python sessions.
- Moved over requests to sessions for consistent headers, proxies and truststore
- Fixed auth using sessionAuth for both sessionAuth and keyAuth
- Fixed functions that pre-populated payload to accept params
- Datafeed now utilizes session user to determine userId
- Update to ProxyURL. No longer prepend http://
- Changed config.json for Proxy to mirror Java SDK
- The updates in this release may break your existing bot implementation. Please ensure you review and test against this client prior to deployment in Production.
- Extensively renamed client libraries to follow PEP8 Python standard. Naming now follows "snake_case" convention.
- Bug fixes.