Skip to content

serverless.md

Latest commit

 

History

History
156 lines (135 loc) · 5.52 KB

serverless.md

File metadata and controls

156 lines (135 loc) · 5.52 KB

Setup - AWS Lambda

These instructions are for AWS. The bot has not been tested under Azure/Google functions, and may require minor modification to work in those environments.

Screenshot of chat with Dalibot

Create the Lambda function

Go to the Lambda console

  • Create a function
    • Function name: dalibot
    • Runtime: the Python version that matches your build environment (check with python -V or python3 -V)
    • Architecture: arm64
    • Hit Create Function
  • You should now be in the function on the Code tab. Scroll down to Runtime settings, and Edit
    • Handler: bot.lambda_handler
    • Hit Save
  • Go to the Configuration tab
    • Go to General configuration and Edit
      • Timeout: 1 min
      • Hit Save

Create the API Gateway

Go to the API Gateway console

  • Build an API of type REST API
    • API Name: dalibot
    • Hit Create API
  • From the Action menu, Create a method of type POST
    • Lambda function: dalibot
    • Hit Save.
  • Under your POST method > Integration Request:
    • Expand HTTP Headers and add a header named X-Amz-Invocation-Type mapped from 'Event' (with quotes). This tells Lambda to process incoming requests asyncronously, preventing duplicate notifications from Telegram.
    • Expand Mapping Templates and add a mapping template for Content-Type application/json. Scroll down and enter the following mappings and Save. This allows the bot to route the request appropriately, and read Telegram's auth header:
{
    "method": "$context.httpMethod",
    "uri": "$context.path",
    "Content-Type" : "$input.params('Content-Type')",
    "body" : $input.json('$'),
    "headers": {
        #foreach($param in $input.params().header.keySet())
            "$param": "$util.escapeJavaScript($input.params().header.get($param))" #if($foreach.hasNext),#end
        #end
    }
}
  • From the Action menu, hit Deploy API
    • Deployment stage: [New Stage]
    • Stage name: dalibot (this name forms your URI)
    • Hit Deploy
    • Take note of the Invoke URL at the top of the stage editor

Prepare the Lambda package

Clone the git repo

sudo $(which apt dnf yum) install git

git clone https://github.com/robdevops/dalibot.git ~/dalibot

Run the build script

cd ~/dalibot

bash build_serverless.sh -p arm64

Complete the config

cd staging

Edit dalibot.ini

Mandatory parameters

Example:

openai_api_key = sk-p999HAfj6Cm1bO00SXgJc7kFxvFPtQ1KBBWrqSOU
telegramBotToken = 4839574812:AAFD39kkdpWt3ywyRZergyOLMaJhac60qc
telegramOutgoingToken = sLnHdQmYoohmysecret7PX5VDM4cPW
telegramOutgoingWebhook = https://xxxxxx.execute-api.us-east-2.amazonaws.com/dalibot

Optional parameters

You can make stderr more verbose:

debug = 1

You can customise the bot command:

telegramBotCommand = dream

You can enable private messaging by listing telegram numeric ids separated by a space:

telegramAllowedUserIDs = 123456789 987654321

Add your config to the package

zip dalibot_arm64.zip dalibot.ini

Upload the package

Back in the Lambda function, go to the Code tab and then upload from .zip file, and choose dalibot_arm64.zip from your staging folder.

Testing and troubleshooting

  • Message the bot, then monitor the logs from Lambda Function > Monitor > Logs, or from the CloudWatch console under Logs > Log groups.
  • The function can also be triggered in various ways, but these will fail at various points because they won't receive the Telegram json payload:
    • From the Test tab in the Lambda function
    • From the POST - Method execution in API Gateway console.
    • With curl, for example:
    curl -iH "Content-Type: application/json" -H "X-Telegram-Bot-Api-Secret-Token: yoursecret" -X POST -d '{"message": {"message_id": 1, "from": {"id": 1, "first_name": "test" }, "chat": {"id": 123, "type": "private"}, "text": "hello" }}' https://xxxxxx.execute-api.us-east-2.amazonaws.com/dalibot
  • If variations don't work and the error contains ImportError: cannot import name '_imaging' from 'PIL', make sure the Python version and CPU architecture under Function > Code > Runtime settings match your build environment.

Amazon API Gateway Resource Policy

(Optional) Improved security by locking down access to Telegram IP addresses. API Gateway Console > dalibot > Resource Policy:

{
    "Version": "2012-10-17",
    "Statement": [
        {
            "Effect": "Allow",
            "Principal": "*",
            "Action": "execute-api:Invoke",
            "Resource": "execute-api:/*/*/*",
            "Condition": {
                "IpAddress": {
                    "aws:SourceIp": [
                        "149.154.160.0/22",
                        "149.154.164.0/22",
                        "149.154.172.0/22",
                        "91.108.4.0/22",
                        "91.108.56.0/24"
                    ]
                }
            }
        }
    ]
}