[REQUEST] support optional CHROME_AWS_LAMBDA_ASSUME_HEADLESS override environment variable

[tmeneau]

What would you like to have implemented?

Support an environment variable (such as CHROME_AWS_LAMBDA_ASSUME_HEADLESS) to optionally short circuit the serverless variable checks in Chromium.headless. For example, this could look like:

  static get headless() {
    if (process.env.CHROME_AWS_LAMBDA_ASSUME_HEADLESS === undefined &&
          (process.env.IS_LOCAL !== undefined || process.env.IS_OFFLINE !== undefined)) {
      return false;
    }
    ...
  }

Why would it be useful?

This would allow users to independently handle weird edge/use cases with minimal changes to their code and/or environments. My own use case is running an app using chrome-aws-lambda in a docker container using the AWS NodeJS Lambda image to locally emulate our production lambda environment (including using the bundled chromium), but overriding the entrypoint to use serverless-offline to mock an AWS API Gateway we have upstream of our Lambda function.

However, it looks like there are a few other edge cases that could also be served by this:

• [BUG] IS_LOCAL=false treated as if it is indeed local #247
• [BUG] headless: false & executablePath: null in AWS Lambda #171

Alternative (Backwards Incompatible) Proposal

You could also make the argument that the problem here is assuming whether the user wants headless or not based on a potential breadcrumb of a specific runtime environment, rather than letting the user specifically define it. While this might be too much of a breaking change at this point, I'd think this would be less confusing for users:

  static get headless() {
    if (process.env.CHROME_AWS_LAMBDA_DISABLE_HEADLESS) {
      return false;
    }
    ...
  }

Users could then explicitly explicitly disable the headless check.