How to deal with spaces using AWS S3 integration in Bitbucket Pipelines?

Oscar Navarrerete Estévez

I am running the aws s3 integration pipeline:

pipe: atlassian/aws-s3-deploy:1.1.0
  variables:
    ...
    EXTRA_ARGS: '--metadata={"style-src":"self inline"}'

It seems the space is not being process properly because I get:

awscli.argprocess.ParamError: Error parsing parameter '--metadata': Invalid JSON: Unterminated string starting at: line 1 column 75 (char 74)
JSON received: {"style-src":"self

 

I have tried different quotes arrangements and also tried the shorthand syntax:

EXTRA_ARGS: '--metadata=style-src=self inline'

The error being:

awscli.customizations.exceptions.ParamValidationError: Unknown options: inline

 

I did simplify the pipeline for readability, but it is longer and all previous arguments in EXTRA_ARGS were accepted, so I do not believe there are other problems besides the space.

Any suggestion is appreciated, thanks in advance.

Oscar Navarrerete Estévez

To whoever may be relevant.

The solution we ended up using in the end is to provide the metadata from an external file and decode it in the pipeline:

- export CSP_POLICY_B64=$(grep -v "^\s*#" default.csp | tr '\n' ' ' | tr -d '\r' | base64 --wrap=0)
- pipe: atlassian/aws-s3-deploy:1.1.0
  variables:
    ...
    EXTRA_ARGS: '--metadata cf-csp-b64=$CSP_POLICY_B64'

 In the above example, a default.csp file has been provided on the root of the project. The contents looks like so:

# This file is parsed by bitbucket-pipelines so it can contain
# commented lines starting with a # and newlines are formatted
# to a space when deploying.
style-src 'self' 'unsafe-inline';