Add env var to disable image parsing

[msabramo]

Added a PROMPTY_CHAT_PARSER_DISABLE_IMAGE_PARSING=true environment variable to disable image parsing in the chat parser.

In my use case, I am using Prompty to send a prompt to the LLM that contains Markdown that should be converted to a different format. I do not need Prompty to parse the images and try to send them to the LLM. In fact, this is inconvenient because I don't want to have to copy the images just so Prompty can read them.

An environment variable seems kind of janky perhaps, but it's the best solution I could come up with that doesn't require a lot of changes, because there's a fairly deep call stack so in order to pass a parameter, it would require modifying a lot of code. I'm open to other ideas though.

[sethjuarez]

This is a good idea - let me think about how to do this in a more declarative way.

[msabramo]

Bump

[taylorn-ai]

Could this please be reviewed? This is actually causing a problem for me right now.

@sethjuarez @weastler @markwallace-microsoft

[taylorn-ai]

I have made some changes, but I haven't put in a PR yet though, as the newer version of Prompty is quite different, and seemingly inconsistent. For example, this is in the chat.prompty test file:

inputs:
  messages:
    type: array
    items: 
      type: object
      properties: 
        role: { type: string }
        content: { type: string }

And if I try to use this example, I get an exception as it doesn't expect items to be there...

Also when _validate_inputs is called, it's not checking the required value, it just checks if there is a default value, if not, throws a ValueError. So it's hard for me to test properly...

[taylorn-ai]

OK fixed a few things PR #274. @sethjuarez - could you please review?

[sethjuarez]

Thank you for your contribution! We've migrated Prompty to v2 on main — the codebase has been significantly restructured with AgentSchema-based types, scoped @prompty/* npm packages, and new runtimes.

The v1 code is now preserved on the v1 branch for maintenance. If your change is still relevant for v1, please re-open against the v1 branch. For v2 contributions, please open a new PR against main.

Apologies for any inconvenience, and thanks for your patience!