install playwright for crawler agent working after re-create

[isthaison]

crawler agent working after re-create

[KevinHuSh]

Appreciation!
I did not find the place where we used this 'playwright'. Could you specify what kind issue you want to resolve?

[isthaison]

https://github.com/unclecode/crawl4ai package https://github.com/infiniflow/ragflow/blob/main/pyproject.toml#L29

crawler component in agent won't work after being initially launched. Thank you for your interest

[KevinHuSh]

CI fail.
I don't think it should be installed with pipx, should be?

[isthaison]

Let me check again

[isthaison]

Hi @KevinHuSh. It seems that CI has been working correctly.

[yuzhichang]

What exact issue does this PR fix?
With the change, building image will download ~300MB from https://playwright.azureedge.net. This slows down the build and enlarges the image.

[isthaison]

[isthaison]

test.json
agent test. With the current image, the crawler is not really working

[yuzhichang]

The python package selenium requires chromium, so we installed one at /usr/local/bin/chrome. See Dockerfile line 124.
It's better for playwright to share the same one chromium via env PLAYWRIGHT_BROWSERS_PATH.

[isthaison]

I got it, so I'm going to close this request, can you help me customize the environment variable so that the component can work as soon as it's started?

[yuzhichang]

That's only my theory. You need to research where and how to customize the environment variable.

[KevinHuSh]

300MB is too much for one component.
Not every one needs that extra 300MB.
We're gona figure it out
Hope for your understanding.

[isthaison]

I got it, let me see if I can customize the environment to work.

[isthaison]

It seems like @yuzhichang is not effective.