🚀 Feature Request: Pages publish should support --site-exclude/--site-include

[mcclure]

Describe the solution

Context: I have a static web page I am attempting to migrate from Amazon S3 to Cloudflare Pages. With Amazon, there was a simple program I could run on a directory and it would do something like an rsync (IE upload changed files only) of the directory to the server. CloudFlare pages appears to prefer a situation where I push a git repo and CloudFlare checks out the git repo and runs a "builder". I would prefer to just upload you the files. I have a source control repo for the site but it is not practical to give you access to it.

Problem: I have hit a blocker because wrangler pages publish appears to only be able to upload complete directories. However, I have a .hg (source control) subdirectory in the directory. That should absolutely not be uploaded.

The wrangler publish command, for whatever "workers" are, has a convenient --site-exclude command which addresses my need exactly. However it is not supported for wrangler pages publish: https://developers.cloudflare.com/workers/wrangler/commands/#pages

This is a feature the amazon equivalent does have (--exclude) and is a pretty standard feature for most archival/uploading tools (rsync, zip, etc).

Expected behavior: Could wrangler pages adopt a file include/exclude pattern system? Using the tool without this feature is going to be pretty awkward.

[lrapoport-cf]

cc @tanushree-sharma

[Niek]

It's very strange that this is not supported, currently there's only this hardcoded array:

workers-sdk/packages/wrangler/src/pages/validate.ts

Lines 44 to 53 in 0c0374c

	const IGNORE_LIST = [
	"_worker.js",
	"_redirects",
	"_headers",
	"_routes.json",
	"functions",
	"**/.DS_Store",
	"**/node_modules",
	"**/.git",
	].map((pattern) => new Minimatch(pattern));

Having a --site-exclude parameter or an exclude config property support would be great.

[emily-shen]

While this won't be added to Pages, you can customise what is ignored at upload in Workers Assets using an .assetignore file: https://developers.cloudflare.com/workers/static-assets/binding/#ignoring-assets

[Cherry]

This is quite disappointing when Assets isn't yet fit for purpose and migration from Pages. Until static config and a way better DX around building, preview links, etc. I can't recommend anyone migrate to Assets.

Especially when in some cases like the node_modules one, the behaviour is still technically a regression: #3615 (comment)

Should folks consider Pages effectively deprecated at this point?

[emily-shen]

In the interest of not jumping the gun, i'll keep this issue open! Didn't mean to confuse anyone about Pages, just a PSA in case you're in a position to move to Workers Assets, that does currently have a customisable ignore file. Sorry for the confusion 🧡

[nora-soderlund]

It was made abundantly clear that this won’t be added even via a public pull request.

Is that an option again, given there’s a workers implementation that can be extended?

[mcclure]

I still think this is a good idea.

Maybe I should be using Workers Assets? But that seems weird if I really do just want to host static pages. I don't want to make some sort of mistake and wind up getting charged for worker compute when the site was intended to be fully static.