Include first 6 lines of the source code

[nikgraf]

No description provided.

[stereobooster]

Separate page per hook? Then it would be possible to include whole text. For example, https://usehooks.com/useDarkMode/

[nikgraf]

@stereobooster yeah, either a separate page or a section that's collapsed and can be expanded. The issue though is that some hooks might have multiple hundred lines of code. Also is it done on build time or do we fetch it on demand? An iFrame of that Github file could also do it.

[stereobooster]

The issue though is that some hooks might have multiple hundred lines of code.

Let's limit scope. We will care only about hooks :

• implemented in JS as first step, in TS as second step, everything else later
• self contained hooks e.g. we allow imports from packages, but don't allow imports from different files
• one hook per file
• named export, so we can identify hook by use...
• no hooks generators, like generateUseState etc.
• everything that doesn't fit constraints will be disqualified e.g. we will have link, but no code preview

How can we get source, and how we will keep it in sync?

...to be continued

[nikgraf]

sounds good to me!

Regarding how we can get the source. In every Hook object we could link to the raw file. During build time fetch it and include the raw text. We could trigger a build every hour or two to keep it up to date.

What do you think?

[stereobooster]

yes good idea. I don't think we need to trigger every two hours more like once a day - once a week (but this can be configured).

Metadata and information that can be extracted

Github link - stars, forks, open issues, last updated etc
npm package name - latest published version, downloads, date of publishing
link to the raw file - source code, external dependencies, core hooks dependencies, LOC

Script to extract data

• read hooks.json
• convert it to stream https://highlandjs.org/
• filter items with meta-data
• flatmap over stream with async function (fetch)
• log all failures, like 404 or 500
• write output

[stereobooster]

We have link to raw files (for some hooks) and we already fetch source to detect subhooks on each CI build, just in case anybody wants to pick up this task

[aarohmankad]

For files that have the raw source (sourceUrl), I can't seem to just fetch that url.

[dmitryrn]

I would like to improve something here. I already made first-6-lines-source-preview (but it seems a little bit useless). Is this relevant to make PR here?

[dmitryrn]

@stereobooster I guess I’ll pick it up

[nikgraf]

@dmitryrn cool experiment. maybe rather showing the first 6 lines of code it would be good to have some insights on how many lines there are and possibly expand it to see the source?

What do you think? (keep in mind this are rough thoughts and need experimentation)