Rich Text / PDF Render and Use as a Resume

[ctjlewis]

Enough hard work has been put into making this tool that, looking at the output, I would be willing to add it as a supplicant to my resume or even submit in lieu of an actual resume.

Has anyone else considered this? What might this implementation look like, and what small details might be helpful to add to support this use case?

[lowlighter]

I guess the easiest way would be to create multiples jobs like in this workflow using config_output with png or jpg to generate different images from plugins you're interested and embed them on an existing resume.

A more experimental and boldy way would be to start off from a markdown template, fill it up with text and template it with data gathered by metrics and rendered plugins images. With Ctrl + P on a browser you could get a PDF file from generated markdown.
Actually, if we plugged metrics to LinkedIn API to get job data it may be possible to have a fully automated resume builder with day-to-day data 🙂

[lowlighter]

So I digged a bit about an eventual LinkedIn plugin, but it seems impossible to implement unfortunately 😕 (at least atm)

Basically there are no personal tokens so it's required to pass through an OAuth 2.0 token which is a pain to setup when you just want a token for yourself (+ it seems to be valid only for 2 months max so double pain).

Even if you pass through an app, maximum permissions you can get as lambda user is r_liteprofile or r_basicprofile if you're lucky (and you must create a "fictive company"), but to make something usable by metrics r_fullprofile would be required.

It's clearly not designed for individual usage plus documentation seems a bit evasive and there's some broken links a bit everywhere on the developer portal 😞 I understand that they're not giving access easily to whole userbase but kind of bummer that you can't even get back your own data...

A solution based on puppeteer seems impossible too because of the authwall, and if using some kind of hack to log yourself you could just get banned so it's a no-go 😅

Best that can be done is to setup an app which has to be approved to get r_basicprofile and get headline which is basically your current job name, but it's not worth the hassle.

I sent a support ticket to LinkedIn to see if there's any way to do it but I'm pretty sure it'll be a negative answer (if I get any)

TLDR; no fully automated resume on GitHub readme profile with LinkedIn for now 🙂

[ctjlewis]

Ahhhh it works! I will work on the template!

Does the README give instructions on how to do a local run? I'd love to just throw my config in and run it locally, tweak the template as I go. I thought yarn start would cover this but not quite.

Will send PR.

---

Footnote: I tried to use nektos/act to debug the workflow locally but it complains about the using: composite flag.

[ctjlewis]

Also, so sorry, but I can't find the example-isocalendar-pdf plugin. Could you help me out? 😅

Also, how would I go about adding the base section? I tried everything from:

embed('base', { base: ... })
embed('isocalendar', { isocalendar: true })

// to:
embed('resume', { base: ..., isocalendar: true })

And have only managed to get the calendar to render. At first I thought example-isocalendar-pdf was its own plugin but that doesn't seem like the case now.

Sorry for all the questions! I will add all this to the docs for you at some point based off whatever notes.

[lowlighter]

Also, so sorry, but I can't find the example-isocalendar-pdf plugin. Could you help me out? 😅

The first argument is a unique string which gets used as filename when the action is run on GitHub (it gets stored in markdown_cache folder). AFAIK it's not possible to upload programmatically to GitHub camo and since base64 sources images don't render in markdown, metrics store images in target repository instead. So as you figured out it doesn't actually matters what you put in there 🙂

Also, how would I go about adding the base section?

I improved the inputs normalization in embed function, when you tested it only supported dot.case (web syntax). Now you should be able to use all syntaxes for base:

• {base:"header, community, ..."} (action syntax)
• {"base.header":true, "base.community":true, ...} (web syntax)
• {base:["header", "community", ...]} (parsed syntax)

Others option should works both in snake_case (action syntax) and dot.case without plugin prefixes (web syntax) 👍

---

Does the README give instructions on how to do a local run? I'd love to just throw my config in and run it locally, tweak the template as I go

Running locally in "action mode" has not been implemented properly 🙂

Currently the start script make it run in web server mode because historically metrics was a web service first and later switched to an hybrid app. There is still a "hacky way" to do it which is used in tests/metrics.test.js:

metrics/tests/metrics.test.js

Lines 11 to 27 in 5b63ecb

	//Github action
	const action = yaml.load(fs.readFileSync(path.join(__dirname, "../action.yml"), "utf8"))
	action.defaults = Object.fromEntries(Object.entries(action.inputs).map(([key, {default:value}]) => [key, value]))
	action.input = vars => Object.fromEntries([...Object.entries(action.defaults), ...Object.entries(vars)].map(([key, value]) => [`INPUT_${key.toLocaleUpperCase()}`, value]))
	action.run = async (vars) => await new Promise((solve, reject) => {
	let [stdout, stderr] = ["", ""]
	const env = {...process.env, ...action.input(vars), GITHUB_REPOSITORY:"lowlighter/metrics"}
	const child = processes.spawn("node", ["source/app/action/index.mjs"], {env})
	child.stdout.on("data", data => stdout += data)
	child.stderr.on("data", data => stderr += data)
	child.on("close", code => {
	if (code === 0)
	return solve(true)
	console.log(stdout, stderr)
	reject(stdout)
	})
	})

Basically it parses action.yml (for defaults values), convert inputs to environment variables (e.g. INPUT_*) and then spawn a child process which acts the same as a "GitHub runner".

Tests are only run in dryrun so there is actually no output. On GitHub runners, files get written to /renders which is mounted to /metrics_renders folder. We could edit this so when run locally a CLI option lets it write somewhere else 🤔 ?

metrics/source/app/action/index.mjs

Lines 280 to 286 in 5b63ecb

	if (dryrun)
	info("Actions to perform", "(none)")
	else {
	await fs.mkdir(paths.dirname(paths.join("/renders", filename)), {recursive:true})
	await fs.writeFile(paths.join("/renders", filename), Buffer.from(rendered))
	info(`Save to /metrics_renders/${filename}`, "ok")
	}

I suppose this could be used as a workaround for now since composite actions are not supported yet by nektos/act as per nektos/act#607. A PR has been opened for it but not sure when it's going to be merged

Sorry for all the questions! I will add all this to the docs for you at some point based off whatever notes.

No problems! Hope it helped 🙂

[ctjlewis]

It definitely helped! I'm making progress, and got the PDF to render the sections I want.

Final thing: Would there be a way to inherit the top-level settings? Can this template read from the same variables in the workflow file that the profile render does? I think if it could inherit the defaults and allow for overrides with a pdf_ prefix, i.e. pdf_base: ['header', 'activity', ...] (my profile by default does not include header), this would be pretty much close and shut on features and would be good to document and roll out.

Once you can configure what goes on the PDF with arbitrary precision, and the template inherits the workflow settings by default, this is pretty much implemented. Also really, your work here has been absolutely first-class.

[lowlighter]

Would there be a way to inherit the top-level settings?

I edited a bit the embed function so it inherits all top-level options by default (it makes more sense anyway) except "enable switches", so everything will be pre-configured but you can still choose when to display each plugin in template.

with:
  base: header, repositories
  plugin_isocalendar_duration: full-year

<!-- Isocalendar plugin, defaulting to "full-year" as per top-level definition -->
<%- await embed(`isocalendar`, {isocalendar:true}) %>
<!-- Isocalendar plugin, overriden to "half-year" -->
<%- await embed(`isocalendar-half`, {isocalendar:true, isocalendar_duration:"half-year"}) %>

<!-- Set base to "true" to default to top-level definition -->
<%- await embed(`base`, {base:true, notable_repositories:false}) %>
<!-- Or override base definition -->
<%- await embed(`base-custom`, {base:"community, activity"}) %>

I don't think pdf_ options are needed with this inheritance system, since it's probably easier to edit the template directly

Also really, your work here has been absolutely first-class.

Thanks a lot for the compliment 🙂

[ctjlewis]

I got this mostly working and it's pretty beautiful (GitHub PDF preview kinda messes it up but it looks fine on download). With a little HTML in the markdown, we can force page breaks, which is how this is blocked out (otherwise it just runs out into the next page):

<div style="break-after: always; page-break-after: always;"></div>
<div style="padding-top: 3vh;"></div>

One issue I think I found: I can't override activity_days, I tried 0 and 365 days and can't get the window to change from 14 days, sadly. In this preview you'll notice the activity page renders the last two weeks of events, but activity_days: 365 is set in the embed.

But it looks great honestly. The final thought I'm having is, can we get these to render as actual anchor links? The markdown links seem to work out of the box (which is amazing) but the GitHub metadata (like "repositories", links in activities sections) do not.

Also, finally, is there a way for me to see these console.debug statements and so on, even if I can't run it locally? Super sorry for the slew of questions again lol.

[ctjlewis]

Incredible. Thank you so much. A few days ago I started getting "an error occurred" for activity so I think the window is too wide.

[lowlighter]

If you set debug: yes on your workflow I can take a look at what causes the error if needed

[ctjlewis]

I merged from upstream and now the activities page is blank when I comment out the overrides, and the +- lines changed also errored:

https://github.com/ctjlewis/ctjlewis/blob/master/resume-template.markdown.pdf
https://github.com/ctjlewis/ctjlewis/runs/2558007075?check_suite_focus=true

[lowlighter]

I've pushed 421a770 to fix activity plugin:
https://raw.githubusercontent.com/lowlighter/ctjlewis/master/resume-template.markdown.pdf

As for lines and empty stats in community sections I'll check later (I forked your repositories for easier debug) 👍

[ctjlewis]

Thanks a bunch Simon. No rush on any of this, the activities section def used to work, but it seemed frozen to the default lookback. I tried to tinker around and couldn't get it unstuck sadly.

[lowlighter]

The upcoming config_display for v3.10 may be quite convenient for this feature as it'll allow to take advantage of the larger width of a pdf page and will feel more natural 🙂

[jayvdb]

StackOverflow used to have an "Import from Linkedin", but it was removed because of Linkedin TOS. c.f. https://meta.stackoverflow.com/questions/312851/feature-request-import-cv-from-linkedin

While not as "good" as Linkedin, StackOverflow has a "developer story" feature which could be used as a data source. I couldnt see if this has an API (it doesnt appear in https://api.stackexchange.com/docs) , but the HTML source is quite basic to parse (see source of https://stackoverflow.com/cv/jayvdb for example). Ah, they have hresume support .. ? c.f. https://meta.stackexchange.com/questions/28375/should-careers-stackoverflow-com-support-hresume/32216

[lowlighter]

This looks interesting, may be worth to dig deeper 👍