Support sub-thread starting points

[mwarkentin]

Not sure if I've never tried this before, or if Slack just changed something, but recently I copied a link to a message within a thread (I didn't care about anything before it in the thread) and using this link on gh-slack didn't work:

❯ gh slack --limit=1000 https://*****.slack.com/archives/C0220QQNUHE/p1684440714803369?thread_ts=1684440607.895759&cid=C0220QQNUHE | pbcopy
[1] 24189
zsh: no matches found: https://*****.slack.com/archives/C0220QQNUHE/p1684440714803369?thread_ts=1684440607.895759
[1]  + exit 1     gh slack --limit=1000

When I grabbed the first message in the thread the URL didn't have the query params and worked:

❯ gh slack --limit=1000 https://*****.slack.com/archives/C0220QQNUHE/p1684440714803369 | pbcopy

Ideally this would grab the contents of the thread starting from that point, but I'd also be ok if it just ignored the query params and pulled the whole thread.

[rneatherway]

Hey @mwarkentin, just got back from a trip. The only case I am completely sure that gh-slack works in is when reading a whole thread. Reading a partial thread (your case) or reading a unthreaded messages from a channel both have oddities.

That being said, I think to test properly you need to quote the URL because it contains an ampersand. Right now it's being interpreted by zsh as a request to background the gh slack invocation (hence [1] 24189). This happens to me with curl pretty regularly when I forget to quote the URL.

[mwarkentin]

I'll try that and let you know, thanks!

[mwarkentin]

Only took me 9 months, but I tried it with quotes and doesn't seem to help :D

states/base/files on  master [?] via 🐍 v3.11.7 (.ops-venv) on ☁️  [email protected](us-central1)
❯ gh slack --limit=1000 'https://sentry.slack.com/archives/C06MA1ST3B5/p1709648390444389?thread_ts=1709320542.716369&cid=C06MA1ST3B5'

states/base/files on  master [?] via 🐍 v3.11.7 (.ops-venv) on ☁️  [email protected](us-central1)

[rneatherway]

Hi @mwarkentin, I think the issue is down to the parsing of the permalink. Here is an example link for the first message in a thread:

https://sanity-io-land.slack.com/archives/C9Y51FDGA/p1709663536325529

No query parameters!

One for the third message in a thread:

https://sanity-io-land.slack.com/archives/C9Y51FDGA/p1709664320730949?thread_ts=1709663536.325529&cid=C9Y51FDGA

This code doesn't look at the query parameters at all:

gh-slack/cmd/gh-slack/cmd/read.go

Lines 42 to 55 in 86f738f

	// https://github.slack.com/archives/CP9GMKJCE/p1648028606962719
	// returns (github, CP9GMKJCE, 1648028606.962719, nil)
	func parsePermalink(link string) (linkParts, error) {
	result := permalinkRE.FindStringSubmatch(link)
	if result == nil {
	return linkParts{}, fmt.Errorf("not a permalink: %q", link)
	}
	return linkParts{
	team: result[1],
	channelID: result[2],
	timestamp: result[3] + "." + result[4],
	}, nil
	}

That means that it's the same as just requesting all the top level messages in the channel after the time of the original message in the thread.

I think just using the thread_ts query parameter if it exists would fix the issue. You can validate that by trying:

gh slack api get conversations.replies -f channel=C06MA1ST3B5 -f ts=1709320542.716369

and seeing if the expected messages come out.

[rneatherway]

@mwarkentin I've put together a fix, which was a little more complicated than I thought, over at #60. Could you give it a try please? If you check out that PR then you can run:

go run ./cmd/gh-slack/ --limit=1000 'https://sentry.slack.com/archives/C06MA1ST3B5/p1709648390444389?thread_ts=1709320542.716369&cid=C06MA1ST3B5'

[mwarkentin]

@rneatherway thanks! I will try to give that a shot tomorrow!

[mwarkentin]

Confirmed fixed over here: #60 (comment)