Charcoal HQ

11:00 PM

there we go...

github.com/Charcoal-SE/SmokeDetector/commit/…

@ArtOfCode How hard would it be (theoretically) to add chat report message ID as a key to whever posts are stored in the MS DB?

Thomas Ward

In case you care, @ArtOfCode

SmokeDetector

CI on 49bd688 succeeded.

ArtOfCode

@quartata non-trivial but not impossible

quartata

Although wait there is one message per room...

ArtOfCode

11:01 PM

@QPaysTaxes what's the << operator?

NobodyNada

@QPaysTaxes Should be, except maybe if there's an overflow

@ArtOfCode bit shift

Right, but I don't know if they'd give the same result

SmokeDetector

[ SmokeDetector | MS ] Pattern-matching website in answer: Bayesian network for classification using PyMc or PyMc3 by arsenyinfo on datascience.SE

fp- by NobodyNada

NobodyNada

actually...hmm, this looks fishy.

Never mind; it's the same answer, just reported a whole bunch of times (every time Community bumps it)

quartata

@ArtOfCode Would it be harder to map multiple message IDs (if the report was posted in multiple rooms) to the same post?

I think that would require a separate table perhaps

I don't know the layout

ArtOfCode

either a new table or three new columns

still possible, but what do we get out of it

quartata

11:10 PM

Post placeholder -> edit placeholder to add report -> send to MS -> when a reply command is done simply take the chat ID and use it to retrieve whatever from MS

I'm trying to minimize what Smokey stores locally

ArtOfCode

why do we need to?

quartata

Well we need some form of new ID since by-post causes problems with edits. And I'd like to keep MS sending async rather than waiting for the MS post ID

SmokeDetector

CI on ac93a3d succeeded.

Thomas Ward

@ArtOfCode i just noticed I wasn't being PEP484 compliant with the post object lol

!!/pull

SmokeDetector

[ SmokeDetector ] SmokeDetector started at rev ac93a3d (Thomas Ward: Type hinting, IDE suppressions, PEP484 compliance in Post object.) (running on ArtOfCode/EC2)

Restart: API quota is 19065.

ArtOfCode

11:13 PM

@quartata is it that much of a problem?

the only time it's really a problem is when a post gets reported twice in a short (10m or less) timeframe, which is... rare.

NobodyNada

it happens whenever a user who vandalized their post is blacklisted, then the vandalism is rolled back

quartata

@ArtOfCode I think it is, yeah. Feedback on one can end up on the other.

^^

SmokeDetector

[ SmokeDetector | MS ] Username similar to website in answer: How to manual call agent to agent using Vicidial Agent API by VICIdial on stackoverflow.com

CI on b437448 succeeded.

ArtOfCode

45 secs ago, by ArtOfCode

the only time it's really a problem is when a post gets reported twice in a short (10m or less) timeframe, which is... rare.

SmokeDetector

naa- by NobodyNada

quartata

11:14 PM

41 secs ago, by NobodyNada

it happens whenever a user who vandalized their post is blacklisted, then the vandalism is rolled back

ArtOfCode

happens occasionally on vandalism, yes, but the keyword is occasionally

quartata

Which has happened twice in the past week

Thomas Ward

@SmokeDetector that doesn't need immediate pulling by the way - it's only necessary with the next pull since it only messes with a single comment and nothing functional.

quartata

It's still a weakness either way. Post URLs are not unique to a revision

And since you said "maybe we should totally rewrite smokey"... we have an opportunity to change that

Thomas Ward

i think we also said "You're free to try and undertake the task"

i highly doubt we're going to do a Smokey rewrite until we get everything we need as features, and then be 'done' for the most part :P

that said a Smokey rewrite would do wonders for my head lol

OH that reminds me i wonder how many instances of str() we have...

ArtOfCode

11:16 PM

@ThomasWard shrug on the other hand, Smokey development is never done. We might as well start a rewrite

quartata

I have started a rewrite.

Really just breaking ground and making stubs.

ArtOfCode

@quartata cool - want a repo on the org?

Thomas Ward

give standard Smokey people access too lol

quartata

@ArtOfCode Not yet.

@ThomasWard I'll implement the bare minimum of what I was thinking and put it up

Thomas Ward

@QPaysTaxes i'm excluding tests, but i find 11 instances with GitHub search (I'm also looking for case specific)

i've done a buttload of work on the Post object so if we can keep that that'd be great heh

meh was just curious

if we can make everything Unicode i'll be happy xD

ArtOfCode

11:19 PM

@quartata I assume you're using Py3?

quartata

@ArtOfCode Yes.

ArtOfCode

good

Thomas Ward

yay py3 lol

@quartata my Post object should already be Py3 compliant, I'll run it through Py3 inspections.

SmokeDetector

[ SmokeDetector | MS ] Bad keyword in body, bad keyword in title: Would working in a laboratory that produces dietary supplements reflect poorly on me in the future? by Krispyz on workplace.SE

NobodyNada

@SmokeDetector why

SmokeDetector

11:21 PM

Title - Position 53-64: supplements
Body - Position 565-576: supplements, Position 1254-1265: supplements

NobodyNada

:D

SmokeDetector

Conflicting feedback on How to manual call agent to agent using Vicidial Agent API.

tpu- by NobodyNada on How to manual call agent to agent using Vicidial Agent API [MS]

fp feedback on autoflagged post: Would working in a laboratory that produces dietary supplements reflect poorly on me in the future?

fp- by NobodyNada

NobodyNada

How on earth did I tpu- that?

@SmokeDetector f

wait, what?

Thomas Ward

no idea @NobodyNada but you did

quartata

@SmokeDetector autoflagged?

SmokeDetector

11:22 PM

@quartata That post was automatically flagged, using flags from: MAR.

NobodyNada

I must have done with the edge of my finger

quartata

@M.A.R. What is your threshold

ArtOfCode

Paging @M.A.R., you have an invalid autoflag that needs retracting

@quartata 175

quartata

Yikes.

ArtOfCode

@QPaysTaxes because Smokey autoflagged it successfully

@quartata nowt wrong with that

I wouldn't use it myself, but nowt wrong with it

quartata

11:23 PM

@ArtOfCode No I'm not blaming them. But goes to show how two words can really skyrocket the reason weight...

ArtOfCode

@QPaysTaxes nah, we never auto-nuke posts

aye

nah, that's just about the flag itself

Speaking of, I should make that fp on autoflagged message actually ping the relevant people

Thomas Ward

oopsies

quartata

@ThomasWard ???

Thomas Ward

i broke it

@quartata for the record, if you are doing a Py3 refactor, and intend to use my Post object, let me know, I'll give you a py3 version. Maybe.

Because it depends on a few other things but eh.

ArtOfCode

though actually, half the flaggers are never in here so... meh

Thomas Ward

11:26 PM

@QPaysTaxes lies, PURE UNADULTERATED CHAOTIC CHAOS is delicious.

quartata

Anyways I probably will need it. My intent is to make it so that every rule is whole_post

Thomas Ward

@quartata mmkay, I'll have to yank some parsing out of the parsing system currently and then move things to Py3 for that

but that should then work. GlobalVars, though, i can't 'erase' so mugh brains chaos souls_of_the_damned

ArtOfCode

Okay, while we're at it, what other terrible design decisions are there in Smokey that we should fix?

quartata

Which GlobalVars do you need?

Thomas Ward

@ArtOfCode check_if_spam -> test_post -> why,reasons should be simplified lol

@quartata um...

quartata

11:28 PM

Some of those may be stored in my shiny secret store

Thomas Ward

@quartata well obviously #1 is the CHQ chat socket.

I have a feeling that part of Smokey will remain the same

the rest is GlobalVars.parser.unescape

ArtOfCode

yeah, we'll still use ChatExchange

Thomas Ward

and that's it.

@ArtOfCode i meant init for that, etc. too

which would just be copypaste

ArtOfCode

Worth making config JSON instead of weird-format?

Thomas Ward

OH wait i can do some things

@ArtOfCode ehh, config is only a handful of lines, we could probably store as flat-JSON but I'd want a setup wiz. script for that

@quartata I can move the unescape stuff into Post direct, so no need for secret store things

it's just an HTML parser

quartata

11:30 PM

import smokey_chat; smokey_chat.send_and_forget(<msg>)

ArtOfCode

It'd be pretty nice if we had dynamic queue sizing

J F

@quartata Semicolon‽

Thomas Ward

@quartata cant if we need specific rooms.

quartata

@ThomasWard That's a different method.

Thomas Ward

also consider three separate chat systems

so three separate chat sockets

quartata

11:32 PM

I have written a few things

Thomas Ward

and some are CHQ-only reports

@quartata link :P

ArtOfCode

correction most

Thomas Ward

@ArtOfCode this will either blow up Smokey or not.

quartata

send_and_forget is more like logging. It only sends here by default.

ArtOfCode

@ThomasWard it'd be a nice feature, though - adjust queue sizes for the day based on yesterday's reports or something

quartata

11:33 PM

It could probably use a better name.

Thomas Ward

@ArtOfCode I meant this commit i'm making

I'm basically making Post have its own HTML Parser for the parsing :P

so we only need to depend on the CHQ chat socket in Post, and nothing else (because it needs it for a "Failed to Parse" case and reporting in here)

ArtOfCode

It'd just be creating the CSV's I made the other day in memory and applying the data to queues

@ThomasWard BS4?

quartata

BS4 is used by CE so we already have it. I'd prefer that.

Thomas Ward

@ArtOfCode acronym.expand("BS4")

ArtOfCode

that was my thinking

@ThomasWard I'm sure this call was language.expand_acronym last time you used it :P beautifulsoup4

Thomas Ward

11:36 PM

ehhhh

SmokeDetector

CI on 8722b06 failed.

Thomas Ward

foo what did I do

ArtOfCode

it blew up

Thomas Ward

@ArtOfCode well...

class acronym(language):
    def expand(self, stringdata):
        return self.expand_acronym(stringdata)

class extension ;)

ArtOfCode

yeah yeah

Thomas Ward

11:39 PM

well that CI is because I typoed.

whoops

ArtOfCode

of course

Thomas Ward

@ArtOfCode does BS4 have individual .unescape functions?

ArtOfCode

@ThomasWard for what?

NobodyNada

@ArtOfCode HTML unescaping, I'd guess

SmokeDetector

CI on f559c02 succeeded.

ArtOfCode

11:40 PM

@NobodyNada yeah, but individual whats?

NobodyNada

(unless you're asking why)

quartata

@ArtOfCode Entities?

I guess?

Does "SmokeDetector started at rev" get broadcast to all rooms?

Undo

Just here

Thomas Ward

@ArtOfCode we are unescaping titles apparently

and bodies

during init of the post object from the response

NobodyNada

@Undo \o

@quartata no

quartata

11:46 PM

Define unescape

Thomas Ward

self._title = self._parser.unescape(response["title"])
self._body = self._parser.unescape(response["body"])

Undo

o/

Thomas Ward

@quartata ask HTMLParser().unescape(somedata) what it does?

quartata

If you mean unescaping backslashes there's a thing for that

Thomas Ward

IDK what exactly it does

quartata

11:46 PM

@ThomasWard Ah, that's entities

Thomas Ward

where somedata is where it's pulling entities from the API response

or the JSON

i'd love to ditch HTMLParser for BS4, but we need to make sure that we still get the unescaping of data when we get the response from the SE API

it's also used elsewhere but meh

in spamhandling with logging, and in parsing.py

so meh

ArtOfCode

@ThomasWard if you're looking at spamhandling:125, just get rid of that log if you like

Thomas Ward

@ArtOfCode also 198 then

:P

SmokeDetector

[ SmokeDetector | MS ] Blacklisted user: How to find the area of a region? by anon420 on math.SE

Andy

@QPaysTaxes Eh?

Thomas Ward

11:53 PM

@ArtOfCode at the core though I still need to create an unescape function

possibly for BeautifulSoup as an extension

or just create an unescape class somewhere heh

ArtOfCode

@QPaysTaxes I also believe Shog shot that post down

Mods' opinions are valuable in elections

Andy

Seems familiar, but I don't recall context.

Undo

@QPaysTaxes Wait, what? I don't see that in the mod agreement o.O

Andy

Anyway...just check that nothing was in flames. I'm back to my vacation. :)

ArtOfCode

@Andy Context: posted by DVK after that SciFi election.

Undo

11:57 PM

It just says I can't "store or copy this information". Doesn't say I can't lie about it!

honorable lies only, though.

ArtOfCode

manually

Undo

@QPaysTaxes Yeah, but I think this is lawyery talk, not programmer talk

ArtOfCode

yeah, somehow I think SE recognises that they don't need to be that pedantic ;)

Andy

@ArtOfCode ah.

ArtOfCode

No more explanation required :)

Thomas Ward

11:59 PM

don't tempt me

i'll go into legalese spam time

Transcript for