Tavern on the Meta

ProgramFOX

6:11 PM

SPAM meta.stackexchange.com/questions/249915/…

Doorknob

kaboom

Jan Dvorak

At least sh.st seems to be gone

Doorknob

Hey @Undo, can SmokeDetector have regexes for "bad keyboard in body"?

I guess it can, because it can detect repeating characters and whatnot.

user259867

@Doorknob It does have them. Check the source.

Kevin Brown

@Doorknob Pretty sure that's a yes

ProgramFOX

6:14 PM

@Doorknob Yes, it can, and it already does.

user259867

In the file findspam.py the rules with body: true apply to post bodies.

Doorknob

Shouldn't the "bad keyword in title" hooks also apply to the body, then?

Pretty sure either "muscle" or "testosterone" (or both) are blacklisted, but SD didn't catch that spam.

ProgramFOX

@Doorknob They do, but for body fetching, all posts are added to a queue so title and body are not checked at the same time.

Doorknob

Ah, right. Okay then.

ProgramFOX

Also relevant:

in Charcoal HQ on The Stack Exchange Network Chat, yesterday, by hichris123

I have a feeling Smokey is missing a post sometimes when it fetches from the API (i.e. it tries to fetch 2 posts, but sometimes gets two and sometimes only gets one).

and further messages

cv-pls too broad stackoverflow.com/questions/28660985/…

err... wait, link not on my clipboard anymore

hichris123

6:18 PM

What are we closing now? Yourself? :P

ProgramFOX

:p

hichris123

@ProgramFOX Although IIRC for meta posts it checks on every post.

ProgramFOX

oh, right

Then perhaps the post wasn't included in the API data yet.

hichris123

Probably.

Which means we need to find a good way to do what I was saying.

Or find a different way to make sure we get all posts.

ProgramFOX

yeah

Jan Dvorak

6:21 PM

or just convince the devs SE needs an official modding API with a realtime answer feed

hichris123

I wish.

Doorknob

And while we're at it, I want a unicorn and a million dollars.

user259867

yesterday, by Tim Stone

There's up to 1 minute of lag

user259867

Bodyfetcher.py waits 30 seconds at present.

ProgramFOX

in Charcoal HQ on The Stack Exchange Network Chat, yesterday, by hichris123

Perhaps, but I don't want to make it too long as then the spam might already be gone.

Jan Dvorak

6:23 PM

Or we could just ~~blackmail SE~~ give up spam hunting until we get proper tools

ProgramFOX

Waiting 60 seconds should probably be feasible though. What do you think @hichris123?

Doorknob

@ProgramFOX Is one minute really too long to ask for?

hichris123

@FamousBlueRaincoat My idea was to put posts back in the queue if we don't get them from the API, but then we'd have to figure out how to do that. :P

@ProgramFOX How about add that for now, and we can find a long-term solution later.

ProgramFOX

good idea

hichris123

@Doorknob That's 60000 ms!

Doorknob

6:24 PM

@hichris123 Or, 1/60th of an hour! :P

Come on, is 1/86400th of a day really too much? :P

hichris123

Think what we could be doing in that amount of time.

ProgramFOX

!!/rev?

SmokeDetector

@ProgramFOX e8ee117 (ProgramFOX: Wait 60 seconds. --autopull)

ProgramFOX

done

Doorknob

yay

user259867

6:26 PM

The little summary that comes from websocket (first ~200 chars of the question) often contains those bad keywords already. If one is really impatient, that could be checked immediately.

hichris123

@FamousBlueRaincoat Oooh, good point.

Doorknob

How about we all fly to New York and threaten random SE staff until someone makes it so that the entire body is sent with the websockets?

@FamousBlueRaincoat I just noticed how interesting that URL is.

hichris123

Hmm, so that's called bodySummary - what do you think, @ProgramFOX?

ProgramFOX

I think it's a good idea.

hichris123

@ProgramFOX I'll put it in.

ProgramFOX

6:30 PM

k

user259867

@Doorknob Interesting? It's a public Dropbox folder because I'm lazy.

ProgramFOX

lol, the tool says "You're welcome!" are my wise words.

Doorknob

@FamousBlueRaincoat What? That's an imgur URL.

Click the reply arrow

yesterday, by Famous Blue Raincoat

user259867

Oh, QQQvQ.png name... perhaps that's interesting...

user259867

@ProgramFOX That's a hint to delete a chatty comment...

ProgramFOX

6:38 PM

heh :p

Kevin Brown

I noticed some of my "words of wisdom" were auto-generated comments of the VLQ queue

user259867

@KevinBrown The wisdom tends to be wiser on meta sites.

hichris123

status-completed

user259867

That was a pretty easy modification...

hichris123

Yup. :)

Kevin Brown

6:42 PM

@FamousBlueRaincoat Ah, much better :)

Braiam

6:54 PM

offensive right? askubuntu.com/a/588624/169736

SmokeDetector

[ SmokeDetector ] Repeating characters in body: How to print a this diamond pattern in java? by Yash Jain on stackoverflow.com

61612

False positive

Kevin Brown

@SmokeDetector The answer there could probably be improved

hichris123

@61612 And probably because bodySummary doesn't include HTML. :/

SmokeDetector

[ SmokeDetector ] Repeating characters in body: loop test strings to see if input is valid for C program input by jmunsch on unix.stackexchange.com

hichris123

6:59 PM

Ugh, yeah.

@ProgramFOX still around?

user259867

Another of the same. Yes, it's a problem. Suppressing repeated-chars check specifically for that thing seems nontrivial.

hichris123

@FamousBlueRaincoat Yeah. I was thinking of stripping code blocks... but I'm not sure if it gives us Markdown.

user259867

@hichris123 It's plain text.

user259867

One solution (kludge) would be to run the repeated char regex on bodySummary and to not send it if matches.

hichris123

Yeah... hm...

user259867

7:01 PM

Because repeated chars in the first 200 characters of a question are probably fp.

user259867

May be better to strip them away with replace, actually.

hichris123

Someone may have replaced their question with aaaaaaaaaaaaaaaaaaaaaaaaaa though.

user259867

@hichris123 So that would wait 60 seconds to be detected in the normal way.

hichris123

True.

SmokeDetector

[ SmokeDetector ] Offensive answer detected: Installing XRDP on Cent OS 6.5 by Asshole on unix.stackexchange.com

user259867

7:11 PM

@SmokeDetector gone

ProgramFOX

@hichris123 yep

hichris123

@ProgramFOX We're now catching repeated characters from bodySummary even though they're in code blocks.

Although it looks like the websocket gives plain-text, so...

ProgramFOX

Plain text? Hmm...

In that case, we could add an extra parameter to check_if_spam and if that's false, don't check for repeated chars in the body.

hichris123

We could...

SmokeDetector

[ SmokeDetector ] All-caps title: PHP SEARCH NEED HELP PLEASE by maiko on stackoverflow.com

ProgramFOX

7:21 PM

I'll add a thing to do that now.

61612

cv-pls the above

> - The search party is on its way now. Please stay calm. - Thankyou

ProgramFOX

7:46 PM

!!/rev?

SmokeDetector

@ProgramFOX b16c899 (ProgramFOX: Exclude some regexes from the body summary)

ProgramFOX

@hichris123 ^

Got to go now, cya!

Mooseman

0

Q: Please make [document.ready] a synonym of [document-ready]

Both tags have the same meaning, although only one question is tagged with both.

nicael

lifehacks.stackexchange.com/review/suggested-edits/2827

lulzedit

Mooseman

@nicael I've been removing the tag: markdown from excerpts...

NAA stackoverflow.com/a/28662258/1234256

delete-vote-pls stackoverflow.com/q/5525959/1234256

hichris123

7:53 PM

@Mooseman why?

Kevin Brown

@Mooseman Definitely shouldn't be reopened, I just pushed it out of that queue.

SmokeDetector

[ SmokeDetector ] Repeating characters in body: Last line is wrong when splitting a file with awk by user40129 on unix.stackexchange.com

61612

False positive again...

user259867

Let's find the bug!

user259867

8:10 PM

(False if body_is_summary is True and rule['body_summary'] is False else True) looks suspicious

Kevin Brown

Looks somewhat reasonable without context

hichris123

@FamousBlueRaincoat What is it?

user259867

That one was okay, actually.

user259867

This looks like the bug:

hichris123

Yeah, that's it.

user259867

8:14 PM

The last parameter of check_if_spam should be True in this case.

Braiam

@nicael huh? Why it says that Mooseman is the author?

user259867

But the line (False if body_is_summary is True and rule['body_summary'] is False else True) ought to be expressible in a nicer form: it amounts to (NOT body_is_summary) OR rule['body_summary']

Braiam

cv-pls stackoverflow.com/q/27229312/792066

cv-pls stackoverflow.com/q/9263920/792066

ugh stackoverflow.com/q/14684507/792066

Kevin Brown

@Braiam Just going through [musicxml]?

Braiam

@KevinBrown actually no, linked post from a closed question

hichris123

8:26 PM

@FamousBlueRaincoat Done.

nicael

@Braiam cuz he is the creator of the first excerpt. To be honest he is just the first filler of this except, the creator is Community, but this is how the things work :D

Kevin Brown

@Mooseman Looks like you're getting rejected on that one

Mooseman

approve-pls stackoverflow.com/review/suggested-edits/7126728

approve-pls stackoverflow.com/review/suggested-edits/7126727

Braiam

that reject reason for both are totally BS

how something can be made easier to read, if it didn't exist?

or harm?

nicael

0

A: How do I fill a hot water urn that doesn't fit in the sink?

This is really simple, but if you urn is very big, it is quite difficult to do :D take a smaller volume fill it with water from your tap pour the water from the smaller volume into your urn. Another variant: take a large plastic bottle (e.g. 1.5L+) cut the bottom and put the bottle in this ...

How do you like my freehand? :D

Kevin Brown

8:33 PM

+1, great drawing

SmokeDetector

[ SmokeDetector ] Repeating characters in answer: Bypassing Transistors with a Switch, problems? by user68375 on electronics.stackexchange.com

61612

Yay, true positive

user259867

Or any large plastic bottle with the base cut off... — RedSonja Feb 2 at 14:45

nicael

Wait, bad :( But she didn't specify how to cut. And provided no freehand.

Braiam

mm... wut?

hichris123

8:40 PM

I really think they've gone off the deep-end...

user259867

That whole tag... judaism.stackexchange.com/questions/tagged/purim-torah-in-jest

2

SmokeDetector

8:56 PM

[ SmokeDetector ] All-caps title: PLEASE HELP W EASY MATH QUESTION by Sam on math.stackexchange.com

[ SmokeDetector ] Repeating characters in body: firebase promises, (data), and index values of asArray by Chris L on stackoverflow.com

Kevin Brown

@SmokeDetector Quality issues on that question

Braiam

9:21 PM

'Rating: 1/5. Room filled to brim with semen, and when front desk clerk opened mouth to talk, bedbugs poured out.'

Hobo Sapiens

cv-pls ugh! Ugh! Ugh! stackoverflow.com/q/28663418/1864610

hichris123

@HoboSapiens I tried... and then remembered I was out of close votes for the day.

Hobo Sapiens

lol - I hate it when that happens

user259867

@HoboSapiens gone

Hobo Sapiens

@FamousBlueRaincoat 6 minutes - not bad.

hichris123

9:36 PM

cv-plz stackoverflow.com/questions/28663530/…

user259867

10:08 PM

You ever accidentally spend 2 hours working on something before checking to see if the APIs even support it?

Tweeted by jc4p on February 22, 2015 at 8:26 PM

2

Mooseman

10:48 PM

Good. It's not just me.

user259867

@SmokeDetector fp

SmokeDetector

@FamousBlueRaincoat Registered question as false positive and added title to Bayesian doctype 'good'.

user259867

11:14 PM

@SmokeDetector fp

SmokeDetector

@FamousBlueRaincoat Registered answer as false positive.

[ SmokeDetector ] Repeating characters in body: db2 - export into File by tiago on stackoverflow.com

user259867

@SmokeDetector fp

SmokeDetector

@FamousBlueRaincoat Registered question as false positive and added title to Bayesian doctype 'good'.

[ SmokeDetector ] Bad keyword in answer, Offensive answer detected: What if an asteroid the size of the moon hitting earth by user73788 on physics.stackexchange.com

user259867

@SmokeDetector tpu

SmokeDetector

@FamousBlueRaincoat Blacklisted user.

Mooseman

11:32 PM

cv-pls opinion stackoverflow.com/q/28664688/1234256

Kevin Brown

@Mooseman closed

Thoughts on stackoverflow.com/questions/28204978/… being on-topic or not?