Page MenuHomePhabricator

IABot adds invalid accessdates
Closed, ResolvedPublic

Description

Hello the bot appears to be creating random accessdates when it completes the references, many of which are invalid. As an example on this edit https://en.wikipedia.org/w/index.php?title=Intersex&diff=851790773&oldid=850975552 it gives an accessdate of 20 October 2028 for one of the references. It should only give valid accessdates, leave the field blank or put in the date of the run.

Event Timeline

Cirdan renamed this task from Creating invalid accessdates to IABot adds invalid accessdates.Jul 24 2018, 7:40 PM
Cirdan edited projects, added InternetArchiveBot (v2.0); removed InternetArchiveBot.
Cirdan updated the task description. (Show Details)
Cirdan moved this task from Unsorted to Bugs on the InternetArchiveBot (v2.0) board.

A slightly different situation adding a second accessdate to a Citar web template that already has an acessodata parameter. See this edit https://en.wikipedia.org/w/index.php?title=Brazilian_Navy&diff=851815567&oldid=850721929

More invalid accessdates this time dates before wikipedia was created. There are 3 invalid dates in this edit https://en.wikipedia.org/w/index.php?title=WMSV&diff=852359185&oldid=849608373

Some of those dates are getting erroneously transferred from the snapshot date.

No idea where the 2028 date come from.

The old access dates issue is easily fixed as those are just being copied from the archive timestamps. I can't replicate what's causing the 2028 timestamp. My only suggestion is to fix these timestamps if you come across them in the DB. There is however a different bug I found that I can't explain.

I’ve added a sanity check to the subroutines to look for these odd dates. It shouldn’t be happening anymore as of beta7, however I couldn’t find the problem to replicate this to begin with.