harvest_template.py should import item of redirect, not item of target page
Open, LowPublicBUG REPORT
Actions

Assigned To

Authored By

	JAnD
	Jul 1 2022, 7:10 PM

Description

List of steps to reproduce (step by step, including full links if applicable):

pwb.py harvest_template -prefixindex:Wikipedie:Pískoviště -template:"Hlavní článek" "1" P301 -lang:cs

there are three redirects:

redirect (without wikidata item) to another article
redirect to section of another item (both redirect and target have own wikidata item)
redirect to another article (both redirect and target have own wikidata item)

2 and 3 are marked as Wikidata redirect with template Q16956589

What happens?:
Script imports Q144, Q11567 and Q1471

What should have happened instead?:

script should import Q144, Q9165172 and Q472766

Details

	Subject	Repo	Branch	Lines +/-
	[IMPR] Only follow redirects in harvest_template.py if no wikibase item exists	pywikibot/core	master	+74 -19

Customize query in gerrit

Related Objects

Mentioned Here: T54564: Allow sitelinks to redirect pages to fix the 'Bonnie and Clyde problem'

Event Timeline

JAnD created this task.Jul 1 2022, 7:10 PM

Restricted Application added subscribers: pywikibot-bugs-list, Aklapper. · View Herald TranscriptJul 1 2022, 7:10 PM

JAnD renamed this task from harvest_template.py should import target of redirected page to harvest_template.py should import item of redirect, not item of target page.Jul 1 2022, 7:15 PM

Maintenance_bot added a project: Pywikibot.Jul 1 2022, 7:29 PM

Xqt added a subscriber: Multichill.Jul 1 2022, 10:24 PM

Cannot follow this issue and need any help. Let's explore the last added Q1471 which is intended to be Q472766:

The starting link text is Sequana and HarvestRobot._template_link_target found [[cs:Sequana]] as linked_page
[[cs:Sequana]] is a redirect and the method above uses the redirect target which is [[cs:Saina]]
The corresponing wikibase item of [[cs:Saina]] is Q1471.
[[cs:Sequana]] indeed has its own wikibase item which is Q472766

This is a bit confusing to me but looking forward the behaviour must be cleared:

When should the redirect page be used?

always; there is no default. create a wikibase item if it does not exist.
if a wikibase item not exists, otherwise use the current behavior as default
if a wikibase item not exists or there is no __STATICREDIRECT__ given for the redirect
if a wikibase item not exists or there is not a wikidate redirect template inside the redirect page (Q16956589), otherwise use the current behavior as default
Other behaviour
One of 1-4 above but with an additional bot option
never; the current behaviour is the right one

Xqt triaged this task as Medium priority.Jul 2 2022, 3:00 PM

Xqt claimed this task.Jul 2 2022, 3:56 PM

Change 810472 had a related patch set uploaded (by Xqt; author: Xqt):

[pywikibot/core@master] [IMPR] Only follow redirects in harvest_template.py if no wikibase item exists

https://gerrit.wikimedia.org/r/810472

gerritbot added a project: Patch-For-Review.Jul 2 2022, 4:01 PM

Xqt added a subscriber: Legoktm.Jul 3 2022, 6:58 AM

I keep this open to discuss whether harvest_templates should follow redirects under the condition

there is no __STATICREDIRECT__ magic word
there is not WD template Q16956589
the new behaviour should only work with a new option

Change 810472 merged by Xqt: