User talk:Matěj Suchánek

Jump to navigation Jump to search

About this board

Maundwiki (talkcontribs)

The description "Wikimedia disambiguation page" should not be merged into a non disambiguation page. For example {{Q|1661989}}. See also [[User talk:Pasleim]] Merge of objects. I assume your code also should be fixed. Maundwiki (talk) 22:42, 16 June 2021 (UTC)

Matěj Suchánek (talkcontribs)
Maundwiki (talkcontribs)

Thanks I did not look back that far. ~~~~

Reply to "Merge of objects"
JiriMatejicek (talkcontribs)
JiriMatejicek (talkcontribs)

P.S. Informace u te fotky uz jsem kdysi opravil, ale az ted jsem si vsiml, ze ma jeste propojeni na Q31831638 v zalozce 'structured data'. Zapsal jsem i tam spravnou budovu, tak mozna se to tim spravi.

Matěj Suchánek (talkcontribs)

Dobrý den, jak je uvedeno ve zdroji, robot ji importoval v rámci obousměrné synchronizace seznamu s Wikidaty z jeho tehdejšího obsahu. K přidání došlo 20. 11. 2020 v 04:54, ze seznamu jste ji odstranil tentýž den až ve 14:25. Obrázek jsem znovu odstranil z položky Wikidat, robot by už teď neměl mít důvod ho tam znovu vrátit. (Díky za ten postřeh se strukturovanými daty na Commons, nicméně s těmito údaji robot nepracuje.)

Reply to "Q31831638"
CasserandeLucide (talkcontribs)

Hello, I have been assigned, by my employer, with the task of making edits/updating a WikiData record, as part of a wider Wikidata project. But MatSuBot has kicked protected page/lock status into gear. Please advise. Thank you.

Matěj Suchánek (talkcontribs)

Which items do you mean? I don't know what you mean, my bot is not supposed to "kick protected page/lock status into gear" in any way.

CasserandeLucide (talkcontribs)

I confused MatSubot with MsynABot - basically the problem is, a page that I need to continue editing appears locked to further edits, not sure how to become an 'autoconfirmed user' or how to access 'Template:Edit Request' as per action notation in History. I'll contact MsynABot if necessary. Sorry again to have bothered you!

Matěj Suchánek (talkcontribs)

Understood. Per Wikidata:Autoconfirmed users, you will (automatically) become autoconfirmed after 4 days and 50 actions. As for the edit request, you simply have to include {{edit request}} with your request on the talk page.

Gymnicus (talkcontribs)

Why did you delete this data object Q106538782? I just wanted to expand the “empty” object. In addition, the object had also received a link from me shortly before your deletion, which resulted in a structural need for the object.

Matěj Suchánek (talkcontribs)

Undeleted. Was deleted according to the request for deletion, but I didn't notice the new link.

Adding missing labels in en/pt/etc?

7
Mike Peel (talkcontribs)
Matěj Suchánek (talkcontribs)

Apparently not. The bot queries recent changes and tries to match on the edit summary the following way:

AND rc_source IN ('mw.edit', 'mw.new')
AND (rc_source = 'mw.new'
 OR comment_text LIKE '%wbsetsitelink-add%'
 OR comment_text LIKE '%wbeditentity-update%dded%[[%:%]]%'
 OR comment_text LIKE '%wblinktitles-connect%'
)

Basically, it catches all wbsetsitelink and (rare) wblinktitles calls. When wbeditentity is used, I try to find "Added/added ... [[some:interwiki link]]. Perhaps I should extend it to Add/add(ed) ... sitelink?

Mike Peel (talkcontribs)

Ah, I see. I've been using targetpage.editEntity(data, summary=u'Add '+prefix+'wp sitelink') - I'll change that now to targetpage.editEntity(data, summary=u'Added [['+prefix+':'+page.title()+']]') .

Mike Peel (talkcontribs)
Mike Peel (talkcontribs)
Mike Peel (talkcontribs)
Matěj Suchánek (talkcontribs)

The last one is returned from the query, so this should work. The previous two are not, as you have already found out. But since you originally used "add", it might be natural to have the bot match on the imperative as well.

Infovarius (talkcontribs)

Can you please provide tool for undoing your edits about merges? Like this. KrBot usually did such edits through EditGroups.

Matěj Suchánek (talkcontribs)

Hey, I didn't get to implementing it yet. For now, just tell me what should be reverted and I'll have my bot do it.

Matěj Suchánek (talkcontribs)

Starting tomorrow, the bot will register each item pair separately in EditGroups.

Reply to "Undoing merges"

Propojení položek na Wikidata

7
Unpocoloco (talkcontribs)

Ahoj, dobrý den. Přicházím s prosbou o vysvětlení a provedení sloučení položek.

Provedl jsem sloučení dvou položek, které podle mě řeší stejné téma. Jedna má cca 30 mezijazykových odkazů na různé wiki, druhá pouze na češtinu. Sloučení bylo následně revertováno, že bylo chybné. Jde o položky Q339180 a Q69796120 - ta druhá je rozcestníkem, a první ne. Možná to byl důvod vrácení zpět. Předem díky.

Matěj Suchánek (talkcontribs)

Je to zřejmě tak.

Jedna věc je, že to sloučení do druhé položky překopírovalo všechny popisky (z vlastní zkušenosti vím, jak zdlouhavé je takové případy uklízet). Ta druhá je zásada, že položky označené (šablonou, kouzelným slovem) jako rozcestníky mají zůstat oddělené. Ale třeba by nevadilo, kdyby se ten odkaz jenom přesunul z jedné položky do druhé (podobně je na tom i link na iswiki)...

Unpocoloco (talkcontribs)

Ahoj. Bylo by tedy řešením, že bych českému heslu přidal trošku nějakého úvodního textu a odstranil z něj šablonu a kategorii Rozcestník? Následně i u Q69796120 také odstranil tu položku "Rozcestník" a následně ji začlenil do Q339180? Jsem přesvědčený, že by cs.heslo nemělo být osamocené.

Matěj Suchánek (talkcontribs)

Jo, i to je možnost. Ovšem z položky Q69796120 bych ten odkaz jenom vyjmul a jinak položku nechal být, případně předal k odstranění.

Unpocoloco (talkcontribs)

Ahoj. Tak jsem to provedl tak, jak jsi psal. Odkaz na české heslo z Q69796120 vyjmul a vložil ho do Q339180. S Q69796120 jsem nic dalšího nedělal. Připadá mi jako takový pohrobek bez užitku, ale zde na Commons se bojím s tím něco dělat. Díky za pomoc.

Matěj Suchánek (talkcontribs)

Nejsme na Commons, ale na Wikidatech. Díky za provedení, nezbylo než položku smazat.

Unpocoloco (talkcontribs)

Ano, Wikidata, ne Commons. Překlep. I tak jsem v krocích zde nejistý. Díky za dotažení.

disambiguators and prefixes depending on P31

3
Jura1 (talkcontribs)

I suppose your bot adding labels based on sitelinks already does some of it.

Can you remove disambiguators directly also for:

If it works out, I can expand the list. BTW for TV seasons, it can keep it as there the part in parenthesis is generally not a disambiguator (sample fixed since).

For eswiki/Spanish labels, could it remove "Anexo:" directly for the following types:

The last one was discussed here.

Matěj Suchánek (talkcontribs)

I try to avoid hardcoding language-specific strings and item identifiers to code. The bot will remove these disambiguators when there is a match with the description (because then it's likely the information is redundant).

Jura1 (talkcontribs)

I understand.

The removal in these cases generally makes sense, but for Q5, I think it can always be removed. There is now a bot task that cleans these up ;)

Also, typically it would be problematic for TV seasons, but that might just be an exception.

Reply to "disambiguators and prefixes depending on P31"
Rei Momo (talkcontribs)

Good evening, dear Matěj Suchánek.

This little message is to thank you to put new Languages in Wikidata about "La vita comincia ogni mattina".

Greetings From caselle Landi, Italy.

VIGNERON (talkcontribs)

Hi,

What do you think of adding Lexemes (page_namespace == 146) on the Special:AbuseFilter/49? Is it a good idea? And if so, could you edit the filter? (for the context, I just removed some self-referencing on Lexemes, there is some false-positive but probably no more than for the items).

Cheers,

Matěj Suchánek (talkcontribs)

Hi. It is probably a good idea if it happens frequently (well, we probably do not know that). Though I do have one concern to share: what about linking from/to an embedded form/sense?

VIGNERON (talkcontribs)

Hi,

I'm not sure it's very frequent but FYI, I corrected 10 today.

Yes, that's a good point. I'm not sure to fully understand the pattern: could it be exactly the page_title? (to exclude the form and sens)

Matěj Suchánek (talkcontribs)

Well, for items we have to match just the numeric part. For lexemes, I can see we can match the whole page_title. But that wasn't my point. What I actually meant was link from a form/sense to the "parent" lexeme. (Maybe we could also check self-links for forms and senses, but I doubt it's possible.)

VIGNERON (talkcontribs)

Oh ok. References from form or sens inside lexeme does happens a lot, it's may be seen as selference but it's not, and it's common and on purpose (see this diff Special:Diff/1147794431 on Lexeme:L69#P1343 for an example), so it should not be taken into account by this filter.

Matěj Suchánek (talkcontribs)

Well, what you are demonstrating is a link from the "parent" to the form/sense. But forget it, that you cannot make up an example is probably a justification for not ignoring it, at least for now. I will take some time and then change the filter.

VIGNERON (talkcontribs)

Oh, you meant the other way around, sorry. No indeed we don't do that. In fact, with a SPARQL query, I only found one example Poulsen (L411879) which is a probably mistake.

Matěj Suchánek (talkcontribs)

Done.