Wikipedia talk:PetScan

Latest comment: 28 days ago by Kostja in topic PetScan not working

Bug or what: randomly adding "®exp_filter" into fields edit

Sometimes PetScan adds ®exp_filter into text fields after I hit the "Do it!" button. Those can be found in random fields, like in Linked from or Site links text fields, sometimes in multiple fields and other times just in one field. Any idea what's causing this odd behaviour? Everytime this happens, I'll have to reload a fresh PetScan page to make PetScan work properly because those "®exp_filter"'s break the search. 87.95.206.253 (talk) 17:31, 20 January 2021 (UTC)Reply

I've seen ®exp_filter appear in the "Link to a pre-filled form for the query you just ran" URLs, causing them to appear in the form when the external link is followed. Certes (talk) 17:40, 20 January 2021 (UTC)Reply
I have the same problem. It is annoying. How can I urge this problem to be solved? --Herzi Pinki (talk) 10:09, 6 October 2022 (UTC)Reply
PetScan accepts a URL parameter called regexp_filter. "Link to a pre-filled form..." inserts &regexp_filter=value into the URL it generates. (value is usually blank.) This works perfectly when the link is clicked. However, such links are often pasted into wikitext or HTML. Some browsers interpret the string "&reg" as an HTML entity for the registered trademark symbol, even when not followed by a semicolon, so &foo=bar&regexp_filter= becomes &foo=bar®exp_filter=, appending the unwanted text to the preceding parameter and usually causing it to make an unwanted appearance in one of the input boxes.
Could we change the parameter name to something that doesn't begin with reg? Of course, regexp_filter= should also be accepted for backwards compatibility, but if a replacement parameter name not beginning with reg can be offered in "Link to a pre-filled form" then the problem should be solved. Certes (talk) 23:37, 1 February 2023 (UTC)Reply

Other discussions about this longstanding annoying bug:

2001:14BA:9C35:6600:193F:16A0:BF8A:43A1 (talk) 18:55, 8 July 2023 (UTC)Reply

I have replaced 'regexp_filter' with 'rxp_filter', which should prevent the issue. 'regexp_filter' is still supported for backwards compatibility. --Magnus Manske (talk) 10:55, 22 December 2023 (UTC)Reply

Great news; thank you! Certes (talk) 10:58, 22 December 2023 (UTC)Reply

No result for source categories edit

Hi, I'm new to Petscan, just tried using it this evening and I got the following error: "No result for source categories".

This error happened with one of the example queries as well: https://petscan.wmflabs.org/?psid=552

Any ideas what I've done wrong? audiodude (talk) 01:25, 16 July 2023 (UTC)Reply

...and it's working again! audiodude (talk) 01:38, 16 July 2023 (UTC)Reply

Petscan displays blank page, unusable edit

https://petscan.wmflabs.org/ just displays a blank page. Taking a peek at the HTML, I see this likely culprit: <body style="display:none">. Reported on Discord, affects at least 2 people. –Novem Linguae (talk) 00:03, 26 September 2023 (UTC)Reply

I think it's probably unrelated to that CSS, as that line seems to have been there since the tool was created (initial commit).
However, in the browser-console I see ORES is mentioned, which makes me wonder if phab:T347344 is involved? I've filed a separate task at phab:T347367 (as it's a different error-message) in case it is related. Quiddity (talk) 00:52, 26 September 2023 (UTC)Reply
It affects me too, and has been thus for a few days now. I didn't think to examine the CSS; I just assumed its server was down. Certes (talk) 10:03, 26 September 2023 (UTC)Reply
As Quiddity mentions, display:none has been in the repo forever, so is probably normal. JavaScript probably changes that to display:block when it is ready to show the app. The cause of the current blanking is probably ORES-related errors messing up the JavaScript before it reaches the code that shows the app. phab:T347344 is fixed or will be fixed shortly, but there is also phab:T347317 and phab:T347367. These are all ORES-related and are all probably what is breaking PetScan, so all of them probably need to be fixed. Looks like the machine learning team is working on everything except maybe phab:T347317, which may require a code change to PetScan (@Magnus Manske) –Novem Linguae (talk) 16:08, 26 September 2023 (UTC)Reply
Just reported at User talk:Evad37/rater.js#Missing ORES predication that ORES for Rater is now working. Thought to share here. PETSCAN remains a blank page. Regards, JoeNMLC (talk) 20:38, 26 September 2023 (UTC)Reply
Good news: PetScan actually works, as long as you go into the browser's object inspector and remove the display:none property manually before using both the query entry screen and the results. Certes (talk) 21:00, 26 September 2023 (UTC)Reply
All fixed now. Yaay! -MPGuy2824 (talk) 03:04, 28 September 2023 (UTC)Reply
Seems to be down again - I just tried it, no luck. --Ser Amantio di NicolaoChe dicono a Signa?Lo dicono a Signa. 07:18, 29 September 2023 (UTC)Reply
Scratch that - seems to have been a momentary blip. --Ser Amantio di NicolaoChe dicono a Signa?Lo dicono a Signa. 07:32, 29 September 2023 (UTC)Reply
Might have been phab:T347665. All of Toolforge was down tonight for awhile I think. –Novem Linguae (talk) 10:05, 29 September 2023 (UTC)Reply

Discussion at WT:CATP § Alternatives to PetScan? edit

  You are invited to join the discussion at WT:CATP § Alternatives to PetScan?. Best, ‍—‍a smart kitten[meow] 09:32, 19 December 2023 (UTC)Reply

"No result for source categories" edit

When using PetScan, I will intermittently get the error No result for source categories. There doesn't seem to be anything I can do to control this (at least, that I'm aware of) - sometimes a query will work, and sometimes it won't. I just wanted to check if this was a known/reported issue. (pinging Magnus Manske)

All the best, ‍—‍a smart kitten[meow] 12:52, 10 January 2024 (UTC)Reply

A similar problem was reported above. Certes (talk) 13:49, 10 January 2024 (UTC)Reply
Are you following an old link to a pre-filled PetScan query? A bug, now fixed, put the spurious text "®exp_filter" into a random(?) field on one of the tabs. Blanking that field may solve the problem. Certes (talk) 13:53, 10 January 2024 (UTC)Reply
@Certes: I'm not, no - this is happening on newly-created queries. Best, ‍—‍a smart kitten[meow] 13:54, 10 January 2024 (UTC)Reply
@Certes and meow: Here are some "clues" that seem to make PetScan work better without failing off. 1. At top Tab ↹ forward thru fields to Categories, usually Depth 0 (zero). 2. At Categories, hand-type the cat. as copy/paste seems to often not work. SOmetimes replace those spaces with "underscore" (_). 3. Then "Do it!". 4. If it runs Ok, optional, go back and make Output changes (plain text, sorted, etc) and run again. Bottom-line, while "buggy" maybe these work-arounds can help. Good luck, JoeNMLC (talk) 00:28, 11 January 2024 (UTC)Reply
When copy-pasting within Wikipedia, I sometimes get unwanted invisible characters such as a zero-width space on my clipboard. I'm not sure where they come from – possibly search results. If typing rather than copy-pasting works, this may not be a PetScan problem. If the text is still on the clipboard, pasting it to somewhere like http://qaz.wtf/u/show.cgi might be instructive. Certes (talk) 09:50, 11 January 2024 (UTC)Reply
@Certes and meow: More: - Another clue, whenever I launch PetScan without the / after the .org it always runs Ok. Tested on 3 completely different (new) categories with good results. Cheers! JoeNMLC (talk) 01:54, 12 January 2024 (UTC)Reply
That's weird. Could your browser be auto-filling a field with unhelpful text, perhaps thinking PetScan wants a line of your address or other irrelevant information? You could examine the "Link to a pre-filled form for the query you just ran" and see if it contains any unexpected filters that would limit the search unhelpfully. Certes (talk) 11:21, 12 January 2024 (UTC)Reply
@Certes - Here's that URL as I'm totally clueless of what to look for. https://petscan.wmflabs.org/?before=&source_combination=&links_to_all=&format=html&manual_list_wiki=&minlinks=&search_max_results=500&cb_labels_any_l=1&categories=Orphaned_articles_from_January_2024&manual_list=&cb_labels_yes_l=1&edits%5Bflagged%5D=both&labels_no=&since_rev0=&smaller=&templates_yes=&output_limit=&project=wikipedia&language=en&show_soft_redirects=both&templates_no=&wpiu=any&edits%5Banons%5D=both&links_to_no=&ores_prediction=any&cb_labels_no_l=1&langs_labels_no=&active_tab=tab_categories&ns%5B0%5D=1&show_disambiguation_pages=both&sitelinks_no=&search_wiki=&langs_labels_any=&sparql=&larger=&langs_labels_yes=&ores_type=any&referrer_url=&edits%5Bbots%5D=both&output_compatability=catscan&interface_language=en&ores_prob_to= It's a plain PetScan, depth-0, cat-Orphaned_articles_from_January_2024 (copy/pasted). So hope this helps. JoeNMLC (talk) 15:51, 12 January 2024 (UTC)Reply
Nothing there looks suspicious. That URL also produced "No result" for me. I also tried typing "Orphaned articles from January 2024" into a blank PetScan form and still got "No result". Same for 2023, so it's not using an out-of-date cache. Then I messed about with a few things randomly and it started working ... so I went back and tried your version again, without my "improvements", and it works too (718 results). It seems to be an intermittent bug, which is often the hardest sort to fix. Certes (talk) 16:05, 12 January 2024 (UTC)Reply

PetScan currently down edit

Just leaving a notice here that PetScan currently seems to be down, displaying the error This web service cannot be reached. Please contact a maintainer of this project. This has been reported in the GitHub repo. Best, ‍—‍a smart kitten[meow] 19:40, 26 January 2024 (UTC)Reply

Greetings @A smart kitten - Thank you for reporting. I use PetScan almost daily & it's frustrating without. I had searched for where to report without any luck. Do you know of any alternate tool like PetScan? (Category filter, etc.) Regards, JoeNMLC (talk) 16:05, 27 January 2024 (UTC)Reply
@JoeNMLC: I don’t personally know of any, unfortunately (expect for potentially crafting database queries by hand using Quarry). I’m therefore hoping that this is just a temporary issue :) Best, ‍—‍a smart kitten[meow] 16:31, 27 January 2024 (UTC)Reply
Update @A smart kitten - PetScan up & running again. Cheers! JoeNMLC (talk) 19:45, 27 January 2024 (UTC)Reply

PetScan timeouts edit

Once a month, I try to run a simple PetScan query- categories=List-Class articles, depth=3, namespace=Talk. The idea that I want a count of all talk pages in a "List" category. This usually takes several dozen tries at minimum to actually return results; this month I'm getting nothing but "Io(Io(Custom { kind: UnexpectedEof, error: "connection closed" }))" as the response. I'm guessing that this is because it returns around 300,000 results, but I don't know. The thing is, I only need the count, not the page details, so is there a better way to run the query that will actually finish? --PresN 15:21, 1 February 2024 (UTC)Reply

PetScan not working edit

When trying to make any request, I get instead the following error: "Reqwest(reqwest::Error { kind: Request, url: Url { scheme: "https", cannot_be_a_base: false, username: "", password: None, host: Some(Domain("en.wikipedia.org")), port: None, path: "/w/api.php", query: Some("meta=siteinfo&action=query&format=json&siprop=general%7Cnamespaces%7Cnamespacealiases%7Clibraries%7Cextensions%7Cstatistics"), fragment: None }, source: hyper::Error(Io, Os { code: 104, kind: ConnectionReset, message: "Connection reset by peer" }) })". Is there any way to fix this? Kostja (talk) 12:40, 2 April 2024 (UTC)Reply

It works for me on a random query. (I didn't test it earlier.) Certes (talk) 16:29, 2 April 2024 (UTC)Reply
Works for me, too. I use it about 2x a day, and it hasn't failed for several weeks now. -- Michael Bednarek (talk) 03:29, 3 April 2024 (UTC)Reply
It started working again after a few hours. Kostja (talk) 06:45, 5 April 2024 (UTC)Reply