Commons:Village pump/Technical/Archive/2021/01

From Wikimedia Commons, the free media repository
Jump to navigation Jump to search

Error: 502, Server Hangup

I am having trouble uploading a new revision of a PDF file at the size of 45 MB. After some time I will get either a time-out error or a server hangup message. The file is still way below the maximum size of 100 MB, so this is quite annoying. De728631 (talk) 18:19, 1 January 2021 (UTC)

Apparently this is T247454 at Phabricator. bigChunkedUpload did the trick for me, but this bug needs to be fixed. De728631 (talk) 20:30, 1 January 2021 (UTC)

File counts in subcats of Category:1880 by month by country

While perusing Category:Canada by month by year, I noticed that Category:1880 in Canada by month was listed as having 29 files in addition to its subcategories. As a metacat, it should have none, so I loaded the category to subcat the files, but there were none. This led me to Category:1880 by month by country, which lists 29 files for all "1880 in X by month" categories (except Germany, which is the most recent subcategory of the bunch), despite those categories containing no files, only subcategories. I suspect this is some stray leftover from a database or other action, though the relevance of the number 29 escapes me. I tried purging the cache for several of the subcategories, but the issue persists. Mindmatrix 14:41, 3 January 2021 (UTC)

I've scanned other sub-categories of the form Category:YYYY by month by country, where YYYY represents a year. The same problem occurs for the following:
Note that this is also true of some of their subcats; for example the province and Category:2016 in Canada by month by city subcats of Category:2016 in Canada by month. Again, the most recent subcats therein have no such issue (for example, Category:2016 in Quebec City by month in Category:2016 in Canada by month by city).
This issue can be noted by comparing, for example, Category:Canada by month by year and Category:United States by month by year, and noting those subcategories which are listed as containing files. Mindmatrix 14:59, 3 January 2021 (UTC)
I think this is covered by phab:T247187. --ghouston (talk) 01:18, 4 January 2021 (UTC)

Finding Creative Commons YouTube videos

Hi. I tried to use the Creative Commons filter on YouTube to find an image that could be used here, but it absolutely does not seem to work. Is there a way to get it to work, or if not an alternate search engine for CC videos? I apply it but the videos that show up are not labeled as Creative Commons anywhere. DemonDays64 (talk) 06:20, 8 January 2021 (UTC) (please ping on reply)

@DemonDays64: You can use the filter under the searchbar in YouTube after you searched on a term. You get a result like this: [1] under filter you see 'Creative Commons'. DutchTina (talk) 22:52, 8 January 2021 (UTC)
Please notice you have to click the option after every search, even when it is enabled at the moment you search for a new term. DutchTina (talk) 22:54, 8 January 2021 (UTC)

DepictBot?

I'm certain something like this has been proposed before, so I was mainly looking to see if there's documentation of a proposal like this. Has anyone ever proposed a bot that would add wikidata:Property:P180 to files in a category on Commons? So, to take a recent example that I've worked on, a bot would assign wikidata:Q627098 as the value of wikidata:Property:P180 to every file in Category:Mike Lee. I've been manually inputting a ton of metadata lately and I was just thinking there would have to be a better way—seems like there must be a way to assign metadata by exploiting the pre-existing category infrastructure. Thanks for any guidance. AleatoryPonderings (talk) 06:13, 9 January 2021 (UTC)

A fully automated Bot was disliked because of many potential false statements. There are some users running Bots with checked batch tasks and there are tools like AC/DC and SDC tool --GPSLeo (talk) 09:43, 9 January 2021 (UTC)
@GPSLeo: Thanks! AleatoryPonderings (talk) 15:47, 9 January 2021 (UTC)

Using a single page from a pdf

Is there a simple way to use a single page of a pdf file on Commons? I am looking at File:Actes de la société d'histoire naturelle de Paris - tome premier, premiere partie (IA actesdelasociete1117soci).pdf, and would like to use page 175. I can see how to link to the whole pdf, but not how to link to one page. Of course I can download the whole pdf, extract the page and upload it, and I will if necessary, but I wonder if there is an easier way... Thanks Kognos (talk) 20:41, 9 January 2021 (UTC)

Do you mean like [[File:Actes de la société d'histoire naturelle de Paris - tome premier, premiere partie (IA actesdelasociete1117soci).pdf|page=175|thumb]]? --HyperGaruda (talk) 21:09, 9 January 2021 (UTC)
Simple when you know how! Thanks, that does it. Kognos (talk) 23:35, 9 January 2021 (UTC)

15:40, 11 January 2021 (UTC)

Suggestion to improve notifications for village pumps or new uploads

Hi all. Happy New Year to everyone editing or reading Commons and sister wikis. To make it easier to receive notifications about new discussions at village pump, I think a facility should be offered with these features:

  • notification of each new edit, instant (already available via watchlist with optional notification to email)
  • notification of each new topic, but not each new edit, instantly, delivered to talk page or to email
  • daily or weekly digest, delivered to talk page or to email

The latter two could be done via a script or a bot. Please let me know if you want it written, or it already exists.

This facility could be documented and linked to from {{Welcome}}, and the headers of the Village pumps.

Similar kind of notifications could also be enabled for newly uploaded images, with the contributors selecting a category (topic or place) to be notified on.

Best regards, --Gryllida (talk) 08:26, 4 January 2021 (UTC)

@Gryllida: The village pump part closely relates to the goals of the Talk pages project. @Whatamidoing (WMF): FYI. Maybe these have already been proposed? I don’t remember so, but I may have forgotten it. —Tacsipacsi (talk) 00:19, 9 January 2021 (UTC)
Thanks. I'm going to ping @PPelberg (WMF) and @JKlein (WMF) so they can see the proposal directly, too. Whatamidoing (WMF) (talk) 22:35, 12 January 2021 (UTC)

Admin help requested

On this image:

Could someone please delete the earlier version? Thank you.

Evrik (talk) 19:08, 15 January 2021 (UTC)

Done. Evrik, please don't overwrite one file with another; the proper solution was splitting the history and putting the earlier version back to its original name. See COM:OVERWRITE for more details. Nyttend (talk) 15:51, 18 January 2021 (UTC)

16:08, 18 January 2021 (UTC)

Problem with Commons's search index and OCR text

The Wikimedia Foundation's Search team has found that they are currently unable to keep up with the rate that the commonswiki_file index is growing due to large uploads of PDFs containing OCR text. This problem will be immediately addressed by placing a default 50kb maximum limit on the amount of file text (including OCR text, but excluding metadata and wikitext) that is indexed for search. This will reduce the commonswiki_file size by 77%. More information is available on Phabricator.

Note that this will not inhibit the continued ability to upload documents with OCR text; rather it just means search will not use this text for indexing. Search will continue to index the wikitext and metadata/structured data, and no change to those systems are required to resolve this issue.

Community Feedback

While there is no future work currently planned following this change, the Search team wants to make sure they can continue to support the community and the work you do. It would be helpful for the team to hear from you about any potential needs you have that would warrant us investing resources into investigating another solution:

  • Are there any specific needs or use cases you have that require indexing the entire file text on Commons uploads?

On behalf of the Search team, thanks for your time and I look forward to hearing about potential uses for complete OCR indexing. Keegan (WMF) (talk) 20:25, 19 January 2021 (UTC)

From what you've written here, can we presume the whole OCR will still be in the metadata record, hence accessible via the API?
Having the OCR available means it is possible to continue with Common searches that pick up keywords to aid with curation or categorization. On the IA books project page, you can see how the API searches based on the full OCR were used to find copyright infringements, this was without using the search engine itself. It would be great to have a solution that means these types of search and filter remain possible and hopefully easy, even if not in the "main" search engine. Without this, categorization and discovery of the million PDFs recently uploaded may become a lot harder or at least less reliable.
BTW, glad you aren't asking me to stop, the uploads are actually deliberately at a slow rate over several months. -- (talk) 20:43, 19 January 2021 (UTC)
Thanks for the question . I’m Mike (he/him), and recently joined the Search Platform team as the Product Manager. I confirmed with my team that there will be no effect on your copyright infringement use case (as outlined in https://commons.wikimedia.org/wiki/User_talk:Fæ/IA_books#Automatic_detection_of_possible_copyright_issues), as the desired metadata is coming from MediaWiki core, rather than the search index. MPham (WMF) (talk) 22:11, 19 January 2021 (UTC)
Good. In which case, though losing the complete search is a drag, the alternative of slowly checking metadata is a possible work around. But this would make searching a million documents turn from moments using search, to, hm, around 46 days to generate a report presuming my home laptop never crashed, my wifi never dropped out, and using one processing thread. I can imagine faster methods, but they would take significant time to work out. -- (talk) 22:15, 19 January 2021 (UTC)

Can't upload

For more than an hour, I've been trying to upload some files, but I'm always getting the same standard error:

Error: Our servers are currently under maintenance or experiencing a technical problem. Please try again in a few minutes. See the error message at the bottom of this page for more information. Request from - via cp1089.eqiad.wmnet, ATS/8.0.8 Error: 502, Server Hangup at 2021-01-20 20:21:01 GMT

Seems like the receive-an-upload servers aren't working at all? I got this error a couple of hours ago, but then it let me upload File:Goode Oakland Methodist.jpg at 19:07 server time; however, since then it's never produced anything except this error. Nyttend (talk) 20:24, 20 January 2021 (UTC)

PS, at 21:11 I was able to upload another image, File:Parrish Chapel Methodist Church.jpg, but since that time I've tried eighteen further uploads, and they all produced the same error message. Nyttend (talk) 21:35, 20 January 2021 (UTC)

18:29, 25 January 2021 (UTC)

Checkmark This section is resolved and can be archived. If you disagree, replace this template with your comment. DannyS712 (talk) 01:58, 19 February 2021 (UTC)

Adding automatically to category if metadata fits

Hallo lieber Wikimedianer,

gibt es eine Möglichkeit, es so einzustellen, dass eine hochgeladene Datei automatisch einer Kategorie zugeordnet wird, wenn ein Wert in den Metadaten stimmt? Wenn also ein Foto mit einer Belichtungsdauer von 1/100s erstellt wurde, automatisch der Kategorie Category:Exposure time 1/100 sec zugeteilt wird?

Vielen Dank und Grüße, --PantheraLeo1359531 😺 (talk) 17:25, 24 January 2021 (UTC)

Hey dear Wikimedians,

is it possible to add an uploaded file to a category automatically, if the metadata fits? If I upload a file with an exposure time of 1/100s, it should automatically be categorized to Category:Exposure time 1/100 sec. Is there a tool for this?

Thank you and greetings, --PantheraLeo1359531 😺 (talk) 17:25, 24 January 2021 (UTC)


PDF Preview distorted

Why is the preview of the PDF Die_kaiserlichen_Privilegien_der_Universit%C3%A4t_Marburg_-_Eine_academische_Rede.pdf distorted the way it is. The original upload is o.k. The File is part of a project on Wikisource. Help and explanation would be appreciated.--Jürgen Nemitz (HSP) (talk) 21:59, 24 January 2021 (UTC)

upload not working

Special:Upload continues giving me an error while uploading; Upload Wizard works instead, but I need to upload a new version of an existing file, so I have no way to do that. Can someone help me? Thank you!--ValeJappo (talk) 15:59, 26 January 2021 (UTC)

Opened a task, phab:T273032.--ValeJappo (talk) 09:16, 27 January 2021 (UTC)

css guide

"Hey how can one change the font-family on his/her own common.css?Any pdf or tutorial about using css here would be appreciated.Thanks. — Preceding unsigned comment added by NairobiPapel (talk • contribs) 15:00, 30 December 2020 (UTC)"

Copied from "Commons:Help desk". --Donald Trung 『徵國單』 (No Fake News 💬) (WikiProject Numismatics 💴) (Articles 📚) 11:59, 27 January 2021 (UTC)

@NairobiPapel: hey, you can copy and paste as follows
body{
	font-family: sans-serif; /*change sans-serif with another font family*/
}
More informations about font families --ValeJappo (talk) 12:57, 27 January 2021 (UTC)

The thumbnail is a different color from the file image. The file image matches the image at the source (New York Public Library digital collection). I tried downloading the image again from the source and re-uploading it, but the thumbnail is still the wrong color. Thank you for any help you can give me. Vzeebjtf (talk) 12:59, 27 January 2021 (UTC)

This may be because the source file is using a different color space. I don't know how MediaWiki handles that for thumbnails though. pandakekok9 04:18, 2 February 2021 (UTC)