Wikipedia talk:WPCleaner/Archive 2017

Edits to Books namespace
✅

- I received a nice note on my talk page from stating: The tool being used by WikiCleaners on books should not be used (as used here, here and here). Please click onto Help:Books/for experts, then scroll on down to Allowed syntax, where further on down you will come to Formatting an article title, which reads "You can use a pipe to add formatting to a title." This is the reason the tool should not be used, as it eliminates formatting altogether." I did not notice that WPCleaner was making these automatic formatting changes. Could you please update WPCleaner so that it does not make these changes in the Book namespace, so we can continue using WPCleaner to fix ambiguous links?  Thanks!  GoingBatty (talk) 13:08, 5 September 2017 (UTC)
 * I don't understand the sentence "it eliminates formatting altogether", because for example this version of Nina Simone still have the titles in italic (the formatting has not been removed, but simply moved around the link) :  (which gives Nina Simone and Her Friends) has been replaced by   (which gives the same formatting Nina Simone and Her Friends). I have modified WPC to replace such links only in the main namespace but I don't see any problem with the modifications in other namespaces... --NicoV (Talk on frwiki) 16:33, 5 September 2017 (UTC)


 * The links work fine, but in order for books to be printed out when somebody's ordering one, all italics, "quotation marks", bold-type, even punctuation, everything concerning that link itself, must be within the brackets for it to succeed as it should be when printing, otherwise it won't work. As shown above with Nina Simone, one way works for books, the other doesn't:
 * Wrong format:  – works like this for articles, but not for books.
 * Correct format:  – unlike articles, this is the way that it's set up for books.
 * Best, --Discographer (talk) 17:01, 5 September 2017 (UTC)

Improvements to error 521?
✅

I've made some additions and improvements to error 521 so it catches more date errors, especially those caught by Category:CS1 errors: dates. Would it be possible to update the error so it catches more issues? (e.g. years with more than 4 digits, months > 12, days > 31) Thanks! GoingBatty (talk) 14:20, 27 November 2017 (UTC)
 * Thanks for the suggestions GoingBatty. I think I can add the following rather easily (the problem will be to find the time to do it...):
 * Check that day is between 1 and 31.
 * Check that month is between 1 and 12.
 * Check that year has no more than 4 digits. For this one, I wonder if there are cases where year could have more than 4 digits (far away future) : what do you think ?
 * --NicoV (Talk on frwiki) 14:40, 27 November 2017 (UTC)
 * Looking at WikiProject Check Wikipedia/Translation, error 521 covers citation templates and infoboxes. I can't envision any situation where a year > 2100 would be appropriate.  GoingBatty (talk) 14:52, 27 November 2017 (UTC)
 * GoingBatty. I added the extra verifications, but I can't release the new version yet. You can check if they work after seeing WPCleaner downloading a new version. --NicoV (Talk on frwiki) 18:43, 27 November 2017 (UTC)
 * Seems that it's now marking valid dates as errors (e.g. List of best-performing Latin songs in the United States). GoingBatty (talk) 04:40, 29 November 2017 (UTC)
 * GoingBatty. There were 2 problems, one in my code with month names, and one in the configuration file. Should be fixed. --NicoV (Talk on frwiki) 07:01, 29 November 2017 (UTC)
 * Oops! I accidentally deleted the "END" in the configuration file when removing the Persondata records.  Sorry about that, and thanks for fixing the problems!  GoingBatty (talk) 15:14, 29 November 2017 (UTC)
 * WPCleaner isn't picking up "2017-18-25" as an invalid month in Aviva Kempner. Any thoughts?  Thanks!  GoingBatty (talk) 05:36, 2 December 2017 (UTC)
 * I still haven't released the new version, so unless you're compiling it yourself from the source code in GitHub, the new verifications are not available yet... I have troubles with my code signing certificate right now, so I'm unable to release a new version. --NicoV (Talk on frwiki) 08:41, 2 December 2017 (UTC)

WPCleaner is significantly slow
Today, I am facing a lot of technical issues with WPCleaner. Firstly, it has significantly slowed down. Secondly, it is not cleaning up some of the errors. For example: at Checkwiki Task ID:44, it says that WPCleaner cleans up the bold headings but today, it hasn't been doing what it says and it hasn't been doing that task. Same goes to:, , , , , , , , , , and most of the time. It says that WPCleaner can perform all these tasks but it fails to make a single edit and fails to fix the problems in each task where it says that WPCleaner can correct the problem. Pkbwcgs (talk) 20:00, 10 December 2017 (UTC)
 * Hi Pkbwcgs. Can you tell me exactly what you are doing ? What features are you using ?
 * For example, I tried the bot tools on 5 pages for #44, all have been fixed automatically: 1999–2000 FIBA Korać Cup, 2008 Chatsworth train collision, A3 (musician), Abdu Rahiman Nagar, Abu Dis.
 * I hope I will be able to release a new version with many enhancements in the near future, but it depends on the renewal on my code signing certificate. --NicoV (Talk on frwiki) 07:13, 11 December 2017 (UTC)
 * Pkbwcgs, one explanation for doing nothing and being slow is usually that MediaWiki returns an error when WPC tries to save a page. Could you check the logs in the Java console (see explanations on how to display it) if there are any errors? --NicoV (Talk on frwiki) 08:29, 11 December 2017 (UTC)
 * It is still slow even after enabling 'show console'. Even to load up three pages, it takes at least two to twenty minutes. Pkbwcgs (talk) 16:06, 11 December 2017 (UTC)
 * Pkbwcgs, showing the console doesn't change anything except to give access to WPCleaner logs. Could you start WPC again, load the 3 pages, and paste here the logs that are displayed in the console ? I tried myself with the English wikipedia for a dozen random pages, and each page is loaded in no more than a few seconds. Does it happen whatever method you're using (full analysis, CW edit...) ? --NicoV (Talk on frwiki) 16:33, 11 December 2017 (UTC)
 * Where can I locate the WPCleaner logs? Pkbwcgs (talk) 16:39, 11 December 2017 (UTC)
 * I am also saying, it is when I press the 'load pages' button, it is taking ages to load the pages. Pkbwcgs (talk) 16:43, 11 December 2017 (UTC)
 * It also says:

Pkbwcgs (talk) 16:55, 11 December 2017 (UTC)


 * Pkbwcgs, logs are in the Java console. For example, if I do the following:
 * Run WPCleaner and log with my account on enwiki
 * Click on the button "Project Check Wikipedia"
 * Select only error #44
 * Clear the Java console (to remove all the logs from the beginning)
 * Click on "Load errors 44"
 * Click on "Load pages" with 10 errors selected. It took about 30s to load the 10 pages with the following log
 * If I close all the pages, and do it again it takes less than 5 seconds.
 * I believe the problem due to checkwiki being slow on tools : the first time, there were 3 pages among the 10 were WPCleaner couldn't find the problem, so it asked checkwiki for confirmation (the calls to checkarticle.cgi in the log), and then after confirmation that there was no more error, it asked checkwiki to remove the page from the list (the calls to checkwiki_bots). The second time, as the 3 pages has already been remove from the list, it was a lot faster. I tried the same kind of access several times and each time I got the same result: slow if WPC has to ask checkwiki, fast otherwise. I will ask in the Check Wiki project. --NicoV (Talk on frwiki) 17:12, 11 December 2017 (UTC)
 * It is also saying
 * It is also saying

and doing nothing. Pkbwcgs (talk) 17:20, 11 December 2017 (UTC)
 * Pkbwcgs It can be the same problem: when you fix an error, WPC also notifies checkwiki that the error is fixed (with the checkwiki_bots.cgi call). Can you do the following : display the Java console and WPCleaner window side by side, and when you save a page, WPCleaner will write several lines in the Java console, I need to know which one is taking a long time (if it's checkwiki_bots.cgi line, then it's a slowness on checkwiki's side I believe). --NicoV (Talk on frwiki) 17:25, 11 December 2017 (UTC)
 * I made a trial with 23 articles on error #64 (link equal to linktext). The time I pressed 'load pages' was 17:11 (GMT). At 17:18, the pages have loaded up and at 17:25, I managed to go through all the pages. It took 14 minutes in total. Also, it is taking longer to write the lines on the Java console. After this test, the results were:

I have been experiencing more problems with the system writing the results in the Java Console as it took ages. Rate is 1 edit per 20-30 seconds. Pkbwcgs (talk) 17:32, 11 December 2017 (UTC)
 * Thank for the log Pkbwcgs. From your contributions, I see that WPCleaner is editing at a rate of 4 pages a minute, which is normal for a non-bot user (WPCleaner throttles edit speed to a maximum of 4 edits per minute). It has been like that for ages, because such repetitive tasks should preferably performed with a bot account. You should ask for a bot flag if you want to use WPCleaner at full speed. --NicoV (Talk on frwiki) 20:00, 11 December 2017 (UTC)
 * The thing is, I took 100 pages which took approximately 30 minutes to load up which is a bit strange. Although with this rate, it is still quite a problem to load up pages onto the system. Pkbwcgs (talk) 20:02, 11 December 2017 (UTC)
 * Yes Pkbwcgs, there's still the problem with the load time, which I believe is due to CheckWiki on WMF Labs. I'm running my bot to mark pages as fixed when they already are, it's currently going through error #47. Can you reload a list with error number below #47 and see if loading time has improved ? --NicoV (Talk on frwiki) 20:05, 11 December 2017 (UTC)
 * No, it hasn't and WPCleaner doesn't fix #47. Pkbwcgs (talk) 20:09, 11 December 2017 (UTC)
 * Pkbwcgs, here's what I understand about the problems you're reporting:
 * Loading pages seems slow: when I try to load the first 10 pages for error #47, WPCleaner loads them in less than 2 seconds with the version currently released (slower with my development version)
 * Saving pages seems slow: for non-bot users, WPCleaner throttles edits to a limit of 4 per minute, it's its normal behavior to prevent users without the bot flag to overflow watch lists
 * Error #47 isn't fixed: WPCleaner reports error #47 and suggests a fix, but as it is not always the correct one, it's not automatically applied. For example, on 1998–99 Miami Heat season, the  in the middle of the NBA game log is detected, but WPCleaner doesn't know what the correct fix is (it only suggests to delete it).
 * --NicoV (Talk on frwiki) 07:42, 12 December 2017 (UTC)
 * I would like to fix error #16 using WPCleaner but it wouldn't fix any of them. Why is this the case? Pkbwcgs (talk) 13:50, 17 December 2017 (UTC)
 * Hi Pkbwcgs. For #16, I know there are some situations where removing the control code could lead to problems (I think especially with RTL/LTR codes and non-ASCII alphabets), but I've never been able to have a precise explanation about the case. So WPCleaner doesn't always fix #16, it depends on the control character found an the other characters around it. Can you try opening a few pages from #16 list in WPCleaner and see what it says: does it find the error? what is the control character that is found? are there any non-ASCII characters around it? --NicoV (Talk on frwiki) 15:54, 17 December 2017 (UTC)
 * It managed to fix unicode control characters at in article Akong Rinpoche but I don't which Unicode Control Character this is. Pkbwcgs (talk) 16:16, 17 December 2017 (UTC)
 * Pkbwcgs. For the example you provided, as the characters around the control characters are only letters or a square bracket, WPC can fix the problem automatically. It's rather for errors that WPCleaner doesn't fix that I'm interested in the informations: in WPCleaner, select the #16 error, right on the error detected in the text and you should get a contextual menu, the first item should be the control character that has been found. --NicoV (Talk on frwiki) 16:20, 17 December 2017 (UTC)
 * According to Check Wikipedia (where it displays the invisible unicode control character), it is unicode control character  and   which WPCleaner is not fixing. Pkbwcgs (talk) 16:23, 17 December 2017 (UTC)
 * It is able to fix Unicode Control Character . Pkbwcgs (talk) 16:26, 17 December 2017 (UTC)
 * Sometimes, it also not fixing  unless it is in a link. Pkbwcgs (talk) 16:29, 17 December 2017 (UTC)
 * Pkbwcgs. 00A0 is a "Non Breaking Space" and 2004 is a "3 per EM space". WPCleaner can't decide by what they should be replaced: a regular whitespace character,  for 00A0, remove it for 2004.
 * 200E is a "Left to Right mark", which is only replaced when surrounding characters are safe. --NicoV (Talk on frwiki) 16:31, 17 December 2017 (UTC)
 * What is ? It is also being fixed on some occasions. Pkbwcgs (talk) 16:34, 17 December 2017 (UTC)
 * Pkbwcgs. They are the Unicode value of the characters. 200B is Zero Width Space. --NicoV (Talk on frwiki) 16:36, 17 December 2017 (UTC)
 * Thanks. This has been fixed in . I will get back to you when I spot more unicode control characters this WPCleaner is not fixing. Pkbwcgs (talk) 16:38, 17 December 2017 (UTC)
 * What is  and  ? WPCleaner is not fixing these Unicode Control Characters. Pkbwcgs (talk) 17:28, 17 December 2017 (UTC)
 * There is a bit of a problem. The limit is only 1,000 pages to load up so I can't load all 8,000 pages on the system. How can I change this limit? I have tried at the settings in the 'limits' section but it didn't work. Pkbwcgs (talk) 17:41, 17 December 2017 (UTC)
 * Just came across  which WPCleaner is not fixing. What is  ? Pkbwcgs (talk) 19:23, 17 December 2017 (UTC)
 * After today's clean up of Unicode Control Characters, I found:


 * - WPCleaner is not fixing this at all
 * - WPCleaner is not fixing this at all
 * - WPCleaner is not fixing this at all
 * - WPCleaner is not fixing this at all
 * - WPCleaner is occasionally fixing this
 * - WPCleaner is not fixing this at all and majority of articles have this character
 * - WPCleaner is fixing this most of the time but not always
 * - WPCleaner is not fixing this at all
 * - WPCleaner is not fixing this at all
 * - WPCleaner is not fixing this at all
 * - WPCleaner has fixed errors but this has been found to only one page
 * - WPCleaner is not fixing this at all
 * - WPCleaner is not fixing this at all

This is out of around 4,250 pages that I reviewed. Pkbwcgs (talk) 20:50, 17 December 2017 (UTC)
 * Pkbwcgs. Please, see Unicode for the meaning of each code. It's normal that WPCleaner is not fixing everything. It's either:
 * A human decision has to be performed: for example for 00A0 (non-breaking space), should the non-breaking space be kept ? If no, it can be replaced by a regular whitespace character; If yes, it should be replaced by . Only a human can decide, that's why WPCleaner give both suggestions, but doesn't apply any automatically (right click on the reported error to see the suggestion and choose).
 * There's a doubt: for example 200E (Left to right mark), where I don't know if it's required or not in some situations (next to non-ASCII character): users have to check that removing the control code is harmless. As I wasn't able to get a proper explanation on what cases can be done automatically or not, WPCleaner doesn't remove it automatically when there are some non-usual characters around it. For the moment, usual characters are "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyzáàâäåãÀéèêëÉíìîïĩóôöōŌúùûüýćčńňšź0123456789 []|{}<>,.!?;:--–=+*#/%'\"«»\n\t";
 * The hard limit is set to 1000 pages, because Labs interface wasn't working very well with big numbers. Errors in the first 1000 pages have to be fixed to get the next elements in the list.
 * By the way, if you don't have the bot flag, and you run automatic fixes with WPCleaner, especially on minor errors, you risk getting some remarks from people because they will get a lot of modifications in their watchlist. --NicoV (Talk on frwiki) 07:10, 18 December 2017 (UTC)
 * I already moved on from Unicode Control Characters. I am now fixing issues with square brackets and headings. Pkbwcgs (talk) 15:34, 18 December 2017 (UTC)

Replacing typos
I was gonna use WPCleaner to replace an incorrect term on a bunch of articles, but I can't seem to get error 501 to load at all, unless I'm doing something wrong. Pariah24 ┃ ☏  20:53, 8 June 2017 (UTC)
 * Hi Pariah24 . What do you mean by loading error 501 ? Errors below 500 have a list generated by CheckWiki project, but errors above 501 don't, so it's not possible to load a list of pages for an error above 500. Best way to do that, would be to generate the list of pages through an other way (asking a bot, performing a search, analyzing a dump file...) and provide this list to WPCleaner. Does that answer your question? --NicoV (Talk on frwiki) 12:34, 28 June 2017 (UTC)
 * I tried to make my own template with custom typos for WPCleaner, but it was too much for me to figure out. It's okay though; I have AWB rights now and I just used a regex with the find & replace function and took care of the issue. Thanks for your response. Pariah24 (talk) 02:40, 19 July 2017 (UTC)

Line break tag
WPCleaner is replacing the tag  by. JFG mentioned, that  creates trouble with syntax highlighters. According to the help page WP:LINEBREAK# , only   should be used. Can you use that tag in WPCleaner? --GünniX (talk) 08:13, 25 June 2017 (UTC) instead of , but I don't know if it's wise. What do you think ?
 * Is this correct? Many people reacted when I propsoed that. maybe wants to comment on that. -- Magioladitis (talk) 08:52, 25 June 2017 (UTC)
 * That's crazy. We are not keeping invalid syntax, for the sake of a syntax highlighter. Fix the syntaxhighlighter instead. Also that advise in WP:LINEBREAK is just bad and misguided. valid is valid, invalid is invalid. —Th e DJ (talk • contribs) 12:17, 27 June 2017 (UTC)
 * Hi GünniX, Magioladitis, Th e DJ
 * First, I think that W3C has done a poor job by promoting things like  in HTML5, which means adding special handling for only a few tags... Second, I really believe that the syntax highlighter should be modified to work correctly with   as W3C won't change its standard: I and others have already asked for that (see here), but the developer of the syntax highlighter is unwilling to modify the gadget to work properly with.
 * In WPCleaner, I could easily use
 * --NicoV (Talk on frwiki) 12:30, 28 June 2017 (UTC)
 * I think WPCleaner should behave according to the Wikipedia standards. Currently WP:LINEBREAK#   says: ″The MediaWiki software converts valid forms like,  and  to .″ The section was changed yesterday. It would be good to wait for further discussions in that section. After the discussion is finished, WPCleaner should be customized according to that section. --GünniX (talk) 18:35, 28 June 2017 (UTC)

Unreadable text
I have a problem with links to disambiguation pages that are marked by a disambiguation template. Even in the latest version (just downloaded and installed) is hows up as a golden/yellow text on a white background. Hardly readable for me. Is there a setting somewhere that I have to change to make it better to read? Firefox 54.0.1. The Banner talk 13:53, 31 July 2017 (UTC) I am cleaning up very old templated links (now November 2012), so I see them often.
 * Hi The Banner. From the main window, you have access to Options (Alt + O), and you should find several option in the formatting tab. --NicoV (Talk on frwiki) 17:00, 31 July 2017 (UTC)
 * But which one? The formatting menu has no options for text in gold or yellow, so I have no clue what to change. The problem is there with texts like (d. 18 September 1412, Rouvres). When checking with WPCleaner Rouvres will become yellow. This is from the article Margaret of Bavaria. The Banner talk 19:47, 31 July 2017 (UTC)
 * Hi The Banner. It's the "Help requested" item normally. On my computer, it's a Yellow foreground. --NicoV (Talk on frwiki) 05:58, 2 August 2017 (UTC)
 * It solved one part of the problem, the link in the text itself. But in the left hand menu it is still a yellow text on a white background. Something to fix in the next update? The Banner talk 09:34, 2 August 2017 (UTC)

Two-factor authentication
Seems that WPCleaner doesn't support two-factor authentication for those Mediawiki users who have it turned on (mostly admins). Is that in fact the case? -- Fuzheado &#124; Talk 14:26, 22 September 2017 (UTC)
 * Hi Fuzheado, check WPCleaner/FAQ: I haven't tested it myself, but it's supposed to work. --NicoV (Talk on frwiki) 10:42, 24 September 2017 (UTC)