Wikipedia:Bots/Requests for approval/Dark Shikari Bot


 * The following discussion is an archived debate. Please do not modify it. Subsequent comments should be made in a new section.

User:Dark Shikari Bot

 * What: The purpose of this bot is to do WikiProject maintenance for whatever WikiProject thinks its services are useful. WikiProject Anime and Manga will be the first beneficiary if this bot is approved.  Of course, all its actions will have to be approved by any WikiProject before it is used on that WikiProject, as while in many cases they may be very useful, some WikiProjects may have different sorting systems that this bot should not mess with.
 * Why: I have noticed that WikiProjects run into a few problems in terms of raw sorting. First of all, there are often a huge number of stubs (hundreds or thousands) that belong in the category of a WikiProject (with the appropriate template in the discussion page) but have not been included.  A bot could easily do this.  There are other similar things, of course, that the bot could also do.
 * Exactly What:
 * Put all "anime and manga stubs" not in WikiProject Anime and Manga into Wikiproject Anime and Manga as Stub-Class.
 * If WikiProject Anime and Manga lists a stub as anything but stub-class, remove the stub tag as long as the article is more than 1000 characters long. If it is shorter, do nothing, as it could be an actual stub. The 1000-character cutoff isn't an arbitrary cutoff, but simply a safety feature to avoid a really short article that was accidentally listed as a non-stub in the article rating scale to stop being called a stub.
 * For all projects with their own rating system, turn all GA-class articles in those projects that aren't listed as GA or better into GA-class articles under that project (this will start with WikiProject Anime and Manga only).
 * It uses PyWikipedia Framework. It won't be a server hog because it will be run manually (i.e. only when I tell it to), so instead of patrolling recent changes for hours on end it will simply patrol the categories when told to, probably weekly or every few days.
 * There are thousands of articles here that need proper sorting, and a bot in my opinion is the best way to do it. In addition, if there is some sort of mistake (i.e. a B-Class article that's currently a stub that gets unlisted as a stub when it shouldn't be), it isn't devastating: a 99% success rate would add far more good than it would bad, and I highly doubt it would have any real failures, unless some fool ran around rating Stubs as B-Class articles. Dark Shikari 10:26, 18 July 2006 (UTC)


 * I don't see a problem, as long as it's being sponsored by a Wikiproject, and the exact details of what it is going to do are approved here in advance. We don't generally give a "whatever you need it to do" clearance; you tell us specifically what it will do, and we approve that. If you add something, you drop a note here saying what you're adding, and we say yea or nay. I see no problem why it can't have a trial period once it's set up and has a specific list of tasks. Essjay   ( Talk )  00:41, 19 July 2006 (UTC)
 * Its not really "sponsored" just yet: I'm asking what people think of it on the WikiProject talk page. So far I've received no disapproval--they've also made suggestions as to what I can and can't do in regards to sorting using the bot.  You can find the discussion so far here.  How much approval should I look to get before I set the bot into motion? Dark Shikari 01:09, 19 July 2006 (UTC)
 * Dear God no!!!! You don't seem to realise that stubs and Stub-Class articles are two completely different things! The terminology is admittedly extremely confusing (and the sooner something is done about it, the better). Also you clearly don't understand that length is only a minor consideration when it comes to working out what a stub is. An article swith one line of text followed by a list of 50 examples, or one line of text followed by a large table, is definitely a stub and likely well over 1000 characters. This is why stubs are sorted by hand, rather than by some automated method. Having the bot run in this way could well reverse much of the work of the Stub sorting wikiproject. Grutness...wha?  01:29, 19 July 2006 (UTC)


 * (Just for clarity) Hence why I said "as long as it's being sponsored by a Wikiproject, and the exact details of what it is going to do are approved here in advance." My take would be that the Wikiproject people would be quick enough to know what should and shouldn't be done, and that as long as specific changes are set out here before they are done, there is plenty of time for approval. Just saying that for the sake of clarity on my earlier comments. Essjay   ( Talk )  01:42, 19 July 2006 (UTC)

I have similar concerns to Grutness. The whole concept of a "stub class article" is poorly defined at best: at a deletion discussion on one such category, one WP1.0ist argued to keep them separate specifically on the grounds that "Stub class" is not the same as "stub";  another wanted to keep, on the basis that they were essentially the same. This needs to be much more clearly defined before a bot goes around making sweeping changes based on an assumption one way or the other. I see no evidence of support for this at the indicated Wikiproject, and the following comment sounds like opposition, or at least a reservation, to me: "It probably should go case-by-case (which I guess isn't what you want to hear for a bot, huh)." I'm especially opposed to stub-tag removal by bot;  if a B-grade article is still tagged as a stub, there's clearly been a snafu someplace, since appropriate tagging and categorisation should be required to get it to that standard:  much better to detect these by category intersection sorts of report generation, and have someone fix them manually. Stub-tagging might be more reasonable, but again, it's been specifically claimed by one person that some "Stub class" articles are not stubs, so this would require clarification and refinement; it would also have to be done very carefully to make sure that the "Stub class" tag, and the "stub" tag have the same scope by topic, and that no more specific tag applied instead. (For example, doing this to apply a single stub type to the whole of WPJ Albums, or WPJ Military History, would be a disaster.) Alai 04:06, 19 July 2006 (UTC)
 * I hope nobody minds my ill-informed opinion, but to incorporate Alai's suggestion into the bot sounds like a benefit. Without doing any editing, the bot could see how many articles have been snafu'd in such a manner and generate a report to somebody's userpage. It almost sounds like this should be a generic bot that Project leaders could download and configure for their specific purposes, or you could keep it proprietary and write all those little sub-bots yourself. ;) Xaxafrad 04:20, 19 July 2006 (UTC)
 * Well I had been informed by many that stub-class articles and stubs were in fact the same thing, and that anything above a stub-class article could not possibly be a stub. I had also been told that anything under the category "anime and manga stubs" most certainly belongs in Wikiproject Anime and Manga as a stub-class article.  So who's right?  I'm now getting confused... Dark Shikari 09:48, 19 July 2006 (UTC)
 * And in addition, wouldn't the most recent assessment of an article always be trusted? It seems as if most of the stub categorization was made months or even years ago, and hundreds of articles are still marked as stubs that have expanded far more since then, and have much higher ratings within the WikiProject.  Are you saying that WikiProject ratings are totally useless and cannot be used to justify removing a stub tag?  The WikiProject members would disagree. Dark Shikari 10:00, 19 July 2006 (UTC)
 * Oh, and also, is there anything wrong with simply having the bot shove all the Anime and Manga Stubs into the Wikiproject? There's a huge number that aren't part of the WikiProject, and nobody seems to have a problem with assigning them to the Project. Dark Shikari 13:03, 19 July 2006 (UTC)
 * If "Stub class article" and "stub" really are the same, then we're back with my original concerns, i.e., why have two sets of categories for the same thing, thereby duplicating work, and causing exactly the sort of inconsistency described? See for example this discussion, in which the matter is made...  well, rather opaque, actually.  One claim seems to be that a "long but useless" article would be "Stub class", but not a "stub" (though personally I would say that a textually long article can still be a stub, so even that's far from clear either way).  It's possible one implies the other, but not vice versa, for example.  No, I'm not saying assessment ratings are totally useless:  are you saying stub tags are totally useless?  Members of WSS would disagree.  An automated rule that simply assumes which of the two is incorrect is highly problematic, either way.  An assessment that ignores the presence of a stub tag is either a) using different criteria for the two, or b) failing to leave the article in a state consistent with said assessment;  a person has made them inconsistent, and a person (same or otherwise) should make a judgement as to how to make them consistent.  People are stub-tagging things all the time, I see no reliable basis for assuming that the assessment rating is either more reliable, or more recent.  I have no objection to the application of WPJ-tags by bot, if the associated WPJ has expressly agreed to the basis this is being done on, in their particular case.  Alai 14:58, 19 July 2006 (UTC)

Thanks for the input, Alai. I've made a revised set of things the bot could do:
 * Put all "anime and manga stubs" not in WikiProject Anime and Manga into Wikiproject Anime and Manga as Stub-Class. There's no way for the bot to figure out if they deserve a higher rating than Stub-Class, so its fair to start them off there.
 * For all projects with their own rating system, turn all GA-class articles in those projects that aren't listed as GA or better into GA-class articles under that project (this will start with WikiProject Anime and Manga only). Do the same with FAC articles that aren't listed as FACs under the WikiProject system. Dark Shikari 15:15, 19 July 2006 (UTC)
 * That seems completely fine, no objections to those tasks. Obviously in each case consultation with the wikiproject, and confirmation with them of the scope of the articles they're "adopting" thereby would be indicated.  Alai 02:59, 21 July 2006 (UTC)
 * Seems fine to me as long as there is agrement at the project. Voice -of- All  17:14, 29 July 2006 (UTC)


 * Before I run the bot, I want to see what Wikipedians think of another idea I had in mind (possibly a second thing the bot can do). How about a routine that checks through Wikipedia articles and fixes redirects: i.e. if an article links to "Blue", when the actual article is "Blue (colour)" and there is a redirect between the two (not a disambig), the original article will be changed to link to "Blue (colour)."  The appearance of the page will remain unchanged: only the link in the code will be changed.  This would probably slightly lower the load on Wikimedia servers by lowering the number of redirects.  In addition, its always struck me as somewhat odd to have articles link to redirects, as redirects make more sense as methods of catching various user-inputted article names and sending them to the right article.  This especially applies to older articles that link to a main article which has now been moved.  What do you all think?  Da rk Sh ik ar i  talk / contribs 19:41, 31 July 2006 (UTC)
 * Doesn't seem such a good idea to me. There's often been complaints about "needless" edits to eliminate redirects, and the opinion has been expressed that they cause more load issues than they solve, due to edits being much (much, much) more expensive than page fetches.  Alai 17:36, 4 August 2006 (UTC)
 * Figured. Seems like a reasonable idea, but below the surface there are hidden problems, I guess.  Or else someone else would have already done it ;) —  Da rk Sh ik ar i  talk / contribs 19:05, 5 August 2006 (UTC)
 * Actually, it's a feature I've been working on quite some time ago. I think that a massive fixup of links on a page is a good idea. My MiszaBot has even done a few such edits in the past (examples on its userpage). As soon as I rewrite the bot to use the pywikipedia platform and make sure it conforms with Redirects, I'll propose it here, since I now have the ability to set it up as a regular service. Misza 13 T C 21:00, 5 August 2006 (UTC)
 * Like has been mentioned, there is a problem changing redirects just for the sake changing redirects, but if your bot is making another edit ANYWAY, then I have no problem with changing redirects to point to the correct article. — Ram-Man (comment) (talk) 01:41, 2 September 2006 (UTC)


 * Now that this is under the Awaiting Operator Response category, I'm going to talk with the Wikiproject folks to ensure that they think this bot is fine. — Da rk Sh ik ar i  talk / contribs 13:36, 16 August 2006 (UTC)
 * This bot s not in the awawiting operator repsonse section (generally used to note that an application is expiring), although as stated above, there certainly are some details to work out. —  xaosflux  Talk 21:53, 19 August 2006 (UTC)
 * The single purpose has been decided and approved by the WikiProject. The bot will patrol the stub category and add unrated, standard WikiProject templates to the talk pages of each of the articles. —  Da rk Sh ik ar i  talk / contribs 22:07, 28 August 2006 (UTC)
 * I have completed the code and run the bot supervised through the first few articles in the category as a test and proof-of-concept. Its edits can be shown here. —  Da rk Sh ik ar i  talk / contribs 00:03, 29 August 2006 (UTC)
 * For those interested, I have uploaded my code here. — Da rk Sh ik ar i  talk / contribs 00:07, 29 August 2006 (UTC)
 * I'm concerned about a bunch of those edits. For example here and here.  In both of those cases, you added a template when one already existed, thus creating a duplicate template.  Another user had to delete the duplicated template. "" was changed to "" and then the other user changed it to "" . — Ram-Man (comment) (talk) 01:30, 2 September 2006 (UTC)
 * Yup, I got that message. It was a bug in my code, and is being fixed.  That's why we have this process ;).  I will continue running small, limited tests until all bugs are ironed out. —  Da rk Sh ik ar i  talk / contribs 02:13, 3 September 2006 (UTC)
 * Two bugs have been fixed: see the Bot's userpage. I'm running it again on a small number of pages. —  Da rk Sh ik ar i  talk / contribs 10:58, 3 September 2006 (UTC)
 * I ran it on another batch of pages and it seems to be working fine. See the userpage for the changelog.  What is the next step in the approval process?  Oh, and a second question: if the bot makes a mistake and screws up a bunch of talk pages, can Vandalproof (which I use) revert all the changes?  I've seen the feature, but never used it, so don't really know how well it works. —  Da rk Sh ik ar i  talk / contribs 11:11, 3 September 2006 (UTC)
 * Is this bot Emergency shutoff compliant, and/or Stop when a user leaves a message on it's talk page? Betacommand (talk • contribs • Bot) 22:47, 10 September 2006 (UTC)
 * How would I go about doing that? I figure for the user page trick, I should simply have it check the user page every few iterations of its main loop. —  Da rk Sh ik ar i  talk / contribs 23:38, 11 September 2006 (UTC)
 * Add Emergency-bot-shutoff to the Bots userpage. Betacommand (talk • contribs • Bot) 23:42, 11 September 2006 (UTC)


 * Done. Editing the code now to check the user talk page. (By the way, is simply storing the talk page when it starts and checking further downloads of the page against the original copy a good way to do it?) — Da rk Sh ik ar i  talk / contribs 23:46, 11 September 2006 (UTC)

''Program their bot to stop editing if someone leaves a message on its talk page. This can be checked by looking for the "You have new messages..." banner in the HTML for the edit form'' Betacommand (talk • contribs • Bot) 00:01, 12 September 2006 (UTC)
 * My program uses the PyWikipedia framework though, which fetches the Wikipedia code behind the page, not the HTML itself, which doesn't contain the template. I'm testing out the talk-page-get message now. — Da rk Sh ik ar i  talk / contribs 00:08, 12 September 2006 (UTC)
 * Ugh... problem. Pywikipedia caches all pages I fetch, so that when I fetch it again, it doesn't re-download the page. Is there any way to avoid this? — Da rk Sh ik ar i  talk / contribs 00:23, 12 September 2006 (UTC)
 * Try to append the URL with the current date (down to the second) with something like "&tstamp=TIME". That is what I use for VoABot II to load its settings panel. Voice -of- All  21:06, 3 October 2006 (UTC)


 * Symbol support vote.svg Bot trial run approved for the duration of two weeks. Try to see if you can find a way to fix the cache issue, but test it out in the meantime.  Voice -of- All  23:41, 22 September 2006 (UTC)


 * What is the status of this bot? -- RM 12:42, 12 October 2006 (UTC)


 * I noticed a problem with the bot in its current trial run, but otherwise it seems to be working well. In the case I just cited, an existing template (Anime) was not being replaced by this newer template. Once this issue is resolved, the bot can be approved. I'm not sure what happens in cases like this one where different Wikiprojects assess articles. Do they have to agree or are they independent? -- RM 02:16, 13 October 2006 (UTC)


 * According to your talk page the above problem should have been fixed, but wasn't apparently. Please let us know what the status of the bug is. -- RM 02:17, 13 October 2006 (UTC)
 * I'm pretty sure the bug was fixed. I've had it recognize basically every possible permutation of the A&M template. I also haven't run it since the beginning of October, when I tested it on the trial period. Can you give me an example of the current bug? — Da rk Sh ik ar i  talk / contribs 10:10, 13 October 2006 (UTC)


 * This bug right here. Maybe you fixed the code since then? On your talk page here you said that you fixed the problem in early september. This bug occurred in october. -- RM 11:24, 13 October 2006 (UTC)
 * Yeah, that was a minor bug that I caught quickly, reverted, and fixed in a few minutes. The exactly reason was because the first letter in the template's name was capitalized, which threw off the bot's code. It hadn't been programmed to deal with that at that time. — Da rk Sh ik ar i  talk / contribs 13:17, 13 October 2006 (UTC)

Approved. This bot is useful and the problems have all been addressed to my satisfaction. The bot shall run with a flag. -- RM 18:05, 13 October 2006 (UTC)


 * The above discussion is preserved as an archive of the debate. Please do not modify it. Subsequent comments should be made in a new section.