This is an archive of past discussions. Do not edit the contents of this page. If you wish to start a new discussion or revive an old one, please do so on the current talk page.
The Communications committee is working on coordinating the use of sitewide notices for Wikimedia projects. This is needed for the upcoming fundraiser, as well as other potential announcements relevant to all projects, such as Wikimania or technical issues like single login.
It's anticipated that we will use a bot to help update these messages, which due to the nature of the pages being edited will need to have universal sysop privileges. Rest assured that we won't be using it for other sysop functions.
All done, except for a few random ones, I think my bot ignored them because the image was is transcluded from a template or something. (p.s. I just added functionality to WP:AWB to get a list of articles from an image page, it will be in the release version soon). Martin15:46, 4 June 2006 (UTC)
Almost certainly an authorised bot becoming logged out, this is a problem with the pywikibot software. Martin23:07, 6 June 2006 (UTC)
It's obviously using the interwiki module from pywikipedia. It's not that uncommon for interwiki links to be added anonymously, a lot of users don't bother creating accounts in other languages. The only way this could be a big deal is if the links were incorrect. Ideally, we'd have one screen-name to rule them all, but then, in a perfect world, the interwiki links would be automatically generated by the software as well. In the meantime, let's not block people who creating any actual disruption. — Jun. 6, '06[23:17] <freak|talk>
You really should talk to the bot owner, but it appears to have tagged them as they were fair use without a rationale, which is surely correct behaviour. Martin22:04, 7 June 2006 (UTC)
I've noticed some editors and their accounts and bots have been doing away with good edits, improvements of the page. What they are doing is not harmless to the encyclopedia. They seem to be trying to keep what they wrote as the top page, thus preventing any possible improvements.--Chuck Marean23:32, 7 June 2006 (UTC)
Any specific examples? Both of the bots I see regularly reverting vandalism have a >99.9% success rate. (ESkog)(Talk)03:34, 8 June 2006 (UTC)
FWIW, when I looked at Tawkerbot2's accuracy rate I ended up with with 5 samples of ~53 reverts, 266 total (random time, go down a contrib list of 100) assuming and and I rejected with over 97.5% confidence. Some of the times it was reverts to vandalized versions, others reverting someone uncensoring song names or something, sometimes reverting someone that claimed to be the subject trying to 'delete' the article when it probably should have been deleted (in one case it reverted blanking of unsourced claims of someone molesting children, thrice). I don't think that >99.9% claim is that accurate. Kotepho04:58, 16 June 2006 (UTC)
I'd be interested in specific examples as well. Bots are not intended to prevent forward progress in articles, but rather to thwart silly vandalism. When examples are brought forward, the bot authors make their bots better. That seems goodness to me. ++Lar: t/c03:37, 8 June 2006 (UTC)
He has no examples. The only time one of his edits was reverted by a bot was when he reverted a page to its 2003 version in order to make a point. He seems to think that anyone reverting his edits must be guilty ofmalfeasance. I fear he thinks that because I occasionally used the rollback button when he was being especially recalcitrant, I must be a bot. Sorry you guys have to deal with this. - EurekaLott11:32, 8 June 2006 (UTC)
Hi all. I am thinking about setting up a bot to work on archived pages. Not a bot to do the actual archival of the pages themselves, but one for archival maintenance, such as adding archive header templates (for example, see Talk:Main Page/Archive 1).
This bot would be human assisted. Basically, I think I would feed it a list of pages and it would make sure they all had the proper headings. Although there are quite a few archival pages, I don't think this would be a server hog, as it could be run at set intervals.
Would a bot like this be approved? I could write a basic javascript for my monobook.js that would let me basically do the exact same thing, but I'd have to sit there and click through the pages instead of letting the bot click them for me. ~MDD469617:17, 12 June 2006 (UTC)
As long as its human assisted and it goes on pages people want edited I'd have no problems approving such a bot -- Tawker18:56, 12 June 2006 (UTC)
The Bureaucrats have a special MediaWiki interface through which they can assign a certain user (the bot's account for example) to a special user group named "bot" and that's it - the user's edits no longer show up in Special:Recentchanges. What talk page did you mean? This is a talk page. Misza13TC11:05, 15 June 2006 (UTC)
A bot (or perhaps a script, or some other tool) would be very useful to the growing amount of people (myself included) who are interested in studying Wikipedia. I'd very much like to see and use a tool that would look at the history of any article (including a talk page!) and:
generate a list of people who have edited target article
generate an information feedable to a statistical analytical program (comma-separated values format is quite simple and popular) which would tell:
how often each individual edited this article
when did he edit it
how much new content has he changed
was the edit marked as minor
was the edit summary used
was the edit a vandalism (possibly using parts of Tawkerbot2)
Portions of this information can be obtained via the toolserver, (Example). Perhaps it would be better to request expansion of those tools. — Jun. 19, '06[20:13] <freak|talk>
Thanks for the info. This is definetly a good start. Unfortunately the toolserver seems to analyze all namespaces save the Wikipedia namespace (and as far as I can tell the toolserver doesn't display results for anything other then articles). Where should I request expantion of it?--Piotr Konieczny aka Prokonsul PiotrusTalk22:43, 22 June 2006 (UTC)
Request for approvals page
Would this work better with sub pages like AfD? It would bmake archiving eaier. RichFarmbrough 15:13 20 June2006 (GMT).
Please delete this account, the bot spams user pages, doesn't identify correctly tagged images (apparently looking at templates instead of categories), and it doesn't wait to let users fix their errors. The owner is unresponsive to complaints on the bot's talk page. -- Omniplex13:03, 23 June 2006 (UTC)
Agreed that OrphanBot is problematic. It's attempting to do unattended something which really should not be being done without manual supervision - namely accusing (or stopping just short of accusing) large numbers of users of plagarism. That it looks neither at the image itself nor at the description (except for the template) is a severe technical limitation, most likely not fully resolvable. For this reason, it should not be running unattended for pretty much the same reasons that we traditionally have opposed running spellcheckers unattended.
That other bots have been reverting some of its edits as vandalism, while marginally amusing, is telling. If it's too time-consuming for one person to manually oversee the activity of this 'bot, given the number of pages it's attacking, then the task needs to be split between multiple people - not blindly left under control of an unattended automated process which neither understands plaintext in the image descriptions nor is able to make any determination of content by viewing the image itself. --carlb17:12, 22 August 2006 (UTC)
Crypticbot - now orphaned?
It seems that Crypticbot is now rather orphaned, as its master User:Cryptic(User talk:Cryptic) hasn't edited for several months. It continues plodding on its way, doing useful things it appears, and hasn't particularly misbehaved AFAIK. However, without a master it can't be adapted to changing situations.
Cryptic just showed up to reply to a "your bot is broken" message. I left him a message that if he is going to "stay with the project" at least as a bot operator, that I will unblock the bot account. — xaosfluxTalk00:32, 3 July 2006 (UTC)
Hi, I think the list of bots is becoming too big, and should not be maintained on this page. I propose we split it up and move the list to another page. --Yurik22:07, 7 July 2006 (UTC)
I moved the section, you are welcome to split it up :) The page uses a number of sub-page templates, so don't forget to make proper redirects. --Yurik20:46, 19 July 2006 (UTC)
Unicode problem
I am writing a bot to run on Hindi Wikipedia. I am using python wikimedia framework. I am getting an error, "UnicodeDecodeError: 'utf8' codec can't decode byte 0xff in position 0: unexpected code byte" when executing the following code,
f = codecs.open('hi-towns.csv', 'r', config.textfile_encoding)
x = f.read().decode('utf8')
Project Page Entry clarification - spiders and dynloads
Point number 5 and 6 below "When naming your bot" seem to not really belong to that section. The information they contain seems unrelated to naming ones bot and seems to be nothing one would expect under this headline. Also, if there exists something laid out as a policy in its own right on these subjects, there might be a good point in pointing to that policy, and doing that early on the project page instead of hiding it in the place it is now.
--62.134.227.1401:15, 6 August 2006 (UTC)
AWB Spellcheck - assisted bot?
I want to make a lot of AWB spellchecking edits, and so as not to crowd my contrib list and make more than the 1-2 edits per minute, I want to make a bot account. As always it will be human-assisted, and I'll monitor each and every change (AWB shows me the diff preview) but since I'll be making a lot of edits, a bot account makes sense. Is this the place to apply for approval for such a venture, or is there some policy that easily defines what I should and shouldn't do with AWB? --Draicone(talk)06:55, 7 August 2006 (UTC)
Just create a second account and make it a AWB account no need to get a Bot status it is still just you behind the controls it doesnt matter about getting it approved. Just be careful it meets WP:BOT#Spell-checking_botsBetacommand06:32, 8 August 2006 (UTC)
I have recently gotten the approval to run and have received a bot flag for my bot user:BetacommandBot. during this process I noticed that the procedure for getting the preamble for running a bot can a extreamly difficult and time consuming process. I would like to make a suggestion. Split the page Wikipedia:Bots/Requests for approvals into five six pages.
Bots that are being discussed
Bots that have a aproved trial run
Bots that are approved and waitng a flag
Bot that are approved and wish to get approval for another task
Bot disscusion archives
and disscusions that have stalled and are over60 days old without any edits
I have no problems with the first four; the archives should be, well, archived, and so should anything more than a month old. If the bot operator can't be bothered to come back and say "Uh, hello?" they're probably not that dedicated to having their bot, and we need people who are going to be dilligent with regard to thier bots. Essjay (Talk • Connect)05:17, 8 July 2006 (UTC)
This seems to imply that quite a number of recently-archived discussion with no approval, disapproval, or pending questions from the AG will presently be getting unarchived and having "uh, hello" suffixed to them, somewhat defeating the point of archiving them in the first place. This might more usefully have been done on a somewhat more gradual and selective basis. Alai03:23, 11 July 2006 (UTC)
The vast majority of the 90+ requests that I archived had not been commented on in months. Months. I don't see a problem, in the least, with removing requests that are doing nothing but taking up space and making the page unmanageable. The archiving has been done for several days now, and I've yet to see a single old request pop back up. If a large number of them pop back up, my first question is going to be "Where have you been for the past three months, and will you be dropping out again, leaving questions about your bot unanswered for three months or longer?" Until a problem surfaces from it, I say it was a good solution. Essjay (Talk • Connect)08:23, 11 July 2006 (UTC)
please let me clarify i didnt want them it be unachived, just put on a seprate achive somthing like WP:BOT/no consensus. as for the other approvals i suggest something like WP:BOT/Approved & WP:BOT/denined or something like that so it would be easier to find a bots status. Betacommand16:24, 11 July 2006 (UTC)
To give due credit, E. seems to have sped things up enormously, so I'll file the above under the general heading of "startup costs" of new system, and "all's well that ends well". I just rather had visions of endless "churning" of requests without resolution... Alai17:12, 13 July 2006 (UTC)
Running bots
I'd love to learn a programming language and run a bot... but I don't want to download programming software. Is there an alternative to downloading software? --Gray Porpoise15:50, 11 August 2006 (UTC)
Well, or access to webserver running a compiler. Programming languages like Perl are compiled at run-time so you don't need a compiler on your own machine, just notepad, and the webserver compiles the code for you - PocklingtonDan11:48, 7 December 2006 (UTC)
Many users apparently having problems with cydebot actions
I, and apparently, many other people are having problems with the actions of User:Cydebot (see User_talk:Cydebot, especially sections 38, 41, and 43, though there may well be additional notes on this page from dissatisfied users.) I wonder if someone could block its actions and suggest that it is changed to merely put a note on people's user pages asking them to manually edit the pages, as this seems to be the simplest option. I have had to revert my page, and have yet to work out what I must do to successfully achieve what the bot was trying to do. Thank you. DDS talk22:23, 11 August 2006 (UTC)
Thanks for the response and the actions. I noticed that complaints are still being made by people on the talk page of the bot, and looking back over the previous messages, it seems that similar problems were being complained about since May of this year at least. It appears that nothing has been done with respect to those previous requests, and so I wonder whether this bot could simply be blocked unless or until its owner replies speedily and change its behaviour almost immediately. DDS talk10:23, 14 August 2006 (UTC)
User:Cydebot, despite being blocked once, has now been unblocked and is still apparently doing things it is not supposed to do, judging by fresh comments made on User_talk:Cydebot. Can something not be done to fix this once and for all? Is this the correct place to draw someone's attention to it who can act to fix the problem, since Cyde seems not to respond to any calls to fix his bot? DDS talk16:00, 20 September 2006 (UTC)
I am not sure this is the right place to bring this up. I am not happy with some recent actions of Cydebot. The bot's action basically made the articles orphans (without categories). Without the categories, there is no way to get to these articles. This is like destroying constructive work of others and make the encyclopidia building process difficult. The bad part is that the bot did not inform any one about this. I had to find this on my watchlist. I am not for this bot empyting out categories. Can someone please stop this bot from doing this? Can someone put this job, "Emptying out deleted category" to a vote? - Ganeshk(talk)16:35, 20 September 2006 (UTC)
Ganeshk's problem appears to have been resolved by Betacommand. It had to do with a redlinked category added to an article, and deciding which extant category a nonexistant category might have been a subcat of is usually beyond bots. I tried to solve the other thing (about the removing cabinet of norway's cat) and I cannot figure out what happened unless it was something similar (cab of norway was redlinked when cydebot was given the task and had since been created and populated). Cab of norway was only created September 6th or so. Syrthiss18:05, 20 September 2006 (UTC)
Please, go ahead. I won't do anything but uncontroversial standard CFD work until I talk to Tawker at least (in case this is about something else). I must admit, his block reason didn't exactly give me any information about what was going on :-( Cyde Weys18:33, 20 September 2006 (UTC)
It's possible that it is some bot that periodically loses its "logged in" status, but it is vital to be able to monitor the bot, so it can't be run anonymously. The rambot has accidentally run as an IP before, although that was back when most people knew what it was. But it should have been blocked then so that RC patrol can verify that it isn't a vandal or a malfunctioning approved bot. The speed of additions is irrelavent. Of course on a practical note, an anonymous bot has no bot flag, so it will uncessarily clutter the RC. All this is previous precedent to support blocking. But you are correct: It is good that it seems to have a slow pace. — Ram-Man(comment) (talk)13:32, 1 September 2006 (UTC)
I've unblocked the IP, since it is an approved bot, albiet occasionally getting logged out. Looking at the recent archives, this problem has occurred before without any apparent resolve, so we may want to consider a more permanent solution. Update: It's not on the list of Registered bots, but as it's only a manual bot I won't mess with blocks again unless someone else find a problem. — Ram-Man(comment) (talk)16:21, 1 September 2006 (UTC)
Regarding my comment on speed, this "bot" is making less than 1 edit per day, that's hardly enough to need to go through the bot process, in fact if they would have not put the word "robot" in their edit summary, would we even be having this disucussion? — xaosfluxTalk16:34, 1 September 2006 (UTC)
Wouldn’t it be very likely that Ram-Man was talking about the contribs of Thijs!bot and not the IP? Anyway, I don’t see a problem anymore; the link between bot and IP is made and a small note on the IP talk page is sufficient. --Van helsing07:46, 5 September 2006 (UTC)
The Red Button
...I know this may sound kind of stupid, but I always wonder what the red button on the bot page is for, it says it'll block the bot, but mabye... will it block the person who presses it? (please respond on my talk page.) Tinlv718:34, 5 September 2006 (UTC)
Is it possible that we could get some useful categorisation of the bots? I'd be interested in seeing them categorised by programming language (so people can find similar examples) and open sourceness (likewise looking for code to use), and also if there's some way to categorise their general wikipedia functions.
I guess it might be useful to have different categories for Pywiki or AWB bots etc. though AWB is generally not used as a bot specifically, so it might get a bit confusing. Categorisation by task could be pretty useful. Martin14:20, 7 September 2006 (UTC)
I second this - it would be really useful to see eg all Perl bots so that bot developers using common languages could cooperate/query one another directly etc. - PocklingtonDan11:47, 7 December 2006 (UTC)
Wiki Bot Communication
How does a Wiki Bot communicate with English Wikipedia? Does it request pages like any browser with HTTP, or does it use another protocol? Thanks in advance. -ENIAC (Talk) (Current Projects) 12:01, 8 September 2006 (UTC)
I propose that we add a statement to this policy that all bot edits (and edits made with any software assistance) must abide by all existing guidelines and common practices. The reason is that I occasionally notice bots/javascript etc. edits that change articles in a manner - that while not necessarily wrong - is not consistent with the normal wiki-style. Normally this is limited to minor formatting things, but it is still needlessly annoying. Martin21:11, 20 September 2006 (UTC)
I am going to add this ammendment now, as I think it is uncontroversial, largely because it is simple commen sense. Martin20:17, 24 September 2006 (UTC)
Overactive unlinking bot.
Greetings; There is an apparent job or bot that is unlinking of “common words” which seems to be running overboard in a number of instances; it is delinking items that while on their face may seem "common" may in fact be apropos to the article. Also, in general I prefer to see and abundance of linking, makes for richer exploration. I have seen some errors where it has removed the link around a year but truncated the value (1993 became 993). The bot appears to be operated by User:Colonies_Chris and the changelog url is [4]. Thank you for looking at this. Bdelisle08:41, 21 September 2006 (UTC)
Are they required to have a bot flag to make multiple edits per minute? And it is debatable whether "Overlinking" applies in a blank sense to all links that they have on their list. Ansell 10:43, 21 September 2006 (UTC)
Obviously there are times when a link to "Saturday" is in context, it is up to the editor to make this judgement. Martin10:47, 21 September 2006 (UTC)
Spam reversion bot
On another wiki where I'm a sysop under a different username, I've got a bot that looks for spamming, then reverts it with the edit summary of (Spam reverted - AUTOLINKREMOVE - Please do not mass-add external links!)
The interwiki bots don't seem to mention rearrangements just additions modifications and removals, that edit it seems just involved rearangements. Plugwash23:39, 20 October 2006 (UTC)
It is time we added a new section specifically dealing with assisted bots/software (WP:AWB etc. ). Though the spirit of this policy is obvious, I would like to express some principles clearly:
Assisted bots are defined as any software that allows rapid editing of articles.
Typical assisted jobs include those which are repetitive, but need some human interaction e.g. Disambig repair, re-categorisation.
Assisted bots don't necessarily need bot approval, though some software has built in approval detection, whereby approval from an admin is required (developers are encouraged to build in approval mechanisms).
A separate account is advised if many edits are going to be made.
Always make extra sure there is consensus before making a large series of edits.
I don't think any of these points are new or remotely controversial, so I'll add them pretty soon. thanks Martin06:34, 7 October 2006 (UTC)
I agree with this and there is nothing controversial here. I would say though that in some cases it may make sense to get a bot flag if there are a large enough number of edits that are not one-time-only. It isn't clear from the above when such a bot should go through the process. Perhaps "when in doubt, go through the process". -- RM19:25, 7 October 2006 (UTC)
My idea: Create a bot that checks the new changes and finds those with multiple exclamation marks (!!!!!!!!!11). 50% of all the vandalism I've seen has that. --Jinxs12:25, 13 October 2006 (UTC)
I have noticed that perhaps six out of seven anonymous users who leave comments on talk pages do not sign their posts properly. I have usually added the {{unsigned}} message after those posts when I have encountered them. However, this could be a job for a bot: scan the Recent changes list limited to the Talk space, and if a comment is made by an IP-address, check it for a signature and add one if necessary. Of course logged in users also forget the signature sometimes, and those could be checked too, if it doesn't take too much resources. Alternatively only check those users that have not created an user page yet, they are often new to Wikipedia and do not know about signing their posts. Is anyone with the skill/equipment up to this? --ZeroOne (talk | @) 11:03, 17 October 2006 (UTC)
Tricky for sure, but it might be able to be pulled off. I would expect it to go through a number of iterations correcting for mistakes before it was fully successful though. -- RM12:58, 17 October 2006 (UTC)
I thought about this before, and I realized it would be very hard to make it accurate. For example what about when someone adds a new paragraph to a message they already wrote, or altered a peice of common text on a talk page. I could not think of a way to avoid false positives. HighInBC(Need help? Ask me)15:59, 11 November 2006 (UTC)
Since the bot would scan the recent change list anyway it could look at only those edits that were first edits to that page by that user, that would surely rule out most false positives. In any case, I think there is a bot that does something similar already. It is called User:HagermanBot - PocklingtonDan15:07, 7 December 2006 (UTC)
Spellchecking
I noticed that it is stated that no spellchecking bot could ever function without assitence. Now I understand the fundemental faults with spell checkers, lord know I have tried to depend on them in the past due to my deplorable spelliong. But it does not seem accurate to say it is not technically possible to automatically correct spelling.
What about a bot that only changed diferentiatetodifferentiate? To say it is not technically possible to create a spellchecking bot is just wrong. The scope simply needs to be limited.
What needs to be done for such an automated bot(which I for one am not skilled enough to create) is to provide a list of words and their mispellings where there is no possible alternate meaning to the word. Such as sofisticated turned into sophisticated, no possible error in mistaking it for another word. Not trying to change the world/wiki, just pointing out what I see, wheeeee. HighInBC01:01, 18 October 2006 (UTC)
What would the bot do when it encountered a quotation from a book where the original author spelled it as "diferentiate"? --Carnildo06:13, 18 October 2006 (UTC)
Ohhhh, ya got me. I was wrong, the page is right. No automated system could account for that. I retract my contention. HighInBC06:17, 18 October 2006 (UTC)
There's a simple solution to that - any quotation that includes a misspelling or grammatical error should include [sic] after the error. This is something editors should be doing independent of any consideration of spell-checking bots. I was interested in creating a bot much like HighInBC's suggestion, although the use of "predominately" (which is not a word, but a common misspelling of "predominantly") was my target. If Carnildo's objection is the only one, I don't see it as a real problem. | Mr. Darcytalk18:18, 6 November 2006 (UTC)
For every solution built, someone builds a better idiot. What about someone who cites from a book as "predominately" [sic] or any of a thousand other variants? Automated spellchecking is FTL. That's why semi-automated corrections (via AWB) are allowed, but fully automated "corrections" are not. --Connel MacKenzie - wikt16:00, 11 November 2006 (UTC)
Likely not logged in authorized but but we really have no clue short of checkusering the IP which is a privacy vio -- Tawker03:47, 21 October 2006 (UTC)
Looks like it's been very low volume, and not recent. If it starts going bezerk, it can be blocked, you can report it here, or even on WP:AIV if its really crazy. — xaosfluxTalk04:01, 21 October 2006 (UTC)
I've spent hours scraping around and I can't find the API? Where is the documentation that says "This is what you need to send" and "This is what you get back". I don't want wrappers or dll's or libraries, I just want to know what strings of bytes I need to send, where to send them and the format of the replies. TIA for any assitance. 87.112.20.15208:52, 11 November 2006 (UTC)
No good, that is just gobbledegook to me since I don't know php. It also does not explain how to write pages.
I'm looking for the definition of the actual API. Not a wrapper or example in some particular language (unless there's a c++ example).
Something along the lines of {Open port 80 and send "title=Wikipedia_talk:Bots&action=edit§ion=21&data=the new data"}, the returned data will be {format of returned data}. 87.112.20.15209:48, 11 November 2006 (UTC)
It appears the "api" is only for reading information at the moment, it seems to make actual edits you still have to use the interface designed for web browsers, that means you need to read up on enough html to find and parse a form and enough http to post the results of editing that form back to the server. Plugwash13:39, 11 November 2006 (UTC)
OK, thanks, that's easy. I thought that was deprecated though, as it's effectively the same thing as screenscraping as far as server load is concerned. 87.112.20.15215:05, 11 November 2006 (UTC)
Yeah, nobody said it was pretty ... although there is now a query.php, which offers more direct access to the anti-vandalism bots than having to just use the standard web interface that all of the humans do. --Cyde Weys17:36, 16 November 2006 (UTC)
Do Wiktionary bot approval requests go on the same page as Wikipedia ones?
OK. I was dealing with a vandal, as was another user, and the bot reverted an edit and provided the vandal with a generic warning.(given test1, then test2, then generic)
I gather you can call the package to log in, read and write pages, but there is no documenation with the package and I can't find any here. —The preceding unsigned comment was added by 87.112.74.253 (talk • contribs) 16:43, 15 November 2006.
How can you contact the people responsible for the Python bot framework
I've been trying to find out how to use it but either no one knows, or they're not letting on. 'Someone' must know how to use it from other languages. It claims on the framework page 'we will welcome you'. Quite frankly it's about as welcoming as a rabid doberman :-) 87.112.15.1409:01, 16 November 2006 (UTC)
Yes, but as the poster says, it's 'partial'. And that is something of an understatement. I had a look at the sourceforge page but trying to go to the documentation just took me around in circles. I'll try and post to the mailing list but it seems a very odd way of finding out the basics. —The preceding unsigned comment was added by 87.112.15.14 (talk • contribs) 15:50, November 16, 2006 (UTC)
I would agree that user documentation of this framework is rather weak. IMHO, it's best used by developers, especially if they know Python. RedWolf16:38, 16 November 2006 (UTC)
Well I've developed several thousand lines of robot software and I'd like to save a few days (weeks) trying to work out the http input required by the servers as, of course, this isn't documented anywhere. I've been told that it can be used from other languages and the Python documentation supports this view, but of course that means that you need the function names, call argument list and returns. Surely not too much to ask? 87.112.15.1416:53, 16 November 2006 (UTC)
Your best bet is to just code up what you want as another bot using the pyWikipediaBot framework in Python. There's also a Perl bot framework out there if you're interested, though it's not as fully-featured. Other than that, I wouldn't really recommend trying to use the frameworks from other languages. Python isn't that bad, and you should be able to do anything you need using it. --Cyde Weys17:35, 16 November 2006 (UTC)
The problem is that I was told this was the best way to proceed (by someone who seems to have done it themself) and having checked the Python documenation to ensure it was callable from C++ I've now got rather a lot of time invested in C++ robot code which I'd really rather not just scrap. Given the fact that 'bots are quite slow moving it's even possible to communicate by reading and writing files on the disk - not very elegant, but possible. 87.112.15.1418:32, 16 November 2006 (UTC)
the basic idea is simple, request the edit page, parse the html to find the form, change the stuff you need to change and then submit the result over http, your http implementation will also need to be able to handle cookies so that it can log in. Plugwash16:47, 18 November 2006 (UTC)
The problem is simple - "refdeskbot" is archiving too soon. I have raised this point Wikipedia talk:Reference desk#refdeskbot also see User talk:87.102.21.223 - the problem is that active discussion edits are not shown in the edit history (it uses tranclusion apparently). Bot operator martin is clearly aware what the problem is but will not change it's operation without some clear mandate. Perhaps some sensible person could give him a push in the right direction. Thank you.87.102.21.19003:05, 18 November 2006 (UTC)
Just to say, the bot operator (martin, above) is me, and I'm only realy happy to make archival changes at the command of consensus at WT:RD, as the bot serves them (and the help desk). Thanks, Martinp2314:55, 19 November 2006 (UTC)
Archiving project talk page
Is there a bot that automatically archives Wiki-project talk pages?
Werdnabot would probably do it if requested. --ais523 13:15, 4 December 2006 (UTC)
Fair use bot
I have an idea for a bot: one that checks if fair use images are being used outside of article namespace (WP:FUC criterion #9). I've come upon fair use images on article talk pages, image pages and image talk pages (linking to other versions), WP:BJAODN, user pages, user talk pages and even a template.
It doesn't necessarily have to remove the images, it just needs to notify a human editor of the problem. --Oden13:12, 5 December 2006 (UTC)
Would like to add a new section to this page, or a new page altogether
As someone who was a short while ago looking to start out on the path to setting up a bot on wikipedia, it struck me that the pages on bots (below) contained no help information on getting started with your bot, but rather focused solely on policy.
The main Wikipedia:Bot page should really be named Wikipedia:Bot Policy, sicne that is overwhelmingly what it covers. This is important, but when people are looking to set up a bot it seems to me we shuold be able to give them a page within the botnav tree of Wikipedia:Howto Contribute with a Bot or similar (the name probably stinks, i know). It could outline the various languages available, outline issues that everyone writing a bot is going to bump into and need to know about (edit tokens etc), have links to set code snippets for certain tasks etc, as well as being the obvious place for bot developers to post quesries (on the accompanying talk page) relating to bot development - currently such questions would be scattered between this page (which is really about bot policy and the bot requests page (which again, isn't really suitable).
It could have FAQs on "how do I know if another bot alread does this", "how do I get my bot logged in to make edits" etc etc etc
Obviously I'm open to refinement of this idea, but just as there is a Wikipedia:Contributing_to_Wikipedia I feel there shold be something for bot developers similar to the helpful "here's how you get started" page for human editors. Anyone else feel the same way? Is there any backing for this idea? PocklingtonDan16:43, 8 December 2006 (UTC)
OK, I have created the new Wikipedia:Creating_a_bot and linked to it fromt he botnav template. I would appreciate any help others could give in expanding it, its very new and stub-by at the minute and really needs some good work on it to make it useful. I will leave moving Bot -> Bot Policy for a short while - PocklingtonDan21:15, 9 December 2006 (UTC)
Writing Bots
What language does one use to write bots. I am thinking about learning how to write a bot so I can write one for the purpose of trolling and cleaning up Category:Lists of ambiguous human names by converting entries without the last name first structure to take on that structure. Is that something that a bot could do? Is that likely a bit much for a first time bot maker? TonyTheTiger22:56, 8 December 2006 (UTC)
If you are new to bots or new to programming generally, you're going to find it a steep learning curve to write a genuinely useful new wikipedia bot. If you have programming knowledge, then go for it. If nt, you might be best either running a clone of an existing bot or else using a customised version of an existing tool such as the auto wikipedia browser. You can write a wikipedia bot in almost any programming language. If you do start from scratch, you are unlikely to be running a bot from the dedicated wikipedia toolserver at first. You will therefore have to either:
Write a bot in a client-side language such as Javascript that you can operate through your browser. or
Write a bot in a sever-side lanuguage such as PerlorPython that runs on a web server that you have access to.
I have just created a null template {{nobots}}. The idea is that it can be placed on a page to discourage bots, much in the same way that robots.txt works. Whether it is honoured will depend on the bot, the name space and potentially arguments. E.g {{nobots|except=WerdnaBot}} might be used on a user talk to allow only WerdnaBot to edit it, but AntiVandalBot might still decide to override. Possibly {{nobots|theseones=SmackBot,Rambot}} would allow any bots execpt SB and RB. Comments? RichFarmbrough, 21:24 12 December2006 (GMT).
Well, really late to the game, I don't think it would be the best idea in the world for AVB. Our nice friendly vandal could post the no-bots tag and voila, AVB wouldn't revert - which would be a bad thing....-- Tawker19:20, 22 March 2007 (UTC)
I'm working on a new Java-based bot framework that I'm currently calling "jwikibot". The intention is to write a library that uses the MediaWiki API for as much as possible and fills in the gaps with legacy methods like XHTML parsing until the API supports the added functionality. I am planning for it to have functionality comparable to pywikipediaorperlwikipedia once it is more developed.
Currently, I have prototyped read-only code for using Special:Linksearch to build a report of links to lyric sites. I have an unapproved bot account called Dillonbot, but I have been running the parsing code without logging in because it is not yet doing any writing.
I'm not in a hurry to let this code start editing yet, but I was wondering what my next steps should be once I get to that point. Should I seek approval for the account for general framework testing using user subpages, or will I need to specify an actual task (probably also using user subpages)? Mike Dillon05:30, 22 December 2006 (UTC)
You can and should request approval for the bot to perform the specific task of testing a bot. That's fine. Plus, you may get java bot programmers (such as myself) to help out. If you don't seek approval, some admins would block the bot on sight. -- RM13:23, 22 December 2006 (UTC)
Also, the question wasn't so much whether I should seek approval, but whether it was appropriate to ask for approval for something as general as "testing". Thanks for your response. Mike Dillon16:03, 22 December 2006 (UTC)
I couldn't find a task approval for this particular task (removing YuoTube links) for this bot, or for any other bot, in the usual places (Wikipedia:Bots/Requests for approval, etc). Just leaving a note here so that bot people can keep an eye on this. Barberio, who notified about it on WP:VPP, said he was trying to contact the bot operator. --Francis Schonken00:24, 28 December 2006 (UTC)
Bot commands
How do you command your bot? For example, as per a new agreement made at WT:NYCS I originally requested approval for my bot to replace templates such as {{NYCS 7}} with {{NYCS service|7}}. If the template is a rush template, such as {{NYCS 7 rush}}, then I want {{NYCS service|7|<7>}} to replace it. If {{NYCS time}} is followed by it, such as {{NYCS 7}} followed by {{NYCS time|1234}}, then I want {{NYCS service|7||1234}} to replace it. If {{NYCS time}} is followed by a rush template, such as {{NYCS 7 rush}} {{NYCS time|1a2a3c}} then I want {{NYCS service|7|<7>|1a2a3c}} to replace {{NYCS 7 rush}} {{NYCS time|1a2a3c}}. If they are shuttle templates, such as:
{{NYCS S 42nd}} replaced by {{NYCS service|S - 42nd Street Shuttle (New York City Subway service|S}}
{{NYCS S Franklin}} replaced by {{NYCS service|S - Franklin Avenue Shuttle (New York City Subway service|S}}
{{NYCS S Rockaway}} replaced by {{NYCS service|S - Rockaway Park Shuttle (New York City Subway service|S}}
I'm a bit confused. I thought you were just doing this in AWB. The trouble with what I'm seeing above is that it requires the bot to know to replace a shortened station name with a longer station name. That's basically manual work, unless I'm reading this wrong. Could you clarify what you're trying to do? alphachimp.01:43, 28 December 2006 (UTC)