Jump to content

Wikipedia talk:ReFill/Archive 1

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia
Archive 1Archive 2Archive 3Archive 4

Publisher field

Can you add an option (defaulting to off) to fill in the "publisher" field with the URL's second and top-level name (e.g. cnn.com, microsoft.com, ap.org, etc.)? It's not perfect, but it might be better than nothing. I realize it's not entirely straightforward, but you can use the Public Suffix List to deal with two-level domains (.co.uk, .com.br, etc.). --Ahecht (TALK
PAGE) 16:19, 30 September 2014 (UTC)

 Doing... Good idea. Zhaofeng Li [talk... contribs...] 00:23, 1 October 2014 (UTC)
I was about to ask for this feature, but Ahecht has made the request already.--DThomsen8 (talk) 22:05, 6 October 2014 (UTC)
Thanks! --Ahecht (TALK
PAGE) 01:13, 7 October 2014 (UTC)
@Ahecht:  Done Apologies for the late response. I'm currently busy with my studies so things can be slow. It's now available as the Use the base domain name as work when this information cannot be parsed option. Zhaofeng Li [talk... contribs...] 11:08, 17 October 2014 (UTC)

Automatic updating of accessdate

It has been reported that the script automatically updates |accessdate=, but this parameter should not be updated unless the editor actually reads the source and verifies that it still supports the text in the Wikipedia article. If the report is true, it would seem the script either should not update the accessdate parameter, or should provide a manual option to indicate, on a source-by-source basis, whether to update the parameter. Jc3s5h (talk) 16:50, 6 October 2014 (UTC)

@Jc3s5h: Now it doesn't add accessdates unless the Do not add access dates option is unticked. Zhaofeng Li [talk... contribs...] 11:09, 17 October 2014 (UTC)

Hi, Zhaofeng Li, just wanted to inform you there is a typo in the text: Sorry for the outages earlier. Blame Zhaofeng for playing around with the server configuations. :(
configuations to be changed into configurations. Lotje (talk) 09:03, 15 October 2014 (UTC)

 Fixed I've removed the message since it's already been a few days since the incident. Zhaofeng Li [talk... contribs...] 11:12, 17 October 2014 (UTC)

Catch templates with Missing or empty |title= field in other CS1 templates than {{cite web}}

The tool doesn't seem to be catching references using CS1 templates other than than {{cite web}} where a citation template has been used but no title has been specified. For example, the All in the Family article had the following citation: <ref>{{cite news| url=http://insidetv.ew.com/2013/06/01/jean-stapleton-dies-at-90/ }}</ref>, which the tool didn't catch because it used {{cite news}} instead of {{cite web}}. --Ahecht (TALK
PAGE) 15:18, 17 October 2014 (UTC)

Good idea. Will do it later. Zhaofeng Li [talk... contribs...] 07:18, 19 October 2014 (UTC)

Suggestion: The following reference(s) could not be filled:

From time to time Reflinks gives the message "The following reference(s) could not be filled: <reason>" where <reason> may be bad URL (401), No title (202), etc.

How about an empty field after each such message into which the editor can enter the missing info. Also a "rerun" button so Reflinks can use the editor supplied info to do its job properly.

For bonus points, make clickable the URL of the page with the missing info, 401s excepted.

Great tool, by the way, this Reflinks.

-Arb. (talk) 11:29, 18 October 2014 (UTC)

Nice one! Added to to-do list. Zhaofeng Li [talk... contribs...] 07:19, 19 October 2014 (UTC)

"Installing" the wmflabs version on my "Tools" menu in the left margin

I'm so glad to have discovered this! How do I add the "official" wmflabs version - https://tools.wmflabs.org/fengtools/reflinks/ - to the "tools" menu on the left of my screen? For the old version I had several lines of code in my common.js page. Roger (Dodger67) (talk) 16:40, 13 September 2014 (UTC)

@Dodger67: Just add {{subst:iusc|User:Zhaofeng Li/Reflinks.js}} to your common.js. You also get the option to install it via Greasemonkey, see User:Zhaofeng Li/Reflinks#Toolbox link for details. Thanks for using the tool! Zhaofeng Li [talk... contribs...] 00:14, 14 September 2014 (UTC)
Thanks! Roger (Dodger67) (talk) 06:07, 14 September 2014 (UTC)
Is there a way to add the "test" version to the Tools menu? --Ahecht (TALK
PAGE) 14:58, 17 October 2014 (UTC)
@Ahecht: Yeah, but the test version may conflict with the stable one. Just replace Reflinks.js with Reflinkstest.js in your common.js. I just did a major restructure, and many things may not work. Please report any bugs you've found, thanks! Zhaofeng Li [talk... contribs...] 07:18, 19 October 2014 (UTC)
Using the test version didn't fix my issue apparently. When I go to the tool directly I see the option to use the base URL as the work, but I don't see that option if I click (options) on the tools menu. --Ahecht ([[User_talk:|TALK
PAGE]]) 17:33, 23 October 2014 (UTC)
@Ahecht: It should work when you have only the test version in your common.js. Also, try clearing your browser cache. Zhaofeng Li [talk... contribs...] 00:07, 26 October 2014 (UTC)

Can you make your tool works in ”quick mode” on other wikipedias too, not only on en.wp? (without copy-paste, just in 1-2 clicks). --XXN (talk) 18:56, 14 October 2014 (UTC)

@XXN:  Doing... Yeah, of course. I'm doing some major refactoring at the moment, will do it later. Zhaofeng Li [talk... contribs...] 11:11, 17 October 2014 (UTC)
@XXN: This now available for a couple of wikis on the test version. Could you try it out? Thank you. Zhaofeng Li [talk... contribs...] 07:21, 19 October 2014 (UTC)
Tested on wmflabs page for frwiki. Works. Can you add rowiki to select option please? Thank you. For the case when tool script is imported and is used directly from a page of a specific wikipedia i think you can make tool to work for every wikipedia by grabbing wiki's subdomain name identifier via regex ([a-z-]+). For this page it's a bit more work. XXN (talk) 20:54, 19 October 2014 (UTC)
@XXN: Okay, rowiki is added. Automatic wiki detection is coming soon. Zhaofeng Li [talk... contribs...] 00:11, 26 October 2014 (UTC)
Thank you. --XXN (talk) 00:19, 26 October 2014 (UTC)

Easy change

Making the article name entry field longer would be an easy and helpful change, so editors can see the whole name being entered.--DThomsen8 (talk) 02:13, 29 October 2014 (UTC)

Great suggestion! An easy thing to do, yet yielding significant usability improvements. — Dsimic (talk | contribs) 00:58, 31 October 2014 (UTC)
@Dthomsen8 and Dsimic:  Done Additionally, the field now auto-resizes itself it the name entered gets too long. I've also merged the new version with a revamped code structure and multi-wiki support in (It's been sitting in the test version for ages), please notify me if anything breaks, thanks! Zhaofeng Li [talk... contribs...] 11:31, 31 October 2014 (UTC)
Thank you for doing that so quickly, it looks and works great! — Dsimic (talk | contribs) 17:48, 31 October 2014 (UTC)

When I try to access Reflinks, I get "Internal error: The URI you have requested, /fengtools/reflinks/, appears to be non-functional at this time." --Jax 0677 (talk) 23:04, 4 November 2014 (UTC)

Something is wrong

@Zhaofeng Li:,@Dispenser: the reflinks tool doesnt add the Date anymore just title and publisher. Weird and kind of annoying :) Take a look at Isabel Adrians article for an example and here. I did press away "Do not add access dates" so that is not the problem. Cheers. --BabbaQ (talk) 16:10, 31 October 2014 (UTC)

Same thing for me. I'll add access dates manually for now. Thanks for resurrecting reflinks! – Muboshgu (talk) 16:24, 31 October 2014 (UTC)
That's by design. If you didn't actually access the URL today it shouldn't be adding today's date. If you want it to add the date, simply uncheck "Do not add access dates" in the options. --Ahecht (TALK
PAGE) 20:50, 31 October 2014 (UTC)
That is my point I have unchecked it. And it still does not add the date. Otherwise I would not have asked :) I guess it is the same for user Muboshgu.--BabbaQ (talk) 20:58, 31 October 2014 (UTC)
Technically the tool is "accessing" the web page to fill in its contents. I don't think we're expected to go back into article histories to find what day the bare url was actually added to the article. – Muboshgu (talk) 15:05, 1 November 2014 (UTC)
@BabbaQ:  Fixed It was a bug and it's now fixed, see User talk:Zhaofeng Li#Reflinks: Access dates. Zhaofeng Li [talk... contribs...] 09:10, 1 November 2014 (UTC)
It's not loading for me at the moment, but I'll try it again later. – Muboshgu (talk) 15:05, 1 November 2014 (UTC)
It's working! Thanks! – Muboshgu (talk) 16:16, 1 November 2014 (UTC)
@Muboshgu: Actually, you are expected to manually verify whether the webpage still supports the statement. It's true the tool accesses the page, but there is no way for it to check if it still supports the corresponding on-wiki material. Zhaofeng Li [talk... contribs...] 01:43, 2 November 2014 (UTC)
@Zhaofeng Li:. Some sort of Internal error at the moment with the tool.--BabbaQ (talk) 12:54, 2 November 2014 (UTC)
@Zhaofeng Li:, I probably sound like a sour-puss but it has been impossible to get to the Reflinks page all day. It does not upload. Cheers :)--BabbaQ (talk) 22:02, 8 November 2014 (UTC)

@BabbaQ: Sorry for the all the problems. Looks like the server was running out of sockets, again. Trying to investigate into it... Zhaofeng Li [talk... contribs...] 00:05, 9 November 2014 (UTC)

@Zhaofeng Li:, that is OK. Technical things are never easy :) The tool seems to be down completely at the moment...--BabbaQ (talk) 19:07, 10 November 2014 (UTC)
@Dispenser: perhaps you could take a look at the problem.--BabbaQ (talk) 22:17, 11 November 2014 (UTC)

Server is giving error and no services are working. Giving out database errors. Please fix. Thanks

Seems to be working now, and Pingdom didn't log anything last night. Moreover, Reflinks doesn't use databases at all. Zhaofeng Li [talk... contribs...] 22:32, 19 November 2014 (UTC)
Uhh, the original one by Dispenser gives out database errors. That version doesn't work, and is not expected to work again. Use this one instead. Zhaofeng Li [talk... contribs...] 01:53, 20 November 2014 (UTC)

Bug report!

Hello Zhaofeng! I'm here to report a bug in your javascript to access the tool. I've mw.loader.load()ed your script in my vector.js and the script loads fine. When I clear my error console after the page is completely loaded and I click on (options), I get a $().name is not a function error that kills the script. That error is on line 84 of your script as follows: 

    </form>\
</div>" );
    if ( !rlIsWatching() ) {
        var nowatch = $( "<input>" ).name( "nowatch" ).type( "hidden" ).value( "y" );
        $( "#reflinks-form" ).append( nowatch );
    }
    $( "html, body" ).animate( {
        scrollTop: $( "#reflinks" ).offset().top - 10
    }, 250 );
}

I'd be happy to provide any specs you need to fix this bug. Thanks! — {{U|Technical 13}} (e • t • c) 04:43, 29 November 2014 (UTC)

@Technical 13:  Fixed that embarrassing bug.e5d899 Gonna get some solid knowledge of the jQuery API soon. Thanks for the report! Zhaofeng Li [talk... contribs...] 05:38, 29 November 2014 (UTC)
@Zhaofeng Li: - right now when you have typed in the name of the article you want to be linked and you press Fix page you get a message Error! Not found. --BabbaQ (talk) 00:15, 30 November 2014 (UTC)
@BabbaQ: The tool is working for me (both through the main web interface and the toolbox script). Could you try again? Zhaofeng Li [talk... contribs...] 02:40, 30 November 2014 (UTC)
My wrong @Zhaofeng Li:. It is not on the English version it is on the Swedish Wiki refs fix that it does not work. --BabbaQ (talk) 12:03, 30 November 2014 (UTC)

In this edit, Reflinks apparently wrapped an existing URL in a {{cite web}} template, but it missed a step. Any existing vertical bars (pipes), i.e. the "|" character, need to be converted to %7c in order to be used inside a citation template. Can you please adjust the Reflinks code if it does not already make this change? Thanks.

P.S. Pipes in parameters other than the URL parameter need to be converted to | in order to be used inside a citation template. – Jonesey95 (talk) 02:18, 1 December 2014 (UTC)

@Dispenser: Since this is apparently an edit made with your version of the tool, I hope you can take a look.
Anyway, @Jonesey95:, a similar problem exists in my rewrite (It replaces the pipes incorrectly), will fix it soon. Zhaofeng Li [talk... contribs...] 04:44, 1 December 2014 (UTC)
 Fixed Zhaofeng Li [talk... contribs...] 13:40, 1 December 2014 (UTC)
Thanks! – Jonesey95 (talk) 16:09, 1 December 2014 (UTC)

Unnamed discussion

Hi User:Zhaofeng Li, I used the magical tool. It is just great. Thank you! Lotje (talk) 06:43, 12 September 2014 (UTC)

Enjoy! Please report any problem you encounter on my talk page, thanks. Zhaofeng Li [talk... contribs...] 13:06, 12 September 2014 (UTC)
Thank you for this one :) It is really helpful. Though it does not combine references, but that is just a small thing. :). Thanks again.--BabbaQ (talk) 23:42, 13 September 2014 (UTC)
I'm pretty sure that there's actually a bot (I forget the name of it off-hand) that will go around & combine references that are exactly duplicated within an article more than once. I don't know how often the bot roams around Wikipedia, but I would expect it to find recently duplicated references within a few days of their insertion into a particular article. Guy1890 (talk) 00:38, 14 September 2014 (UTC)
Sadly, it seems there are no bot for combining refs. Because I have not seen any such bot doing anything on any of the articles that I have created.--BabbaQ (talk) 17:55, 14 October 2014 (UTC)
The bot that I was thinking of earlier is apparently User:Yobot, but I'm not sure exactly how that bot picks out which articles to edit. Sometimes I see it clean up after adding some exactly duplicated citations to a particular article, and sometimes it doesn't seem to make it around to certain other Wikipedia articles. Guy1890 (talk) 04:48, 15 October 2014 (UTC)
I guess User:BattyBot also does some of these same kind of fixes as well. Guy1890 (talk) 14:54, 19 October 2014 (UTC)
@Guy1890 and BabbaQ: Part of AutoWikiBrowser's general fixes is to combine duplicate references. While many users and bots such as Yobot and BattyBot perform these general fixes when doing other tasks, I'm not aware of any bot that combines duplicate references as its primary task. GoingBatty (talk) 15:03, 19 October 2014 (UTC)
I think such a bot would come in handy for many users if one was created. @GoingBatty:--BabbaQ (talk) 15:12, 19 October 2014 (UTC)

Old concerns

Hi.

I see that Reflinks is re-activated. This both good news and bad news. You see, I've once nominated an article for WP:FA (and succeeded, going through the brutal process and coming out alive!) and I can tell you that Reflinks generates less than desirable output. Here are some of my main concerns.

  • Incorrect population of |publisher=: As it is visible in revision #636168922, Reflinks has placed Jwz.org, foldoc.org, stallman.org, Xemacs.org and tux.org into |publisher=, which is wrong. They must go into |website=.
  • Please populate |website=: Domain names such as Jwz.org, foldoc.org, stallman.org, Xemacs.org and tux.org should go into |website= (or its alias, |work=, if you prefer}}.
Real-world examples

To give you a better idea of what is |website= and what is |publisher=, is a common list of websites used in the computing articles and the corresponding |website= and |publisher= parameters:

Source website |website= or |work= |publisher=
pcmag.com PC Magazine Ziff Davis
pcworld.com PC World IDG
cnet.com CNET CBS Interactive
download.com Download.com CBS Interactive
arstechnica.com Ars Technica Condé Nast
theverge.com The Verge Vox Media
lifehacker.com Lifehacker Gawker Media
msdn.microsoft.com MSDN Microsoft
technet.microsoft.com TechNet Microsoft
blogs.msdn.com/b/b8/ Building Windows 8 Microsoft
blogs.msdn.com/b/e7/ Engineering Windows 7 Microsoft
  • Converting some {{Cite web}} to {{Cite news}}: To put it short: Just don't do it. Why? Please read below.
Why not convert between {{cite web}} and {{cite news}}?

Effective 2014, with the deployment of Module:Citation/CS1, the differences between {{Cite web}} and {{Cite news}} is almost completely eliminated, except for the following:

  1. {{Cite news}} puts the name of the publisher into parentheses while {{Cite web}} does not
  2. {{Cite web}} throws an error if |url= is not supplied.

Other than this, both templates accept the same parameters, even those that their documentation does not mention. So, the only thing that converting {{Cite web}} to {{Cite news}} does is to make a WP:FA nomination crash and burn because the citation styles are not consistent. Whether a page entirely uses {{Cite news}} or entirely {{Cite web}} is now a matter of local consensus.

  • Capitalizing and de-capitalizing template name's first letter: Just don't! There is no shortage of scripts and tools that change "Cite web" to "cite web" and vice versa while both are equally valid. Doing so only increases the size of diffs. According to a footnote in MOS:STABILITY it must not happen.

Best regards,
Codename Lisa (talk) 17:33, 1 December 2014 (UTC)

@Codename Lisa and Technical 13: Thanks for the comments. However, those problems only exist in the original Reflinks written by @Dispenser:, not in my rewrite. My rewrite only generates an empty |publisher= parameter (And yes, I can improve my rewrite with your table), and does not change anything in references that are already filled. Zhaofeng Li [talk... contribs...] 22:34, 1 December 2014 (UTC)
revision #636168922 shows your rewrite also puts website domain names into |publisher= instead of |website=. Thanks. Best regards, Codename Lisa (talk) 23:41, 1 December 2014 (UTC)
@Codename Lisa: There's never code that populate the publisher field in my rewrite. There's a non-default option to include the domain name, but only in |work= (See this edit). Although it looks strange, my best guess is that @Derek R Bullamore: did that manually. Zhaofeng Li [talk... contribs...] 01:29, 2 December 2014 (UTC)
(edit conflict) I am unable to reproduce the above edit using Zhaofeng Li's reflinks. Codename Lisa, if you copy the old XEmacs page to a sandbox page and run reflinks on it, does it add the |publisher= parameters? When I do it, Reflinks says there are no proposed changes. It is useful to remember when looking at Reflinks edits that humans can modify the article beyond the Reflinks-proposed changes while preserving the Reflinks edit summary. – Jonesey95 (talk) 00:35, 2 December 2014 (UTC)
You are right: I had forgotten there is an interactive mode too that allows editors to modify things manually. Setting up a test case now...
I apologize for this oversight.
Best regards,
Codename Lisa (talk) 08:26, 3 December 2014 (UTC)
Test completed. I've fed rv 636168218 to the Reflinks via copy and paste. I am updating my request above.
Thanks for the taking the time to clarify my confusing. My mistake. Sorry.
Best regards,
Codename Lisa (talk) 08:47, 3 December 2014 (UTC)
@Codename Lisa: Thanks for the feedback. Actually, the tool already tries to obtain |work= from the metadata on the web pages. There is an option called Use the base domain name as work when this information cannot be parsedWhat an ugly name... which does exactly what you want. Should this be made default? Also, the test version now tries to fill in |publisher= according to a hard-coded list as well, as there isn't a reliable way to obtain this information from the pages themselves. By the way, there isn't a separate "interactive mode" in my rewrite: Editors always can edit the result wikitext manually to fix the errors before it's saved. Zhaofeng Li [talk... contribs...] 09:54, 3 December 2014 (UTC)
@Zhaofeng Li: Although not all pages use it, you could look for OGP metadata and grab the content of the <meta property="og:site-name" content="Example" /> tag. --Ahecht (TALK
PAGE) 15:27, 4 December 2014 (UTC)
@Ahecht: It already does that. Zhaofeng Li [talk... contribs...] 03:36, 5 December 2014 (UTC)

As a note, you can manually feed raw wikitext to the tool. Just click Show advanced input on the web interface. Zhaofeng Li [talk... contribs...] 10:10, 3 December 2014 (UTC)

Hi. Yes, I think either this check box must be the default or the user must prompted for one. And no, its name is not ugly. Something that baffles me however, is the effect of "Do not expand references with a captioned external link only" check box. Best regards, Codename Lisa (talk) 11:47, 3 December 2014 (UTC)
@Codename Lisa: Okay, will make it default soon. The baffling option, if selected, will make the tool refrain from changing references like [URL title]. Any suggestions for a better name? Zhaofeng Li [talk... contribs...] 13:35, 3 December 2014 (UTC)
The "domain name as work" option is now enabled by default. Zhaofeng Li [talk... contribs...] 14:07, 3 December 2014 (UTC)
Well yes, the name is partially problematic. "Captioned external link" is ambiguous. "Bare link" means this: "http://www.example.com". "Plain link" means this: "[http://www.example.com]" or "[http://www.example.com Example.com website]". Which one do you mean?
My suggestion would a list box titled "What to expand" instead of a checkbox; the list box should preferably contain four items:
  • Bare links only (e.g. http://www.wikipedia.com)
  • Bare links and unnamed plain links (e.g. [http://www.wikipedia.com])
  • Bare links and all plain links (e.g. [http://www.wikipedia.com Wikipedia home page])
Best regards,
Codename Lisa (talk) 14:03, 4 December 2014 (UTC)
@Codename Lisa: The second "captioned" one. The list box idea sounds good, will research in that direction. Zhaofeng Li [talk... contribs...] 03:36, 5 December 2014 (UTC)
The third issue, Capitalizing and de-capitalizing template name's first letter was a very big problem. Imagine this: An article had two bare links and 120 properly referenced one. Only these references were hand-coded. Now, some lazy guy run Reflinks on it to repair the remaining two. His fix log, however, would have had 120 changes which crowded the diff and were absolutely unnecessary. I agree that change between the capitalization is absolutely bad. Fleet Command (talk) 16:01, 9 December 2014 (UTC)
  • You do realize that if there are hundreds of lower case template calls, that the server (because of forced capitals) has to convert the lowercase calls to the uppercase pagename (where the template is actually stored) which increases server load, right? I'd think removing this additional server load (which depending on how many calls there are may or may not be a negligible difference in load times) would be a good thing. I do it automatically on most of my edits anyways. Perhaps the best way to tackle this is to offer a checkbox that will "capitalize template calls" or something. I wouldn't want this feature to go away. — {{U|Technical 13}} (e • t • c) 16:38, 9 December 2014 (UTC)
@FleetCommand, Technical 13, and Codename Lisa: Again, there's no such feature. It's never the intended effect. Reflinks skips any reference it doesn't recognise. It's either a) the editor making the edit changed them manually, or b) you hit a bug (Make sure you can replicate it). Please clarify. Zhaofeng Li [talk... contribs...] 02:22, 10 December 2014 (UTC)
@Zhaofeng Li: FC's message said "was". So, I guess he is saying "thank you". (Okay, the second part of my message is absolutely a matter of WP:AGF but still, his message said "was".)
@Technical 13: This argument is false because I know for a fact that the conversion has absolutely zero effect. You are essentially refactoring the factually inaccurate argument of the proponents of changing redirects to direct links: Reducing load. It is like arguing that killing ant reduces the load off of Atlas (mythology). I am afraid you and those proponent have no idea how colossal the server load is.
Best regards,
Codename Lisa (talk) 10:42, 12 December 2014 (UTC)
  • Lisa (remembered to leave the d off this time :D), it's much more like doubling pennies or a ton of feathers weighs as much as a ton of bricks than killing ant reduces the load off of Atlas (mythology). This debate however, whether they should be capitalized or not, is mostly one of personal preference and no more am I going to convince you to not use lowercase than are you going to convince me to not change them all to uppercase. — {{U|Technical 13}} (e • t • c) 12:32, 12 December 2014 (UTC)
@Technical 13: Resolving a Wikipedia page take exactly 44 SQL queries for the server and first caps and first small both weight the same for it. In addition, if you read the footnote in MOS:STABILITY, you will find that stubbornness and ignorance often come hand in hand; so I think you'd rather not go the fall way that other have already gone. If it is personal preference, then ArbCom says don't change existing instances. Fleet Command (talk) 22:48, 14 December 2014 (UTC)
  • FleetCommand If that was the only change I was making, you'd have a point. Otherwise, there are community policies such as Cosmetic changes should only be applied when there is a substantial change to make at the same time. that say that making such changes is perfectly fine as long as it is not the only changes being made. Happy editing! — {{U|Technical 13}} (e • t • c) 23:15, 14 December 2014 (UTC)

Bug report - Billboard template

Your tool doesn't seem to be able to convert the template Template:BillboardURLbyName into a full reference. See this diff where I had to put in the URL manually to get the tool to pick up the reference. All the template did was show the URL. It didn't provide any more info which a cite web template would. Dismas|(talk) 09:51, 17 December 2014 (UTC)

@Dismas: It's an edge case, and the tool cannot parse templates without some major restructuring at the moment (Keep in mind that the tool can only see the source template code, not the resulting URL). I'd say correct them manually by now, since I don't think there are too many of them. By the way, the template is intended to be used as the |url= parameter of {{cite web}}. Zhaofeng Li [talk... contribs...] 10:15, 17 December 2014 (UTC)
I see. Thanks. Dismas|(talk) 10:16, 17 December 2014 (UTC)

Changing ref names to a colon and number

I don't like it changing meaningful ref names to a name consisting of a colon and a number, eg, ":1". Is this feature and its rationale documented somewhere? Nurg (talk) 00:38, 18 December 2014 (UTC)

(talk page watcher) @Nurg: See User talk:Zhaofeng Li#Combine duplicate references? for a discussion on this topic. GoingBatty (talk) 04:27, 18 December 2014 (UTC)
thanks. Nurg (talk) 10:02, 18 December 2014 (UTC)

"Author! Author!"?

Add blank |author= and |author= fields if the information is unavailable When the corresponding metadata is unavailable, insert blank |author= and/or |date= for filling in manually.

Shouldn't that second "author" be "date"? --Thnidu (talk) 17:33, 18 December 2014 (UTC)

@Thnidu: Nice catch! Fixed. There are also a few others whose captions don't match the actual ones on the tool page. Will deal with them later. And of course, you can also edit it if you have time. Thanks. Zhaofeng Li [talk... contribs...] 22:53, 18 December 2014 (UTC)

Author (2)

Hello Zhao, I tried your bot a few times, but I don't like the fact that the title and author are left out. Why is a publisher drawn in, and a lot of times left empty? The tribute to the author and the complete title of the webpage should be saved, not the other way around. Can you fix that? Taksen (talk) 07:18, 27 December 2014 (UTC)

@Taksen: The publisher field is left empty for convenience when it cannot be found (It currently gets the data from a hard-coded list, which support only a few sites), so that it can be filled manually. The author field is only available on certain pages which contains metadata. I don't understand what you mean by "the title is left out". Zhaofeng Li [talk... contribs...] 08:25, 27 December 2014 (UTC)
I appreciate your work, but have to find out what should be changed manually in order to keep to full title and the author's name is provided. Here you can see what happened [1]l. Sometimes the author is added, sometimes left out. The title is often cut. In other cases your bot found a title. In any case could you add an empty field for author, just like you did for the publisher? So the user can fill in the empty ... I hope I made myself clear enough.Taksen (talk) 09:28, 27 December 2014 (UTC)
@Taksen: There is an option labelled Add blank metadata fields when the information is unavailable which does exactly what you want. I'm working on a feature that will let you save your preferred options. As for the title, the tool automatically strips the site name from the end of the titles which is not supposed to be there (|work= would be a better place for this). But things can go wrong when the title on your browser bar (i.e. <title>) doesn't contain the actual title at all (for example, The Voice of Russia pages). In this case, you'll need to fill in the title field manually. Zhaofeng Li [talk... contribs...] 10:10, 27 December 2014 (UTC)

Bad titles: "Cookies must be enabled." and "No cookies"

Hi Zhaofeng Li! Could you please adjust your Reflinks tool to not create references when the title is "Cookies must be enabled." and "No cookies"? (e.g. try Inga Peulich) The original Reflinks gives a warning stating "Unusable web page title". Thanks! GoingBatty (talk) 16:04, 3 January 2015 (UTC)

Good idea. Will do when I'm free. Thanks. Zhaofeng Li [talk... contribs...] 09:38, 4 January 2015 (UTC)
@GoingBatty:  Done The test version now performs this check. Zhaofeng Li [talk... contribs...] 14:33, 10 January 2015 (UTC)

"Use the base domain name as work when this information cannot be parsed" not working

@Zhaofeng Li: The "Use the base domain name as work when this information cannot be parsed" option doesn't seem to be working, either in reflinks or reflinkstest. The test version seems to be parsing the base domain name correctly and using it as the <ref name=________>, but it isn't inserting it as work=________. --Ahecht (TALK
PAGE) 16:36, 14 January 2015 (UTC)

@Ahecht:  Fixed It's now on the test version which will go stable in a few days. Thanks for reporting. Zhaofeng Li [talk... contribs...] 02:47, 15 January 2015 (UTC)
Thank you --Ahecht (TALK
PAGE) 17:09, 15 January 2015 (UTC)

Problem with watchlist

@Zhaofeng Li: Everything is working fine, except that now the reflink tool is removing the articles that I use the tool on from my Article watchlist. I have to press "Watch this article" button everytime before saving to keep the article on my Watchlist.--BabbaQ (talk) 22:34, 19 January 2015 (UTC)
@BabbaQ:  Fixed Look like I wasn't awake enough when writing the code. Could you confirm? Zhaofeng Li [talk... contribs...] 00:57, 20 January 2015 (UTC)

When I try to access reflinks, I get "504 Gateway Time-out nginx/1.5.0" --Jax 0677 (talk) 21:29, 27 January 2015 (UTC)

While it's down, you can use User:Dispenser/Reflinks as an alternative. I've just used it so it appears to be fine. MadGuy7023 (talk) 23:53, 27 January 2015 (UTC)
 Fixed See below. Zhaofeng Li [talk... contribs...] 03:25, 28 January 2015 (UTC)

No webservice

I'm getting a no webservice error when I try to use reflinks . -- t numbermaniac c 00:56, 28 January 2015 (UTC)

Didn't know about that. Thanks. -- t numbermaniac c 02:48, 28 January 2015 (UTC)
 Fixed I've restarted the server. Let's see how long it can last... Zhaofeng Li [talk... contribs...] 03:25, 28 January 2015 (UTC)

Bookmarklet

the following bookmarklet works for me in Firefox: 

javascript:if(typeof%20wgEnableAPI=='undefined'){alert('This%20bookmarklet%20only%20works%20on%20Wikimedia%20pages')}else{xhr=new%20XMLHttpRequest();xhr.onreadystatechange=function(){if(xhr.readyState==4){eval('pages=('+xhr.responseText+').query.pages');for(var%20page%20in%20pages)edittoken=pages[page].edittoken;if(edittoken=='+\\')if(!confirm('You%20are%20not%20logged%20in,%20continue?'))return;location.href='https://tools.wmflabs.org/fengtools/reflinks/result.php?defaults=y&nowatch=y&wiki='+wgContentLanguage+'&page='+encodeURIComponent(wgPageName);};};xhr.open('GET',wgScriptPath+'/api.php?action=query&prop=info&intoken=edit&titles=M&format=json',true);xhr.send(null);}

it's basically an adaptation of the one for the old reflinks. feel free to use/share however you like. Frietjes (talk) 23:50, 29 January 2015 (UTC)

Adding in javascript, ads, spam and other fun-filled oddities

See here. Bgwhite (talk) 06:51, 1 February 2015 (UTC)

@Bgwhite: It's dumping the entire source of the page into the title parameter. Looking into it. Zhaofeng Li [talk... contribs...] 08:07, 1 February 2015 (UTC)
For those interested, the page is using <title> in a weird way (deliberately, I think, to combat bots):
<title
>World Religions</title
>
Which confused the HTML5 parser I'm using. Surprisingly, such usage is valid. I'll see if I can fix that when I get home. I'll also add a length limit to the tool to prevent errors like this. Zhaofeng Li [talk... contribs...] 08:17, 1 February 2015 (UTC)
 Fixed Patch made and applied. Zhaofeng Li [talk... contribs...] 04:03, 2 February 2015 (UTC)

Hi Zhaofeng Li! Since the original Reflinks is now working at http://dispenser.homenet.org/~dispenser/view/Reflinks and http://dispenser.homenet.org/~dispenser/cgi-bin/webreflinks.py you might want to consider renaming your tool to avoid confusion. GoingBatty (talk) 16:12, 3 January 2015 (UTC)

It's actually been working for quite a while, though intermittently (see the top of the page). I'll consider renaming the tool to reduce confusion and to make its purpose obvious for new users. Zhaofeng Li [talk... contribs...] 09:40, 4 January 2015 (UTC)
@Zhaofeng Li and GoingBatty: That's really important. There could be a contest or something, to name it. Refminer, refgnomer, refiner, dunno. But much more importantly, I'd like to know what plans there are to merge with core infrastructure like Citoid. I really hope you guys aren't all reinventing the wheel any more than the original reflinks forces people to do by unbelievably withholding its source. What plans are there? Good job, man. Thanks. — Smuckola (Email) (Talk) 00:27, 2 February 2015 (UTC)
@Smuckola: I will start a naming contest soon. About Citoid, it's a Node.js-powered API that can generate citation metadata from URLs. the tool is now flexible enough to use Citoid as the back-end to get metadata, we just need to implement CitoidLinkHandler as a drop-in replacement of the current StandaloneLinkHandler which does all the heavy lifting ourselves. Zhaofeng Li [talk... contribs...] 04:11, 2 February 2015 (UTC)
@Zhaofeng Li: Wow, you're super cool, man. I was wondering if Citoid even had a GUI other than the whole-article visual editor, which I never use. Is that presently the only one, until you migrate this one to it? Do you think your backend has some functionality to offer them, or is one of them clearly superior? — Smuckola (Email) (Talk) 04:44, 2 February 2015 (UTC)

Replacing Greek with Chinese.

When I run Reflinks on Alpha Phi Omega, in the second cite, it indicates that AΦΩ should be replaced with A桅惟 for the title. I can go in and change it back by hand of course, but that one seems wierd.Naraht (talk) 18:53, 31 January 2015 (UTC)

It's the character set detection problem described on User talk:Zhaofeng Li#Problems running Reflinks on sites with Arabic titles. Could you try out if it's fixed in the test version? Zhaofeng Li [talk... contribs...] 03:39, 1 February 2015 (UTC)
Seems OK in the test version.Naraht (talk) 20:32, 5 February 2015 (UTC)

Error with naming refs

@Zhaofeng Li: When I run reflinks on https://en.wikipedia.org/w/index.php?title=87th_Academy_Awards&oldid=645949913 I got a strange result. The following reference was used twice in the article:

<ref>{{cite web|title=Brit Phenom Rita Ora To Perform At The Oscars® |url=http://www.oscars.org/news/brit-phenom-rita-ora-perform-oscarsr|accessdate=January 28, 2015|publisher=Oscar.com|date=January 28, 2015}}</ref>

The first time it was used it has name=auto added, the second time it had name=auto1 added, and then the entire original reference got copied to the bottom, after the categories. If the intention is to consolidate refs, shouldn't the first instance just had name=auto added, the second instance changed to <ref name=auto />, and nothing copied to the end of the articles after the categories? --Ahecht (TALK
PAGE) 21:54, 6 February 2015 (UTC)

 Fixed See below. Zhaofeng Li [talk... contribs...] 02:27, 7 February 2015 (UTC)

Problem

Zhaofeng Li when using Reflinks it seems to be adding a cite to the bottom of the page [2],
Thanks, –Davey2010Talk 00:31, 7 February 2015 (UTC)

@Davey2010 and Ahecht:  Fixed Good catch! Zhaofeng Li [talk... contribs...] 02:26, 7 February 2015 (UTC)
Ah brilliant thank you :) –Davey2010Talk 02:29, 7 February 2015 (UTC)

apostrophe

If you've unchecked 'Do not add access dates' and there's an apostrophe in the name of the article, only the part before the apostrophe gets passed

@Fuddle: Can't reproduce it. Could you give me an example of such article? Zhaofeng Li [talk... contribs...] 12:42, 3 February 2015 (UTC)
You've Got Mail sends me to https://tools.wmflabs.org/fengtools/reflinks/result.php?page=You
@Fuddle:  Fixed1a48e4 Thanks for the report. On an unrelated side note, please consider signing your posts. Zhaofeng Li [talk... contribs...] 05:46, 7 February 2015 (UTC)

Feature

Another feature request. Can we have a checkbox (enabled by default) to provide a named reference? <ref name="TitlePublisher01" How about a link to autopopulate a search box for the book's title at worldcat.org? It'd be nice if it could grab the ISBN and OCLC and such from worldcat, or at least jumpstart my search there. Ultimately, we'd want a fully featured GUI that's aware of all of the 'cite' templates and their metadata options, and descriptions as mouseover popup texts.— Smuckola (Email) (Talk) 04:46, 2 February 2015 (UTC)

Sometimes while I'm editing wikitext, I suddenly discover that I need to use reflinks. Is it possible for the link on the lefthand toolbar to launch the reflinks page in a new tab, importing the contents of my existing editing buffer? As it stands, it's not a big hardship because I copy and paste the existing buffer, open a new tab, click the 'reflinks' bookmark button atop my browser window, paste the article's wikitext into reflinks, configure and run reflinks, copy the results back to my original buffer, and usually disregard giving any credit to reflinks in the edit summary. Thanks! — Smuckola (Email) (Talk) 20:24, 5 February 2015 (UTC)
Added to todo list. Thanks. Zhaofeng Li [talk... contribs...] 02:40, 7 February 2015 (UTC)
How about a checkbox option to fetch the Wayback archive with deadurl=[yes/no]? I may have suggested this before but I don't see it. — Smuckola (Email) (Talk) 20:08, 6 February 2015 (UTC)
@Smuckola: Thanks for all those suggestions. I'm planning to add an interface similar to the "interactive mode" in the original tool, which allows you to change individual references easier. About WorldCat, it's presently not possible to fetch data from it as the API is restricted to "qualifying institutions"(1). I'll try to implement the Wayback look-up feature after I get the interactive interface done, and it will be integrated into the interface. Zhaofeng Li [talk... contribs...] 02:40, 7 February 2015 (UTC)
@Zhaofeng Li: API? I didn't know about that. I was expecting a little dab of screen scraping anyway, manually (manly, amirite) forming a form submission. ;) — Smuckola (Email) (Talk) 19:27, 7 February 2015 (UTC)
Yes, that's another way to do it. Zhaofeng Li [talk... contribs...] 01:30, 8 February 2015 (UTC)
@Zhaofeng Li: I was basically just thinking of a one-click button to launch a new tab at the appropriate search results at worldcat. It would save the user a copy and paste and such, if it just opened a tab containing a url like this one. It would be too chaotic to actually screen scrape, because worldcat's database isn't even normalized. It has so many duplicates and one-offs that I can hardly imagine how even the API could function unless you already had the OCLC or ISBN. — Smuckola (Email) (Talk) 01:50, 8 February 2015 (UTC)
@Zhaofeng Li: Are you planning on migrating this app over to nest within Visual Editor's UI? So it can go mainstream? Are you working with them or the Citoid guys at all? I don't see any automated tools there except for maybe one url at a time. And I don't see any way to spontaneously switch between wikicode editing and Visual Editor, so that would be one reason to need this as an ancillary app so far.  :-/ — Smuckola (Email) (Talk) 19:36, 7 February 2015 (UTC)
@Smuckola: There's no plan to integrate it into VE at the moment, but this can be a good idea. I'm not currently working with WMF people, however, I'll try to get in touch with them later. Since VE is powered by Parsoid which can translate pages between HTML and wikitext bi-directionally, it wouldn't be hard to integrate some of the wikitext-based tools into VE. Zhaofeng Li [talk... contribs...] 01:30, 8 February 2015 (UTC)

Erroneous removal of reference

Thank you for this very useful script.

However, in this edit, it removed the entire {{cite book|last=Noszlopy|first=George T.|authorlink=George T. Noszlopy|title=Public Sculpture of Birmingham including Sutton Coldfield|publisher=Liverpool University Press|year=1998|series=Public Sculpture of Britain|volume=2|isbn=0-85323-682-8}} reference. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 10:09, 10 February 2015 (UTC)

(talk page watcher) @Pigsonthewing: It looks like it tried to combine references by replacing <ref name="Noszlopy">{{cite book|last=Noszlopy|first=George T.|authorlink=George T. Noszlopy|title=Public Sculpture of Birmingham including Sutton Coldfield|publisher=Liverpool University Press|year=1998|series=Public Sculpture of Britain|volume=2|isbn=0-85323-682-8}}</ref> with <ref name="Noszlopy"/> - note the extra slash - but it should have left the first instance of the reference alone. GoingBatty (talk) 03:45, 11 February 2015 (UTC)
@Pigsonthewing:  Fixed Thanks for reporting. Zhaofeng Li [talk... contribs...] 08:55, 11 February 2015 (UTC)

# error

I have just recently begun using your tool, but I am given # to replace any content with . or otherwise. It can be viewed here. I have since then reverted my edit. Thank you. Callmemirela (talk) 01:15, 22 January 2015 (UTC)

@Callmemirela: If you are using IE, the problem is caused by the Anti-XSS filter. Try using another browser. I've added a note for IE users. Zhaofeng Li [talk... contribs...] 03:05, 22 January 2015 (UTC)
(talk page stalker) @Callmemirela: You can disable the Internet Explorer Anti-XSS filter by going to "Internet Options", going to the "Security" tab, clicking on "Custom level...", scrolling to near the bottom of the list, and selecting "Disable" under "Enable XSS filter". --Ahecht (TALK
PAGE) 20:43, 12 February 2015 (UTC)

Removal of base named ref

Looking at this edit, I understand compressing the second named ref, but why compress the first, leaving no base and causing an error? This was using Google Chrome Version 40.0.2214.94 m on Win7 and the Reflinks js. Thank you.   — Jeff G. ツ (talk) 06:16, 3 February 2015 (UTC)

@Jeff G.:  Fixed Please test and confirm. Thanks for the report! Zhaofeng Li [talk... contribs...] 10:07, 3 February 2015 (UTC)
Tested and confirmed, thank you.   — Jeff G. ツ (talk) 23:57, 14 February 2015 (UTC)

Problems combining refs

I'm very grateful that you've resurrected this app and have been supporting it. I'm writing because the feature that combines references has problems. First, it changes the reference's name if the reference already has one; this may have been done to more easily deal with same references having different names. It wouldn't be a major problem except that reflinks will also steal previously used names from unrelated references, potentially leading to name conflicts. Please compare existing reference names before creating new ones, and consider using the existing name when dealing with a named ref that is to be combined. —Ost (talk) 16:26, 11 February 2015 (UTC)

Not to keep making complaints, but there is an additional problem when using Template:reflist to define references. The reference needs to be defined within the template and not be defined elsewhere within the article. —Ost (talk) 18:07, 11 February 2015 (UTC)
@Ost316: Could you give an example where it occurs? It should reuse the name if one of the duplicates is named (e.g. <code><ref>blah</ref><ref name="test">blah</ref></code> where it'll reuse the name "test") and add numbers to end of an auto-generated name if it's already used elsewhere. Zhaofeng Li [talk... contribs...] 01:47, 12 February 2015 (UTC)
If you run [3] through the tool, you'll see that it it takes the "WFPL-bikepark" reference and renames it "WFPL", which is an existing reference name. It's defined in the reflist template and by your comment below, that is perhaps the cause. —Ost (talk) 20:22, 12 February 2015 (UTC)
@Ost316: You've found a bug. The tool was trying to reuse name=WFPL-bikepark, but parsed the name as "WFPL". Fixed the regex. Zhaofeng Li [talk... contribs...] 02:02, 13 February 2015 (UTC)
THanks for looking into it and fixing! —Ost (talk) 13:36, 19 February 2015 (UTC)
About reflist, the tool currently has no idea where the references should be placed, and will fix the references in situ. Hopefully I'll find a way to deal woth this. Zhaofeng Li [talk... contribs...] 02:04, 12 February 2015 (UTC)
I would think (without looking at any of the code) that you could look for reflist and fix those before parsing the rest of the article, but that may be overly optimistic and simplistic. —Ost (talk) 20:22, 12 February 2015 (UTC)

Suggestions for improvement

Thank you for restoring this tool. It is very useful, but I think it can be made more useful if the following changes were made:

  1. Allow us to select which ref changes we want to continue with after the tool has made its determinations. Start with everything checked, and let us uncheck the ones we don't want to use.
  2. I've noticed that URLs ending in .aspx aren't handled by the tool.
  3. Somehow determine which cites are better done with "cite news", or let us select an entry to convert from cite web to cite news (or the other common cite templates).
  4. It would be nice if this could be run on a section or an individual reference at a time. This is because of some articles with a lot of references to fill in.

Stevie is the man! Talk • Work 21:32, 21 February 2015 (UTC)

(talk page watcher) @Stevietheman: The original Reflinks will let you do #3, and will allow you to set a limit for the number of references as a workaround for #4. See WP:REFLINKS for the new URLs. GoingBatty (talk) 22:16, 21 February 2015 (UTC)
Thanks, @GoingBatty:. When using this link as directed to on that page, it's not giving me the options you describe. What am I missing? Stevie is the man! Talk • Work 02:51, 22 February 2015 (UTC)
As for #4, if you go to https://tools.wmflabs.org/fengtools/reflinks/ and click "Toggle advanced input", you can copy and paste the text of a single section of a page. --Ahecht (TALK
PAGE) 03:10, 22 February 2015 (UTC)
Try using the User script, and change limit=20 to the maximum number of referencs you want Reflinks to expand. GoingBatty (talk) 03:43, 22 February 2015 (UTC)
Thanks to you both. I'll test both options to see if they fit my needs. Stevie is the man! Talk • Work 04:27, 22 February 2015 (UTC)

Dutch/English description when using reFill

Hi, Zhaofeng Li, if you take a look at the 3 last edits of mine, making use of the wonderful reFill tool, you will notice the description is in Dutch. Does that mean we will be able to use this tool on the Dutch wikipedia somewhere in the future? That sounds wonderful. Lotje (talk) 16:38, 26 February 2015 (UTC)

@Lotje: Thanks to frankgeerlings (I don't know his username here), now there are Dutch translations for the tool! Actually, you can already use the tool on Dutch Wikipedia. Just select "nl" from the drop-down list on the web interface. The toolbox link should be compatible with Dutch Wikipedia as well. Try following the same instructions on User:Zhaofeng Li/reFill#Toolbox link and it should add a quick link to the tool in the sidebar. Dutch translations may not show up the first time you use it (The translations are not yet loaded), but it should afterwards. Zhaofeng Li [talk... contribs...] 17:12, 26 February 2015 (UTC)

Time Out

My guess is that the article is too long, but I cannot figure out how to run it in sections on Recognition of same-sex unions in Mexico. Cannot seem to run the program as it always times out. Any suggestions? Thanks! SusunW (talk) 00:16, 27 February 2015 (UTC)

@SusunW: Unfortunately, there is currently no way to run the tool automatically against a specific section, but you can copy a single section manually to the web interface (head to toollabs:refill/, click on "Toggle advanced input" and paste the section into the box. Currently, the time limit is set to 10 minutes. I'm considering extending the limit after a progress bar is implemented. Sorry for the inconvenience. Zhaofeng Li [talk... contribs...] 02:47, 27 February 2015 (UTC)
By the way, Tool Labs is down again at the moment. :/ Zhaofeng Li [talk... contribs...] 02:47, 27 February 2015 (UTC)
Thanks for the quick response. I don't think it has anything to do with it being down at the moment, as I've tried for several days. To "paste a section" are you saying the title of same, i.e. Recognition of same-sex unions in Mexico#Mexico City (Federal District), or the entire verbiage in the section? I'll give it a go tomorrow and see what happens, if you can clarify for me which way to input it. SusunW (talk) 02:52, 27 February 2015 (UTC)
@SusunW: You can also try WP:Reflinks by using http://dispenser.homenet.org/~dispenser/cgi-bin/webreflinks.py?page=Recognition_of_same-sex_unions_in_Mexico&client=script&citeweb=on&overwrite=simple&limit=10&lang=en (The "limit=10" can be adjusted to indicate how many URLs you want it to process at once. Hope this helps! GoingBatty (talk) 03:37, 27 February 2015 (UTC)
Lovely! Thank you. This was one of the first articles I ever edited and it is obviously an on-going situation of change. Once the number of links go to be so daunting, I realized I would probably never go back and convert them all. Someone said try this little program, so...here I am. Appreciate the help. SusunW (talk) 04:13, 27 February 2015 (UTC)
  • FYI, I took a crack at using these tools to try & edit the article in question. I have also noticed in the past that when an article has a very large number of references to process that the "reFill" tool sometimes times out after a good long while. Guy1890 (talk) 04:33, 27 February 2015 (UTC)

reFill

Hi, Zhaofeng Li, can you add support for the nl.wikipedia? Thanks a lot. Lotje (talk) 05:18, 28 February 2015 (UTC)

@Lotje: It's already supported. Just select "nl" instead of "en" in on toollabs:refill/ (besides Page name) and it should work. For quicker access, add this to your common.js on nlwiki:

mw.loader.load( "https://meta.wikimedia.org/w/index.php?title=User:Zhaofeng_Li/Reflinks.js&action=raw&ctype=text/javascript" );

And it will add link labeled reFill in the sidebar.
Zhaofeng Li [talk... contribs...] 09:09, 28 February 2015 (UTC)

your reffill tool isnt ready for prime time

Zhaofeng Li I dont know if nobody ever told you , but people use your tool to insert title and work, thinking all is cool. But your tool does not even find dates for the refs or the publisher. Dates, come on ! who cares about the accessdate when a date is present? the author and publisher are as important as work. what's worse is your tool overwrites existing entries of parameters it doesnt collect. And people use it by the hundreds right now. That's NUTS! --Wuerzele (talk) 06:51, 2 March 2015 (UTC)

Look at this edit today, Catlemur unfortunately destroyed ALL the dates of the 10 references he meant to fix. He overwrote the dates of publication (DOP) with accessdates. The date of the publication ALWAYS trumps the accessdate. Accessdate is ONLY interesting if there is no DOP. I explained to him that the ref template needs to be opened completely to see author and date entry boxes. I asked him, to reinsert all the dates and authors he deleted. what did he do ? he reverted himself/ the use of your tool. I think your tool needs to undergo some more development before you can let it loose--Wuerzele (talk) 23:26, 23 February 2015 (UTC)--Wuerzele (talk) 07:04, 2 March 2015 (UTC)

@Wuerzele: Firstly, sorry for the trouble. Let me start by explaining why the tool removed the dates. The dates are incorrectly placed as the link caption (i.e. inside the brackets) while metadata is supposed to be placed outside the link syntax. As a result, the tool went on happily to replace the reference, thinking that it consists of a single link only. Maybe it's good idea to detect such incorrect usage and skip those references. And the tool does parse dates, when the webpage includes appropriate microdata structures. About publisher, it's a difficult thing to parse since the standarised way to include publisher information isn't widely used. The tool intentionally leaves a blank |publisher= for editors to fill in. I fully agree that my tool isn't perfect, but it's getting better over time as bugs are gradually uncovered by users. Thank you. Zhaofeng Li [talk ♦ contribs] 04:52, 3 March 2015 (UTC)
Zhaofeng LiThanks for your reply. first: what do you mean by "Maybe it's good idea to detect such incorrect usage and skip those references".
FYI I am NOT using your tool. I dont like it, I am complaining to you, and I dont you think you understood that in the observed case, the ref was off WORSE rather than before ! So I am the wrong address for explanations how to use it.
you need to disseminate this info to your users - like at the point of dissemination of the tool, and to users who dont know the above, like Catlmur and who knows how many more times this has wreaked havoc.
I honestly think your tool needs to undergo more development and the responsible thing is to stop its use.--Wuerzele (talk) 22:17, 5 March 2015 (UTC)
(talk page watcher) @Wuerzele: I agree that the references were worse after the use of the tool than before. Since you don't use the tool, you may not be aware that at the very top it states "You are responsible for every edit you make. Please double-check the edit before saving!" While it's great that Zhaofeng Li has created and continues to maintain this tool, no tool will ever be able to account for all the creative ways that editors have written references. I suggest you also contact the person who made the edit, so they can learn from their mistake. It would have been better for the editor to not save their edit and report the issue here. Thanks, and happy editing! GoingBatty (talk) 22:55, 5 March 2015 (UTC)
GoingBatty you appear not have read my original post. Of course I contacted the person. thanks for your "suggestion", man. i' m not dumb. plse look at the reaction after above described/inserted edit and see for yourself. sarcastic edit summary in a reversal. that's from a person that saw or didnt see teh caveat which you stencilled out in yellow. why are you telling ME that? look, if kids that cant swim jump in a pool with some boards labelled use at your own responsibility i go to the kid, ok, but I also go the person that put the boards out there. you may not have seen what I ve seen: the people that use this tool (self-selected for their indolence, we might agree?) appear often not to look at what they created, in my experience. I've checked their edits and edit summaries and I invite you too. a techno solution without quality control isnt worth the electrons I m typing here.
Batty before snippily suggesting and replying again, please check the situation, investigate the good faith efforts even of the one who's critical and think first. Thanks ! And happy editing !! Yeah. --Wuerzele (talk) 23:28, 5 March 2015 (UTC)
@Wuerzele: I checked User talk:Catlemur before making this post, but I failed to check Talk:Unexploded ordnance. I apologize for my mistake. GoingBatty (talk) 00:54, 6 March 2015 (UTC)
(talk page watcher) @Wuerzele: I won't speak for Zhaofeng Li, but when he said "Maybe it's good idea to detect such incorrect usage and skip those references", I interpreted that to mean that he thought it might be a good idea for a future version of the reFill tool to detect and skip such references, not that he thought that you should detect and skip them. --Ahecht (TALK
PAGE) 23:43, 5 March 2015 (UTC)
@Ahecht and Wuerzele: That's what I mean. The tool will be able to detect such incorrect usage (having markup inside the brackets) and skip those references. Thank you for discovering this problem. And to make myself clear, I firmly believe that problems are discovered and fixed through usage. It is under active development and will be this way, and I have no intention to disable it at the moment for this edge case. Zhaofeng Li [talk ♦ contribs] 04:15, 6 March 2015 (UTC)

What does the information after filling in the references mean?

For example, this is a comment the tool makes: Filled in 5 bare reference(s) with reFill (d5c4e5e). What does (d5c4e5e) mean? Is it necessary to include it with the comment? I normally delete the () part because I am not sure what it means.

@Frmorrison: It's the commit ID, which marks the version of the tool being used. You can remove it as you like, but keeping it intact may help debugging. Zhaofeng Li [talk ♦ contribs] 16:30, 6 March 2015 (UTC)

reFill page goes blank when I type in "The Colourist"

When I try to add refs to The Colourist, the page now goes blank. Also, you may wish to start archiving your talk page. --Jax 0677 (talk) 03:18, 6 March 2015 (UTC)

@Jax 0677: The tool works on both the current page and the old revision before Reflinks/reFill was used. It's a pity there isn't a decent logging system for the tool so that errors can be automatically collected. There is the server error log, but it lacks sufficent context to be actually helpful. Zhaofeng Li [talk ♦ contribs] 17:00, 6 March 2015 (UTC)
And about archiving, yeah, the page is getting a bit long. I've set up automatic archiving. Thanks for the note. Zhaofeng Li [talk ♦ contribs] 17:02, 6 March 2015 (UTC)

Fill in fields given URL params

Hi, any way you could make it so that a url like https://tools.wmflabs.org/refill/?page=Google&wiki=es would fill in the corresponding fields on page load? This would then make it worthwhile to add to MoreMenu, where there would be a link on any given mainspace page to reFill with the fields filled in for that page/wiki. Thanks! — MusikAnimal talk 15:56, 6 March 2015 (UTC)

@MusikAnimal: It's already possible. Try https://tools.wmflabs.org/refill/result.php?page=Test&defaults=y&wiki=en Zhaofeng Li [talk ♦ contribs] 16:24, 6 March 2015 (UTC)
This is great, but what if I wanted to link to the form with just the basic fields filled in so that the user could decide on the other options before executing the tool? — MusikAnimal talk 16:29, 6 March 2015 (UTC)
On second thought this is fine. They'll want the default options most of the time, and the purpose of the direct link is to make it quick and easy. Thanks for the awesome tool! — MusikAnimal talk 16:34, 6 March 2015 (UTC)
@MusikAnimal:  Done anyway, as this is a useful feature to have. Zhaofeng Li [talk ♦ contribs] 16:51, 6 March 2015 (UTC)
Also  Done adding reFill to MoreMenu, which is a gadget (the "add Page and User dropdowns" under Appearance). Best — MusikAnimal talk 17:03, 6 March 2015 (UTC)

ReFill incorrectly merged empty references

ReFill incorrectly merged empty references in this edit. It should either ignore them or convert them into closed reference tags. – Jonesey95 (talk) 19:54, 3 March 2015 (UTC)

 Confirmed Will do it later today. Zhaofeng Li [talk ♦ contribs] 04:17, 6 March 2015 (UTC)
Thanks. You're the best. Here's an ASCII barnstar for you: * – Jonesey95 (talk) 17:26, 6 March 2015 (UTC)
@Jonesey95:  Done It now turns named references which are empty into self-closing tags, and skips other ones. Zhaofeng Li [talk ♦ contribs] 00:37, 7 March 2015 (UTC)

Bad title added for New York Times articles

It looks like this script is being blocked by the New York Times, so titles are filled in as "Log In - The New York Times". See here and here for examples.

Is it possible to fix this somehow? Once it is fixed, please post here so that I can post a bot request to remove these errant titles, which currently appear in about 70 articles. Thanks. – Jonesey95 (talk) 19:53, 6 March 2015 (UTC)

@Jonesey95:  Fixed New York Times links are supposed to be blacklisted (see WP:REFILL#Frequently asked questions) and the blacklist isn't working. I've fixed it. Sorry for the problem it's causing. Zhaofeng Li [talk ♦ contribs] 01:05, 7 March 2015 (UTC)

A new name for the tool

Hello, everyone. The tool is being renamed to address a couple of issues:

  • The original tool by Dispenser has been restored on the author's own server. Having two tools with the same name is confusing.
  • The current name (Reflinks) doesn't get the purpose of the tool (expanding bare references) across very well. As a compensation, there is an example shown on the right of the main interface.

Some ideas have been suggested at Village pump earlier, and you can discuss or vote for your favourite name in the corresponding section below. You can also propose new names, of course! Thank you for your interest. Zhaofeng Li [talk... contribs...] 02:06, 8 February 2015 (UTC)

Update The tool has been renamed to reFill. Thank you for your participation! Further comments are welcome. Zhaofeng Li [talk... contribs...] 10:55, 26 February 2015 (UTC)

CiteExpand

This one is coined by GoingBatty.

zLinks, zLinkRefs, or zLinkrefs

Coined by Dsimic

Refill

Short form of reference filling, coined by Chongkian.

Handbanana

Coined by Catlemur.


@Zhaofeng Li: You put a notice of this discussion on the main reflinks page, but it would probably get more input if you also added a note on the test version too... Just an idea. EoRdE6(Come Talk to Me!) 22:27, 15 February 2015 (UTC)

@EoRdE6: 👍 Like Good idea! It's now added. Zhaofeng Li [talk... contribs...] 00:04, 16 February 2015 (UTC)

@Zhaofeng Li: Can I suggest you also update the edit summaries, which still point to User:Zhaofeng Li/Reflinks. EoRdE6(Come Talk to Me!) 03:21, 28 February 2015 (UTC)

@EoRdE6: {{done}) Thanks! Zhaofeng Li [talk... contribs...] 05:08, 28 February 2015 (UTC)

Accessdate

Hi. Thanks so much for maintaining this awesome tool. Would it be possible to add an optional |accessdate parameter? Many thanks, Ostrichyearning (talk) 19:33, 16 March 2015 (UTC)

|access-date= is the date on which the editor who added the reference saw the article and verified that it contained the content that is being cited. Editors who use ReFill are typically modifying references that were added some time ago, so adding |access-date= when ReFill is run would not make sense. If you actually go look at the cited source to verify that it contains the referenced content, you can add |access-date= yourself. Having it as a feature of ReFill would introduce too many inaccurate |access-date= values. – Jonesey95 (talk) 19:57, 16 March 2015 (UTC)
@Ostrichyearning: This is already available, but not enabled by default. To enable it, uncheck the "Do not add access dates" option on toollabs:refill. I'm working on a feature that will let you save your options. Zhaofeng Li [talk ♦ contribs] 01:34, 17 March 2015 (UTC)

Display of Refill in Tools?

Hello Zhaofeng Li, has something changed with the Refill tab in "Tools" in the left sidebar? Currently it displays [appname] for the tool and [label-gadgetoptions] for the options. The tool link still works (aside from the label) but the options link does nothing, when I click on it. In my common.js I use importScript( 'User:Zhaofeng Li/Reflinks.js' );// Backlink: User:Zhaofeng Li/Reflinks.js; which worked OK until yesterday (or 2 days ago, not sure). GermanJoe (talk) 12:03, 20 March 2015 (UTC)

@GermanJoe:  Fixed It was due to a bug in the new version of the Intuition framework that powers the translations. I've patched it. To reload the translations, click on [label-gadgetoptions], wait a while and refresh the page. Sorry for the trouble. Zhaofeng Li [talk ♦ contribs] 13:38, 20 March 2015 (UTC)
No worries, was only a minor issue :). Thank you for the quick fix. GermanJoe (talk) 13:42, 20 March 2015 (UTC)

Incorrect title found

This happened on Oleg Bryzhak. The title of the ref was "В авикатастрофе на юге Франции погиб оперный певец Олег Брияк" but for some reason reFill filled it in as "Âàâèêàòàñòðîôå íà þãå Ôðàíöèè ïîãèá îïåðíûé ïåâåö Îëåã Áðèÿê".  Liam987(talk) 00:33, 25 March 2015 (UTC)

@John123521 and Liam987: Looking into it. Zhaofeng Li [talk ♦ contribs] 02:05, 25 March 2015 (UTC)
@Liam987: Apologies for being slow. The patch is now live on the test version. Could you try it out to see if it works? Thank you. Zhaofeng Li [talk ♦ contribs] 00:51, 30 March 2015 (UTC)
@Zhaofeng Li: It worked with the ref that was the problem before, though I haven't tried it with any others.  Liam987(talk) 01:19, 30 March 2015 (UTC)
@Liam987:  Fixed After some testing, I decided to push the fix to the stable version. Please let me know if anything goes wrong. Thanks! Zhaofeng Li [talk ♦ contribs] 13:48, 30 March 2015 (UTC)

Question

@Zhaofeng Li:, I have a question: why, for example in this article, when trying to get rid of the reds in the reference section by removing the German citation templates, the tool does not pick up certain urls, such as:
1) http://www.urbandiscipline.de or
2) http://kunstverei.m61s11.vlinux.de/wordpress/?p=432
as per the DAIM version? Thank you for your time. Lotje (talk) 11:30, 1 April 2015 (UTC)

Detect citation type

Hi there. I don't know how much logic reFill has to detect the type of citation template it needs to apply. Video-only sites like youtube need to have {{cite av media}}. BTW, your user page still calls it "reflinks". :) Thank you. — Smuckola(talk) 16:09, 12 March 2015 (UTC)

@Smuckola: The stable version currently generates {{cite web}} template only. I've done some work on the test version, and it now guesses the type of the referenced page from the URL according to the domain name (a hard-coded list is used). I will fix my user page later. Tha ks for the note! Zhaofeng Li [talk ♦ contribs] 08:13, 13 March 2015 (UTC)
The feature is now available on the stable version. Zhaofeng Li [talk ♦ contribs] 14:18, 20 March 2015 (UTC)
What about using a separate plain-text file rather than having it hard-coded in the code? There's probably a good list of newspapers out there that can already be used and not have to be hard-coded in. The Haz talk 17:06, 7 April 2015 (UTC)

Can ReFill remember settings for options?

Firstly, thank you for a brilliant tool! Every time I use it, I open the Options and tick "Add blank metadata fields when the information is unavailable". Is there any way I can get it to remember that setting, and have it ticked by default? --Gronk Oz (talk) 07:09, 1 April 2015 (UTC)

@Gronk Oz: Sorry for the late response. This is a planned feature, and should be available soon. Zhaofeng Li [talk ♦ contribs] 22:49, 7 April 2015 (UTC)
@Zhaofeng Li: - that's brilliant, thanks. There is no hurry, it's more that I was curious if there was such a facility already and I just couldn't see it. All your hard work, and your ReFill tool, are appreciated!--Gronk Oz (talk) 23:28, 7 April 2015 (UTC)

access date, author fields

Hi! Thanks for this amazing tool. I use it all the time. Would it be possible to insert blank fields for date/author, like currently exists for publisher? These would be helpful to add manually, when they exist but are not pulled. This is not a huge request, but just an idea. Thanks again! —МандичкаYO 😜 19:15, 8 April 2015 (UTC)

Never mind, I see it's already there under the "Add blank fields" ... Duh. Thanks! —МандичкаYO 😜 05:50, 9 April 2015 (UTC)

Use Zotero backend?

Citoid uses Zotero to scrape metadata from 700+ different websites and create citations for them. It'd be cool if reFILL used the same backend so that it could get better metadata about the pages it is scraping. It seems like this is at least theoretically possible (because we're making the Zotero server available as a public API) but I don't know the details. —Luis (talk) 15:59, 7 April 2015 (UTC)

@LuisVilla: An experimental instance of the tool using Citoid is available at toollabs:refill/citoid (for quite some time, actually). Being a proof-of-concept, it only fills the title at the moment, but it shows the possibility of having a complete Citoid integration in the future. Frankly speaking, however, I think Citoid is inferior to reFill's standalone scraper in a number of ways. Citoid fills the URL as the title when a page isn't accessible (i.e. 404, 403, etc), but there is no way to get the details of the failure (HTTP code returned and such) via the API. And Citoid seems to parse fewer metadata fields than reFill's scraper for many sites aside from academic journals.
The current proof-of-concept Citoid integration is done with CitoidLinkHandler, which is a drop-in replacement of StandaloneLinkHandler, the tool's default scraper. In this way, switching to Citoid is as easy as changing a configuration variable. See [4] and [5].
Zhaofeng Li [talk ♦ contribs] 12:27, 9 April 2015 (UTC)
I was curious about your statement that "Citoid seems to parse fewer metadata fields" (certainly possible, in my limited experience with Citoid), so I tested a bunch of most-cited sites and put together this page. Looks like on the most commonly-cited dozen-or-so sites Refill is better on 5, Citoid is better on 5, and they are about the same on 2. I've filed several bugs against Citoid to improve it based on the experiment, and I figured you might want to see the data too, so here it is. This is obviously still a small sample size, so don't take it as gospel, but seemed worth playing with. —Luis (talk) 02:20, 12 April 2015 (UTC)
I should add that obviously I think it'd be great if we were able to work together on shared backends that did an awesome job on the top sites and also a solid job on the long tail, instead of two backends that each had some shortcomings/strengths. But understand if you want to keep plugging away :) Relatedly, have you filed a bug against Citoid about the API shortcomings you mentioned? —Luis (talk) 03:16, 12 April 2015 (UTC)
@LuisVilla: Thanks for taking time to carry out such a comprehensive test. A few notes:
  • About the blacklisting of webcitation.org, it's because that there is no useful information on WebCite pages (captioning it with "WebCite query result" is worse than keeping the original bare URL IMO). The relevant code reFill uses to enhance Wayback Machine (archive.org) references is available here (also noted at phab:T95389).
  • For |publisher=, I have to say that there's currently no way to reliably parse the information. There is a standard but it's not used on any major site that I know of. reFill cheats a little bit by looking it up in a hard-coded list, which is obviously inadequate. reFill automatically adds a blank |publisher= field for editors to fill out if it fails to get the information from the list.
  • Regarding the strange author information parsed on IMDB, you've found a bug. The tool mistook the author of a user review as the film's.
  • I'm curious as to why Citoid/Zotero is able to parse New York Times without being subject to the "10 free articles per month" restriction.
I believe that Citoid is the way to go, both to maintain consistency across tools and to reduce development costs. I hope we can work together to improve it, but I have fairly limited knowledge on Node.js and related tools. I'll report the API shortcomings later. Thanks. Zhaofeng Li [talk ♦ contribs] 06:21, 12 April 2015 (UTC)
@Zhaofeng Li:
  • Hrm, we should figure out who to talk to at WebCitation - I bet that's something WMF can get fixed - we surely must have contacts there.
  • Urgh, messy. But maybe Citoid should learn from that hardcoded list, at least ...
  • re IMDB, the really interesting thing is that it changed - I saw at least two different answers; maybe new reviews got added while I was testing.
  • I asked yesterday about NYT; apparently we have an API key for that?
@Mvolz (WMF): is probably the person to talk to if you'd like to figure out how to work together; I'm just a messenger :) —Luis (talk) 16:33, 18 April 2015 (UTC)

Lior Schleien#References

Zhaofeng Li, sorry to trouble you, but can you take a look and tell me how to fix this? Thank you for your time. Lotje (talk) 08:17, 17 April 2015 (UTC)

@Lotje:  Fixed Sorry for being so slow. The patch has been merged into the stable version. Zhaofeng Li [talk ♦ contribs] 10:41, 30 April 2015 (UTC)

Mojibake problem

Looking at Nissan Note and it seems like it can encore Japanese correctly.--John123521 (Talk-Contib.) RA 14:29, 22 March 2015 (UTC)

@John123521: Already done Apologies for the late response. This is a known bug, and a previous patch appears to be working for this case. Zhaofeng Li [talk ♦ contribs] 00:49, 30 March 2015 (UTC)
Was this fixed for the non-test version? It filled in mojibake for 3 titles from this Thai news site when using ReFill for Politics of Thailand in this edit for me, which I had to manually correct. Googol30 (talk) 00:07, 2 April 2015 (UTC)
@Googol30: Looks like the news site is using some exotic encoding which PHP's built-in encoding converter does not support. I've made the fix on the test version. Could you give it a try? Zhaofeng Li [talk ♦ contribs] 00:03, 5 April 2015 (UTC)
Merged into the stable version. Zhaofeng Li [talk ♦ contribs] 11:08, 30 April 2015 (UTC)

Watch pages Comment

I like using the reFill tool to fix bare citations, however most of the time I don't want to add the fixed page to my watch list. Can this checkbox be changed to not selected? --Frmorrison (talk) 20:13, 5 May 2015 (UTC)

@Frmorrison: Try enabling the toolbox script. When you launch the tool via the link in the Tools sidebar, the checkbox will be selected only when you are already watching it. Zhaofeng Li [talk ♦ contribs] 00:49, 6 May 2015 (UTC)

Wikiquote

I would like to use this on Wikiquote, but since Wikiquote does not use ref tags, I do not want to remove duplicates. Can we do that? bd2412 T 13:48, 22 April 2015 (UTC)

Converting various characters to "#"

Please take a look at what happened here. I'm not sure whether that's actually reFill-related, but it looks that way since non-reFill edits by the same user to the same page didn't show the same problem. Huon (talk) 14:30, 14 May 2015 (UTC)

@Huon: That's not a refill bug per-se, it's Internet Explorer's Cross-site scripting filter trying to be "helpful". You disable the filter by going to →"Internet Options"→"Security"→Custom level..., scrolling to near the bottom of the list to "Enable XSS Filter", and choosing "Disable". You can also work around the filter by copying everything in refill's "New Wiki Markup" box and manually pasting it into Wikipedia or by using another browser such as Mozilla Firefox or Google Chrome. --Ahecht (TALK
PAGE) 15:09, 14 May 2015 (UTC) (talk page watcher)
@Zhaofeng Li: Is it worth petitioning the powers that be to have Wikipedia serve the X-XSS-Protection: 0 HTTP response header to avoid this problem (see here), or is there a more elegant way to avoid this from reFill's end? --Ahecht (TALK
PAGE) 18:04, 14 May 2015 (UTC)

Request for bs.wikipedia.org

Hey, saw the FAQ and it said to request here. Basically, I generally use the reFill's manual option and for the most part it works fine, but the date format needs to be changed. So, could you add support for bs. and also make the accessdate or date things go in the format of DD. MM. YYYY whereas there are no zeroes in front and no full stop after the year (for example January 9th, 2015 would be "9. 1. 2015")? Thanks a bunch. -- Srđan 💬  16:33, 4 June 2015 (UTC)

Edit summary

Hello, the edit summaries look broken to me, why is there an empty () at the end? https://tools.wmflabs.org/sigma/summary.py?server=enwiki&max=500&ns=&nosect=on&name=3gg5amp1e&search=with+%5B%5B%3Aen%3AWP%3AREFILL%7CreFill%5D%5D It seems to have broken/changed on April 30th because it had some strange set of numbers and letters before that, but I have no idea what those are either. 3gg5amp1e (talk) 14:47, 8 June 2015 (UTC)

brilliant

This is the most BRILLIANT tool! I have always thought these bare web addresses were unsightly. Thank you Zhaofeng for making this helpful utility. Alec Station (talk) 12:39, 22 June 2015 (UTC)

Use on sister sites

Hi, and thanks for this great tool!

Is there any chance that this tool can be expanded to the sister sites (eg. Wiktionary)? Thanks, Mdann52 (talk) 14:52, 7 August 2015 (UTC)

Use of "Wiki"

Can you please change the button from "Preview / Save on Wiki" to "Preview / Save on Wikipedia". Unless this services other wikis – other sites using wiki software – this teaches people the same malaprop we explain not to use. See also Wikipedia:Don't abbreviate "Wikipedia" as "Wiki"! and {{notwiki}}. Thanks--Fuhghettaboutit (talk) 13:45, 16 August 2015 (UTC)

ReFill

@Zhaofeng Li:, there seems to be a problem with this wonderful tool of yours leads me to wikitech.wikimedia.org Thank you for your time. Lotje (talk) 05:57, 18 August 2015 (UTC)

Refill not working fully

Hi there,

Thanks for the fantastic, time-saving tool that you have developed; it truly is incredible. However, I was trying to edit this page with Refill, but it just doesn't seem to detect the bare URLs (of which there are many) in the references. Any ideas why? Thanks, Rambunctious Racoon (talk) 11:21, 18 August 2015 (UTC)

@Rambunctious Racoon: take a look at the history of the Representation of the European Commission in Germany. Hope that answers your question. Lotje (talk) 12:03, 18 August 2015 (UTC)
@Lotje: I see that it was translated from the page in German; does this affect Refill? Otherwise I'm not sure what you meant. Rambunctious Racoon (talk) 13:02, 18 August 2015 (UTC)
@Rambunctious Racoon: to enable refill, follow the steps as I did in for example the Manuela Kasper-Claridge article. After that, refill does what it is good at: being perfect. Lotje (talk) 13:09, 18 August 2015 (UTC)
@Lotje: Oh, I see now! I hadn't realised that this was necessary, thanks very much for taking the time to show me! Rambunctious Racoon (talk) 13:12, 18 August 2015 (UTC)
@Rambunctious Racoon: you are most welcome. Glad I could be of any help. btw, the username you picked is awesome. Lotje (talk) 13:19, 18 August 2015 (UTC)

Processing error (Fetching error)

What is this error actually caused by? Trying to reFill Uday Kotak as a test. --NeilN talk to me 13:02, 2 September 2015 (UTC)

In my experience, usually this means that the urls in question are actually dead links (like this one apparently is). I also don't think that this tool here works well with pdf files. Guy1890 (talk) 04:05, 3 September 2015 (UTC)

NY Times = blacklisted?

Hi Zhaofeng, thanks for your work on reFill--I use it constantly! Very helpful. Had a quick question I thought you might be able to answer. I recently ran it and got errors:

The following reference(s) could not be filled:
  • [http://www.nytimes.com/2012/09/19/nyregion/ad-demeaning-muslims-to-appear-in-new-york-subway.html] Blacklisted
  • [http://www.nytimes.com/2012/07/21/nyregion/mta-violated-rights-of-pro-israel-group-judge-says.html] Blacklisted

I don't see any NY Times pages blacklisted. Any thoughts about this? Thanks again, Cyphoidbomb (talk) 14:24, 3 August 2015 (UTC)

@Cyphoidbomb: The NY times is blacklisted from the tool because it is behind a paywall after a certain number of pageviews. See User:Zhaofeng_Li/reFill#Frequently_asked_questions. --Ahecht (TALK
PAGE) 14:34, 3 August 2015 (UTC) (talk page watcher)
Gaah! Shoulda thought of that. Not just the paywall thing, but the FAQ! Thanks Ahecht as usual. :) Cyphoidbomb (talk) 14:38, 3 August 2015 (UTC)
Any idea how the old reflinks got around that? --NeilN talk to me 12:53, 2 September 2015 (UTC)
(talk page watcher) @NeilN: The original Reflinks doesn't create citation templates for nytimes.com either. You can try it on To Kill a Mockingbird for an example. GoingBatty (talk) 03:06, 4 September 2015 (UTC)

Option to disable blank publisher field

There doesn't seem to be any way to disable the addition of the blank publisher field when that field cannot be filled in for a citation. Could you please add a checkbox for this? Mdrnpndr (talk) 16:37, 14 September 2015 (UTC)

ReFill will not run due to expired security certificate

Starting today, when I try to run ReFill it gives the following error message: 

Your connection is not private
...
This server could not prove that it is tools.wmflabs.org; its security certificate expired 1 day(s) ago. This may be caused by a misconfiguration or an attacker intercepting your connection.

--Gronk Oz (talk) 15:41, 15 September 2015 (UTC)

@Gronk Oz: This is a known toollabs issue: T112608. --Ahecht (TALK
PAGE) 15:53, 15 September 2015 (UTC) (talk page watcher)
Thanks, @Ahecht:. Is there somewhere can I monitor, so I know when it is ready to use again? (I could just keep trying, I suppose, but I hope there is a better way...) --Gronk Oz (talk) 16:01, 15 September 2015 (UTC)
@Gronk Oz: You should be able to bypass that error in your browser and continue using the tool. If not, you can monitor that Phabricator ticket (where they expected it to be done at around midday PDT. -Ahecht (TALK
PAGE) 16:04, 15 September 2015 (UTC)
checkY @Ahecht:, it is fixed now! --Gronk Oz (talk) 16:09, 15 September 2015 (UTC)

Removal of ref text

Hi. In this edit, the tool removed two instances of ref text URL http://www.biography.com/people/marilyn-manson-504532/videos/marilyn-manson-meeting-twiggy-ramirez-17527875547#awesm=~oIwmYBaH5P7jmr from Line 32, leaving named but empty ref "biography.com" and one named referral thereto. I compensated and expanded that ref in the following two edits.  :(   — Jeff G. ツ (talk) 07:41, 4 October 2015 (UTC)

Is there any scope for the tool to also convert basic external links included in references to cite web? I don't know about other editors, but I for one am put off tidying up pages that have a lot of basic links that could be expanded with more metadata, as it's quite labour-intensive to go through long swathes of basic links and convert them. It needn't even attempt to convert those that have other metadata already in the reference tag (such as accessed date or publisher), but could simply convert entries that have the ref tags up against the square brackets.

For instance, 

<ref>[http://www.google.com Google homepage]</ref>

would convert to 

<ref>{{cite web|url=http://www.google.com|title=Google homepage}}</ref>

but 

<ref>[http://www.google.com Google homepage], ''Google''. Accessed 10 October 2015.</ref>

could just be included in the "could not be filled" exception report at the bottom (as currently exists for fetching errors and other issues).

reFill is already one of my favourite tools, but I feel this would make it invaluable. Thank you for your time! — Sasuke Sarutobi (talk) 12:56, 10 October 2015 (UTC)

See this diff. Lots of #'s were added, I've no idea why, and I didn't notice it on the diff either. Strange! --  Kethrus |talk to me  11:19, 31 October 2015 (UTC)

Possibly this. – Jonesey95 (talk) 13:45, 31 October 2015 (UTC)
Hmm, I've been using Edge 12 the past few days (I have the choice of IE or Edge, I chose edge because it's a big improvement). I wasn't using VE either. It may still be a glitch within Edge 12 that hasn't been fixed from IE 11. Strange. --  Kethrus |talk to me  14:35, 31 October 2015 (UTC)
@Kethrus and Jonesey95: It is because of the XSS filter employed on IE/Edge. For more details, see this post on MSDN. I have added a notice for users of those browsers to the tool.81c1b2 I haven't used Edge personally, and doesn't know if there is a way to turn the filter off, like on IE 10. Zhaofeng Li [talk ♦ contribs] 09:04, 16 November 2015 (UTC)

Afrikaans Wikipedia

Can you please add support for the Afrikaans Wikipedia - af.wikipedia.org? Best regards. Naudefj (talk) 07:17, 18 October 2015 (UTC)

@Naudefj:  Done I'm really sorry for being slow. It's now added.6feb3a If you have time, please consider joining the Transifex team to help translating the tool to Afrikaans. Thanks very much! Zhaofeng Li [talk ♦ contribs] 09:18, 16 November 2015 (UTC)

Problem title generated

Loking at this edit there is a problem with the generated title which is "Check Browser Settings" can this be avoided and a proper title be used. Keith D (talk) 23:08, 29 September 2015 (UTC)

@Keith D: Sorry for the late response. It was due to the anti-bot mechanisms employed on the website. I will see what I can do to improve the situation, while playing by the site owner's rules. Zhaofeng Li [talk ♦ contribs] 13:41, 16 November 2015 (UTC)

Ludicrous titles

I see that in this edit, two of the generated title parameters were "title=404 Error – The New School in New York City" and "title=An Error Occurred Setting Your User Cookie". If this is an inevitable consequence of the automated pocess which the tool uses, there surely ought to be a warning to editors using the tool that they need to check the output. Powerful tools are dangerous if not used carefully. - David Biddulph (talk) 17:53, 23 September 2015 (UTC)

(talk page stalker) This warning already exists. At the top of the reFill preview page, it says "You are responsible for every edit you make. Please double-check the edit before saving!" in orange text. – Jonesey95 (talk) 21:00, 23 September 2015 (UTC)
To be fair editors should automatically check what's been done before smacking Save page but that's just my opinion. –Davey2010Talk 21:51, 23 September 2015 (UTC)
@David Biddulph: The edit was made in May. For the record, when I tested the tool against the link just now, it skipped the reference correctly. And I agree that editors should double-check the changes instead saving blindly. Zhaofeng Li [talk ♦ contribs] 14:08, 16 November 2015 (UTC)

Spamming

Some single-use spamming accounts have just started using this tool to add links to commercial websites such as spamblackfridaymattress.org; spamwhatsuppworkout.com to various articles, e.g. [[6]], [[7]], [[8]], [[9]]. . . Mean as custard (talk) 20:32, 18 November 2015 (UTC)

@Mean as custard: Thanks for the heads up. Apparently, those links were added manually after the tool is run, and I believe they should be handled the same way as other spam. Also, I see strong indication of sockpuppetry here. Zhaofeng Li [talk ♦ contribs] 02:27, 19 November 2015 (UTC)

YouTube

Hi there! I'm glad that Refill does format youtube links as {{cite av media}} but unfortunately it says work=YouTube. It should say via=YouTube. YouTube doesn't create, author, or produce virtually any of its content; it just delivers it. YouTube isn't created by the content producer; it's just utilized by them for the delivery of their own content. It's similar to an archive such as Highbeam. Thank you very much, sir! — Smuckola(talk) 07:37, 9 November 2015 (UTC)

@Smuckola: The change4e1e85 is now live on the test version. Could you test it out? Thanks for your suggestion! Zhaofeng Li [talk ♦ contribs] 13:01, 16 November 2015 (UTC)
@Zhaofeng Li: You'd better believe it, sir! I often use the test version, so I'll let you know. — Smuckola(talk) 13:12, 16 November 2015 (UTC)
@Zhaofeng Li: It hasn't been working. It's still doing work=YouTube and it's also doing work=twitter which is the same problem. And worse, it automatically adds work= to everything. A website is not automatically a work, but a delivery method for via= if anything. But still, it can't discern that automatically either, and there's no need for it to try. So now I have to have a sed command to remove all the work=. Thanks. — Smuckola(talk) 08:12, 23 November 2015 (UTC)

Feature request: total reformatting

Hi there. Would it be easy to add an optional feature, activated by a checkbox, where Refill can just forcibly reformat all citations in an article? Even if they're not bare URLs, they could still be malformatted or otherwise obsolete. So Refill could simply strip out all metadata except each URL and start over. It seems to my lowly mind that this would be easy, but I apologize if it is not. And again, please add a checkbox for American date formats. Thank you for your consideration and for all of your efforts. — Smuckola(talk) 17:34, 14 November 2015 (UTC)

This sounds like automation of disruptive editing. Date formats should stay consistent with the existing citations in the article, and there's no need for a checkbox if the tool can do that. LeadSongDog come howl! 19:40, 2 December 2015 (UTC)

Filling in External links, as well as References

I imagine this has probably been asked and answered a hundred times, so please feel free to just point me to the discussion. Would it be feasible to have ReFill populate bare URLS in External links, as well as References? Currently if there are a lot of bare URL links there, I often copy them to my Sandbox, then convert them all to Refs so I can use ReFill to do all the hard work. Then I copy them back, removing the <ref> tags. It would be lovely to bypass all the copying back and forth.--Gronk Oz (talk) 08:41, 23 November 2015 (UTC)

Yeah and it could have additional checkboxes there, like for "format as named reference (<ref name=)" and "format as {{URL}}". — Smuckola(talk) 08:46, 23 November 2015 (UTC)
@Gronk Oz, Smuckola. No. If you look up Wikipedia:Manual of Style#External_links, it will refer you to Wikipedia:External links#External links section which explicitly states "citation templates were not designed for use in the External links section" and "editors who use citation templates in this section should be careful to ensure the resulting description is appropriate for an external link". So while citation templates are not forbidden, they are often contrary to the notion that external links present a "concise description of the contents and a clear indication of its source" because these are "more important than the actual title of the page, and access dates" and other details provided for references (quotes from previous sections). External links are fundamentally different from references. References are meant to be verifiable and reliable. Extra details help in that regard by helping a person locate the source or spot questionable sources at a glance. External links by their nature already give the source and the standard for inclusion is not the same as for sources, so the extra information is often just clutter. Jason Quinn (talk) 23:24, 2 December 2015 (UTC)
@Jason Quinn: I know. That's exactly why I said what I said, which is the solution. — Smuckola(talk) 23:39, 2 December 2015 (UTC)
Jason Quinn, thanks for taking the trouble to explain that. The distinction has always been a bit unclear to me, so now I will follow up on the links you provided and get a better understanding.--Gronk Oz (talk) 01:09, 3 December 2015 (UTC)

Work values

In this edit the values used for the title, work, and publisher could have been much better. Rather than

  • "Chagas drug, fexinidazole, goes into Phase II trials - Outbreak News Today". outbreaknewstoday.com.

it should have been

Certainly, any instance of {{|}} in the title should be a red flag. LeadSongDog come howl! 19:50, 2 December 2015 (UTC)

@LeadSongDog: Right, the algorithm reFill uses to strip the site name from the title is naive. It could use some work. Meanwhile, I've made the tool more conservative by not adding the domain name into |work= by default. Zhaofeng Li [talk ♦ contribs] 03:00, 4 December 2015 (UTC)

citoid?

Hi, thank you for this wonderful tool! I'm curious -- does it use any of the underlying technology from citoid or is it built on its own foundation? -Pete (talk) 02:26, 12 September 2015 (UTC)

Is there any documentation you can point me to of what technology underlies this tool? What code base allows it to infer meaning from various sites' presentation of bibliographic info? -Pete (talk) 21:17, 18 September 2015 (UTC)
@Peteforsyth: Sorry for such a late response. The tool normally uses its own foundation to parse metadata from the webpages, and an experimental version with rudimentary Citoid integration is available for testing. The complete code is available at GitHub, and I will elaborate a little bit on how its own parsers work.
The tool has a chain of metadata parsers, each of which is tasked with parsing a specific kind of metadata (like Schema.org, OpenGraph and Meta tags), or improving the metadata (i.e. doing case-by-case adjustments to fit Wikipedia's needs) retrieved from the parsers before it in the chain. After the parser chain has been successfully executed, a citation generator will generate a cite template or a plain CS1 citation, based on user preference. Zhaofeng Li [talk ♦ contribs] 13:25, 16 November 2015 (UTC)
@Zhaofeng Li: Hi. I would test that, but it looks like the link is broken. Did it move? Thanks. — Smuckola(talk) 23:46, 2 December 2015 (UTC)
@Peteforsyth: He said the citoid link is back up again, FYI. — Smuckola(talk) 12:25, 5 December 2015 (UTC)

scowiki

Can you add support for scowiki? --AmaryllisGardener talk 03:24, 4 December 2015 (UTC)

@AmaryllisGardener: Sorry for the slow response. I've added experimental support on toollabs:refill/test. Once the translation is finished and tested, it can move on to the stable version. Thanks very much for your help! Zhaofeng Li [talk ♦ contribs] 03:25, 7 December 2015 (UTC)
@Zhaofeng Li: How about id.wikipedia.org please? HaEr48 (talk) 01:50, 9 December 2015 (UTC)
@HaEr48:  Done on the test version. Translations may be required to make the tool more useful there, though (localized template parameters, etc.). Thanks! Zhaofeng Li [talk ♦ contribs] 12:40, 9 December 2015 (UTC)

Blank publisher from Bookmarklet

Until recently, I don't believe that the bookmarklet was creating blank publisher parameters. It seems like it is now. How do I flag the tool to "Use the base domain name as work when this information cannot be parsed" when being called from the bookmarklet? —Ost (talk) 19:10, 10 December 2015 (UTC)

@Ost316: It was caused by a change to the default configuration. I've updated the bookmarklet to make things clearer. To revert to the old behavior, add &usedomainaswork=y to the end of options so that becomes:
javascript:options='defaults=y&nowatch=y&usedomainaswork=y';location.href='https://tools.wmflabs.org/refill/result.php?wiki='+wgContentLanguage+'&page='+encodeURIComponent(wgPageName)+'&'+options;
Thanks! Zhaofeng Li [talk ♦ contribs] 08:34, 11 December 2015 (UTC)
Thank you! —Ost (talk) 19:23, 11 December 2015 (UTC)
Retrieved from "https://en.wikipedia.org/w/index.php?title=Wikipedia_talk:ReFill/Archive_1&oldid=1137433890"