Eisspeedway

User talk:Dispenser/Reflinks/Archive 1

Archive 1Archive 2Archive 3

Two suggestions

  1. Check your named reference code. It did not work for me in the Old revision of Eastern Mediterranean University.
  2. Put the citation in the right place (that is, after the punctuation, except for dashes), in cases where the editors have been careless.

--Adoniscik(t, c) 20:50, 7 August 2008 (UTC-4)

  1. Will incorporate the better version from DumZiBot sometime in the future.
  2. Initially, I used regexes from AWB, but found abuse of people using only to make white-space changes. Rewrote it and might be incorporated in AWB's general fixes.
Dispenser 17:11, 21 September 2008 (UTC-4)

Error

It skips anything after a & (example) LegoKontribsTalkM 21:42, 18 September 2008 (UTC-4)

 Fixed That was a rather interesting error in my logic:
- return [s and '-'+s for s in urllib.unquote(os.getenv("QUERY_STRING", '')).replace('=', ':').split("&")]
+ return [s and '-'+urllib.unquote(s) for s in os.getenv("QUERY_STRING", '').replace('=', ':').split("&")]

Dispenser 22:29, 18 September 2008 (UTC-4)

Bug report

Try running reflinks [sic] on Illegal drug trade as of my last edit and seeing what goes wrong with the bare links. --Adoniscik(t, c) 13:28, 17 October 2008 (UTC-4)

 Fixed An unescaped pipe was the culprit on a new common fix. — Dispenser 14:34, 18 October 2008 (UTC-4)

When I attempt to use reflinks, I get redirected to reflinks-svn.py --Adoniscik(t, c) 00:51, 26 October 2008 (UTC-4)

I moved the URL today to webreflinks.py to better fit with the naming scheme of the tools. So links will need to be updated. — Dispenser 02:14, 26 October 2008 (UTC-4)

encoding issue ?

When i try to save changes made in checklinker for an article with utf8 characters, I get an error from mergeChagnes.py

It reads:

Python traceback
Abdullah Öcalan

Traceback (most recent call last):
  File "mergeChanges.py", line 238, in <module>
    main()
  File "mergeChanges.py", line 66, in main
    text = page.get()
  File "/home/dispenser/public_html/cgi-bin/wikipedia.py", line 176, in get
    raise NoPage('Illegal character in %s!' % self.aslink())
NoPage: <unprintable NoPage object>

— Preceding unsigned comment added by TheDJ (talk • contribs) 15:50, 30 October 2008

 Fixed and now works in IE6 too. — Dispenser 11:40, 7 November 2008 (UTC-5)

Python exception and a feature request

Hi, reflinks is a great tool. I came across this exception and it seemed worthwhile to report it:

Python traceback
<type 'exceptions.IndexError'>	Python 2.5.2: /usr/bin/python
Wed Nov 5 21:04:03 2008

A problem occurred in a Python script. Here is the sequence of function calls leading up to the error, in the order they occurred.
 /home/dispenser/public_html/cgi-bin/webreflinks.py in ()
  883         try:
  884                 wikipedia.startContent(form=False)
  885                 main()
  886         finally:
  887                 wikipedia.endContent()
main = <function main at 0xb7a488>
 /home/dispenser/public_html/cgi-bin/webreflinks.py in main()
  878                 generator = pagegenerators.NamespaceFilterPageGenerator(generator, namespaces)
  879         bot = ReferencesRobot(site, generator, always, limit)
  880         bot.run()
  881 
  882 if __name__ == "__main__" and wikipedia.handleUrlAndHeader():
bot = <__main__.ReferencesRobot instance at 0xb89098>, bot.run = <bound method ReferencesRobot.run of <__main__.ReferencesRobot instance at 0xb89098>>
 /home/dispenser/public_html/cgi-bin/webreflinks.py in run(self=<__main__.ReferencesRobot instance at 0xb89098>)
  443                                 continue
  444                         
  445                         new_text = commonfixes.fix(new_text)
  446 
  447                         for match in linksInRef.finditer(new_text):#wikipedia.removeDisabledParts(new_text)):
new_text = u'The Prince A960, also known as the "China Phone,...ded to end of article by script-assisted edit-->\n', global commonfixes = <module 'commonfixes' from '/home/dispenser/public_html/cgi-bin/commonfixes.pyc'>, commonfixes.fix = <function fix at 0xb6c488>
 /home/dispenser/public_html/cgi-bin/commonfixes.py in fix(text=u'The Prince A960, also known as the "China Phone,...ded to end of article by script-assisted edit-->\n', verbose=True)
  405                 elif yearRef and refname+yearRef.group(1):
  406                         refname+=yearRef.group(1)
  407                 elif pageORef and refname+pageORef.group(1):
  408                         refname+=pageORef.group(1)
  409                 elif yearORef and refname+yearORef.group(1):
pageORef = <_sre.SRE_Match object at 0xb866b0>, refname = u'totobay', pageORef.group = <built-in method group of _sre.SRE_Match object at 0xb866b0>

<type 'exceptions.IndexError'>: no such group
      args = ('no such group',)
      message = 'no such group' 

Secondly, I was wondering whether it would be possible to have an option to always make reflinks edits marked as minor? What do you think? Thanks Rjwilmsi 16:11, 5 November 2008 (UTC-5)

Oh yes, and thirdly, have an option not to watch the page. Thanks Rjwilmsi 18:12, 5 November 2008 (UTC-5)
 Fixed bug (forgot capturing parenthesis), added a default minor options, and see #Unwatching. — Dispenser 11:40, 7 November 2008 (UTC-5)
Thanks. Rjwilmsi 14:18, 8 November 2008 (UTC-5)

Another python exception

Traceback
<type 'exceptions.NameError'>	Python 2.5.2: /usr/bin/python
Sat Nov 8 19:15:10 2008

A problem occurred in a Python script. Here is the sequence of function calls leading up to the error, in the order they occurred.
 /home/dispenser/public_html/cgi-bin/webreflinks.py in ()
  883         try:
  884                 wikipedia.startContent(form=False)
  885                 main()
  886         finally:
  887                 wikipedia.endContent()
main = <function main at 0xb8c488>
 /home/dispenser/public_html/cgi-bin/webreflinks.py in main()
  878                 generator = pagegenerators.NamespaceFilterPageGenerator(generator, namespaces)
  879         bot = ReferencesRobot(site, generator, always, limit)
  880         bot.run()
  881 
  882 if __name__ == "__main__" and wikipedia.handleUrlAndHeader():
bot = <__main__.ReferencesRobot instance at 0xb86098>, bot.run = <bound method ReferencesRobot.run of <__main__.ReferencesRobot instance at 0xb86098>>
 /home/dispenser/public_html/cgi-bin/webreflinks.py in run(self=<__main__.ReferencesRobot instance at 0xb86098>)
  443                                 continue
  444                         
  445                         new_text = commonfixes.fix(new_text)
  446 
  447                         for match in linksInRef.finditer(new_text):#wikipedia.removeDisabledParts(new_text)):
new_text = u'{{Infobox Greek Dimos\n| name = Oropos\n| na...opos]]\n[[it:Oropos]]\n[[la:Oropus]]\n[[ro:Oropos]]\n', global commonfixes = <module 'commonfixes' from '/home/dispenser/public_html/cgi-bin/commonfixes.pyc'>, commonfixes.fix = <function fix at 0xb6c488>
 /home/dispenser/public_html/cgi-bin/commonfixes.py in fix(text=u'{{Infobox Greek Dimos\n| name = Oropos\n| na...opos]]\n[[it:Oropos]]\n[[la:Oropus]]\n[[ro:Oropos]]\n', verbose=True)
  204                 # Remove all non approved attributes
  205                 for m in re.findall(r' (\w+)(="[^"\']+"|(?= ))', attr):
  206                         if attrib.group(1).lower() not in htmlattrs:
  207                                 attr = attr.replace(m.group(), '')
  208                 
attrib undefined, global htmlattrs = ('title', 'align', 'lang', 'dir', 'width', 'height', 'bgcolor', 'clear', 'noshade', 'cite', 'size', 'face', 'color', 'type', 'start', 'value', 'compact', 'summary', 'width', 'border', ...)

<type 'exceptions.NameError'>: global name 'attrib' is not defined
      args = ("global name 'attrib' is not defined",)
      message = "global name 'attrib' is not defined" 

Thanks Rjwilmsi 14:19, 8 November 2008 (UTC-5)

Me being lazy and not testing again. Also, don't mark your posts as minor, I hide minor edits on my watchlist. — Dispenser 13:00, 20 November 2008 (UTC-5)

List of deists

This doesn't seem to work on List of deists. --Closedmouth (talk) 02:10, 9 November 2008 (UTC-5)

It's designed to avoid converting lines starting with *#;:= . However, I've removed the condition of <=30 autonumbered link on a page. — Dispenser 10:21, 9 November 2008 (UTC-5)
I didn't even think of that. Is there a way to exempt pages like that one? Like if there's no external links section or something? I dunno. --Closedmouth (talk) 08:21, 10 November 2008 (UTC-5)
I really don't like hacking around too much with the common fixes, the are other times when people put links into bulleted lists. Anyway, it isn't too hard to encodes the links by hand with <ref>? — Dispenser 13:00, 20 November 2008 (UTC-5)
Yeah but I'm really lazy. --Closedmouth (talk) 01:21, 21 November 2008 (UTC-5)

Open up WikEd and use the following regular expression (Remember to check /R/):

Regex:   (.{4,}?)(\s*)\[(\w+://[^\[\]<>\s"]+)\s*\](?!.*</ref>)
Replace: $1<ref>$3</ref>$2

Each run will replace a single instance. — Dispenser 10:29, 21 November 2008 (UTC-5)

No Longer Working Properly?

I had been using Checklinks successfully, but as of today the little boxes to click on to see the details no longer do anything. Has something changed? --Burntnickel (talk) 09:03, 12 November 2008 (UTC-5)

 Fixed syntax error broke it. — Dispenser 12:42, 12 November 2008 (UTC-5)
Thanks! --Burntnickel (talk) 14:16, 12 November 2008 (UTC-5)

Bug report and feature request

I really like how you now allow the user to choose which template to use! It does contain some bugs, though; the fields themselves are not inserted inside the <ref> tag. Surely this should be easy to fix.

Please could you also change the behavior of the URL to toggle the iframe status between open/close? As it is, it keeps on adding new iframes every time you click on the link. It would be more intuitive if it closed the iframe after clicking on the link if one iframe were already open. --Adoniscik(t, c) 15:35, 15 November 2008 (UTC-5)

plus Added The insertion will place where the cursor lays, but Firefox will attempt to put it before |accessdate= or the last }}. I've gone ahead and implemented the iframe toggle. I need still to make a writeup on how to use drag'n'drop. However, there are still serious bugs where it will desync the textarea. — Dispenser 04:06, 16 November 2008 (UTC-5)

Feature request: add title to cite web where URL but no title

I was wondering whether it would be possible to include logic in reflinks to add the title to a reference within a {{cite web}} template (and other citation templates?) where the URL is given but the title is missing or null? Thanks Rjwilmsi 12:36, 17 November 2008 (UTC-5)

I wont implemented until parameter filling is better and I might have to steel some of DOI bot's source code. — Dispenser 13:00, 20 November 2008 (UTC-5)

In Masoumeh_Ebtekar there's a link to [1]. Reflinks found a title for this page but it was a load of accented characters and Unicode. I replaced the title before saving the edit, but wonder if reflinks could catch such errors? Thanks Rjwilmsi 07:41, 18 November 2008 (UTC-5)

Reflinks uses the same statistical based encoding determination system that Firefox uses. However, the problem it runs into is an invalid character (&#128;) encoded into the page. So it fails the UTF-8 test and chardet is run and determines with a confidence of 45.7% that the character encoding is ISO-8859-2. — Dispenser 13:00, 20 November 2008 (UTC-5)

Error message

<class 'sre_constants.error'>, just now, on every link from the BEL link. //roux   02:24, 23 November 2008 (UTC-5)

Me too. ·:· Will Beback ·:· 03:04, 23 November 2008 (UTC-5)
 FixedDispenser 13:40, 23 November 2008 (UTC-5)
Thanks! Have I told you how damn much I love this tool, btw? I think it should be included in the MediaWiki software. //roux   14:35, 23 November 2008 (UTC-5)
You maybe interested on the mailing list discussion about a MediaWiki strategy document. — Dispenser 16:52, 23 November 2008 (UTC-5)

type 'exceptions.TypeError'

Traceback
A problem occurred in a Python script. Here is the sequence of function calls leading up to the error, in the order they occurred.
 /home/dispenser/public_html/cgi-bin/webreflinks.py in ()
 1163         try:
 1164                 wikipedia.startContent(form=False)
 1165                 main()
 1166         finally:
 1167                 wikipedia.endContent()
main = <function main at 0xb89578>
 /home/dispenser/public_html/cgi-bin/webreflinks.py in main()
 1158                 generator = pagegenerators.NamespaceFilterPageGenerator(generator, namespaces)
 1159         bot = ReferencesRobot(site, generator, always, limit)
 1160         bot.run()
 1161 
 1162 if __name__ == "__main__" and wikipedia.handleUrlAndHeader():
bot = <__main__.ReferencesRobot instance at 0xbcc200>, bot.run = <bound method ReferencesRobot.run of <__main__.ReferencesRobot instance at 0xbcc200>>
 /home/dispenser/public_html/cgi-bin/webreflinks.py in run(self=<__main__.ReferencesRobot instance at 0xbcc200>)
  720 
  721                         if commonfixes:
  722                                 new_text = commonfixes.fix(page, text=new_text)
  723 
  724                         for match in linksInRef.finditer(new_text):#wikipedia.removeDisabledParts(new_text)):
new_text = u'[[Image:Music_easel2.jpg|250px|thumb|Buchla Musi...acturing companies]]\n\n[[sv:Buchla Synthesizers]]\n', global commonfixes = <module 'commonfixes' from '/home/dispenser/public_html/cgi-bin/commonfixes.pyc'>, commonfixes.fix = <function fix at 0xb775f0>, page = <wikipedia.Page object at 0xb97e10>, text undefined

<type 'exceptions.TypeError'>: fix() got multiple values for keyword argument 'text'
      args = ("fix() got multiple values for keyword argument 'text'",)
      message = "fix() got multiple values for keyword argument 'text'"

Thanks Rjwilmsi 04:08, 29 November 2008 (UTC-5)

 Fixed, another weekend of development bug :-/. — Dispenser 11:49, 29 November 2008 (UTC-5)

Feature request: redirection detection

Please help users detect and correct redirected links. This could be as simple as adding another line in the metadata section below the iframe to the effect that "This URL redirects to ...", or it could be something more proactive with automatic correction with a yes/no prompt. --Adoniscik(t, c) 14:39, 30 November 2008 (UTC-5)

I take it that you've completely missed the Checklinks tool? Bypassing HTTP redirects is not productive since older URL have more copies in the Wayback Machine and webmasters tend to keep redirects up. — Dispenser 17:35, 30 November 2008 (UTC-5)
You guess wrong (about checklinks, at least). Webmasters do not maintain redirects indefinitely; only for long enough for users to update their bookmarks. --Adoniscik(t, c) 17:44, 30 November 2008 (UTC-5)
That's the way its suppose to work in theory. The reality is web browsers silently go to the new page, bookmarks aren't updated (I suppose Firefox 3 could do it now), and it's additional work to remove redirects from .htaccess with no benefits. When a webmaster decides to break legacy redirects they tend to break it for all versions not just those previous to the last. MediaWiki implements redirects close to the intentions of the HTTP specification, purposely limited due to client interface workings.
Now looking at what each tool does: Checklinks deals with analysis of URLs; Reflinks deals with meta data addition. So I'm marking this as no Declined as it falls outside the scope of the tool. — Dispenser 12:36, 1 December 2008 (UTC-5)

Hi there. when you run the reflinks tool it replaces the parameter publisher with 'agancy' for the cite news template, though the cite news template has it coded correctly as agency. great tool by the way i use it almost every day, Tom B (talk) 15:03, 3 January 2009 (UTC)

 Fixed. — Dispenser 21:08, 3 January 2009 (UTC)

2 requests

  1. Sometimes I'll find that a page has been redirected, or there was a small error in the URL in the article. Would it be possible to run the tool on a link in-place on the tool results page, rather than having to put in the new link, save, and run the tool again?
  2. Instead of having the tool add <references />, could it add {{reflist}} instead? // roux   editor review12:39, 1 December 2008 (UTC-5)
Please don't add {{reflist}}; that's for articles with lots of citations. Now if the bot had to add a references tag, it is fair to assume that there are few citations. --Adoniscik(t, c) 14:02, 1 December 2008 (UTC-5)
Why not? References don't need to be the same size as the rest of the text, and very few pages use anything other than reflist. Using the templated version also means that changes can be implemented sitewide, if needed, with a single edit. // roux   editor review 14:09, 1 December 2008 (UTC-5)
If we had wanted to use reflist universally then we would do that, but we don't. Why use small type where it is not needed? --Adoniscik(t, c) 14:16, 1 December 2008 (UTC-5)
It's my understanding that reflist supersedes references/. Uniformity in pages, as much as we possibly can, is essential to dispensing information. Partly because it trains people to recognise different types of information quickly, and partly because people are generally more trusting of uniformly-presented data; it looks like more care has been taken. // roux   editor review 14:20, 1 December 2008 (UTC-5)
Please can you point me to the policy article that backs your first sentence? --Adoniscik(t, c) 14:22, 1 December 2008 (UTC-5)
I don't remember where, and I could be wrong, which is why I said "is my understanding" and not "is the case, per XYZ." I'm not going to get into ridiculous policy arguments anyway, I'm talking about basic usability and transference of information. // roux   editor review 14:26, 1 December 2008 (UTC-5)
plus Added Well technically there's no consensus of use of citation templates either. So I'm going to implement #2 only when using the citation templates. — Dispenser 18:46, 2 December 2008 (UTC)
Regarding the first point: Checklinks had used webreflinks.py when saving, but people got confused when it didn't automatically go to the diff screen (allowing review of added titles). I've since changed to use the bot version which doesn't need reviewing. You can always run both tools to get the desired effect. — Dispenser 19:37, 1 December 2008 (UTC)
I think I probably didn't explain clearly... let's say I run the tool, and the reference given in the article is example.com/foo. Webreflinks comes back saying the link doesn't work, so I click through and find it's been redirected to example.com/bar. If I want to run the tool on that link, I need to save the new one into the article, and run the tool again. What I'd like to do is type the new link into the review page (before hitting 'show changes), and have the tool parse the link as if I'd submitted it the normal way. Am I making sense? // roux   editor review 19:47, 1 December 2008 (UTC)
That sort of functionality may happen whenever I get around to implementing JSON into webpywikipedia. However, looking at some of your edits you need to be a little more careful about using the values provided by the tool. — Dispenser 18:46, 2 December 2008 (UTC)
Thanks for adding. What values am I using incorrectly? // roux   editor review 18:48, 2 December 2008 (UTC)
I've seen some edits saved with clearly bad info, like dates for authors, publisher/work mixed into the title, etc. Just giving them a glance in the edit boxes to make sure nothing stupendous happened. Of course, for the hard core they can open the article and add the location, add the missing author, choose the right template and verify the source supports the claims. — Dispenser 05:52, 25 February 2009 (UTC)

Doesn't seem to work

I've tried running this program on some pages, and it doesn't seem to work (rather, it seems to do nothing). I just type the name of the page into the window and the bot edits the article, right? Or how dow you use this? The page I want it to work on is this:User:CharlesGillingham/Notes/NY Times on the music business. ---- CharlesGillingham (talk) 02:06, 6 December 2008 (UTC)

There is no bot, you need to save the changes yourself. Hit "show preview" or "show changes" then save using your account. — Dispenser 07:03, 7 December 2008 (UTC)
I'm sorry, pardon me for being so clueless. Can you explain to me step by step how to use this on a particular page? E.g.:
  1. Type the name of the page into the box at Reflinks.
  2. Select "use cite web" or "overwrite" as necessary.
  3. Press the button marked "Fix bare references"
  4. Get the result where?
  5. Edit your page how?
Again, I apologize for being clueless. This tool will eventually be fantastically useful to me, if I can figure out how to use it, and kudos to you for building it. ---- CharlesGillingham (talk) 00:23, 8 December 2008 (UTC)
You may wish to "bot" reflinks (in Pywikipedia) which has less options, but is more automated, though you still need to save the text. I'll or somebody will have to write a tutorial on how to make effective use of the tool at some point in the future. — Dispenser 05:52, 25 February 2009 (UTC)
I'm getting a 404 error on http://toolserver.org/~dispenser/view/Reflinks, that's probably why the tool isn't working for you. Rjwilmsi 08:13, 8 December 2008 (UTC)
The 404 problem was related to the configuration of the webserver as it treated %7E different than the unescaped ~. — Dispenser 05:52, 25 February 2009 (UTC)

exception

<type 'exceptions.UnboundLocalError'>	Python 2.5.2: /usr/bin/python
Wed Jan 7 20:28:42 2009

A problem occurred in a Python script. Here is the sequence of function calls leading up to the error, in the order they occurred.
 /home/dispenser/public_html/cgi-bin/webreflinks.py in ()
 1327         try:
 1328                 wikipedia.startContent(form=False)
 1329                 main()
 1330         finally:
 1331                 wikipedia.endContent()
main = <function main at 0xc59398>
 /home/dispenser/public_html/cgi-bin/webreflinks.py in main()
 1322                 generator = pagegenerators.NamespaceFilterPageGenerator(generator, namespaces)
 1323         bot = ReferencesRobot(site, generator, always, limit)
 1324         bot.run()
 1325 
 1326 if __name__ == "__main__" and wikipedia.handleUrlAndHeader():
bot = <__main__.ReferencesRobot instance at 0xc24248>, bot.run = <bound method ReferencesRobot.run of <__main__.ReferencesRobot instance at 0xc24248>>
 /home/dispenser/public_html/cgi-bin/webreflinks.py in run(self=<__main__.ReferencesRobot instance at 0xc24248>)
 1050                                 print '<dl style="background:#eee; border:1px inset; max-height:40em; overfollow:auto;">'
 1051 
 1052                                 ref.publisherFromDomainAndTitle(domain.search(ref.link).group(2), t)
 1053                                 ## HACK
 1054                                 if ref.publisher in DomainPublisherName:
ref = <__main__.RefLink instance at 0xc618c0>, ref.publisherFromDomainAndTitle = <bound method RefLink.publisherFromDomainAndTitle of <__main__.RefLink instance at 0xc618c0>>, global domain = <_sre.SRE_Pattern object at 0xc460a0>, domain.search = <built-in method search of _sre.SRE_Pattern object at 0xc460a0>, ref.link = u'http://www.hpc.ncep.noaa.gov/research/roth/vaerly20hur.htm', ).group undefined, t undefined

<type 'exceptions.UnboundLocalError'>: local variable 't' referenced before assignment
      args = ("local variable 't' referenced before assignment",)
      message = "local variable 't' referenced before assignment" 

Thanks Rjwilmsi 21:18, 7 January 2009 (UTC)

 Fixed, some time ago. Just noting that it was. — Dispenser 05:52, 25 February 2009 (UTC)

it says unbalanced parenthesis in the code, Tom B (talk) 11:06, 21 January 2009 (UTC)

 Fixed. — Dispenser 05:52, 25 February 2009 (UTC)

errors on Chris Albrecht

Traceback
<class 'sre_constants.error'>	Python 2.5.2: /usr/bin/python
Mon Mar 23 17:06:32 2009

A problem occurred in a Python script. Here is the sequence of function calls leading up to the error, in the order they occurred.
 /home/dispenser/public_html/cgi-bin/webreflinks.py in ()
 1367         try:
 1368                 wikipedia.startContent(form=False)
 1369                 main()
 1370         finally:
 1371                 wikipedia.endContent()
main = <function main at 0xe876e0>
 /home/dispenser/public_html/cgi-bin/webreflinks.py in main()
 1362                 generator = pagegenerators.NamespaceFilterPageGenerator(generator, namespaces)
 1363         bot = ReferencesRobot(site, generator, always, limit)
 1364         bot.run()
 1365 
 1366 if __name__ == "__main__" and wikipedia.handleUrlAndHeader():
bot = <__main__.ReferencesRobot instance at 0xe83c20>, bot.run = <bound method ReferencesRobot.run of <__main__.ReferencesRobot instance at 0xe83c20>>
 /home/dispenser/public_html/cgi-bin/webreflinks.py in run(self=<__main__.ReferencesRobot instance at 0xe83c20>)
 1093                                                 break;
 1094 
 1095                                 ref.setPublisher(domain.search(ref.link).group(2), hints=t)
 1096 
 1097                                 print '<dl class="page_analysis" style="background:#eee; border:1px inset; max-height:20em; overflow:auto;">'
ref = <__main__.RefLink instance at 0xe83488>, ref.setPublisher = <bound method RefLink.setPublisher of <__main__.RefLink instance at 0xe83488>>, global domain = <_sre.SRE_Pattern object at 0xea8960>, domain.search = <built-in method search of _sre.SRE_Pattern object at 0xea8960>, ref.link = u'http://www.variety.com/article/VR1117990473.html?categoryid=18&cs=1', ).group undefined, hints undefined, t = u'[[Variety'
 /home/dispenser/public_html/cgi-bin/webreflinks.py in setPublisher(self=<__main__.RefLink instance at 0xe83488>, domain=u'variety.com', hints=u'[[Variety')
  549                         if partsComparR.sub('', domain.lower()) == partsComparR.sub('', part.lower()):
  550                                 self.publisher = cleanText(part.strip())
  551                 self.fixTitle()
  552 
  553         def setDate(self, list):
self = <__main__.RefLink instance at 0xe83488>, self.fixTitle = <bound method RefLink.fixTitle of <__main__.RefLink instance at 0xe83488>>
 /home/dispenser/public_html/cgi-bin/webreflinks.py in fixTitle(self=<__main__.RefLink instance at 0xe83488>)
  524 
  525         def fixTitle(self):
  526                 self.title = re.sub(ur'(?i)[\s[\]|:»\-(]+%s\W*$'%self.publisher, r'', self.title)
  527                 self.title = re.sub(ur'(?i)^\W*%s[\s[\]|:»\-)]+'%self.publisher, r'', self.title)
  528 
self = <__main__.RefLink instance at 0xe83488>, self.title = u'[[Variety', global re = <module 're' from '/usr/lib/python2.5/re.pyc'>, re.sub = <function sub at 0x7f9b20eb2488>, self.publisher = u'[[Variety'
 /usr/lib/python2.5/re.py in sub(pattern=u'(?i)[\\s[\\]|:\xbb\\-(]+[[Variety\\W*$', repl='', string=u'[[Variety', count=0)
  148     if a callable, it's passed the match object and must return
  149     a replacement string to be used."""
  150     return _compile(pattern, 0).sub(repl, string, count)
  151 
  152 def subn(pattern, repl, string, count=0):
global _compile = <function _compile at 0x7f9b20eba410>, pattern = u'(?i)[\\s[\\]|:\xbb\\-(]+[[Variety\\W*$', ).sub undefined, repl = '', string = u'[[Variety', count = 0
 /usr/lib/python2.5/re.py in _compile(*key=(u'(?i)[\\s[\\]|:\xbb\\-(]+[[Variety\\W*$', 0))
  239         p = sre_compile.compile(pattern, flags)
  240     except error, v:
  241         raise error, v # invalid expression
  242     if len(_cache) >= _MAXCACHE:
  243         _cache.clear()
global error = <class 'sre_constants.error'>, v = error('unexpected end of regular expression',)

<class 'sre_constants.error'>: unexpected end of regular expression
      args = ('unexpected end of regular expression',)
      message = 'unexpected end of regular expression' 

Thanks Rjwilmsi 17:08, 23 March 2009 (UTC)

 Fixed It is now escaped correctly. — Dispenser 03:52, 10 May 2009 (UTC)

Accessdate parameters

I noticed in the edits of some editors that they still use the deprecated date parameters "accessdaymonth" and "accessmonthday". When having a closer look I found that they used Reflinks. Whether these parameters were introduced by this program or by those editors personally, I can not tell. But in any case I would like to request to make sure that you eliminated these parameters from the program.

For your information you might want to have a look at Template:Cite_web#Optional_parameters and Category:Cite web templates using unusual accessdate parameters and see that those parameters are not in use any more and are actively being removed from the system. Sincerely, Debresser (talk) 02:52, 10 May 2009 (UTC)

# Commonfixes.py, 220
    text = re.sub(r'(\|\s*)a[cs]*es*mou*nthday(\s*=\s*)', r'\1accessmonthday\2', text)
    text = re.sub(r'(\|\s*)a[cs]*es*daymou*nth(\s*=\s*)', r'\1accessdaymonth\2', text)
    text = re.sub(r'(\|\s*)accessdate(\s*=\s*[0-3]?[0-9] +(?:Jan|Feb|Mar|Apr|May|Jun|Jul|Aug|Sep|Oct|Nov|Dec)\w*)([^][<>}]*accessyear[\s=]+20\d\d)', r'\1accessdaymonth\2\3', text)
    text = re.sub(r'(\|\s*)accessdate(\s*=\s*(?:Jan|Feb|Mar|Apr|May|Jun|Jul|Aug|Sep|Oct|Nov|Dec)\w* +[0-3]?[0-9])([^][<>}]*accessyear[\s=]+20\d\d)', r'\1accessmonthday\2\3', text)
Red X Won't fix, just correcting accessdate+accessyear and spelling correcting. I'm guess the bot didn't handle this edge case when converting. I've added another fix which will convert accessdaymonth+accessyear into accessdate. — Dispenser 03:35, 10 May 2009 (UTC)
Ok. All's well then. Thanks for the trouble. Debresser (talk) 10:59, 10 May 2009 (UTC)

Two features requests / bugs

Firstly, thanks for this tool - it's an absolute godsend. Couple of things (although I'd like to also chime in for support for {{reflist}} rather than <references/> as in the above threads):

  1. The cleanup code replaces [[file:]] with [[Image:]] - should it be doing that?
  2. When it's only filling in one reference, the summary reads filled in one references plural - should just be "reference" obviously.

Cheers! Chris Cunningham (not at work) - talk 11:32, 14 August 2009 (UTC)

The first one seems to have been coming up a lot lately, so I've commented the code out for now. As for the pluralization it is hard to get this working across multiple languages. There really is no way of specifying singular/plural without creating my own templating language. — Dispenser 05:42, 18 August 2009 (UTC)

Small bug

When it currently rationalises internal links e.g. the [[Commander-in-Chief|Commander-in-chief]] to the[[Commander-in-Chief]], it removes the space just before the rationalisation, as in the example i've just given. pls fix asap, thanks Tom B (talk) 16:23, 15 August 2009 (UTC)

 Fixed I have commented the code since it was also improperly changing the case of links (sometimes being different articles). — Dispenser 05:42, 18 August 2009 (UTC)

Not at all sure this is a bug...

Could be the server(s) is/are just not happy, but I can't run against List of Freemasons, and it has bunch of refs I'd like to clean. :) (Have I mentioned I love this tool? Thanks again!)- sinneed (talk) 00:06, 11 September 2009 (UTC)

At Khalistan movement - instead of getting an overflow, it simply seems to stop. Computers hate me as I have put too many through the recycle crush/shredder :,( .- sinneed (talk) 17:41, 11 September 2009 (UTC)

The script also stops partway through Hip hop music, too. Gary King (talk) 05:40, 9 October 2009 (UTC)

It seems to be working properly now. I left you a long message at http://en.wikipedia.org/w/index.php?title=User_talk:Dispenser/Reflinks&oldid=337193524 with the error message but thought you might not want the full text of the message on your talk page. - Eastmain (talk) 16:31, 11 January 2010 (UTC)

The tracebacks are actually emailed to me ;-). The problem was that the HTTP socket connection to the WMF servers took too long, thus causing it to time out. This happens to be very rare so it doesn't make much sense to program a fancy error message. — Dispenser 16:47, 11 January 2010 (UTC)

work vs publisher

I know it's not 100% clear when to use 'work' vs 'publisher', but could you at least use work for the common newspaper names? I wonder in fact if reflinks should default to work rather than publisher. E.g. this diff. Thanks Rjwilmsi 13:27, 14 January 2010 (UTC)

I was planning on changing it when I got around to reworking the code and pull the information from previously entered publisher (parsed from the database). I'm hesitant to change it without since there is a "signature" to work done by the tool. — Dispenser 18:05, 14 January 2010 (UTC)

107 references

I just used this BOT for 107 references, here are the before and after results. Good job this BOT is great.--intraining Jack In 00:06, 12 March 2010 (UTC)

Doesn't like titles with slashes in them

Hi, I've been trying to use reflinks to fill in the references on R/GA. It does about 5 of them but then stops without properly loading the preview properly. I noticed it does the same when I try to run it on a subpage of my userspace. The rest of the time it is a really useful tool! Thanks for your help. Smartse (talk) 12:08, 4 April 2010 (UTC)

One other little thing, currently links to The Times are marked as dead links even though they are not. I assume it's because their website won't let a bot look at the pages. It would be useful if this could be fixed as it is a widely used source. Btw here's a possible record: 229 links! Smartse (talk) 11:45, 7 April 2010 (UTC)

Quirk report

Great tool! Though I doubt it's a "bug" as such, Reflinks gaffs every time I attempt to run it with article Camp for Climate Action. Perhaps someone else can give this article a whirl, and try to figure out what's preventing it from completing its task. -- WikHead (talk) 19:28, 16 March 2010 (UTC)

Rjwilmsi's looks to have sorted it out. Smartse (talk) 11:46, 7 April 2010 (UTC)

Using it with other wikis than en

Hello,

When using it with other wikis the subject should something like [[en:WP:REFLINKS|Reflinks]], e.g. with the "en:" Great job you are doing, --Christian75 (talk) 20:13, 18 April 2010 (UTC)

IMDB references?

Why does this tool seem to skip references from IMDB? And why has it stopped adding a reference section when one was missing? That was a nice feature. DougHill (talk) 16:54, 4 May 2010 (UTC)

Dates problem

Please note an apparent reflinks problem w/date, described here. Many thanks.--Epeefleche (talk) 07:12, 6 May 2010 (UTC)

Toolbar button?

I used to have a button on my toolbar that I could clikc when on a page, and it would run that article through the Reflinks tool. I had to get a new computer, so I lost that button. Now the instruction page no longer has the button to drag into my toolbar. Can you point me to that again? Or create a button I can put in my toolbar? It was so much more efficient than going to the tool page and manually typing in the name of the article like I have to do now.--2008Olympianchitchat 04:19, 16 May 2010 (UTC)

Its now on the help page, due to the annoyance of changing multiple files. — Dispenser 16:30, 16 May 2010 (UTC)

<sup> tags within <ref> tags

Conversation from my talk page after using the tool on Howard Dean:

Re this edit - don't know how, but you've managed to place new <ref></ref> tags inside existing <ref></ref> tags. Also, what's caused |author=Thursday, January 8, 2004 12:01 A.M. EST? Third, a deliberate external link was reformatted as a ref. I have tried to fix it all up; please check both my fixes and your Reflinks script. --Redrose64 (talk) 21:07, 24 May 2010 (UTC)
The first one is because the original reference used dodgy <sup> tags which the bot automatically converts to <ref> as usually the reference is in the main text not the footnotes. The second one is probably because of the original webpage carrying the wrong coding information. The third one comes because a link that displays just an automatic number is auto-assumed to be a reference from the days before footnotes. I can try passing on your comments to [User talk:Dispenser/Reflinks the bot operator] but some of the problems are rooted in how the page was originally coded. Timrollpickering (talk) 21:25, 24 May 2010 (UTC)

template capitalization

Why does this de-capitalize every {{Cite web}} and others first character? When visiting the template pages this is always capitalized, and visiting the lowercase page it creates a background server side redirect to the capitalized version. Additionally User:Dispenser/Reflinks#Templates shows half capitalized half non, so it doesn't seem to be a consistency matter either. Thanks. Xeworlebi (t•c) 21:02, 3 June 2010 (UTC)

Webreflinks has an error when trying to scan Cello (web browser).Smallman12q (talk) 02:12, 11 June 2010 (UTC)

Fixed, I got the traceback report. Seems that the bit of code was never tested and rarely triggered. — Dispenser 18:12, 11 June 2010 (UTC)

Usually Reflinks works very well. Unfortunately, it errors out when trying to process Night Work (album). Also, some of the authors it suggests for the links on this page are incorrect. GoingBatty (talk) 02:42, 12 July 2010 (UTC)

Known issue with our solaris python installation (core dumps), see TS-385. — Dispenser 15:55, 12 July 2010 (UTC)

Why does this remove the wikilink for publisher? I find it valuable to look up the information for a lesser known publisher. ∴ Therefore cogito·sum 19:00, 12 July 2010 (UTC)

Facebook page title controlled by browser useragent string?

Facebook returned "Niet compatibele browser" to Reflinks in the <title> tag, so Reflinks took that as the article title. Here's the result: (last reference) http://en.wikipedia.org/w/index.php?title=The_Wicker_Tree&diff=prev&oldid=374298058

I wonder if the useragent string can claim Firefox compatibility, while still conveying that Reflinks is a spider? --Lexein (talk) 01:51, 20 July 2010 (UTC)

Deepwaterhorizonresponse.com reports all dates as 2010-05-08

This is probably the best tool I've seen on Wikipedia. Thank you so much! I have run this against several items from http://deepwaterhorizonresponse.com (which is government website for the oil spill) All of them come back as 2010-05-08 regardless of the date (the retrieved date of course is correct). Thanks! Americasroof (talk) 17:39, 9 July 2010 (UTC)

I can confirm having seen this effect. --Lexein (talk) 01:58, 20 July 2010 (UTC)

Error?

In very large pages, or with too many references, can not use "Reflink" and here is very necessary to use this tool. for example in article List of albums released in 2010, "Reflinks" does its job, but can not show preview, so you can not save the changes.
--D6h !? 16:27, 4 August 2010 (UTC)

I've seen it fill in as many as 300 references, but I guess 500 just takes too long (Firefox times out). It's now limited to doing only 200 in a run. Hope that helps. — Dispenser 04:07, 5 August 2010 (UTC)

Another user left clueless.

Could someone please clarify what this is supposed to mean:

WARN: s4 datebase: Server was switched
Applying English Wikipedia fixes
Applying English Wikipedia fixes
Only common fixes for Plant intelligence

Running Reflinks on Plant intelligence, that's the only output I'm getting. pgr94 (talk) 09:39, 6 August 2010 (UTC)

Not sure why it isn't working, but {{cite doi}} and {{cite pmid}} would probably be easier to use than reflinks in that article. Actually, I think the reason it won't work is because the doi and pmid links are redirects, rather than the links to the actual papers, which reflinks can't use. Smartse (talk) 10:19, 6 August 2010 (UTC)
Thanks. I'll try to remember citing that way from now on. pgr94 (talk) 11:30, 6 August 2010 (UTC)

"Fact"

The "fact" template is now "Citation needed". Rich Farmbrough, 04:23, 7 September 2010 (UTC).

Rich, perhaps this problem is part of what I reported in the preceding threat: Reflinks loves redirects over directs. Fleet Command (talk) 13:51, 7 September 2010 (UTC)
No, it's because I haven't updated in a year. Dab solver is much more fun :-). — Dispenser 14:18, 7 September 2010 (UTC)
Okay. I see. Fleet Command (talk) 19:15, 7 September 2010 (UTC)

Tried to run reflinks on above article and I am getting this WARNING : Blacklisted title (hong kong 500 jain - Google Search) any ideas what it means and what I should do? Mo ainm~Talk 19:19, 15 September 2010 (UTC)

Titles which have "Search" and similar in them are blacklisted. Use {{cite book}} with the books.google.com/books?id=XXX URL; to quote a sentences from the source use the quote= parameter. Reflinks does not treat Google Books differently from other sources and will produce generic results. — Dispenser 21:45, 15 September 2010 (UTC)

"I understand that I am excepted to review and correct the machine generated citations before saving" is on tools:~dispenser/view/Reflinks – did you mean "expected" rather than "excepted"? fetch·comms 14:24, 16 September 2010 (UTC)

Yes. I also screw it while programming, it's a muscle memory issue for me. 19:59, 17 September 2010 (UTC)

Hello there, Whenever I try inputting Missing You (The Saturdays song) into WP:REFLINKS this error message (<class 'xml.parsers.expat.ExpatError'>). Any ideas what's gone/going wrong? -- Lil_℧niquℇ №1 | talk2me 23:05, 16 September 2010 (UTC)

The WMF's API server were out yesterday. They should be working now. — Dispenser 21:02, 17 September 2010 (UTC)

There was a discussion at the bot owners' noticeboard and another at ANI (although it has now been archived) about Reflinks and SmackBot undoing each other. In particular every time Reflinks touches a page it uncapitalizes all of the {{cite}} templates. Could you please make it stop doing that, at least on EN? --Selket Talk 01:32, 23 September 2010 (UTC)

There is already a discussion on this issue. See the discussion two threats above. Fleet Command (talk) 05:17, 23 September 2010 (UTC)

PressTv

Reflinks cannot find the names of articles in Presstv.ir, I used it in an article in which presstv links were not properly cited but it did not find the names of the articles in this website, see [2]. Kavas (talk) 10:09, 30 September 2010 (UTC)

Hi.

I find reflinks to be at odds with other bots. For instance: See this revision.

  1. Reflinks converts "Cite web" into "cite web" while SmackBot converts "cite web" into "Cite web".
  2. Reflinks converts direct links into redirects, while there is another bot (forgot its name) responsible for converting redirects to direct links.

And by the way, why one earth Reflinks tampers with HTML Comments?

Fleet Command (talk) 15:45, 26 August 2010 (UTC)

Commonfixes is the library which makes the insignificant changes. To address your points in that diff:
  1. All the documentation is written in the lowercase, this makes it easier to understand and type. SmackBot is wrong capitalizing them.
  2. Please report this bot for violating WP:R2D. It is helpful and alright the other way around, since it is easier to update a few redirects then to crawl looking for broken section links.
Tpbradbury was also using Advisor. Commonfixes protects comments from unintended changes.
Dispenser 14:09, 7 September 2010 (UTC)
Hmmm. I see! Oh, and perhaps the author of that bot (must check my watchlist to remember its name again) did read this message because I didn't see any case of resolving redirects in article namespace since I started this threat. I'll just check and make sure... Fleet Command (talk) 19:23, 7 September 2010 (UTC)
I think I may have an idea who it was, but consider that operation has practically vanished I don't think it too big of a problem. He was using exception meant for spelling mistakes. — Dispenser 20:56, 7 September 2010 (UTC)
The canonical template name is capitalised, sure it's easier to type lower-case and doesn't matter if people do but to change it to lower-case seems counter productive. I see you are working on commonfixes.py - just apply the shift key judiciously! ;-) Rich Farmbrough, 05:12, 18 September 2010 (UTC).
Agreed. Fleet Command (talk) 12:26, 19 September 2010 (UTC)
FYI, you should see here and here before programming the tool to change the capitalization of templates in either direction. –xenotalk 13:37, 29 September 2010 (UTC)
This is much fuss over nothing important. Both pieces of application must cease changing capitalization. But as I said, I believe when the template is being inserted or completely overwritten, it's better to be inserted in a capitalized fashion. And I am not going to read a topic titled "Please block Rich Farmbrough"; Rich Farmbrough is quite a dear of an administrator in Wikipedia and I have zero tolerance for the words "block" and "Rich Farmbrough" ever coupling together. Point me to a topic from which the question of "Who is to blame?" is completely dropped and the question of "What to do?" is regarded. Fleet Command (talk) 14:44, 29 September 2010 (UTC)
When administrators become recalcitrant and refuse to cease making edits that lack consensus they can and should be blocked just like any other user. Really, the only thing to do is to stop changing the capitalization of templates. I agree if it is a net-new addition of a cite template, one could use capitalization if they desired (though the documentation suggests lowercase). –xenotalk 14:51, 29 September 2010 (UTC)
This Cite web should be fixed to have only one style, now its confusing people. One bot chancing to capital and vice versa --Typ932 T·C 20:16, 30 September 2010 (UTC)

Date Retrieved Now Turned Off?

I just used this and I like the capability that it can now directly save the article. However almost all the functionality is gone (e.g., retrieved date, date of article, source, etc.). Thanks.Americasroof (talk) 14:04, 5 October 2010 (UTC)

Two versions of the script are accessible from tabs above the form: the original pywiki "non-interactive" (or bot) version and a heavily modified "interactive" (or web) version. The interactive version requires human correction of the values since date and author detection often don't work well. However, most users have been blindly saving garbage output. So, I setup sensible defaults for these users to mitigate the harm. — Dispenser 16:33, 5 October 2010 (UTC)
Thanks for the quick response. This is my favorite Wiki tool. I hate citation coding. The retrieved date and website seems like they should be pretty straight forward for the "no brainer" one stop approach though. Again thanks for making my life MUCH easier.Americasroof (talk) 17:10, 5 October 2010 (UTC)

Error

I have tried to use this tool a couple of times today and I am getting the following error

  • Request host: toolserver.org
  • Request path: GET /~dispenser/cgi-bin/reflinks.py

any ideas what is wrong? Mo ainm~Talk 09:58, 11 October 2010 (UTC)

I'm unable to reproduce with Firefox 3.6 and there have been no reports about any issues with the web servers. My guesses as to what's wrong: your ISP mucking your packets, your proxy server is broken, your browser is broken, an extension/addon is going haywire, or you accidentally turned on debugging mode. Next time include the page(s) you tried to run the tool or something that uniquely identifies your requests. — Dispenser 14:08, 11 October 2010 (UTC)
I tried to run reflinks again just now on Thomas Ross Holmes and I have tried it with the latest version of IE, Opera and Firefox and I am getting the exact same error here it is in full,

Internal Server Error (500)

The server encountered an internal error and is unable to complete your request at this time. If the problem persists, please contact the owner of the tool you are trying to use and inform them of this error, quoting the following information:

  • Request host: toolserver.org
  • Request path: GET /~dispenser/cgi-bin/reflinks.py

So do you reckon it is a fault at my end? If so have you any suggestions what I could do to fix it? Thanks Mo ainm~Talk 19:33, 11 October 2010 (UTC)

I have just tried to use if on 2 other article and the tool has worked with no problems I then went back to the article I listed above and I got the error again. Could it just be the article for some reason? Mo ainm~Talk 19:42, 11 October 2010 (UTC)
Same as #Error on Night Work (album) above, there's a hack in webreflinks (interactive version) which skips secure URLs. — Dispenser 19:50, 11 October 2010 (UTC)

Dates formatting

Per MoS dates, the format DD Month YYYY is preferred to the ISO YYYY-MM-DD because it could be confusing to non-experts. Could the script please, please, make it optional to format dates as DD Month YYYY. Every time is use it I have to reformat all the dates, and that is just a wast of time if a click of a button could do it. Sandman888 (talk) Latest PR 07:46, 23 August 2010 (UTC)

Some articles are tagged with {{dmy}} and {{mdy}}. Reflinks may heed these tags. Fleet Command (talk) 15:46, 26 August 2010 (UTC)


Moved from my talk page

Hi Dispenser, thanks for your invaluable tool. I was wondering whether you could make the default date format "DD Month YYYY" as that is vaguely preferred by the Manual of style, see WP:DATESNO?

Also, I don't know if you proposed using the {{DMY}} template, but for me at least, it didn't work. Cheers, Sandman888 (talk) 09:06, 9 September 2010 (UTC)

This is the absolute lowest on the priority list since: i) a date reformatting script is available, ii) the date detection code is pretty bad, iii) the developers should address bugzilla:17905, and iv) the format is unambiguous and multilingual. — Dispenser 18:44, 9 September 2010 (UTC)
Okay, where's the "date reformatting script" you mention above? (in re {{Use mdy dates|date = October 2010}} or {{mdy}} ). --Lexein (talk) 13:58, 12 October 2010 (UTC)
It was Lightmouse's date delinking script, though I'm not sure how well it works now that we've delinked everthing. — Dispenser 21:02, 14 October 2010 (UTC)
Indeed, it didn't work at all. So, "i) a date reformatting script is available" should be stricken, sadly. (plays taps). --Lexein (talk) 00:56, 16 October 2010 (UTC)
If you have added {{dmy}} or {{mdy}} to the article(s) involved I can trivially configure AWB to reformat the dates in citation templates. Just give me a list of articles if you want it done. Rjwilmsi 08:09, 17 October 2010 (UTC)

Cite web or cite web

Hi, could you adjust the reflinks that it would use the Capital version of Cite web which seems to be the right style? Now there is some bots and reflinks doing it different way, also the template Cite web is made with capital C rgds --Typ932 T·C 17:02, 30 September 2010 (UTC)

Seems this isnt so easy, as there isnt clear rule which one is correct --Typ932 T·C 20:11, 30 September 2010 (UTC)
  • Of course it's clear. All you need to do is look at capitalisation style of the name the template itself. End of story. Reflinks should actually not be playing push-pull in this case. It wouldn't be a problem if only newly-inserted citations are in lowercase, it should stop at changing every single instance already in the article. --Ohconfucius ¡digame! 04:12, 21 October 2010 (UTC)

I just noticed that in this edit done with Reflinks, the tool substituted the correct wikilink paths with simpler, redirect pages... Is that done on purpose? I always thought that it was better to use the real #section path instead of redirects in wikilinks... --WikiKiwi (askme) 15:28, 21 November 2010 (UTC)

When an editor renames a section all the links to that section break. I created the rdcheck tool ("Show redirects only" in WhatLinksHere) to assist editors in identifying/fixing broken section links. The code could be extended to check all inbound pages, but it becomes impractical in both bytes and requests with content pages. The operator of User:WildBot which did outbound checking had trouble keeping the bot running at a reasonable pace.
Finding appropriate redirects helps prevent section link rot and makes the wikitext more readable (by unpiping links). I hope to incorporate better code from my dabfix project. Unrelated note: I thought Evangelion names transliterated names from the bible, thus needing disambiguation. — Dispenser 01:27, 23 November 2010 (UTC)

ref=harv

I noticed this recent edit removed |ref=harv from some citation templates. This breaks the wikilink from {{harv}} to {{cite book}}. Is this something that "Reflinks" does? In that case, we need to fix it right away. Thanks. ---- CharlesGillingham (talk) 11:55, 22 November 2010 (UTC)

Testing against commonfixes with no changes. It's likely the editor was using a bad script when converting the ISBNs or was manually removing them. — Dispenser 02:12, 23 November 2010 (UTC)

Refbot

To this article: Alejandro Hamed. Thanks.Lihaas (talk) 19:25, 18 November 2010 (UTC)

You can run the tool by yourself. Open Reflinks page, paste the address (URL) into the textbox, and click run reflinks. — Dispenser 02:16, 23 November 2010 (UTC)

Hello Dispenser, I noticed a minor change in Reflinks. Prior to today Reflinks displayed a status as it converted references in interactive mode. I used it today and the page just continued to load until all of the references were converted. There is no longer a status screen. Also, Reflinks used to convert uncited external links in the body of the article. I found both of these things extremely useful. Did you intentionally remove both of these things, or is this a bug? Thank you for your time. --Alpha Quadrant talk 00:17, 9 November 2010 (UTC)

The tool hasn't been changed since Oct 15. There may have been an issue with an underlying library or JavaScript. I know that Internet Explorer has issues with progressive loading, did you switch browsers by chance? — Dispenser 02:25, 23 November 2010 (UTC)

please check, but in my recent edits using reflinks, it was trying to add

== References ==
<references />

after every completed ref. Materialscientist (talk) 05:15, 23 November 2010 (UTC)

Hi there, I m using reflinks quite along but never came along problem like this, its adding reference tag more than once to article. (plz leave a talkback) KuwarOnline Talk 05:40, 23 November 2010 (UTC)

Fixed. The fun of removing mislabeled cruft. "Special:ParserTests" must have seemed like a good name for isolated inputs (i.e. snippets vs. whole pages) at the time. — Dispenser 06:05, 23 November 2010 (UTC)

I saw this problem while running reflinks on [3]. Wtmitchell (talk) (earlier Boracay Bill) 00:08, 4 December 2010 (UTC)

Skipping issue

When I try to use Reflinks on Holy Terror, Batman!, I get the following message:

Skipping https://twitter.com/FrankMillerInk/status/15220154410
See Issue 7305: urllib2.urlopen() segfault using SSL on Solaris - Python tracker.

I followed the link, but don't understand what this means. Could you please explain what the problem is? Thanks! GoingBatty (talk) 17:57, 7 December 2010 (UTC)

When python tried openning https/secure links, it would unrecoverably crash (core dump). This was an issue in Solaris port as I explained to you above. However, the problem has since been fixed and I've re-enabled https links and removed the message. — Dispenser 01:57, 19 December 2010 (UTC)
Forgot that I asked this question before - maybe because I didn't understand your answer. Thanks for re-enabling the functionality - it works great! GoingBatty (talk) 03:18, 19 December 2010 (UTC)

Running Reflinks on this article revision, I see it renaming a named Ref from "Harrison1" to "autogenerated1", thereby unhooking it from another Ref named "Harrison1" and making it into an orphan. Wtmitchell (talk) (earlier Boracay Bill) 04:53, 19 December 2010 (UTC)

On looking again, I see that article revision presented Reflinks with something unreasonable to deal with. There were three Refs named "Harrison1", all three had non-empty bodies, and the bodies differed from one another. Reflinks compounded the problem by renaming two of the three as "autogenerated1" and leaving the third as "Harrison1". The problem (multiple non-empty named Refs with differing bodies) should be detectable by reflinks -- perhaps it would be better to flag or the problem somehow, to report it if in interactive mode, or just to bail without compounding the problem. Wtmitchell (talk) (earlier Boracay Bill) 05:06, 19 December 2010 (UTC)

First, thanks, reflinks is my favorite tool. Second, please do consider these changes (i) when the reference list was absent in the article, reflinks adds </references> which is nearly abandoned. I would use much more common {{reflist}}. (ii) I find the default setting "add this article to your watch list" unhelpful: janitors like me do not want to watch every article they clean up. Article authors already have it in their watch list. Thanks. Materialscientist (talk) 00:56, 20 December 2010 (UTC)

To change the reference section behavior, please change the noreferences.py in pywikibot/pywikipedia then come back to bug me about updating it. However, we are standardized on the smaller font size, so it might be pointless. As for the second point uncheck "Add pages I edit to my watchlist" in Preferences. — Dispenser 04:59, 23 December 2010 (UTC)
Thanks. Another stupid comment. Reflinks is poor in handling Google books (don't remember about NYTimes), but this tool is better. It would be great to improve this part in reflinks. Materialscientist (talk) 05:04, 23 December 2010 (UTC)

Bug report - spurious insertions of text

Please see this spurious insertion of html comment and this insertion in 'author' field. Thanks. --Ohconfucius ¡digame! 06:51, 20 January 2011 (UTC)

Page incomprehensible

The page with which this talk page is associated is incomprehensible. There is no explanation of what the page is supposed to be, instructions for use, technical reference, or what. There is no indication of how the editor invokes the program. Jc3s5h (talk) 15:44, 12 February 2011 (UTC)

I was never a good writer, but I figured having something available was better than nothing. I have removed most of it leaving only expand templates. — Dispenser 22:47, 12 February 2011 (UTC)
Your writing is at least as good as mine (which, admittedly, isn't saying much). You do good work. Wtmitchell (talk) (earlier Boracay Bill) 21:39, 13 February 2011 (UTC)
I've written some information which is hopefully of some use to n00bs. It would be helpful if someone who hasn't ever used it could take a look and say if anything else needs explaining. I'm also not sure about why the dead links + blacklisted links appear when they shouldn't, so if Dispenser can explain, I'll add some answers! SmartSE (talk) 13:40, 17 February 2011 (UTC)
OK, Smartse's change is a big improvement. However, I tried the option of copying the code (actually, I used
importScript('User:Dispenser/Reflinksjavascript');
with the Vector skin. Now I don't know what to look for in the edit screen to start using the tool (or is it even invoked from the edit screen)? Jc3s5h (talk) 16:58, 17 February 2011 (UTC)
You need to click on the link to get the code - you only copied the location of where I copied the code to earlier. You need to use:
function refcheck()
{
addPortletLink("p-tb", "http://toolserver.org/~dispenser/cgi-bin/webreflinks.py?page=" + wgPageName + "&citeweb=on", "Check Refs");
}
I don't know if the line formatting matters or not.... but you can see how it is in mine: User:Smartse/vector.js and copy that. I think you'll need to WP:PURGE your cache and then a new "Check Refs" button will appear in the toolbox when ever you look at an article (before you click edit). SmartSE (talk) 17:43, 17 February 2011 (UTC)
Thanks. I found I also needed to add the statement about wgNameSpaceNumber. Jc3s5h (talk) 18:52, 17 February 2011 (UTC)

Blacklisted reports

Hello Dispenser. I frequently use Reflinks, and frequently see it reporting blacklisted refs (such as this one), that I usually remove in an effort to keep my edit as clean and trouble free as possible. It's simply a case where I trust that Reflinks wouldn't lie about such a thing. In your opinion, how reliable is the Reflinks blacklisted reports... and what exactly should be done with references that Reflinks reports as such?  -- WikHead (talk) 08:50, 17 February 2011 (UTC)

This definitely doesn't mean that it can't be used as a reference - it's not the same as the spam blacklist - for example I had it appear when trying to fill in a reference to http://www.telegraph.co.uk which should work. It would be useful if Dispenser could explain what it does mean so that users know what to do when it appears. SmartSE (talk) 13:37, 17 February 2011 (UTC)
Thank you kindly for your reply. This is the third time (at least) in the past two weeks, that Reflinks has confused me by reporting government sites (URLs containing .gov) as blacklisted references. Even though this has seemed rather unusual at the time, I've trusted (in good faith) that the tool would not report such a thing in error, and have responded WP:BOLDLY with removal as a result. I'm relieved to know that I can now ignore, or at least take future blacklist warnings with a grain of salt, but it does raise other concerns (as you've mentioned above) about the way these warnings are defined, and how they may be interpreted by other users who have not read this discussion thread. At the very least, the word "blacklist" should perhaps be replaced with a less ambiguous term.  -- WikHead (talk) 21:23, 17 February 2011 (UTC)

I was having some trouble getting it to run on Dragon Quest with 2 references and after some time took the sections out and did a standalone test-run at User:Jinnai/temp. Reflinks does not seem to find any issues, ie it does not recognize the refs are bare urls. I've tried removing {{reflist}}, removing the Cite error: There are <ref> tags on this page without content in them (see the help page). brackets, etc. It continually says it says its fine. In addition, Dragon Quest gives some error messages I don't understand.Jinnai 18:38, 1 March 2011 (UTC)


Hello, I recently noticed that the external link to her AllMusic guide page had disappeared from Joni Mitchell's article. I asked Moxy why he had removed it [4], and he said that the Reflinks bot that removed it automatically. I don't think Reflinks should do this. The AllMusic guide external links are useful to have on musicians pages, just like IMDB external links on actors/actresses pages. Could you fix Reflinks so that it doesn't remove external links? FurrySings (talk) 03:35, 5 March 2011 (UTC)

Sometimes people unfamilar with the Wikisyntax add external links, and then something like this happens:

<ref>[url=http://english.aljazeera.net/video/africa/2011/03/2011331522685587.html Gaddafi's military capabilities], Al Jazeera, 3 Mar 2011</ref>,

or this:

{{cite web|url=http://www.reuters.com/article/2011/04/08/cb-syria-protests-statue-idUSLDE7371LA20110408|publisher=Reuters}}.

(For the record, I fixed both manually)

Could you please add a some function to Reflinks so that these broken links are repaired? Particularly in the latter case, Reflinks should detect these cases, and add the website's title with |title= parameter. --bender235 (talk) 14:16, 9 April 2011 (UTC)

See wrong after vs. before. — Preceding unsigned comment added by Justin Ormont (talk • contribs) 19:16, 21 April 2011 (UTC)

Apparent incompatibility with recent change to Template:Reflist

See this diff. Note the Ref defined within {{reflist}}. Wtmitchell (talk) (earlier Boracay Bill) 20:10, 25 April 2011 (UTC)

Box Office mojo blacklisted?

Specifically, this link which is a valid source for gross revenue on the film in question. I think this isn't the first time I've seen a blacklist report for a Box Office Mojo link so I thought I'd let you know. THe film project uses these links all the time. Is there some specific thing about the link that caused the flag? Millahnna (talk) 00:06, 1 May 2011 (UTC)

Can't get Check Refs to show in toolbox

Anyone know what I'm doing wrong? Have tried repeatedly to add the line here to my monobook.js but no matter what I do, "Check Refs" never appears in the toolbox. (FWIW, "Check Links" has exactly the same problem. I also tried it with the Vector skin (vector.js) with no luck. I have cleared the cache on both browsers I tried this in (Firefox 3.6.16 and latest Chrome) and purged the Wikipedia cache for the page I wanted to use it on as well. I have logged out of Wikipedia and logged back in. I don't have anything blocking scripts and Ad-Block Plus is disabled completely on Wikipedia. I'm out of ideas, and the bookmarklet isn't a good solution because it doesn't follow me around regardless of what computer I'm using. Thanks! B.Rossow · talk 16:55, 10 May 2011 (UTC)

I would much prefer if someone made a page describing how to create their JS toolbox links rather then pasting outdated example code for every tool and function. Anyway, while the code to create the links was defined as a function, it was never called (e.g.: addOnloadHook(refcheck)). I updated and move the code to the documentation page. Also, comment syntax change between languages: HTML/wikitext is <!-- comment -->, JavaScript is /* comment */ or // [until line end], and CSS only is /* comment */. — Dispenser 00:07, 11 May 2011 (UTC)
Thank you! B.Rossow · talk 15:42, 11 May 2011 (UTC)

Instructions?

Where are instructions on how to use reflinks? - Peregrine Fisher (talk) (contribs) 04:09, 1 August 2009 (UTC)

What kind of instruction are looking for? Documentation about what each button does is easiest for me. Should it be a tutorial, background explanation, examples of what and what not to do? I have to admit I'm pretty bad at writing any of this stuff, so it would be welcomed if somebody else could. — Dispenser 05:42, 18 August 2009 (UTC)
I could help, but I'm not sure how to use it. I'm probably missing something fundamental. I tried it recently. I decided to go with what I saw, so I went to http://toolserver.org/~dispenser/view/Reflinks and typed in an article name. I couldn't tell if it was editing the page, or if I was supposed to give it my username, or what, so I just copy and pasted what I saw onto the page and saved it. It worked pretty well, but I get the feeling reflinks is a gadget or a javascript or something that makes it even easier. - Peregrine Fisher (talk) (contribs) 07:30, 18 August 2009 (UTC)
Do you have noscript installed or javascript disabled? Protonk (talk) 08:29, 18 August 2009 (UTC)
I can import javascript into my monobook.js page. I have the "Check external links" link in my toolbox that goes with User:Dispenser/Checklinks. Is there something similar for reflinks? - Peregrine Fisher (talk) (contribs) 17:23, 18 August 2009 (UTC)
I don't know. All I know is that when I use reflinks, it brings me to a page w/ changes preloaded and all I have to do is preview and save them. However, I recall having problems with it because it fed info back/forth between the toolserver and wikipedia.org. When I had NoScript installed I had to whitelist the XSS request. Protonk (talk) 17:33, 18 August 2009 (UTC)
OK. It's probably my browser (Chrome). The save button was grayed out, but no big deal. - Peregrine Fisher (talk) (contribs) 19:05, 18 August 2009 (UTC)

It took me a while but I figured it out. The UI for this tool is not so well designed.

  1. Log in.
    1. Select the Log in link in the top right
    2. Follow the directions
      1. select "Get Credentials"
      2. copy the text starting at "query..."
      3. paste into the text box with the directions(!)
      4. select "Use Credentials"
    3. You should be logged in and now the Save button is enabled, right?
  2. The article title should be in the unmarked textbox at the top of the page to the left of the button that says "Fix bare references".
    1. If it's not then enter it or any article title
    2. Select the "Fix bare references" button.
    3. It will grind away for a bit and then list all the proposed fixes for the article.
  3. Edit any entries that need help or clean up.
  4. Select Save at the bottom of the page.

Jojalozzo 02:46, 26 May 2011 (UTC)

Reflist stipped bare

Reflinks appears to strip {{Reflist}} clean of any column or column-width parameters. {{Reflist|colwidth=30em}} and {{Reflist|2}} become simply {{Reflist}}. Could you perhaps fix the code so that it leaves these alone? Thanks. --Ohconfucius ¡digame! 03:41, 19 May 2011 (UTC)

Just to let you know I appreciate the new user script for Reflinks. --Ohconfucius ¡digame! 05:26, 10 June 2011 (UTC)

How many in one go?

What is the largest number of urls you have ever filled in with Reflinks?

I to had to limit it 200 references at time. Later I introduced a more realistic default limit of 20 so people could take time working on references without fatigued. Anyway, there were 8 edits by 3 people with bigger numbers in the past 30 days. The highest was 142 references which took nine minutes from start to save (the log claims 14.7 minutes for generation). — Dispenser 16:37, 8 July 2011 (UTC)

NoScript problem

I just had my first experience with Reflinks, and it is amazing! I want to make it a regular part of my toolset, but I have a problem. I can't get it to work in my browser of choice, Firefox, apparently due to the NoScript I have installed. In fact, if I hadn't thought to try IE, I wouldn't have learned how great Reflinks is.

Now I know what everyone is going to say is that I need to "whitelist" reflinks. Yeah, I understand that in principle, but I can't figure out how to do that. With Java issues, I just hit that blue "S" in the white circle and tell it to trust the site. But with this problem, I get a bar across that top that reads NoScript filtered a potential cross-site scripting (XSS) attempt from http://toolserver.org. Technical details have been logged to the console.

So I go to the options (no blue S to click on here), and I see this array of tabs that is absolutely beyond my non-techie ability; it's the complete opposite of the simplicity that I have come to expect from NoScript. I really wish I could paste a screen capture here, because the thing is really intimidating. Anyway, I went to the Whitelist tab and added toolserver.org. No dice, even after rebooting. I go to the XSS tab that exists under the Advanced tab, and add ^http://toolserver.org/ to the list there. No change. I can't get it to work, and it's really frustrating me. Can anyone help? HuskyHuskie (talk) 07:08, 10 July 2011 (UTC)

See my help page for instructions. Also, I've created Template:XSS-editnotice which displays on a data-less POSTs (which is the XSS filter kicking in or session restore not keeping the data). — Dispenser 14:21, 10 July 2011 (UTC)
Thanks for the quick reply. I've tried pasting in those lines, but apparently to no effect. Question: After pasting in those lines, do I hit "OK" (which is what I've been doing), "Reload", "Import", or do something else? (It is showing those lines you gave me as saved in there when I go back the next time, so I thought "OK" was the right button to hit. HuskyHuskie (talk) 15:20, 10 July 2011 (UTC)
I've removed the leading spaces that screwed up the matches in the lines. It should work now (tested this time :-). — Dispenser 01:34, 11 July 2011 (UTC)
Works good. Thanks! HuskyHuskie (talk) 02:18, 11 July 2011 (UTC)

IGN

When running Reflinks on references that use IGN.com, Reflinks always returns extra info in the author field which then needs to be manually removed (diff). Is this a problem with Reflinks or something I need to contact the IGN webmaster about. - X201 (talk) 08:25, 8 April 2011 (UTC)

Did you manage to find out anything about this? - X201 (talk) 08:13, 27 July 2011 (UTC)

Hello my friend - was wondering if its possible to get Reflinks to some how make goggle book citations function like at Wikipedia citation tool for Google Books --just a though,,,,Moxy (talk) 00:40, 26 May 2011 (UTC)

Second. --Piotr Konieczny aka Prokonsul Piotrus| talk to me 00:27, 12 September 2011 (UTC)

Why only bare urls?

This tool would be much more useful if it would at least try to cite web/news all links, not only the bare ones. It is annoying that I often have to first strip data, save the page with bare urls, and run the tool to see if it will work. --Piotr Konieczny aka Prokonsul Piotrus| talk to me 00:27, 12 September 2011 (UTC)

Bare URLs template

A suggestion: Reflinks should automatically remove {{Cleanup-link rot}} (and redirects) once all bare URLs are fixed. —bender235 (talk) 17:36, 9 November 2010 (UTC)

I'd like to re-utter that wish, and add another: could you please create a bookmarklet for the interactive version of Reflinks? —bender235 (talk) 18:35, 18 December 2010 (UTC)
Bookmarklets for all tools on the help page, but it needs to be revamped. I had some code for removing the tag; however, IIRC removed it as the checks for to see if cleanup was performed could only be determined after the user saved. — Dispenser 19:38, 18 December 2010 (UTC)
As far as I see there is only a bookmarklet for the automatic reflinks version, not the interactive one. —bender235 (talk) 20:10, 18 December 2010 (UTC)
The bookmarklets were in a table label as [4], [5], [6] and they were too far from the subheading. This was a relic of an older design featuring every permutation possible. I've since reworked the section and upgrade the bookmarklets. Protip: The bookmarklets work on my tools, so you can jump to another tool using them. — Dispenser 00:56, 19 December 2010 (UTC)
Thanks. —bender235 (talk) 00:47, 19 December 2010 (UTC)
Just installed the bookmarklet, and it worked well. Also, one more vote to automatically remove {{Cleanup-link rot}} (and redirects) once all bare URLs are fixed. Thanks! GoingBatty (talk) 23:02, 16 July 2011 (UTC)

Oppose Running Reflinks does not necessarily bring all the citations in an article up to full strength, will all relevant parameters filled in. The {{Cleanup-link rot}} template is about more than just adding "|title=" and dates to a bare URL. The fact that an editor uses the Reflink tool, does not guarantee that he has fixed all the problems with the citations. Lentower (talk) 05:39, 19 September 2011 (UTC)

What the Hex?

It seems that Reflinks replaces six-digit capitalised html hex codes with three-digit lowercase ones. For example, in this edit, Reflinks replaced #FFFFFF with #fff hex colour numbers. There is no noticeable change in the rendered output, but somebody asked me the question, so I'm also curious to know. --Ohconfucius ¡digame! 12:37, 27 September 2011 (UTC)

As seen in this, the formatting offered by REFLINKS for google books is much worse than that offered by the http://reftag.appspot.com app (noted at #Wikipedia citation tool for Google Books and Reflinks incorperated). --Piotr Konieczny aka Prokonsul Piotrus| talk to me 19:27, 5 October 2011 (UTC)

Hi Dispenser,

(I find this tool a useful way to start improving a citation, and when I have the time to do all the research and cut-n-paste, a useful way to get a full citation done.)

The bug: this tool, Relinks, is suggesting to editors to fill in the "|publisher=" parameter wrong. The meta data it is putting there should be placed in the "|work=' parameter.

That is, it is incorrectly adding the dotted name of a web site as
|publisher=en.wikipedia.org (for example), when it should be
|work=en.wikipedia.org (for example).
Clearly, in this case
|publisher=The Wikimedia Foundation, Inc., a non-profit organization.

A web site is a form of media, and in the {{cite family of templates, can be included as the "enclosing"/larger work that the cited web page is part of. It is neither the owner nor publisher of the web page or web site.

Please fix this as soon as possible.

Please consider how to fix the insertions made in error, by this tool in the past.

Please consider changing the CheckLinks tool to detect this error.

Thanks!

Note that the guidelines seem clear about the parameter usage. From Template:Cite_web#Optional_parameters

  • work: In most cases this is the name of the website (as usually given in the logo/banner area of the site), otherwise the site's domain name. If the titled item being cited is part of some other larger work, such as a book, periodical or organizational sub-site (e.g., the law school's section of a university's website system), it is usually better to use the name of that more specific work than that of the entire site. Do not italicize; the software will do so automatically.
  • publisher: The name of the entity that publishes (owns or controls) the website. Commonly, this is a government agency, educational institution, or business. For many websites, the author and publisher are the same, and only one needs to be included in the citation; prefer publisher for organizations, author (or first/last) for individuals. Please note that publisher is not the name of the website; that is the work, except in cases where the business name is identical to the site name. For example, the corporation Walmart is the publisher of the website at walmart.com, which is the work; Amazon.com is the publisher of the website at amazon.com, which need not be specified as the work, as this would be redundant. For republished works, generally use the original publisher information (including location, date, etc., and it is and it is often more appropriate to use {{cite book}}, {{cite news}}, etc. with a |url= parameter instead of {{cite web}}).
  • location: Geographical location of the publisher (or headquarters thereof).

Thank You. Lentower (talk) 22:13, 18 September 2011 (UTC)

This happens when {{Cite web is used. Which is what happens when this tool converts a bare URL to a citation. I have not investigated what it does with the other citation options it offers. Apologies for a less than complete bug report. Lentower (talk) 20:22, 5 October 2011 (UTC)

Dispenser/maintainers:

What's happening here? Is this bug being addressed? Has the bug report even been accepted? I'm tired of having to clean up after this tool and I'd like to know when a fix can be expected. Thanks for any status report you can give us. — JohnFromPinckney (talk) 03:46, 2 October 2011 (UTC)

Len, it's often difficult for a bot to make the distinction between a source that needs to be italicised and one that doesn't. You ought to be using this script to supplement the brilliant Reflinks. --Ohconfucius ¡digame! 02:24, 19 September 2011 (UTC)

Ohconfucius, I'm not talking about presentation here, but content. Editors, using a bot, tool, or not, should be filling out templates with the right content in each parameter. The place to get the presentation right, is in the template code itself. This design separation has been understood for a long time. Probably before Alan Kay and team made it in Smalltalk, and Tim Berners-Lee made in his design of the WWW, separating how a browser and it's user presented/displayed the content described by the HTML.
The Relinks bot is setting a bad example by filling in a parameter with the wrong content. I don't know how to make the case better than I already have. Perhaps if you tell me WHY you think I was talking about font choice, that is presentation of the content in a citation, rather than filling in a parameter properly, I can see how to report this bug better. Did you read the excerpt from the template documentation, I gave above? Have you read the guidelines about citations in general and the {{cite templates? Other questions? Comments? Lentower (talk) 04:10, 19 September 2011 (UTC)
I apologise for making an erroneous assumption. Reflinks is deliberately not a bot. It is a tool which is very useful to editors. I only suggested the formatting issue as one manifestation of the problem. The citation templates and choice of parameters are quite closely related to the format of the rendered output.

I guess you're suggesting that I shouldn't even have to write a script to resolve some of the issues of filling in the wrong parameter. Just how metadata is classified by external websites, and to what extent that data can be correctly mapped onto one of our citation templates is a technical issue for the Dispenser. However, tool documentation makes it abundantly clear that the existence of certain technical issues means the tool should be used carefully and with human supervision. I'd also note that your comment above also implies that |publisher= has to be filled in, whereas it's optional. --Ohconfucius ¡digame! 04:37, 19 September 2011 (UTC)

This tool, Relinks, is suggesting to editors to fill in the "|publisher=" wrong. The meta data it is putting there should be placed in the "|work=' parameter.
Ohconfucius, are you a maintainer of this tool: Reflinks? If not, I really don't need to discuss this further with you. Thanks for your input, but my goal here isn't educating you about this bug, but having the maintainers improve their tool. The script you point at is interesting, but doesn't really speed my editing, as you need to verify the actual original publisher. E.g. even the NY Times re-publishes other publisher's material. I'm sure you and some other editors enjoy using it though. Thanks again. Lentower (talk) 05:04, 19 September 2011 (UTC)

Do we need a new developer?

Seeing no replies to most of my comments above (not counting the thread just above, which I've started right now), I am on the verge of declaring this tool dead and the developer, inactive, and asking somebody at VP to take this over. It's a shame that a tool with such a potential is no longer actively maintained. --Piotr Konieczny aka Prokonsul Piotrus| talk to me 19:27, 5 October 2011 (UTC)

Has anyone does what he asks in the box at the top of the page?
I'm available in ##dispenser (webchat) typically between 15:00 and 4:00 UTC
Maybe checking at different times and on different days? Or checked for recent activity on his other talk pages or his user contributions list? Lentower (talk) 20:25, 5 October 2011 (UTC)
I see that Dispsenser has posted recently on User talk:Dispenser, so I posted a new message there. Hope this helps! GoingBatty (talk) 02:04, 6 October 2011 (UTC)

Dates are not authors

Hi, re this edit (second one down) - it is highly unlikely that the author's name is "Published Friday, Nov 5 2010, 10:06 GMT" - in fact there is a proper author on that web page, it states "By Daniel Kilkelly". How is this happening? --Redrose64 (talk) 11:27, 12 October 2011 (UTC)

It's not a bot - it requires reviewing, so I suggest you drop Danno uk a note and ask them to be more careful. SmartSE (talk) 12:41, 12 October 2011 (UTC)
I realise it's not a bot - but the tool must be using some form of logic to determine which items should go into the various parameters. Have left note with Danno. --Redrose64 (talk) 12:56, 12 October 2011 (UTC)
The web is not semantic yet so no biting the tool devs, please! That is, web developers do not make authorship extraction easy, by any means, even though the tools (tags and metatags) to build such information into websites and pages have existed since 1997-8. The reflinks tool tries its best to decide, based on consistent text positioning and nearby tags, which fields are which. For those websites not hand-coded into the reflinks ruleset, you get a best guess. It is by no means perfect, but it saves WP:Wikignomes an enormous amount of copy/pasting on average. --Lexein (talk) 13:05, 12 October 2011 (UTC)

Simple usage instructions

See the short instruction section on these pages:

Bug report

accessed versus retrieved?

Why does this replace the perfectly acceptable term "accessed" with "retrieved". This would appear to be purely a matter of personal preference. I prefer "accessed" and have used it all my articles. I've never had a problem with it, at FAC or otherwise. AaronY (talk) 09:29, 12 November 2010 (UTC)

Consensus established in the citation templates (and reinforced by most style manuals) is to use "retrieved" instead of "accessed" as the former is clearer and is widely used. SlimVirgin brought this up on the Dab solver talk page. — Dispenser 02:37, 23 November 2010 (UTC)
Do you have a link? AaronY (talk) 02:50, 23 November 2010 (UTC)
I could be mistaken, but I think the link is User talk:Dispenser/Dab solver#Changing access dates to "retrieved"; I think that thread falls well short of establishing consensus. Jc3s5h (talk) 17:32, 25 October 2011 (UTC)
The Chicago Manual of Style uses "accessed". See 16th edition, page 754. Jc3s5h (talk) 17:52, 25 October 2011 (UTC)

Ongoing blacklisting problems

Running Reflinks against Dominance (genetics), most of the NIH links happen to redirect to OMIM.org, run by Johns Hopkins University return a warning of "WARNING : Blacklisted title (redirect)". Example:

Where is the blacklist or whitelist used to determine what is "blacklisted" or not? Or are all redirects blacklisted? How can I instruct Reflinks to expand redirects? --Lexein (talk) 14:09, 2 November 2011 (UTC)

Better?

I'm in the U.S., and http://toolserver.org/~dispenser/cgi-bin/webreflinks.py just redirects to http://toolserver.org/~dispenser/view/Reflinks for me. Seems the same, not "better" as asserted in the edit summary and the "Simplest method" section of the article. Am I missing something? --Lexein (talk) 15:56, 14 November 2011 (UTC)

Concerning WP:Reflinks. When you click the link at the top of the page there (http://toolserver.org/%7Edispenser/cgi-bin/webreflinks.py) you end up at that URL address in your browser's address window:
I believe you are looking at the soft redirect link on that page, and not the actual URL address of the page you are on. Click the link there (http://toolserver.org/~dispenser/view/Reflinks) and you end up at that URL. --Timeshifter (talk) 18:14, 14 November 2011 (UTC)
So the trick is to ignore the giant blue arrow essentially demanding that the user click it. Ignore the shiny shiny arrow. I can't. Must. Click. --Lexein (talk) 19:54, 14 November 2011 (UTC)
lol - I wonder how we can contact the people who edit that toolserver.org page. --Timeshifter (talk) 20:05, 14 November 2011 (UTC)

Perpetual failure to ref up article

I've done everything I could think of, tried running Reflinks on Religions by country. I've tried running it on sections, on the whole document, used interactive mode as well as non-interactive, but to no avail. The ref runs are always incomplete or seem to time out. What to do? --Ohconfucius ¡digame! 16:40, 18 December 2011 (UTC)

Try limiting it to 50 references by using http://toolserver.org/~dispenser/cgi-bin/webreflinks.py?page=Religions_by_country&client=script&citeweb=on&overwrite=simple&limit=50&lang=en Then keep rerunning until you get all the references cleaned up. Good luck! GoingBatty (talk) 20:06, 18 December 2011 (UTC)
It can take 10 minutes for reflinks to finish filling in 100 refs. You can see the refs filling up the page. I believe you can stop at anytime. If all else fails click the stop icon in your address bar. Then go through the process and steps to fill in the refs. Then come back and run it again. Rinse, repeat. :) --Timeshifter (talk) 04:29, 19 December 2011 (UTC)
Thanks, that seems to be doing the trick. I've never seen so many refs in a single article! --Ohconfucius ¡digame! 08:39, 19 December 2011 (UTC)
WP:Reflinks is an amazing and wonderful tool. I was blown away to watch it fill in around 100 references in an article. --Timeshifter (talk) 11:33, 19 December 2011 (UTC)
Did you check each one to verify that it got the work and publisher parameters correct? It usually puts the website in for the publisher, contrary to guidance at the cite templates' documentation. Did you make sure that the date format used by Reflinks (invariably yyyy-mm-dd) matched the date format already established for refs in the article (usually mdy or dmy)? Did it get all of the author names correct (instead of grabbing "Up for discussion" or "Posted Oct. 10, 2011" or something from the page)? 100 at once is indeed really cool, but please don't ignore the manual-checking portion of the task, which is required. — JohnFromPinckney (talk) 19:18, 19 December 2011 (UTC)
I checked only that it was better than before, and did not delete references. Feel free to go back and correct everything. --Timeshifter (talk) 09:10, 21 December 2011 (UTC)

Bug report - ref tags going astray

<ref> tags going astray

Please refer to reference #33 in this edit. — Preceding unsigned comment added by Ohconfucius (talkcontribs) 09:10, 18 October 2011‎

Bug Report

I wanted to test reflinks, so I added a bare URL to Zebra mussel. It didn't find it... The Steve  12:06, 5 January 2012 (UTC)

Internet Archive Wayback Machine titles

Hi! When parsing Webcite urls, like here, the tool sets |title= to "Internet Archive Wayback Machine", which is wrong, because the target pages redirect after a few secs to the proper archived version. This happens a whole lot in different articles. Do note that setting |publisher= to "Web.archive.org" or using main |url= for archived url instead of |archiveurl= is also incorrect, albeit fixable by bot. Regards. —  HELLKNOWZ  ▎TALK 16:01, 6 January 2012 (UTC)

Citation

Hi, I think that using the template {{citation}} would be better for Reflinks, as it is more general that {{cite web}}. Any chance? --KDesk (talk) 18:01, 7 March 2012 (UTC)

Consider honoring "Use dmy dates"

Consider recognising and supporting {{Use dmy dates}} {{Use ymd dates}}, and {{Use mdy dates}}. "Use dmy dates" was present in the article I just ran it on, but that wasn't reflected in the "date=" results. IMHO I see no need to change the "accessdates=" though. Thanks. 67.100.127.7 (talk) 13:15, 16 November 2011 (UTC)

Well, there ought to be an option to set the format for accessdates, too. I'm sick and tired of having to change huge numbers of dates in refs adjusted by reflinks where the new dates are different from the date format used throughout the rest of the article. — JohnFromPinckney (talk) 15:48, 16 November 2011 (UTC)
Get used to being unhappy if you seek perfection on Wikipedia. :) --Timeshifter (talk) 09:12, 21 December 2011 (UTC)

If necessary we can use User:Ohconfucius/script/MOSNUM dates.js to correct the dates. It is not quite as simple as honouring the use xxx templates because some articles have one format for content dates and iso format for accessdates etc, but being consistent with the template if it is present would be better than using a iso dates everywhere. --Mirokado (talk) 11:58, 17 March 2012 (UTC)

Reflist

Hi. This tool only recognize {{reflist}} or in text. Serbian Wikipedia also uses {{([Ии])звори}} and when we use Reflinks, it places -- Bojan  Talk  11:38, 25 March 2012 (UTC)== Reference== at bottom of page. Can You add support for this?

Reflinks appears to be removing wikilinks within refs automatically and without informing the user--see discussion & example here. If this intentional, please indicate policy or guidelines that support this behavior (I did not find any), or please fix. Thanks. --Hobbes Goodyear (talk) 22:21, 31 March 2012 (UTC)

Its considered over-linking. Old discussion at User talk:Dispenser/Archive 2#Links to newspapers in refs (and Lost MOS link). — Dispenser 22:42, 31 March 2012 (UTC)
Hmm, well, if it's only common sources, I guess I don't see a problem, either. Thanks. --Hobbes Goodyear (talk) 22:57, 31 March 2012 (UTC)