Feel free to make requests for various tweaks or more sources to be covered below. Headbomb {t · c · p · b} 08:55, 14 February 2020 (UTC)Reply[reply]

Thanks for making this! I just noticed that Facebook was not listed as an unreliable source. buidhe 14:32, 14 February 2020 (UTC)Reply[reply]

@Buidhe: that's mostly because it wasn't listed in WP:RSPSOURCES. But it's clearly twitter-like, so I'll add it to the generally unreliable. Headbomb {t · c · p · b} 14:46, 14 February 2020 (UTC)Reply[reply]

 Done [1] Headbomb {t · c · p · b} 14:48, 14 February 2020 (UTC)Reply[reply]

Likewise, YouTube is marked as unreliable but similar sites Vimeo, twitch.tv, and Dailymotion are not. Could they also be added? buidhe 03:58, 21 February 2020 (UTC)Reply[reply]

Yup, easily.  Done Headbomb {t · c · p · b} 04:34, 21 February 2020 (UTC)Reply[reply]

I saw your recent post at WT:MED. I followed the first 3 steps of installation, but now what are the steps to actually using the script on an article? Thanks! Sorry if I missed that somewhere. I scanned things, and it did not appear obvious. I tried a couple different things to bypass the cache (I'm currently using google chrome on a chromebook), so I hope that's not the issue. Also, I'm in the middle of rewriting deep vein thrombosis (DVT). Might you be able to scan DVT for any of the potentially questionable sources / journals / publishers you posted about at WT:MED and post any hits on the DVT talk page to make sure I'm providing high-quality information? Thank you for your efforts here. Biosthmors (talk) 15:29, 14 February 2020 (UTC)Reply[reply]

@Biosthmors: basically, you just read the page and it should work right away. You can check Science Publishing Group and you should see 3 red highlighted links or User:Headbomb/unreliable/testcases and you should see things working (at least on some of them, I'm still tweaking to expand things to cover more). Headbomb {t · c · p · b} 15:32, 14 February 2020 (UTC)Reply[reply]

Thanks. It's working. I saw 3 pink-highlighted links at Science Publishing Group. Biosthmors (talk) 15:39, 14 February 2020 (UTC)Reply[reply]

@Biosthmors: In deep vein thrombosis, you should see DOIs that start with 10.2147 (Dove) / 10.4103 (Medknow) / 10.1155 (Hindawi) highlighted in yellow. Keep in mind that those are more grey-area and might even not be problematic to begin with. Headbomb {t · c · p · b} 15:40, 14 February 2020 (UTC)Reply[reply]

Thanks. I like the yellow color. It's subtle, and I missed the highlighting at first because I was scanning for something more obvious. But I like it because it pairs with the level of concern I have about the sources. (I saw 3 highlighted in yellow at DVT.) I don't have any plans to remove them or recite this material at the present, but thanks again for helping me check this out. Biosthmors (talk) 15:53, 14 February 2020 (UTC)Reply[reply]

Totally and utterly awesome. Thank you, Headbomb! SandyGeorgia (Talk) 15:18, 19 February 2020 (UTC)Reply[reply]

At User:SD0001/unreliabe.js.

@Headbomb: You can copy over the changes.

• When I initially wrote this, I mistakenly assumed the code will highlight the whole citation (rather than just the link). Today when I tested this for the first time, I saw this isn't the case. I've made it so now. I think this is desirable?
• I've tweaked the entire structure (clubbed together the regexes with the corresponding CSS styles) which should make this easier to maintain.
• Fixed the bug with "10.1011\/\d+"
• Now text content of all list items will also be checked. This should be good enough to catch stuff in further reading, bibliography sections. I know you asked for searching the whole page text, but I am not sure whether that is necessary. That is quite complicated to write, and by checking each and every text node on the page with each of the dozen-odd regexes, some of which are huge, I suspect there will be an impact on performance.

I haven't tested this exhaustively so let me know if anything isn't working.

SD0001 (talk) 10:05, 16 February 2020 (UTC)Reply[reply]

@SD0001: I'll test and give some feedback shortly. Headbomb {t · c · p · b} 10:07, 16 February 2020 (UTC)Reply[reply]

I prefer the original of only highlighting links, mostly because this can allow for more nuance when a reference has multiple links in it, but also because not everything is within <ref></ref> tags. See User:Headbomb/unreliable#Common_non-problematic_cases for example, but you could also have something like citation with url link to Scribd and a DOI. But here only the url to Scribd is problematic. The 1101 thing seems to work, which is weird because I thought I had tried that myself. Either way, very useful to finally have that fixed. Headbomb {t · c · p · b} 10:10, 16 February 2020 (UTC)Reply[reply]

Testing list content might be a good compromise vs whole page however (whole page would still be preferable though). It doesn't work on the last column of User:Headbomb/unreliable/testcases, but it does work in other lists elsewhere. Headbomb {t · c · p · b} 10:45, 16 February 2020 (UTC)Reply[reply]

@Headbomb: should work now. SD0001 (talk) 10:47, 16 February 2020 (UTC)Reply[reply]

Ok, not on that last column. But if you pull that out of the table, it will work. SD0001 (talk) 10:49, 16 February 2020 (UTC)Reply[reply]

That leads to another bug. We need a different regex for matching "/" when it's in a link vs when its in text. Looking into that now. SD0001 (talk) 10:52, 16 February 2020 (UTC)Reply[reply]

fixed. SD0001 (talk) 10:53, 16 February 2020 (UTC)Reply[reply]

@SD0001: Not in the column. Headbomb {t · c · p · b} 10:54, 16 February 2020 (UTC)Reply[reply]

Ok, that's also fixed. SD0001 (talk) 11:00, 16 February 2020 (UTC)Reply[reply]

@SD0001: Alright, so that's possibly a framework for some Frankenstein solution. What it would need now is some sort of logic that if there's a link you check the link, since that's better and more targeted. But if there isn't, then check the list. Because right now, it will highlight whole citations that aren't problematic [2], instead of just the link [3]. And this should really only be done as a fallback. Headbomb {t · c · p · b} 11:03, 16 February 2020 (UTC)Reply[reply]

@Headbomb: Done. SD0001 (talk) 11:15, 16 February 2020 (UTC)Reply[reply]

@SD0001: That looks really, really promising. Headbomb {t · c · p · b} 11:17, 16 February 2020 (UTC)Reply[reply]

@SD0001: works well, but numbered lists don't seem to work. Headbomb {t · c · p · b} 16:59, 16 February 2020 (UTC)Reply[reply]

@Headbomb: done SD0001 (talk) 18:03, 16 February 2020 (UTC)Reply[reply]

@SD0001: yup. And with this, I believe we have a stable script. Many many thanks. Headbomb {t · c · p · b} 18:05, 16 February 2020 (UTC)Reply[reply]

@SD0001: would there be a way to make use of a JSON-like structure somewhere? Perhaps as a separate subpage?

{
  "Name":    "Publisher",                          // Name of the publisher/journal/website
  "Domains": ["Publisher\.org", "Publisher\.com"], // Regex for URL matches
  "DOI":     ["10\.1234", "10\.4321"],             // Regex for DOI matches
  "Class":   "MEDRS",                              // BL, GUN, PRED, ... to set the CSS classes/colors
  "Note":    "See [[WP:RSPSOURCES#Publisher]]"    // Comment to understand why something is listed, possibly to be displayed as a tooltip
}

(not pretending the above is syntaxically correct, treat this as pseudocode) Headbomb {t · c · p · b} 13:33, 20 February 2020 (UTC)Reply[reply]

Note that I can build that JSON file myself once I have a working example/the syntax for it. Headbomb {t · c · p · b} 00:46, 21 February 2020 (UTC)Reply[reply]

I was initially wanting this to simply makes things easier to manage, but it would open up certain possibilities. Headbomb {t · c · p · b} 13:43, 20 February 2020 (UTC)Reply[reply]

Headbomb, Oh, I would love that. I already have a json file that I use to quickly check on all the online sources in an article. If i could just merge in something like the above in stead of maintaining it myself that would be super! Vexations (talk) 14:38, 20 February 2020 (UTC)Reply[reply]

Prototype JSON file at User:Headbomb/unreliable.json. Headbomb {t · c · p · b} 11:01, 21 February 2020 (UTC)Reply[reply]

I saw your post at the DYK talk page; this looks very useful, and I will install it tonight. Would it be possible to tweak the colour used for "marginally reliable" sources slightly; it's just because the pale yellow is very similar to the colour used in another commonly used script to signify Disambiguation pages, and I must admit I've used it for so long I completely associate that colour with Disambiguation pages now! Cheers, Hassocks5489 (Floreat Hova!) 15:02, 21 February 2020 (UTC)Reply[reply]

@Hassocks5489: I use the same script, but since this doesn't highlight a wikilink, but rather external links and list elements, there's no overlap between the two. There might be customizable colours down the road, but for now this is beyond my coding skills. Headbomb {t · c · p · b} 15:51, 21 February 2020 (UTC)Reply[reply]

Thanks for your reply, and no worries; it won't take long to get used to. Hassocks5489 (Floreat Hova!) 16:07, 21 February 2020 (UTC)Reply[reply]