Do you collect information on whether there are html versions of articles as well as pdf?

Plain HTTP allows a man-in-the-middle attacker to serve a phishing site which might direct user to install a malicious extension or download arbitrary files instead of the actual extension. That is especially problematic for academic personnel because they commonly use a network shared by hundreds of people.

Dear unpaywall team,

Thank you for developing such a nice tool!

I have found a couple of false negatives:

• http://dx.doi.org/10.1016/j.neuroimage.2016.03.054 >> http://www.hal.inserm.fr/inserm-01291748/file/Maumet_NeuroImage_A_Contrario.pdf
• https://doi.org/10.1016/j.neuroimage.2013.04.079 >> http://www.hal.inserm.fr/inserm-00816852/file/Maumet_NeuroImage_2013.pdf

I have checked and the source repository "HAL" (both hal Inria & hal Inserm) seems to be available in your data sources.

Is there anything I can do to help with this issue?

Noticed that when opening the developer console in chrome there are two lines that have been console.logged from unpaywall.js (line 12) in the extensions folder. Is this necessary?

unpaywall: unpaywall is running undefined
unpaywall: got settings [Object]

The American Mathematical Society has a moving wall of 5 years after which all journal articles become freely available. See, e.g., https://www.ams.org/journals/jams/all_issues.html.

It would be great if those articles could be picked up by unpaywall.

(Authors are also encouraged to deposit their final manuscript on arXiv.org but IIUC oadoi doesn't try to find articles in the arXiv heuristically.)

Please send new comments and bug reports to https://support.unpaywall.org or [email protected]. Thanks!

Existing issues have been closed and moved as appropriate.

Links from this site that appear on Google Scholar as plain link do not work when used from a different referrer.

For example:

• google scholar link (from copy and paste): http://www.academia.edu/download/46960023/Fear_of_failure_in_conservation_The_prob20160702-20054-uaiy2r.pdf
• actual url served when clicking above link on google scholar: http://s3.amazonaws.com/academia.edu.documents/46960023/Fear_of_failure_in_conservation_The_prob20160702-20054-uaiy2r.pdf?AWSAccessKeyId=AKIAIWOWYYGZ2Y53UL3A&Expires=1492544676&Signature=KlvxP9jm3x8hDqLt3YJi6G3k3kw%3D&response-content-disposition=inline%3B%20filename%3DFear_of_failure_in_conservation_The_prob.pdf
• using the copy/pasted link directly 404s at Academia.edu

Perhaps academia.edu links shouldn't be used in unpaywall. After all, it is not a dedicated, not for profit self-archiving service but a for-profit social networking site.

update

This is still broken for the example above and I assume for all other papers where the link provided by https://oadoi.org/ is something from Academia.edu. For an app that's clearly been thought out and well-promoted, I'm surprised at the lack of response

tested with: Firefox plugin, version Unpaywall 1.5

record URL: http://www.tandfonline.com/doi/full/10.1080/10508422.2016.1193439

Tested with:

• Firefox plugin, version Unpaywall 1.5 = display in PMC broken, grey icon displayed
• Chrome plugin, version Unpaywall 3.2 = display in PMC works fine, bronze icon displayed

PMC URL: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5484110/

In PubMedCentral:

On publisher website:

https://openaccessbutton.org/?url=http%3A%2F%2Fdx.doi.org%2F10.1287%2Fisre.2015.0618

Trying to access this paper:

https://www.mitpressjournals.org/doi/full/10.1162/NECO_a_00332

the extension gives a correct green icon showing there's a legit OA alternative, but fails to link to the correct address which is

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3472550

and not

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3472550/pdf/

Thanks,
Laurent

Thanks for the great plugin!
There is another OA doi resolver: http://doai.io/

In my experience, this sometimes finds OA PDFs where https://oadoi.org/ cannot find one. Would be great to have the unpaywall button as one aggregator that collects them all.

Example: http://pubsonline.informs.org/doi/abs/10.1287/isre.2.3.173
Redirects to: http://www.jstor.org/stable/23010882

Example: http://www.sciencedirect.com/science/article/pii/S0361368201000277
open url: http://www.academia.edu/download/36140325/1-s2.0-S0361368201000277-main.pdf
Error: Page Not Found. Sorry. We can't find the page you're looking for.

If you are logged into the academia.edu website you do not see the error message (will be redirected to the correct url)

I just installed the new Firefox add-on. It seems to be working pretty well!

However, I wondered if it could be made less intrusive. Since users have have the option of putting the icon in the toolbar anyway, could that change colour and provide access to papers, instead of having the icon drawn on the webpage itself?

I guess making this an option would be one strategy, but I think the toolbar icon should at least provide full functionality, otherwise it's not particularly useful anyway.

Browser: Chrome 56.0.2924.87 (64-bit), Unpaywall 1.6.1

I checked a couple of publisher platforms today, here is a short list of landing pages where the amazing Unpaywall button has not appeared:

• AMS, http://www.ams.org/journals/era/2004-10-13/S1079-6762-04-00136-2/home.html
• APA PsycNet: http://psycnet.apa.org/?&fa=main.doiLanding&doi=10.1037/h0027297
• IEEE XPlore: http://ieeexplore.ieee.org/document/4302745/

... but overall the coverage is pretty impressive for a prerelease! One question out of curiosity: How does the Unpaywall plugin detects article landing pages?

The question arose among us that there was no mechanism to know when the Unpaywall extension is used to direct traffic to our site (we are a publisher). Since the Green Tab only offers a URL, my suggestion is to append a query string to the destination URL in the Green Tab.

Example:

https://www.publishersite.com/article/Complicated-Research.pdf?unpaywall=1

or if there is an existing query string, the code would need to determine this and react accordingly:

https://www.publishersite.com/article/Jurassic-Park-Journal.aspx?vol=12&issue=3&article_id=23677&unpaywall=1

This does not allow the destination site to (easily) track incoming unpaywall references via something like Google Analytics, but it's a good step towards knowing our impact.

If this is something Unpaywall is interested in including, I'll fork, code, and make a pull request.

Hello,
Is there any specific reason for static/node_modules to not be gitignored?

Including node_modules in the repo greatly increases its weight, and as far as I know is considered bad practice.

Fantastic work creating unpaywall! 👏 Also neat that one can differentiate between green and golden open access.

As you know, golden OA can take two forms: open access journals (which is a business model I'd like to support) and hybrid OA, in which subscription journals use both subscriptions and (high) APC as two sources of revenue. I believe the oadoi.org API makes the difference between these two (oa_color": "gold AND is_subscription_journal": true)?

Does that mean it would be possible to make that difference in the unpaywall tab? I'd fully understand if you don't want to indicate that difference, because for the reader of an article it doesn't really matter. But for OA nerds like me, it does. 😄

I find that this extension doesn't work on The Washington Post. Entire articles are behind paywalls and they're using obfuscation techniques to prevent extensions like these from working. I've examined the page source and even if I tried to delete the Paywall Section from the source inspect it still doesn't let you view the article in full, since it's also preventing you from scrolling the page.

Not exactly a problem to the general public, but I notice Unpaywall does interfere the way Firefox parse XML, maybe similar to https://support.mozilla.org/en-US/questions/1074505

Try enable/disable Unpaywall when you open the link below...
http://www.xmlfiles.com/examples/note.xml

example link: http://www.sciencedirect.com/science/article/pii/S1053811916300416

On the https://distill.pub website, unpaywall is finding articles, but as far as I can see none of them are related to the article on the page.

For exemple on this page https://distill.pub/2017/aia/, unpaywall shows up with a green icon, but the linked article (https://arxiv.org/pdf/1609.04468.pdf) doesn't seem to be at all related to the distill page at first, but it is in fact in the references of that page, with a pdf to the article.

Also this page on distill https://distill.pub/2019/activation-atlas/ seems to be working properly (unpaywall just loops back to the same page, I don't know if it's the intended behavior).

Example: https://vivo.tib.eu/fis/individual?uri=http%3A%2F%2Fd-nb.info%2Fgnd%2F123841704

1. Browser: Vivaldi 1.8.770.50 (Stable channel) (32-Bit)

There are links to two DOIs in this page. Unpaywall recognizes only the first one.

2. Browser: Firefox 52.0.1 (32-Bit)

Same page, Unpaywall does not recognize any article.

One may click on a DOI link which uses doi.org to resolve it.

In order to accelerate the pace at which one lands on a OA version of a publication, it could be useful to rewrite the redirection to doi.org to oadoi.org, so that there is no extra need to click on the unpaywall icon.

This may be an opt-in feature. It may also be associated to a slight popup indicating oadoi.org was used to resolve this DOI.

Example: https://www.sciencedirect.com/science/article/abs/pii/S0893395223001412

Some user feedback:

It would be helpful if the plugin would offer the possibility to paste a DOI to be checked explicitly, e.g. because
a. the icon is missing completely (see #24)
b. the display of the icon is delayed (noticed for ACS, BMJ)
c. the information provided is incorrect (see #23) and should be verified

https://github.com/ourresearch/unpaywall/blob/31ce03b59568a12e8a78b162a5c53750d3a342b6/extension/unpaywall.js#L561

You declared pdfUrl as:
var pdfUrl;
so the resultant behaviour when a pdfUrl is not found on the page continues to execute downloadPdf, which unfortunately passes "undefined" as a part of the URL sent in the xhr request. That means providers are probably getting a slew of 404s in their logs.

Could you provide a hook to allow a fall back to university proxies?

Dear Heather & Jason,

is there a data discrepancy between oaDOI and unpaywall?

Example: 10.1017/jns.2017.43

Bronze OA tab:

But the oaDOI API only reports it's closed: http://api.oadoi.org/10.1017/jns.2017.43

{
  "results": [
    {
      "algorithm_version": 1,
      "doi": "10.1017/jns.2017.43",
      "doi_resolver": null,
      "evidence": "closed",
      "free_fulltext_url": null,
      "is_boai_license": false,
      "is_free_to_read": false,
      "is_subscription_journal": true,
      "license": null,
      "oa_color": null,
      "reported_noncompliant_copies": [],
      "url": "http://doi.org/10.1017/jns.2017.43"
    }
  ]
}

... in the meantime, I'll contact CUP and request them to report license information to CrossRef ;)

Thanks for your marvelous work,
inga

Hi,

https://unpaywall.org/integrations states:

the shortcut link https://unpaywall.org/YOUR_DOI automatically redirects to an open version of YOUR_DOI if we have one;

However this syntax doesn't seem to work any longer, e.g. https://api.unpaywall.org/v2/10.3733/ca.2016a0025?email=xxx@xxx returns "is_oa": true, and several oa_locations. But https://unpaywall.org/10.3733/ca.2016a0025 doesn't redirect to any of them.

Thanks & have a nice weekend,
inga

The button is there, but the information about free availability isn't displayed.

Example Record:

Artificial Intelligence, Volume 241, 1 December 2016, Pages 1-44
Integrating social power into the decision-making of cognitive agents(Article)
by: Pereira, G.Email Author, Prada, R.Email Author, Santos, P.A.Email Author View Correspondence (jump link) 
DOI: 10.1016/j.artint.2016.08.003

According to oaDOI, "is_free_to_read" is true: http://api.oadoi.org/10.1016/j.artint.2016.08.003

Hi! Thanks for unpaywall, this is a great tool!

On this manuscript page https://doi.org/10.1038/s41592-018-0235-4 the unpaywall buttton returns: "The Unpaywall extension couldn't find any legal open-access version of this article."

while the manuscript is openly available on biorxiv: https://doi.org/10.1101/306951.

The repo title says "Chrome extension that gives you a link to a free PDF when you view scholarly articles".

It suggests that only Chrome is supported, which is incorrect.

What about changing it "Firefox/Chrome extension that gives you a link to a free PDF when you view scholarly articles"

http://journals.sagepub.com/doi/10.1177/1746847713519386 - displays as being unavailable, but is available via https://www.researchgate.net/profile/Xavier_Minguez_Lopez/publication/260923534_Folktales_and_Other_References_in_Toriyama%27s_Dragon_Ball/links/00b49532c22d2de163000000/Folktales-and-Other-References-in-Toriyamas-Dragon-Ball.pdf

Please fix this and consider using researchgate.net as a source for finding free copies of journal articles. Thank you.

Hi,
I've had a read through unpaywall.js and it seems to me as if most Green repositories will get a Bronze tab on the plugin. To summarise my understanding of how it is working. Having identified a DOI from meta tags, etc.:

• unpaywall queries oadoi by the doi. If the preferred OA source in the API has 'repository' as the host_type, this is considered 'Green'
• unpayall also tries to find a pdf on the current page by looking the the citation_pdf_url meta tag and similar locations. The local pdf source is always set to 'Bronze'
• In deciding on what to display, the getBestOaUrl() and decideTabColor() functions always give preference the local pdf as the source.
• In practice this means that even if I'm on a repository page and oadoi has identified the same page as the best OA source and has correctly identified the host_type as repository (so OADOI considers it green), the tab still displays as bronze because the local pdf path gets preference and doesn't know that it is a repository page

This isn't serious as I'd imagine very few people have the 'nerd mode' enabled, so they'll see green in any case. Just logging here for reference

Example for reference: https://cora.ucc.ie/handle/10468/3163

Microsoft Edge is the second most used browser on the web, making the browsing experience more flexible, offering many more services for users and optimized for Windows devices.

Microsoft Edge supports Chromium extensions, and you can publish your extensions to Microsoft Edge Add-ons website with minimal code changes. Find out more here.

If you have any more queries about porting Chrome extension to Microsoft Edge or anything else, you can contact the relevant teams whose links are given here.

Let me know if this was helpful. I am happy to help you with more details.

When I execute the following request an internal server error 500 occurs:

https://api.oadoi.org/v2/10.1186/[email protected]

Instead I would expect an empty response.

The problem:
Unpaywall does not seem to find the greenOA copies of articles from this journal as accessed at the paywalled version at BioOne

Some explanation:
BioOne assigns DOIs and is in some sense(?) the digital publisher for Bulletin of the American Museum of Natural History , however in parallel to this ALL articles from the Bulletin are also made available at an AMNH IR, immediate (no embargo) greenOA.

e.g.
Voss, R. S. and Fleck, D. W. 2017. Mammalian diversity and matses ethnomammalogy in amazonian peru part 2: Xenarthra, carnivora, perissodactyla, artiodactyla, and sirenia. Bulletin of the American Museum of Natural History pp. 1-118 http://doi.org/10.1206/00030090-417.1.1

is also available at: http://digitallibrary.amnh.org/handle/2246/6783

It seems to be true for every landing page at BioOne that I tried, that Unpaywall does not yet find the free version. If fixed it would add an extra 418 items (I think strictly they are monographs not articles), that are freely available with Unpaywall 😃

Apologies if this already exists--I thought it did, but now can't find any info about it. When the Unpaywall extension shows that no open version of an article can be found, would it be possible to have an option to request the author(s) to send one to the person looking, or suggest that they deposit a legal version in an open access repository? I realize this all depends an the author's email address being discoverable in the metadata.