URLs that do not work correctly in Via or the LMS app

mkdir-washington-edu commented 2 years ago

Bug report form

See: https://hypothes-is.slack.com/archives/C2BLQDKHA/p1642175183152600?thread_ts=1641839979.048300&cid=C2BLQDKHA

tl;dr We see Via users (especially instructors using the LMS app) not always anticipating the issues they may get on URLs they want to annotate. Since this is largely about LMS users the chrome extension and bookmarklet are not viable options.

This issue will be to track "problem URLs"; the reasons might be log ins, our proxy being blocked, content within iframes, and other reasons.

Support folks adding URLs to this form, please list out:

the URLs,
the link to a support ticket where available,
the observed behavior of the site in Via,
if available, a link to a Slack conversation where a developer diagnosed the problem.

Support can help keep this list useful by, of course, vetting these sites and make sure the user is reasonable in what they're linking to. Reasonable would be a site where a user can see the text they would expect students to annotate.

Template for comments: URL:

Support ticket:

Via behavior:

(optional) Slack convo:

janraev commented 6 months ago

URL: https://nces.ed.gov/surveys/pirls/pirls2021/index.asp

Support ticket: https://app.hubspot.com/contacts/6291320/record/0-5/2795447113

Via behavior: The body of text in the webpage does not load.

janraev commented 5 months ago

URL: https://www.colorado.edu/health/blog/cannabis

Support ticket: https://app.hubspot.com/contacts/6291320/record/0-5/2913397121

Via behavior: It takes at least a full minute to see body of texts in the webpage that can be annotated.

There are several console errors. See screenshot below:

robertknight commented 5 months ago

URL: https://www.colorado.edu/health/blog/cannabis

This page inserts a large number of <script> elements in the <head> without async or defer attributes. As a result they block loading of the body. A certain number of these scripts are loaded successfully, but then Via hangs on one of them and it blocks until the request eventually hits a timeout after a minute.

janraev commented 4 months ago

URL: https://www.joanganzcooneycenter.org/wp-content/uploads/2014/10/jgcc_leveluplearning_final.pdf

Support ticket: https://app.hubspot.com/contacts/6291320/record/0-5/2955709528

Via behavior: The PDF loads but the H sidebar doesn't. It appears to be loading something as indicated by the red bar under the address bar, but it does not progress any further.

Possibly the same as https://github.com/hypothesis/product-backlog/issues/1270#issuecomment-2004670707?

robertknight commented 4 months ago

Possibly the same as https://github.com/hypothesis/product-backlog/issues/1270#issuecomment-2004670707?

Yes. In both cases it looks like the PDF is being loaded in the native PDF viewer, instead of our custom annotation-enabled one. This can happen if the "router" in Via detects the content type as HTML and tries to load it as such, but then the content actually turns out to be a PDF.

I notice that fetching this file with cURL returns a 403:

$ curl -i 'https://www.joanganzcooneycenter.org/wp-content/uploads/2014/10/jgcc_leveluplearning_final.pdf'
HTTP/2 403
server: awselb/2.0
date: Fri, 12 Jul 2024 13:38:26 GMT
content-type: text/html
content-length: 118

Spoofing a browser by setting the User-Agent header results in the real PDF being returned:

$ curl -i 'https://www.joanganzcooneycenter.org/wp-content/uploads/2014/10/jgcc_leveluplearning_final.pdf' -H 'User-Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/605.1.15 (KHTML, like Gecko) Version/17.4.1 Safari/605.1.15'
HTTP/2 200
date: Fri, 12 Jul 2024 13:39:08 GMT
content-type: application/pdf
server: Apache/2.4.41 (Ubuntu)

So we should check what user agent we are using for the initial request for the content in Via that discovers what kind of content it is.

janraev commented 4 months ago

URL: https://storymaps.arcgis.com/stories/62f54d97f76f4e3cb50775d309be6cce

Support ticket: https://app.hubspot.com/contacts/6291320/record/0-5/3011890245

Via behavior: Webpage displayed is just blank. Console shows 504 error.

From a partner school, this is what displayed for them in the LMS:

janraev commented 3 months ago

URL: https://manifold.open.umn.edu/read/complete-text/section/66ad2737-0444-4c0c-beda-0bbeee592c38

Support ticket: https://app.hubspot.com/contacts/6291320/record/0-5/3067397767

Via Behavior: Page opens up along with the sidebar. When attempting to annotate, the built-in annotator from Manifold is the only one showing up.

mkdir-washington-edu commented 3 months ago

For https://github.com/hypothesis/product-backlog/issues/1270#issuecomment-2269945443

While barely knowing what I'm talking about, I'd bet that their annotator is swallowing the text selection event so we don't even see that it's happening.

janraev commented 3 months ago

URL: https://www.historians.org/resource/why-study-history/

Support ticket: https://app.hubspot.com/contacts/6291320/record/0-5/3158963352

Via Behavior: It takes a minute before the H sidebar shows up. Similar to https://github.com/hypothesis/product-backlog/issues/1270#issuecomment-2207232861

janraev commented 3 months ago

URL: https://www.abdallahcandies.com/information/7-health-benefits-of-eating-chocolate/

Support ticket: https://app.hubspot.com/contacts/6291320/record/0-5/3159861007

Via Behavior: It takes a minute before the H sidebar shows up. Similar to https://github.com/hypothesis/product-backlog/issues/1270#issuecomment-2207232861

janraev commented 3 months ago

URL:

Support ticket: https://app.hubspot.com/contacts/6291320/record/0-5/3163364468

Via Behavior: It takes a minute before the H sidebar shows up. Similar to https://github.com/hypothesis/product-backlog/issues/1270#issuecomment-2207232861

janraev commented 3 months ago

URL: https://storymaps.arcgis.com/stories/abba237bedbc4e2a98e65654cbfc7e65/ Support ticket: https://app.hubspot.com/contacts/6291320/record/0-5/3163599147

Via Behavior: Page does not load and console shows 504 error.