[Zotero 7 Beta] Problems with the snapshots imported from ScienceDirect

mjthoraval

Here is an example of a snaphost imported from ScienceDirect:
https://s3.amazonaws.com/zotero.org/images/forums/u265723/lxa3apsy4l4f2gzkvl7w.png

During import, I still get a list of errors. For example, for importing this article:
https://www.sciencedirect.com/science/article/pii/S030193222100152X
Debug ID: D1532906549
Zotero 7.0.0-beta.74+2e8073ab9 (64-bit)
Windows 10

A simple click on the snapshot already triggers a lot of errors, even after a fresh restart of Zotero in Troubleshooting Mode:
Debug ID D449557900

[JavaScript Error: "Error in parsing MathML attribute value ‘4.pt’ as length. Attribute ignored." {file: "about:srcdoc" line: 0}]

[JavaScript Error: "Error in parsing MathML attribute value ‘4.pt’ as length. Attribute ignored." {file: "about:srcdoc" line: 0}]

[...]

mjthoraval

If I import from the list of articles in one issue of the journal, I get a different snapshot from saving directly from the article's page.

From the issue page, the snapshot is closer to the original formatting, but it occupies only half of the width available, and it cannot get the outline of the page: https://www.sciencedirect.com/journal/international-journal-of-multiphase-flow/vol/177/suppl/C

https://s3.amazonaws.com/zotero.org/images/forums/u265723/n552b2v835w3jganor4n.png

But saving from the article's page still shows a broken snapshot, but it covers the full width, and is able to extract the outline properly: https://www.sciencedirect.com/science/article/pii/S0301932224001277

https://s3.amazonaws.com/zotero.org/images/forums/u265723/7piadfnf6tojegqrt4hc.png

adomasven

We're investigating the first issue.

The second issue is caused by saving from the search results page, and the snapshot content coming not from the browser where those pages are loaded (because they are not), but by Zotero loading the page in the background and taking a snapshot. We are most likely going to disable saving snapshots from results page completely, as it often produces bad results, and also may be causing users to be marked as bots on some websites. We might be able to provide a workaround for that on Chrome in the future.

mjthoraval

Thank you.