Remove arXiv version number from “file” field (with auto file download disabled)

[BenjaminDAnjou]

When using the arXiv fetcher (with auto file download disabled), the version number still appears in the “file” field. For instance, the “file” field after fetching without file download is:

file = {:http://arxiv.org/pdf/1701.02434v1:PDF},

As a result, the user is not automatically linked to the latest version. To be linked to the latest version everywhere, the link should be

file = {:http://arxiv.org/pdf/1701.02434:PDF},

Would it be possible to remove the version number in the “file” field so that we are automatically linked to the latest version?

Right now, the only way is to manually delete the version number in the “file” field, which is time consuming.

This was tested with JabRef 5.9 on MacOS Ventura 13.2.1.

[Siedlerchr]

Hm, if I remember correctly this should aleady be the case.
Did you use New entry -> By id or did you use the web search?

[BenjaminDAnjou]

This happens both when using the Web Search function and using new entry with arXiv ID type.

This does not happen when using new entry with the arXiv issued DOI because in this case there is no "file" field.

I also should clarify that this was fixed earlier for the "eprint" field. It's only for the "file" field that there is still an issue (with auto-file download disabled).

[koppor]

The file field should never contain a http url. This was a "hack" to enable adding the entry and downloading it after adding to the library. There is something wrong with the download.

Excerpt of the bib entry (retrieved by idfetching arXiv:1701.02434v1)

@Article{Betancourt2017test,
  author        = {Betancourt, Michael},
...
  archiveprefix = {arXiv},
  doi           = {10.48550/ARXIV.1701.02434},
  eprint        = {1701.02434},
  file          = {:http\://arxiv.org/pdf/1701.02434v1:PDF},
  publisher     = {arXiv},
}

The DOI leads to the web site, the PDF should have been downloaded.

If I set the file directory, it works:

@Article{Betancourt2017test,
...
  file          = {:Betancourt2017test - A Conceptual Introduction to Hamiltonian Monte Carlo.pdf:PDF},
...
}

Can you try the latest development versio at https://builds.jabref.org/main/?

[Siedlerchr]

@koppor Nope, it's totally valid to store urls in the file field. We have checks for this e.g. isOnlineLink

[koppor]

@koppor Nope, it's totally valid to store urls in the file field. We have checks for this e.g. isOnlineLink

There should be a cleanup to move this to the url field. Otherwise, this is really strange IMHO.

My main point is that we have auto download in place. Hence, an online link there is an indicator of an error IMHO.

[koppor]

Source: https://docs.jabref.org/finding-sorting-and-cleaning-entries/filelinks

[Siedlerchr]

This is perfectly fine. JabRef can handle http links very well in the LinkedFiles and you can do a right click -> Download

When auto download is not activated (as is here the case) the http link is stored in the file field