Import adds PDF link and URL to file entry

orionpilot · September 8, 2024, 7:10pm

I don’t know when it started, but at some point when importing a new entry (either with the arXiv downloader within JabRef or the browser extension), the file entry contains both the link to the PDF and the URL. On PC this does not matter, but when I try to open the PDF on my iPad using the .bib, the I app am using gets confused as it think its the whole file entry which is the link to the PDF. This is new behaviour. I just want the PDF link in the file entry. I can’t find anything in the settings which would allow be to turn off the URL in the file entry.

Example:

Removing everything after PDF fixes the issue.

Siedlerchr · September 8, 2024, 9:10pm

Hi,

Indeed, this is fairly new and was requested and implemented by @ror3d and refs:

github.com/JabRef/jabref

Re-download missing linked files

opened 01:52PM - 02 Feb 24 UTC

closed 09:37PM - 18 Feb 24 UTC

ror3d

type: enhancement fetcher external files

**Is your suggestion for improvement related to a problem? Please describe.** A…t my work we share a bib file among 2-3 people that we keep updated to keep track of interesting papers, but due to policy, we can't share the actual PDFs. Trying to overcome this, we realized that if a file is downloaded, the original URL it was downloaded from is removed from the entry, so the file can't be easily redownloaded by other people using that link. Moreover, downloading it after the entry has been added seems to require manually downloading and adding it to the entry, which is not ideal. **Describe the solution you'd like** Downloaded files should remember where they were downloaded from, and give a way to re-download them if they are missing from the directory. **Additional context** I am working on a possible solution for this, I will PR when ready.

Maybe we can add a settings switch to toggle this behavior?

orionpilot · September 9, 2024, 12:14am

So that’s the use case for including the link in the file name. Thanks for sharing. This does seem like a good feature to have, I’m just in an edge case where it’s a hinderance.

I fully support adding a switch to toggle this behaviour. Hopefully it’s not too hard to implement.

koppor · September 9, 2024, 9:05am

@orionpilot Which iPad app do you use?

We have a discussion whether we should refactor the file field (Refactor BibTeX file field · Issue #98 · JabRef/jabref · GitHub). TL;DR: The pro side is, the file field could be more clean. The con side is that all aplications need to adapt…

In the concrete issue, JabRef’s behavior can be adapted to write a file field compatible with the iPad app.

koppor · September 9, 2024, 9:05am

Fix by @Siedlerchr: Add pref switch to not store url by Siedlerchr · Pull Request #11735 · JabRef/jabref · GitHub

Siedlerchr · September 9, 2024, 11:51am

@orionpilot You can now try the latest development version:

https://builds.jabref.org/main/

You can disable the feature in the preferences under Web Search “Keep downloaded url”

orionpilot · September 9, 2024, 2:23pm

I use this app. Since JabRef is built on the bib file, I wanted something that also relies on the bib and this app fits that very well. I link the same folder (on the cloud) where my JabRef stores my PDFs and the app can find them. I will note that the app already works if there are more than 1 PDF linked in an entry.

The app does explicitly call out JabRef regarding using the file field to find the attachments.

orionpilot · September 9, 2024, 2:37pm

That was quick! And I’ve tested it and it works as expected!

While testing I’ve found that the app tries to find the PDF under its default name. However when I import the file, it gets auto-renamed to my preferred format. Without the auto-rename this may not have been noticed.

In the example above, the PDF has the name between Full Text PDF and .pdf:PDF. But the app is trying to find the name between articles/<name>.pdf.

Siedlerchr · September 9, 2024, 3:55pm

Okay, which app are you using? I don’t know how it works but jabref file format is pretty much
Description:path to full text.pdf:FileType

orionpilot · September 9, 2024, 5:49pm

It’s called References. More info in this reply.

Siedlerchr · September 10, 2024, 9:36pm

I think you should submit feedback to the developer References | Benjamin Burton (maybe even link to this thread)

Seems like they are not parsing the file field correctly:

This is JabRef’s implementation for writing the file field to bibtex: jabref/src/main/java/org/jabref/logic/bibtex/FileFieldWriter.java at 14ba56d8ce19b80a53b1a22b10bb2d9da783f29a · JabRef/jabref · GitHub

(We also have corresponding tests)
The counterpart would be the parser:

github.com

JabRef/jabref/blob/14ba56d8ce19b80a53b1a22b10bb2d9da783f29a/src/main/java/org/jabref/logic/importer/util/FileFieldParser.java#L52


      
           *     <li>some characters might be escaped</li>
           *     <li>Windows absolute paths might be included without escaping</li>
           * </ul>
           */
          public static List<LinkedFile> parse(String value) {
              // We need state to have a more clean code. Thus, we instantiate the class and then return the result
              FileFieldParser fileFieldParser = new FileFieldParser(value);
              return fileFieldParser.parse();
          }
          
          public List<LinkedFile> parse() {
              List<LinkedFile> files = new ArrayList<>();
          
              if ((value == null) || value.trim().isEmpty()) {
                  return files;
              }
          
              if (LinkedFile.isOnlineLink(value.trim())) {
                  // needs to be modifiable
                  try {
                      return List.of(new LinkedFile(new URL(value), ""));

orionpilot · September 11, 2024, 3:30pm

Understood, thanks for the info! I have submitted feedback to the developer and linked to this thread. I will update once I hear back.

orionpilot · September 12, 2024, 12:17am

The app developer has seen my email and will fix the parsing issue @Siedlerchr mentioned above in the next app update, expected to be released soon. I will update this thread once the update has been published and I have tested it.

orionpilot · September 13, 2024, 8:55pm

References v3.4 released 2024-09-13 fixes this issue. I have tested it and it works. The app can now find the PDF when the URL is included in the file entry. Even with this fix, I am happy that the next JabRef version will include a toggle for this behaviour.

Thank you all for the help and the quick turnaround!

Topic		Replies	Views
Bug Report on associated file path Help	3	461	December 5, 2020
JabRef 4.3 --> JabRef 5 - unwanted/missing behaviour Help	2	511	May 4, 2020
Where to drag PDFs to create new entry Help	0	541	June 2, 2018
How to read bib data for/from a PDF file Help fetcher , entry-editor	8	3075	April 24, 2021
Trying to get a pdf linked to its citation and it's going to "misc"? Help	8	352	August 25, 2022

Import adds PDF link *and* URL to file entry

Related topics

Import adds PDF link and URL to file entry