Invalid DOI should not show "Connection error"

[koppor]

JabRef version

Latest development branch build (please note build date below)

Operating system

Windows

Details on version and operating system

Windows 10

Checked with the latest development build

• I made a backup of my libraries before testing the latest development version.
• I have tested the latest development version and the problem persists

Steps to reproduce the behaviour

1. Open https://link.springer.com/book/10.1007/978-3-319-47590-5
2. Download PDF ("Download book PDF")
3. Go to page 207
4. Select and copy the DOI:
   
5. Open "New entry dialog" and paste DOI:
   
6. See the question mark at the end
7. Click "Generate"
8. Result:
   

Expected behavior:

• The invalid character at the end should be removed internally
• The DOI data should be fetched

Appendix

No response

[TheDeveloperDino]

Hello, is this issue free? I can work on this issue.

[koppor]

Sure! Go ahead!

[koppor]

@TheDeveloperDino You can to test-driven development. A start can be #8124. I think, one should add test cases to org.jabref.model.entry.identifier.DOITest and then work on org.jabref.model.entry.identifier.DOI.

[TheDeveloperDino]

Okey.

[mrcstan]

@TheDeveloperDino, are you still working on this? @koppor , I tried to copy and paste the DOI as described above in Windows 10 but there is no such special character at the end, and no error result. However, I could produce the error if I use invalid characters like question mark at the end of the DOI. I would like to re-define this issue as one caused by any trailing invalid characters at the end of the DOI, and I would like to volunteer to fix it if @TheDeveloperDino is no longer working on it

[jiezheng5]

Hi,

I am a cs student, I repeated the issue in both Windows 10 and Linux Ubuntn OS 20.04. I fixed this issue by adding a method to automatically remove non-valid characters at the end of DOI string and let the DOI constructor call this method. I already verified the solution using both Windows 10 and Linux. I also wrote a ParameterizedTest for the method I created. I just submitted a PR (#8215) for review.

[mrcstan]

@jiezheng5, this is not really in the spirit of open source contribution. You requested to work on this issue after me and we have not heard from @TheDeveloperDino. I have also started working on the issue.

[jiezheng5]

@jiezheng5, this is not really in the spirit of open source contribution. You requested to work on this issue after me and we have not heard from @TheDeveloperDino. I have also started working on the issue.

I actually started working on this issue 2 weeks ago. Then I saw your request, I thought maybe I should also post my progress. Do we have to first get permission before we can work on the open issues? This is my first time contributing to open source, I am sorry that if I have done something wrong.

[koppor]

Example DOI for this behavior: 10.3218/3846-0. It is a valid DOI string - but not not found on doi org: https://doi.org/10.3218/3846-0 returns 404.

When a user fetches the DOI data, "Connection error" is returned. The improved implementation should check for 404 and handle that as "No DOI data existing".

Side note: The DOI is listed at https://vdf.ch/studienarbeiten-e-book.html - and thus users may think, it is a valid one. However, the publisher seems not to have registered it at the official DOI registry.

[koppor]

I fixed this issue by adding a method to automatically remove non-valid characters at the end of DOI string

@mrcstan Could you add a test case to org.jabref.model.entry.identifier.DOITest? I think, the solution is found by working on org.jabref.model.entry.identifier.DOI#DOI_EXP, which seems to include more characters than needed in the resulting DOI?

[koppor]

Then I saw your request, I thought maybe I should also post my progress.

This is actually a good idea. We also outlined that at https://devdocs.jabref.org/contributing#create-a-pull-request:

If you want to indicate that a pull request is not yet complete before creating the pull request, you may consider creating a draft pull request.

Do we have to first get permission before we can work on the open issues?

No, just start working. The general reason is that the probability at 500+ opened issues that two volunteers start working on the same issue at the same time is nearly 0. Maybe in the concrete setting, this issue was a) tagged as "good first issue" and b) at the very beginning of the list.

I am sorry that we have this situation now. I reviewed #8215 and think, it would be good to see an alternative solution.

Proposal:

1. @mrcstan Submits his solution as WIP PR
2. The JabRef developers compare the two solutions and decide which one is closer to a solution based on our 15+ years experience in Java development

Would that be OK?

This is my first time contributing to open source,

Welcome to the open source development world!

I am sorry that if I have done something wrong.

You "just" hit an edge case in the open source world. We are now working to resolve this case. Meanwhile, you could work on another issue, maybe koppor#534?

[koppor]

This PR has two parts:

1. Invalid characters at a DOI --> fixed by DOI determining now uses "DOI.findInText" (and not "DOI.parse") #8227
2. DOI is a valid one, but not returned by the server --> still open

[mrcstan]

@koppor , i just submitted a PR #8228 addressing 1 and 2. I found a case where #8227 did not work and mentioned that in that PR.