Follow-up: #11535

@Siedlerchr, @subhramit please fix this exception, JabRef is not starting 😓

ERROR: Unexpected exception: java.lang.NullPointerException: Cannot invoke "java.io.InputStream.readAllBytes()" because "source" is null
	at org.jabref@100.0.0/org.jabref.logic.citationstyle.CitationStyle.createCitationStyleFromSource(CitationStyle.java:67)
	at org.jabref@100.0.0/org.jabref.logic.citationstyle.CitationStyle.createCitationStyleFromFile(CitationStyle.java:121)
	at org.jabref@100.0.0/org.jabref.logic.citationstyle.CitationStyle.getDefault(CitationStyle.java:136)
	at org.jabref@100.0.0/org.jabref.preferences.JabRefPreferences.<init>(JabRefPreferences.java:728)
	at org.jabref@100.0.0/org.jabref.preferences.JabRefPreferences.getInstance(JabRefPreferences.java:875)
	at org.jabref@100.0.0/org.jabref.Launcher.main(Launcher.java:64)

IntelliJ:

Maybe this is because I didn't include the submodules in my clone?

Seems like, you need to run git submodule init

But we should nonetheless catch this

@Siedlerchr, @subhramit please fix this exception, JabRef is not starting 😓

ERROR: Unexpected exception: java.lang.NullPointerException: Cannot invoke "java.io.InputStream.readAllBytes()" because "source" is null
	at org.jabref@100.0.0/org.jabref.logic.citationstyle.CitationStyle.createCitationStyleFromSource(CitationStyle.java:67)
	at org.jabref@100.0.0/org.jabref.logic.citationstyle.CitationStyle.createCitationStyleFromFile(CitationStyle.java:121)
	at org.jabref@100.0.0/org.jabref.logic.citationstyle.CitationStyle.getDefault(CitationStyle.java:136)
	at org.jabref@100.0.0/org.jabref.preferences.JabRefPreferences.<init>(JabRefPreferences.java:728)
	at org.jabref@100.0.0/org.jabref.preferences.JabRefPreferences.getInstance(JabRefPreferences.java:875)
	at org.jabref@100.0.0/org.jabref.Launcher.main(Launcher.java:64)

Fixed in #11536, thanks for pointing out!

@antalk2 Hey, are you still on GitHub by any chance?

yes

yes

@antalk2 Hey, hey! It's so good to see your reply.
I am an undergraduate student currently contributing for GSoC to JabRef. My project involves work with the OO/LO integration so that CSL styles are now supported.
As you can see, the CSL Styles work now, but I am stuck in a major roadblock trying to get the reference numbers correct for numeric type citations. I tried to do it using reference marks (I know you have already done it for JStyles but I had difficulty adapting them so tried to do them independently from scratch the way Zotero has been doing it): https://github.com/zotero/zotero-libreoffice-integration/blob/main/build/source/org/zotero/integration/ooo/comp/ReferenceMark.java
The current problems that I face are (all pertaining to CSL):

1. The reference marks don't wrap around the citation text, but are inserted after the text.
2. [Major] On restarting JabRef, the numbering starts from 1 again for the same document. It is as if I am not reading the reference marks correctly when a connection is made, and getting the respective citation numbers. I tried to do lazy instantiation and everything, just isn't working. Could there be a problem with how I wrote readExistingMarks() in MarkManager?

If you get even a bit of time, could you take a look at subhramit#12 and let me know what I'm doing wrong?
CC: @Siedlerchr

Question 1: reference marks don't wrap around the citation text

1. The reference marks don't wrap around the citation text, but are
   inserted after the text.

In short: in src/main/java/org/jabref/logic/openoffice/oocsltext/CSLCitationOOAdapter.java

   void insertCitation(XTextDocument doc, XTextCursor cursor, BibEntry entry, String citation)

seems to create an empty reference mark, then insert text into it.
The opposite order: insert text to cursor and create a reference mark around the inserted text is probably easier.

The call ReferenceMark mark = markManager.createReferenceMark(entry, "ReferenceMark");
seems to ignore the XTextCursor cursor passed in to insertCitation.

Background

Both zotero (https://github.com/zotero/zotero-libreoffice-integration/blob/main/build/source/org/zotero/integration/ooo/comp/ReferenceMark.java)
and https://github.com/JabRef/jabref/blob/main/src/main/java/org/jabref/logic/openoffice/backend/NamedRangeReferenceMark.java does some "dancing"
with the cursors, inserting and removing paragraph marks and some other characters.

It has two reasons.

1. The UNO API presents the text and markup as a flat list of
   locations, not a tree of XML-ish nodes. In the flat
   representation an empty cursor points to between two characters.
   An empty text range also points to between two characters.
   The order of a cursor, range-start, range-end is not clear:
   the cursor could be before, inside or after the empty text range.
   Avoiding empty text ranges helps in controlling this order:
   if there is at least one character between to points their order is well defined.

   Attempt to do this seems to be missing from
   src/main/java/org/jabref/logic/openoffice/oocsltext/ReferenceMark.java

2. Controlling where text formatting is "inherited" from when inserting text.
   Text insertion usually inherits formatting from the left (previous character),
   except when inserting at the start of a paragraph (then from the right), and
   when inserting into an empty paragraph (probably from the paragraph mark on the right)

My recollection of what did I try and why.

I was first trying to control (1.) by keeping REFERENCE_MARK_LEFT_BRACKET and REFERENCE_MARK_RIGHT_BRACKET within the reference marks textrange.
Moving over these ZERO_WIDTH_SPACE characters with pushing the cursor arrow keys shows looks like the cursor stopping there, so I tried to minimize the use of these brackets by removing them if
there were enough characters between them.
I was trying to stabilize the reference marks (not deleting and recreating them). The user can
still remove these brackets and the text inside, leaving us an empty reference mark that needs handling.
Later I found that (2.) still needs handling (it probably happened in recreating or refilling the
References) and I started to inject another ZERO_WIDTH_SPACE with the intent to hold (keep) the text formatting applied.
This would allow the user to apply some (probably "direct") formatting to the
range (in particular including the initial ZERO_WIDTH_SPACE). When
we refresh the visible content of the reference mark, the new
content would inherit the formatting from this
ZERO_WIDTH_SPACE. Unfortunately this ZERO_WIDTH_SPACE is not
normally visible to the user, thus hard to reliably include in the range to be formatted "by hand".

Question 2: readExistingMarks

2. [Major] On restarting JabRef, the numbering starts from 1 again
   for the same document. It is as if I am not reading the
   reference marks correctly when a connection is made, and getting
   the respective citation numbers. I tried to do lazy
   instantiation and everything, just isn't working. Could there be
   a problem with how I wrote readExistingMarks() in MarkManager?

On restarting JabRef, the numbering starts from 1 again for the same document.
https://github.com/subhramit/jabref/pull/12/files#

I do not remember having a difficulty in getting and parsing reference mark names.

(Probably not relevant her: If I remember correctly, attempt to create a bookmark with a name already in use results
in a bookmark with a name different from what was requested (some numbering suffix added).
An attempt to create a reference mark with a name already in use results in two reference marks
with the same name, but name-based lookup will only yield one of these.)

If I understand correctly, you encode citationNumber in

public ReferenceMark createReferenceMark(BibEntry entry, String fieldType) 

like this

 String name = CSLCitationOOAdapter.PREFIXES[0] + citationKey + " RND" + citationNumber;

This would look like

"JABREF_{citationKey} RND{citationNumber}"

Parsing:

 private void updateCitationInfo(String name) {
    String[] parts = name.split(" ");
    if (parts.length >= 3) {

I do not see the 3 parts in the

"JABREF_{citationKey} RND{citationNumber}"

format. Does citationKey contain a space? Or a space is missing after "RND"?

Thank you for your invaluable guidance. I will try to tweak the implementation as per your comments.

Does citationKey contain a space? Or a space is missing after "RND"?

Good catch, I was missing appending a random string as the third part for uniqueness.

The code in

 src/main/java/org/jabref/logic/openoffice/backend/NamedRangeManagerReferenceMark.java
 src/main/java/org/jabref/logic/openoffice/backend/NamedRangeReferenceMark.java

seems not to depend on details of what is encoded in the reference mark name and the format of the text in the reference mark.

getFillCursor should provide an XTextCursor strictly within the reference mark, (surrounded by REFERENCE_MARK_LEFT_BRACKET and REFERENCE_MARK_RIGHT_BRACKET), and

The code in

 src/main/java/org/jabref/logic/openoffice/backend/NamedRangeManagerReferenceMark.java
 src/main/java/org/jabref/logic/openoffice/backend/NamedRangeReferenceMark.java

seems not to depend on details of what is encoded in the reference mark name and the format of the text in the reference mark.

jabref/src/main/java/org/jabref/logic/openoffice/backend/NamedRangeManagerReferenceMark.java

Line 30 in 072d7c7

public List<String> getUsedNames(XTextDocument doc)

does not even filter the reference mark names (for jabref-specific pefixes)

jabref/src/main/java/org/jabref/logic/openoffice/backend/NamedRangeReferenceMark.java

Line 202 in 072d7c7

public XTextCursor getFillCursor(XTextDocument doc)

getFillCursor should provide an XTextCursor strictly within the reference mark, (surrounded by REFERENCE_MARK_LEFT_BRACKET and REFERENCE_MARK_RIGHT_BRACKET), and

jabref/src/main/java/org/jabref/logic/openoffice/backend/NamedRangeReferenceMark.java

Line 375 in 072d7c7

public void cleanFillCursor(XTextDocument doc)

public void cleanFillCursor(XTextDocument doc) should clean up the brackets after modifying the content.

Okay, will look into these.
Update: I finally got the parsing correct, thanks to your hints. The marks are being read on restart now :)
The text wrap seems to be tricky, working on that currently.

Maybe this is because I didn't include the submodules in my clone?

Sub modules are "just" in the file .gitmodules (OK and some git internal object storage: https://stackoverflow.com/a/5033973/873282). One cannot omit a single file in a clone. As Chris pointed out: The initialization was missing 😅

Follow-up: #11535

Follow up: #11577

Follow-up: #11712