Show ZWSP (zero width space) as code mining

[marcushoepfner]

Issue: #1002

For more information refer to https://www.compart.com/en/unicode/U+200B To get a ZWSP refer to https://zerowidthspace.me/

[github-actions]

Test Results

   852 files  ±0     852 suites  ±0   53m 43s ⏱️ - 3m 5s
 7 844 tests ±0   7 601 ✅ +1  243 💤 ±0  0 ❌  - 1 
20 058 runs  ±0  19 402 ✅ +1  656 💤 ±0  0 ❌  - 1 

Results for commit a0b6352. ± Comparison against base commit b819cf9.

♻️ This comment has been updated with latest results.

[BeckerWdf]

I did a quick manual test on macOS. I found the following issues:

1. code mining is not displayed emidiately. I have to close and reopen the editor for the code minings to be shown
2. Two ZWSP next to each other are drawn one the (almost?) same location. See line 4 in:

3. When select the "a" and the "ZWSP" in line 2 from the above screenshot to with the keyboard I have to press the right-cursor two times until the caret is behind the "ZWSP".

[BeckerWdf]

I manually tested your change.

the ZWSP is shown correctly:

but when I disable the "show whitespace character" function (e.g. via the toolbar button) the ZWSP code mining still not disappears (even though the other "white space character" stuff does correctly disappear:

[BeckerWdf]

See my comments above

[BeckerWdf]

another topic:
a user can change the color of the code minings in the "fonts and colors" preference page. If you do this you will see a mix of colors:

Maybe this can be improved:

• either by ignoring the users color pref here
• also use the code mining color from preferences for the other "white space character" drawing logic.

[marcushoepfner]

I manually tested your change.

the ZWSP is shown correctly:

but when I disable the "show whitespace character" function (e.g. via the toolbar button) the ZWSP code mining still not disappears (even though the other "white space character" stuff does correctly disappear:

fixed with the last commit.

[marcushoepfner]

another topic: a user can change the color of the code minings in the "fonts and colors" preference page. If you do this you will see a mix of colors:

Maybe this can be improved:

• either by ignoring the users color pref here
• also use the code mining color from preferences for the other "white space character" drawing logic.

I tried to ignore and overwrite the color with the one TextEditor uses for drawing spaces, tabs,...
According to

eclipse.platform.ui/bundles/org.eclipse.jface.text/src/org/eclipse/jface/text/WhitespaceCharacterPainter.java

Line 447 in 2c5b605

fg= fTextWidget.getForeground();

this is SWT.COLOR_LIST_FOREGROUND, which is black.
I even debugged the drawing of a tab and it was colored 0, 0, 0 but appeared grey. I dont get it....

[BeckerWdf]

I just played around a bit with the current state.
I got this:

java.lang.NullPointerException: Cannot invoke "org.eclipse.swt.widgets.Display.getSystemColor(int)" because the return value of "org.eclipse.swt.widgets.Display.getCurrent()" is null
	at org.eclipse.ui.internal.texteditor.codemining.ZeroWidthCharactersLineContentCodeMining.<init>(ZeroWidthCharactersLineContentCodeMining.java:40)
	at org.eclipse.ui.internal.texteditor.codemining.ZeroWidthCharactersLineContentCodeMiningProvider.createCodeMining(ZeroWidthCharactersLineContentCodeMiningProvider.java:108)
	at org.eclipse.ui.internal.texteditor.codemining.ZeroWidthCharactersLineContentCodeMiningProvider.provideCodeMinings(ZeroWidthCharactersLineContentCodeMiningProvider.java:72)
	at org.eclipse.jface.internal.text.codemining.CodeMiningManager.lambda$2(CodeMiningManager.java:194)
	at java.base/java.util.stream.ReferencePipeline$3$1.accept(ReferencePipeline.java:197)

[BeckerWdf]

fixed with the last commit.

Thanks. That looks good now.

[BeckerWdf]

I just played around a bit with the current state. I got this:

java.lang.NullPointerException: Cannot invoke "org.eclipse.swt.widgets.Display.getSystemColor(int)" because the return value of "org.eclipse.swt.widgets.Display.getCurrent()" is null
	at org.eclipse.ui.internal.texteditor.codemining.ZeroWidthCharactersLineContentCodeMining.<init>(ZeroWidthCharactersLineContentCodeMining.java:40)
	at org.eclipse.ui.internal.texteditor.codemining.ZeroWidthCharactersLineContentCodeMiningProvider.createCodeMining(ZeroWidthCharactersLineContentCodeMiningProvider.java:108)
	at org.eclipse.ui.internal.texteditor.codemining.ZeroWidthCharactersLineContentCodeMiningProvider.provideCodeMinings(ZeroWidthCharactersLineContentCodeMiningProvider.java:72)
	at org.eclipse.jface.internal.text.codemining.CodeMiningManager.lambda$2(CodeMiningManager.java:194)
	at java.base/java.util.stream.ReferencePipeline$3$1.accept(ReferencePipeline.java:197)

This is now also fixed. Thanks,

[marcushoepfner]

I just played around a bit with the current state. I got this:

java.lang.NullPointerException: Cannot invoke "org.eclipse.swt.widgets.Display.getSystemColor(int)" because the return value of "org.eclipse.swt.widgets.Display.getCurrent()" is null
	at org.eclipse.ui.internal.texteditor.codemining.ZeroWidthCharactersLineContentCodeMining.<init>(ZeroWidthCharactersLineContentCodeMining.java:40)
	at org.eclipse.ui.internal.texteditor.codemining.ZeroWidthCharactersLineContentCodeMiningProvider.createCodeMining(ZeroWidthCharactersLineContentCodeMiningProvider.java:108)
	at org.eclipse.ui.internal.texteditor.codemining.ZeroWidthCharactersLineContentCodeMiningProvider.provideCodeMinings(ZeroWidthCharactersLineContentCodeMiningProvider.java:72)
	at org.eclipse.jface.internal.text.codemining.CodeMiningManager.lambda$2(CodeMiningManager.java:194)
	at java.base/java.util.stream.ReferencePipeline$3$1.accept(ReferencePipeline.java:197)

This is now also fixed. Thanks,

Is the color of "ZWSP" grey or black in your case?

[BeckerWdf]

I tried to ignore and overwrite the color with the one TextEditor uses for drawing spaces, tabs,... According to

eclipse.platform.ui/bundles/org.eclipse.jface.text/src/org/eclipse/jface/text/WhitespaceCharacterPainter.java

Line 447 in 2c5b605

fg= fTextWidget.getForeground();

this is SWT.COLOR_LIST_FOREGROUND, which is black.
I even debugged the drawing of a tab and it was colored 0, 0, 0 but appeared grey. I dont get it....

I also just debugged this. And I see that the color (for the whitespace chars comes from the styleRange not the textWidget:

eclipse.platform.ui/bundles/org.eclipse.jface.text/src/org/eclipse/jface/text/WhitespaceCharacterPainter.java

Line 449 in 2c5b605

fg= styleRange.foreground;

[BeckerWdf]

Is the color of "ZWSP" grey or black in your case?

black.

[tobiasmelcher]

The WhitespaceCharacterPainter uses in the light theme the black color Color {0, 0, 0, 255} to draw the invisible characters with gc.drawString. It looks greyish because gc.setAlpha(fAlpha); is called beforehand with value fAlpha=80.
Here the location where setAlpha is called.

private void handleDrawRequest(GC gc, int x, int y, int w, int h) {
		int startLine= fTextWidget.getLineIndex(y);
		int endLine= fTextWidget.getLineIndex(y + h - 1);
		if (startLine <= endLine && startLine < fTextWidget.getLineCount()) {

			// avoid painting into the margins:
			Rectangle clipping= gc.getClipping();
			Rectangle clientArea= fTextWidget.getClientArea();
			int leftMargin= fTextWidget.getLeftMargin();
			int rightMargin= fTextWidget.getRightMargin();
			clientArea.x+= leftMargin;
			clientArea.width-= leftMargin + rightMargin;
			clipping.intersect(clientArea);
			gc.setClipping(clientArea);
			if (fIsAdvancedGraphicsPresent) {
				int alpha= gc.getAlpha();
				gc.setAlpha(fAlpha); // <!------------------------------------
				drawLineRange(gc, startLine, endLine, x, w);
				gc.setAlpha(alpha);
			} else {
				drawLineRange(gc, startLine, endLine, x, w);
			}
			gc.setClipping(clipping);
		}
	}

We could do the same in the ZeroWidthCharactersLineContentCodeMining so that all whitespaces are rendered in the same color. Maybe introduce a re-usable in WhitespaceCharacterPainter to retrieve the color config which can then be reused by the code mining.

[tobiasmelcher]

​[c5d21c1] sets the same color in the ZWSP code minings as it is done in the whitespace character painter

[marcushoepfner]

Interesting, the current version in main branch is 3.20

And that is the problem

But the API warning says something about 3.19...

Because we don't increase minor version twice for the same release. It was already bumped in 5f6cae6.

So please undo the version increment in manifest and change since tag to 3.20. After that there will be still API error on org.eclipse.ui.texteditor.AbstractTextEditor.PREFERENCE_SHOW_ZW_CHARACTERS, and you can add a filter for this error in the editor via Ctrl+1.

Also would be good if you could squash all commits to a single one and after that rebase on latest master branch.

I don't see the quickfix/error.
I used 2026-03 RC2 as baseline and did not change any settings. I can even see that org.eclipse.ui.workbench.texteditor already has one filter.

Is there any documentation on how to setup the baseline?
I followed the help, but it did not solve my problem.
Contributing.md does not contain information regarding baseline setup. I wonder why not, as this seems to be an essential step in Eclipse contribution.

[marcushoepfner]

Interesting, the current version in main branch is 3.20

And that is the problem

But the API warning says something about 3.19...

Because we don't increase minor version twice for the same release. It was already bumped in 5f6cae6.
So please undo the version increment in manifest and change since tag to 3.20. After that there will be still API error on org.eclipse.ui.texteditor.AbstractTextEditor.PREFERENCE_SHOW_ZW_CHARACTERS, and you can add a filter for this error in the editor via Ctrl+1.
Also would be good if you could squash all commits to a single one and after that rebase on latest master branch.

I don't see the quickfix/error. I used 2026-03 RC2 as baseline and did not change any settings. I can even see that org.eclipse.ui.workbench.texteditor already has one filter.

Is there any documentation on how to setup the baseline? I followed the help, but it did not solve my problem. Contributing.md does not contain information regarding baseline setup. I wonder why not, as this seems to be an essential step in Eclipse contribution.

@merks Can you help here?

.

[iloveeclipse]

This change introduces memory leak, please check eclipse-jdt/eclipse.jdt.ui#2856.

[BeckerWdf]

@marcushoepfner: Can you pls. have a look?

[marcushoepfner]

It seems that the dipose call on CodeMiningProviders is completely missing from text editor.
Only Java does it for its own providers in JavaEditor.dispose()

[BeckerWdf]

JavaEditor.dispose()

So you say it's not a regression of this PR but this PR only did uncover and pre-existing issue?

[marcushoepfner]

JavaEditor.dispose()

So you say it's not a regression of this PR but this PR only did uncover and pre-existing issue?

Seems so. I have a local solution already and will create a PR later.
Code Mining "experts" need to check it then.

[BeckerWdf]

Code Mining "experts" need to check it then.

We can ask @tobiasmelcher to review it.

[marcushoepfner]

Provided a fix in #3770 .
Let's stop the discussion here and move over to #3770 and eclipse-jdt/eclipse.jdt.ui#2856.

[BeckerWdf]

@marcushoepfner: Should we add this to the New & Noteworthy document?

[marcushoepfner]

@marcushoepfner: Should we add this to the New & Noteworthy document?

Oh yes. Any documentation on how to do that?

[BeckerWdf]

Oh yes. Any documentation on how to do that?

It's about editing https://github.com/eclipse-platform/www.eclipse.org-eclipse/blob/master/news/4.40/platform.md