Repository navigation
JSONObject and JSONArray initial flag for tokener may be flawed #944
Description
From #927:
The boolean isInitial = x.getPrevious() == 0; check can behave in an unexpected way, for example:
JSONParserConfiguration strict = new JSONParserConfiguration().withStrictMode();
JSONTokener tok = new JSONTokener("[]xxx");
// Calling next+back changes "previous"; without these calls the check is performed:
tok.next();
tok.back();
JSONArray a1 = new JSONArray(tok, strict); // No error here, but should be because of extra chars?
This is a far corner case I think, but should be fixed if practical.
Is there a way to capture this in the JSONTokener object?
Activity
Md-Yasir commentedon Oct 11, 2025
I am looking into this @stleary and this is going to be my first open-source contribution. 😙
stleary commentedon Oct 11, 2025
@Md-Yasir This might be a tough issue to start with; there may not be a solution that works without breaking existing functionality.
You are also welcome to start working on some issues that are more approachable. SonarQube has listed > 1000 issues to be addressed. Most of them are in the test files and not a priority. But about 100 issues are in the production code. Most are easy to fix and test. A few will need to be marked as false positive.
All Open issues
To help get you started, this is how previous work has gone:
False positives (Don't work on these type of issues)
Fixed (How similar issues were fixed)
Md-Yasir commentedon Oct 11, 2025
Got it, So should I use Github or sonarqube for the updates? I haven't used sonarqube yet. Thanks
stleary commentedon Oct 11, 2025
All work should be done in GitHub. Just use SonarQube as a reporting tool, to identify the issues to fix.
Md-Yasir commentedon Oct 11, 2025
Understood.. Here is my first pull req
XIAYM-gh commentedon Sep 2, 2026
After viewing the current code in
masterbranch, I found that theisInitialvariable is used to determine whether the object/array is the root, and to determine whether to throw the error.I have an immature idea: why not just use the current index of the tokener?
Hmm, I will test this locally. If it passes the test suite, and hopefully, doesn't break the compatibility, maybe I can open a pull request for reviewing.
XIAYM-gh commentedon Sep 2, 2026
Well, no. This changes the internal behaviour!
I will search for a more reasonable solution...