Found the bug in ASTNode.toFomula()

[tyzerrr]

Hey, all.
I used jsbml-1.6.1, current latest version of jsbml, but I found the method, ASTNode.toFormula() fails to parse the formula.
To be specific, for testcase, I downloaded BIOMD0000000022.xml from BioModels(https://www.ebi.ac.uk/biomodels/search?query=BIOMD0000000022&domain=biomodels), and I wrote the program in C with libsbml.5.18.0 and java with jsbml-1.6.1.
I attached the both result as libsbml-formula.txt and jsbml-formula.txt, so I wanna you'll check these, then you should compare the first line.

libsbml-formula.txt of first line

formula1: compartment_0000003 * (c1 + (B1 + pow(CCn / A1, a)) * s1 / (1 + B1 + pow(CCn / A1, a) + pow(PTn / r1, r)))

jsbml-formula.txt of first line

formula1: compartment_0000003*(c1+((B1+CCn/A1^a)*s1)/(1+B1+CCn/A1^a+PTn/r1^r))

They are completely different. And I checked the related report(Robust oscillations within the interlocked feedback model of Drosophila circadian rhythm., but in this report, result by libsbml is correct.

And, I also checked core/src/org/sbml/jsbml/ASTNode.java.newASTNode, this file was revised 2 years ago, but jsbml-1.6.1 was released in Oct 2022. So even in latest version, jsbml has the bug around ASTNode.toFormula().

Any help, please.

libsbml-formula.txt
jsbml-formula.txt

[luciansmith]

The trick here is that CCn/A1 needs to be in parentheses in the jsbml version of the formula:

i.e. instead of:

CCn/A1^a

it needs to produce:

(CCn/A1)^a

(It could produce the 'pow' version instead, if that's simpler.)

[tyzerrr]

@luciansmith Yes, that is what i wanna say!
jsbml-1.6.1 can't parse this formula correctly, I think.

[luciansmith]

You are correct! Thanks for reporting the issue; hopefully the jsbml team will be able to get to it soon.

[tyzerrr]

I'm gonna to try fix this bug!

[luciansmith]

Good luck! Feel free to post here with questions if you have any.