[Code scan] Reject zero interpolation counts in line-mode KPOINTS

[njzjz]

This issue is a result of a Codex global repository scan.

Line-mode KPOINTS parsing allows nkl[iks] == 0 via assert(nkl[iks] >= 0). Later, each segment divides by nkl[iks - 1], so a zero count causes division by zero. In release builds, the assert is removed entirely.

abacus-develop/source/source_cell/klist.cpp

Lines 440 to 465 in 84ca04b

	for (int iks = 0; iks < nks_special; iks++)
	{
	ifk >> ks[iks].x;
	ifk >> ks[iks].y;
	ifk >> ks[iks].z;
	ModuleBase::GlobalFunc::READ_VALUE(ifk, nkl[iks]);
	assert(nkl[iks] >= 0);
	nkstot += nkl[iks];
	/* ISSUE#3482: to distinguish different kline segments */
	if ((nkl[iks] == 1) && (iks != (nks_special - 1))) {
	kpt_segid++;
	}
	kpt_segids.push_back(kpt_segid);
	}
	assert(nkl[nks_special - 1] == 1);
	// std::cout << " nkstot = " << nkstot << std::endl;
	this->renew(nkstot * nspin); // mohan fix bug 2009-09-01
	int count = 0;
	for (int iks = 1; iks < nks_special; iks++)
	{
	double dxs = (ks[iks].x - ks[iks - 1].x) / nkl[iks - 1];
	double dys = (ks[iks].y - ks[iks - 1].y) / nkl[iks - 1];
	double dzs = (ks[iks].z - ks[iks - 1].z) / nkl[iks - 1];

Relevant code:

ModuleBase::GlobalFunc::READ_VALUE(ifk, nkl[iks]);

assert(nkl[iks] >= 0);
nkstot += nkl[iks];
...
double dxs = (ks[iks].x - ks[iks - 1].x) / nkl[iks - 1];
double dys = (ks[iks].y - ks[iks - 1].y) / nkl[iks - 1];
double dzs = (ks[iks].z - ks[iks - 1].z) / nkl[iks - 1];

Suggested fix:

Validate nkl[iks] > 0 for all segment counts used as divisors and report an input error at runtime rather than relying on debug-only assertions.