test_runner: exclude ignored lines from BRDA in lcov output

[RajeshKumar11]

Summary

When using /* node:coverage ignore next */ comments, the coverage system correctly excluded lines from DA (line coverage) entries in LCOV output but still reported branches as uncovered (BRDA count=0).

This caused branch coverage to show incorrect percentages (e.g., 66% instead of 100%) when the uncovered branch path contained ignored lines, impacting CI/CD pipelines that enforce branch coverage thresholds.

Changes

The fix treats branches as covered when they have count === 0 but contain ignored lines (ignoredLines > 0). This matches the behavior of c8 and ensures ignored code doesn't penalize branch coverage.

Before:

BRDA:4,2,0,0  # uncovered branch at line 4
BRH:2         # only 2/3 branches hit (66%)

After:

BRDA:4,2,0,1  # branch treated as covered due to ignore comment
BRH:3         # all 3/3 branches hit (100%)

Test plan

• Added regression test in test/parallel/test-runner-coverage.js
• Test verifies that BRDA entries don't show 0 for ignored branches
• Test verifies BRH equals BRF when ignored code is not executed

Fixes: #61586

[nodejs-github-bot]

Review requested:

• @nodejs/test_runner

[JakobJingleheimer]

Thanks for this!

I'm not familiar with LCOV itself, but the logic looks right.

[JakobJingleheimer]

nit: no mangling please :)

Suggested change

	assert.strictEqual(brfMatch[1], brhMatch[1],
	`All branches should be covered when ignored code is not executed. ` +
	`BRF=${brfMatch[1]}, BRH=${brhMatch[1]}`);
	assert.strictEqual(
	brfMatch[1],
	brhMatch[1],
	`All branches should be covered when ignored code is not executed. BRF=${brfMatch[1]}, BRH=${brhMatch[1]}`,
	);

[JakobJingleheimer]

nit: no mangling please :)

Suggested change

	assert.notStrictEqual(count, '0',
	`No branch should show 0 coverage when the ` +
	`uncovered path is ignored: ${entry}`);
	assert.notStrictEqual(
	count,
	'0',
	`No branch should show 0 coverage when the uncovered path is ignored: ${entry}`,
	);

[JakobJingleheimer]

Suggested change

	const brfMatch = sourceSection.match(/BRF:(\d+)/);
	const brhMatch = sourceSection.match(/BRH:(\d+)/);
	assert(brfMatch, 'LCOV should contain BRF');
	assert(brhMatch, 'LCOV should contain BRH');
	assert.match(sourceSection, /BRF:(\d+)/, 'LCOV should contain BRF');
	assert.match(sourceSection, /BRH:(\d+)/, 'LCOV should contain BRH');

[JakobJingleheimer]

nit:

Suggested change

	const sourceSection = lcov.split('end_of_record')
	.find((s) => s.includes('source.js'));
	const sourceSection = lcov
	.split('end_of_record')
	.find((s) => s.includes('source.js'));

or

Suggested change

	const sourceSection = lcov.split('end_of_record')
	.find((s) => s.includes('source.js'));
	const sourceSection = lcov.split('end_of_record').find((s) => s.includes('source.js'));

[JakobJingleheimer]

Suggested change

	const branchCount = (range.count === 0 && range.ignoredLines > 0) ?
	1 : range.count;
	const branchCount = (range.count === 0 && range.ignoredLines > 0) ? 1 : range.count;

[RajeshKumar11]

Applied the formatting suggestions, thanks for the review!

[JakobJingleheimer]

Thanks! I'd like to get someone more familiar with LCOV syntax to review/approve before landing.