fix(ENGKNOW-3981): refbases below position 1 returns N instead of throwing

[gmagnu]

Problem

refbases(chrom, pos1, pos2) (and refbases_with_build) fails the whole query with Index -1 out of bounds for length 10000 when pos1 is 0 and pos2 > 0, for example gorrow chr1,0 | calc r refbases(chrom,pos,pos+3). This can happen with rows that carry position 0, such as unmapped variants returned by liftover. refbase/single-base reads at position 0 already returned N.

Cause

In RefSeqFromChromSeq.getBases, the same-buffer check uses (pos - 1) / buffLength, which truncates toward zero, so position 0 maps to buffer 0. The fast path then indexes lastBuff(pos1 - offset - 1), i.e. lastBuff(-1). getBase catches the exception and returns N, but getBases does not.

Fix

Positions below 1 are treated like positions past the chromosome end: each reads as N and the result keeps length pos2 - pos1 + 1.

• getBases emits N for pos1..min(pos2, 0) and continues from position 1, before any buffer lookup.
• getBase returns N for pos < 1 up front instead of relying on the caught exception (which also avoids seeking to a negative offset for positions below -9999).

Positions >= 1 go through exactly the same code as before.

Tests

• UTestRefSeqFromChromSeq.testGetRefbasesBelowPositionOne: getBases(chr17, 0, 4) = NAAGC (fresh and warm buffer), (-5, -1) = NNNNN, (0, 0) = N, getBase at 0 / -1 / -10001 = N, a range from -2 crossing the 10 kB buffer boundary, a range from more than one buffer below 1, and unchanged positive-position reads.
• UTestGenomicFunctions.testRefBasesBelowPositionOne / testRefBasesWithBuildBelowPositionOne: gorrow chr17,0 | calc r refbases(...) / refbases_with_build(...) return NAAG.

The new tests failed before the fix with the index error. :model:test, :gortools:test pass.

Note

A separate pending change for refbases on contigs without sequence (ENGKNOW-3978) edits the same-buffer fast path in getBases and the buffer load in getBase. A trial merge shows RefSeqFromChromSeq.scala auto-merges (this PR only adds early returns at the top of both methods); UTestRefSeqFromChromSeq.java has a trivial conflict because both add a new test method at the same place. Keep both methods.

🤖 Generated with Claude Code

[github-actions]

Junit Tests - Summary

4 912 tests  +3   4 741 ✅ +3   20m 39s ⏱️ - 1m 6s
  509 suites ±0     171 💤 ±0 
  509 files   ±0       0 ❌ ±0 

Results for commit db48da6. ± Comparison against base commit b79342b.