pr: stop dropping lines the columns do not divide evenly

[luantaraschi]

pr in column mode drops whatever the columns cannot divide evenly.

$ seq 9 | pr -t -2 -w 20      # GNU coreutils 9.7
1         6
2         7
3         8
4         9
5

$ seq 9 | pr -t -2 -w 20      # uutils, main at 9875296
1               5
2               6
3               7
4               8

The ninth line is gone. A page holding fewer lines than it has columns comes out empty instead:

$ printf 'a\n' | pr -t -3 | wc -c
2      # GNU
0      # uutils

seq 9 | pr -2 -l 12 ends the same way, on an empty page 3 where GNU prints the ninth line.

It comes down to one line, src/uu/pr/src/pr.rs:1408:

let num_rows = lines.len() / columns;

Integer division truncates, so nine lines over two columns give four rows and the ninth never gets a cell.

Rounding up gets closer but still misses, because GNU balances the columns: each one takes the ceiling of what is still unplaced over the number of columns left to fill, so nine lines over four columns come out 3, 2, 2, 2 rather than 3, 3, 3, 0. I read that rule off the output of /usr/bin/pr rather than its source, then checked it for every column count from 1 to 7 against every input length from 0 to 39, down and across. 560 cases, no mismatch.

Two smaller things came with the fix. The short-file branch was tested before the across-mode one, so pr -a laid a partly filled page out downwards; and the last row of such a page stops part way through, so it needs its line separator without a trailing column separator.

What I ran

All of it in a Debian container where /usr/bin/pr is GNU coreutils 9.7.

I ran 425 combinations of options and input length against GNU, this branch, and a build of main. main agreed with GNU in 138 of them and this branch agrees in 370, so 232 newly agree and none regressed. The comparison ignores column padding: uutils does not match GNU there, and test_columns already carries a TODO about it.

Another 574 combinations went through a line count check, comparing how many input lines survive the round trip. None lost.

cargo test --features pr --no-default-features test_pr passes 83. The four new tests fail on main and pass here, the 79 that were already there are untouched, and no expected fixture changed.

cargo fmt -p uu_pr -- --check and cargo clippy -p uu_pr --all-targets come back clean. clippy -D warnings still trips on the generated uucore/embedded_locales.rs, which fails the same way for a crate I did not touch.

I left the padding alone. GNU pads with tabs and leaves the last column unpadded while uutils pads with spaces, which is a different problem.

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

[codspeed-hq]

Merging this PR will degrade performance by 0.4%

⚠️ Different runtime environments detected

Some benchmarks with significant performance changes were compared across different runtime environments,
which may affect the accuracy of the results.

Open the report in CodSpeed to investigate

⚡ 2 improved benchmarks
❌ 2 regressed benchmarks
✅ 353 untouched benchmarks
⏩ 50 skipped benchmarks¹

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
❌	Simulation	du_summarize_balanced_tree[(5, 4, 10)]	16 ms	16.8 ms	-4.69%
❌	Simulation	du_wide_tree[(5000, 500)]	19.4 ms	20.3 ms	-4.1%
⚡	Simulation	cksum_crc32b	39.7 ms	38.1 ms	+4.19%
⚡	Simulation	cksum_sysv	68.4 ms	66.2 ms	+3.32%

Tip

Investigate this regression by commenting @codspeedbot fix this regression on this PR, or directly use the CodSpeed MCP with your agent.

---

_{Comparing luantaraschi:fix/pr-column-partial-page (2be25be) with main (9875296)}

Footnotes

1. 50 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩

[github-actions]

GNU testsuite comparison:

Skip an intermittent issue tests/date/date-locale-hour (fails in this run but passes in the 'main' branch)
Skipping an intermittent issue tests/cut/bounded-memory (passes in this run but fails in the 'main' branch)
Skipping an intermittent issue tests/pr/bounded-memory (passes in this run but fails in the 'main' branch)
Skipping an intermittent issue tests/tail/tail-n0f (passes in this run but fails in the 'main' branch)
Note: The gnu test tests/cut/cut-huge-range is now being skipped but was previously passing.
Note: The gnu test tests/seq/seq-epipe is now being skipped but was previously passing.