Trim only what you mean.

Select the columns and characters

Copy this synthetic UTF-8 comma CSV with a header. The first ID has one ordinary ASCII space on each edge.

id,team,note
 0017 ,fern,keep  two
0017,fern,keep  two
0017,moss,keep  two

With ASCII edge trim selected for id only, expected resulting records are:

0017,fern,keep  two
0017,fern,keep  two
0017,moss,keep  two

Only U+0020 space and U+0009 tab at selected field edges are removed. Internal double spaces in keep two remain. Unselected fields remain unchanged.

Operation order changes duplicate results

If exact full-record deduplication follows trim, the second record is omitted because all its resulting fields equal the first record’s. Keep the first in source order. The third survives: sharing ID 0017 is not enough when its team is moss.

Expected retained records:

0017,fern,keep  two
0017,moss,keep  two

Without deduplication, all three records remain. Without trim, the first two are not exact duplicates. Count edits on retained and omitted records separately when reviewing a change ledger.

Not all whitespace is ASCII whitespace

A value \u00A00017\u00A0 (escaped notation for non-breaking spaces around the ID, not literal backslash characters) stays unchanged under this rule. So does a field containing only U+00A0; it is not an empty string.

An ASCII-space-only field becomes empty if selected for trimming. In Prooftable’s development build, a selected required check on a retained resulting field then blocks export. Diagnose the character and schema; don’t widen the rule without reviewing the consequences.