Check a sample before importing the whole file.

Create fictional columns id, name and quantity. Use A01/Ana/2, A02/Luis/empty, A01/Ana/2 and A03/Marta/two. The sample contains a complete row, a missing value, a repetition and a quantity written as text.

Keep the original. Before correcting anything, decide whether “two” is allowed or needs review and whether an empty quantity means unknown. Converting it to zero without agreement changes its meaning.

For this exercise, id is required unique text, name is required text and quantity must be a positive integer. These are exercise decisions, not universal rules for every CSV.

Define how to detect repeated rows and what happens when the same id carries different values. An identical repetition and conflicting information require different decisions.

RFC 4180 documents a CSV convention using commas and quoted fields. Check your tools’ actual export and import settings rather than assuming every tool uses the same configuration.

Add a name containing a comma and another with an accented character. Verify that the name stays in one column and characters survive. Also check that identifiers such as 001 retain leading zeros.

Under the proposed rules, A01 is accepted once, A02 awaits a missing quantity and A03 requires explicit correction. Keep a reason for each rejected row so someone can resolve it.

Import the sample into a test destination first. Compare row counts, identifiers and values with expectations. Expand only after explaining each difference, retaining a way to identify records belonging to this import.