Choose a composite CSV key when one identifier is not unique

By Batu · English technical notes

Also published in our Blogger archive.

Quick answer

A tuple is an ordered, hashable Python value, which makes it suitable for a declared composite key.

Example

The loop forms (region, id) for every row. The second EU/7 repeats that exact tuple, while US/7 remains distinct even though its ID is the same. The assertion demonstrates the distinction directly.

import csv, io
rows = list(csv.DictReader(io.StringIO('region,id\nEU,7\nUS,7\nEU,7\n')))
seen = set()
duplicates = []
for row in rows:
    key = (row['region'], row['id'])
    if key in seen:
        duplicates.append(key)
    seen.add(key)
assert duplicates == [('EU', '7')]
print(duplicates)

Expected stdout:

[('EU', '7')]

Reading the result

Do not sort or concatenate components unless the contract calls for it. Sorting changes key meaning, and concatenation can create ambiguous values such as AB/C and A/BC.

Validate that both selected key headers exist before reading data. Otherwise a missing field can turn every tuple into an empty or default-like value and create a misleading duplicate flood.

The order of components appears in the output tuple and should appear in the contract too. A reader must be able to reproduce the same key without guessing field precedence.

Sources

- Python csv module documentation

Prepared with AI assistance. The example uses synthetic data; its stated limits apply.