CODING CHALLENGE · N°59

CSV → Schema Mapper

Medium FDEDataIntegrations

Every customer hands you a messy export. Before it can flow into your system, each row must be coerced to a schema — strings to ints, "true" to booleans — and the rows that don’t fit quarantined, not silently dropped. Build the mapper. Solve it in Python or TypeScript, with hidden tests.

The problem

Implement map_rows(rows, schema). rows is a list of dicts (parsed CSV records) and schema maps each required field name to a type: "int", "float", "bool", or "str". For each row, coerce every schema field to its type. A row is valid only if all schema fields are present and every value coerces; otherwise it counts as an error. Return {"valid": [coerced rows], "errors": count}.

EXAMPLE 1

Input rows = [{"age":"30","name":"al"},{"age":"x","name":"bo"},{"name":"cy"}], schema = {"age":"int","name":"str"}

Output {"valid": [{"age": 30, "name": "al"}], "errors": 2}

row 2 fails coercion ("x"→int), row 3 is missing "age"

CONSTRAINTS

Coerce: int("30")→30, float("1.5")→1.5, bool: "true"/"1"→True, "false"/"0"→False (anything else fails), str: leave as-is.
A missing schema field or a failed coercion makes the whole row an error (not partially valid).
Preserve field order as given in the schema; return the count of error rows, not the rows themselves.

SOLVE IT YOURSELF

Your turn — write it

Edit the stub, hit Run (or ⌘/Ctrl + Enter), and watch the hidden tests. Stuck? the hints are right above and Reveal solution is one click away.

YOUR TASK

Implement map_rows(rows, schema): for each row try to build a coerced record over the schema fields; on any missing field or bad value, count it as an error instead.

HINTS — 4 IDEAS

Write a coerce(value, type) helper that raises/throws on a bad value.
For each row, loop the schema fields in order; if a field is missing or coercion fails, mark the row an error.
Booleans are the tricky type: only a small set of strings are valid; everything else fails.
Return the valid coerced rows and just the count of errors.

CPython · WebAssembly

Approach, complexity & discussion — open after you solve

The approach

Map the incoming CSV’s columns onto your target schema. Match headers by normalized name (lowercased, trimmed, with a table of known aliases), coerce each value to the target type, and report what could not be mapped and which required columns are missing — failing loudly with a clear list rather than guessing.

Complexity

Time O(rows × columns).

Common mistakes

Exact header matching only — real CSVs vary in case, spacing, and naming, so normalize and allow aliases.
Silently dropping columns that did not map instead of reporting them.
Leaving everything as strings instead of coercing to the schema’s types.

Where this shows up

The customer sends a CSV that “almost” matches your schema — mapping and validating it at the boundary is core forward-deployed and integrations work. Turning “the column names don’t match” into an actionable mapping report beats a cryptic crash three layers downstream.

Finished this one? 0 / 75 Challenges done

Explore the topic

See this alongside everything else on the same subject — handbooks, system designs, challenges and tools, in one place.

Forward Deployed Engineering