01Brief05 / 10

Turn CSV rows into records

Write csv_to_records(rows) that takes a list of rows — the first being the header — and returns a list of dictionaries keyed by the column names.

Function	csv_to_records()
Track	Files and formats
Difficulty
Time	~20 min
Concepts	zip header rows refusing ragged data
Cases	9

02The contract

The first row is the header. Every row after it becomes one dictionary.
Values stay strings. Converting types is a separate decision, made later.
Rows keep their order, and each dictionary keeps the column order of the header.
A header with no data rows gives an empty list.
No rows at all gives an empty list — not an error.
A row with a different number of fields from the header is corrupt data: raise ValueError saying which row and what the counts were.

03Why this one

A spreadsheet export is a list of lists, and almost nothing you want to do with it is pleasant in that shape — row[3] tells a reader nothing, and it silently means something different the day somebody inserts a column. Turning the rows into dictionaries keyed by the header is the first thing any data pipeline does. The interesting decisions are what to do with a row that has the wrong number of fields, and whether zip’s famous silence is a feature here or a trap.

—The concept behind it

Modules on pythonmastery.io →

That is the reading half. Read it there if the idea is new, then come back here — explanation is what makes something make sense, and practice is what makes it stay.

—Worth doing first

Group records by a field

Not a gate — start this one whenever you like. These cover ideas it leans on, and a tick means you have already done them.

—What this leads to

These build on what you are about to write here.

your_code.py

05One way to write it

def csv_to_records(rows):
    if not rows:
        return []

    header = rows[0]
    records = []
    for number, row in enumerate(rows[1:], start=2):
        if len(row) != len(header):
            raise ValueError(
                f"row {number} has {len(row)} fields, "
                f"but the header has {len(header)}"
            )
        records.append(dict(zip(header, row)))
    return records

start=2 makes the error message match the file

Line 1 is the header, so the first data row is line 2 — and that is the number somebody will type into their editor to go and look. enumerate takes a start argument precisely for this. An error saying 'row 0' sends the reader to the wrong place, which is worse than no number at all.

The length check is the whole exercise

Everything else is one well-known idiom. The check exists because zip’s silence is a good default for pairing things you control and a terrible one for parsing a file you do not. Knowing which of your tools fail quietly, and putting a guard in front of those, is most of what defensive parsing is.

Both counts in the message, not just the fact of a mismatch

'row 4 has 3 fields, but the header has 4' tells you what happened; 'malformed row' tells you only that something did. The extra few characters are written once and read every time the export breaks.

06When not to write this at all

Do not parse real CSV text with split(',') and this function — that is what the csv module is for, and csv.DictReader does exactly this job including quoted fields containing commas, embedded newlines, escaped quotes and the several dialects spreadsheets emit. This exercise starts from rows that are already split precisely so the quoting problem is out of scope. For anything beyond a few thousand rows of numeric data, pandas.read_csv or polars will be faster and give you types as well. Write this when you have rows in hand and want them keyed.

07Take it further

Add a strict=False option that pads short rows with empty strings instead of raising, and decide which behaviour a nightly import job should default to.
Detect duplicate column names in the header — dict() silently keeps only the last — and decide whether that deserves an error or a rename.
Make it lazy with yield so a million-row export never has to be in memory as a list, and notice what that costs a caller who wanted len().