How can I handle CSV files that use a semicolon instead of a comma?

If you know this file is semicolon-delimited, pass the separator explicitly to Python’s CSV reader. The default delimiter is a comma, so a semicolon file can otherwise be split incorrectly. For file objects, open with newline='' as the CSV documentation recommends. [S1]

import csv

with open("[input.csv]", newline="", encoding="utf-8") as f:
    reader = csv.reader(f, delimiter=";")
    rows = list(reader)

for row in rows[:5]:
    print(row)

Replace [input.csv] with your local filename. This example reads the rows into memory, which is convenient for a small file; for a large file, inspect rows as you iterate instead of collecting all of them.

Print the header and compare the number of fields in each row. If you already know how many columns the file should contain, use that as a check; otherwise, look for consistent row lengths and sensible header names. A quick inspection might look like this:

if not rows:
    print("The file has no rows")
else:
    print("Header:", rows[0])
    print("Column counts:", [len(row) for row in rows[:5]])

For example, if the expected header is ['name', 'city', 'note'], a row like ['Ari', 'Oslo', 'works; remotely'] has three fields. Don’t split each line on ';' yourself; inspect the parsed rows instead. This check is an aid, not proof that the header or every row matches your expectations.

Python’s CSV module provides Sniffer.sniff(sample) to guess a format from a sample. But a guess from a short or ambiguous sample is not confirmation. The documentation’s example samples 1024 characters and still calls for checking the format against reliable knowledge of the file and inspecting parsed rows. [S1]

If you have a reliable file specification or know how it was exported, use that to choose the delimiter explicitly. If you test a guess, treat it as provisional: check the header, several representative rows, and the expected number of fields before using the data. If the evidence conflicts, pause and consult the file’s export settings or its provider rather than silently accepting a guess.

Here is a small pattern for a file whose first row is a header. Set expected_columns to the count you expect, or leave it as None to compare each data row with the header’s width:

import csv

expected_columns = 3  # Replace with your expected count, or use None.

with open("[input.csv]", newline="", encoding="utf-8") as f:
    reader = csv.reader(f, delimiter=";")
    header = next(reader, None)

    if header is None:
        print("No rows found")
    else:
        target = expected_columns or len(header)
        print("Header:", header)
        for line_number, row in enumerate(reader, start=2):
            if len(row) != target:
                print(f"Check row {line_number}: expected {target}, got {len(row)}")

This identifies rows worth reviewing; it does not establish that the chosen delimiter is correct or validate the meaning of each value. If your file has no header, choose an expected column count from a reliable description instead.