CSV

Read and write comma-separated tables with csv.reader, DictReader and DictWriter.

Hi! Python · Level 1: Files and data · Checked against the Python 3.12 documentation · Updated 2026-10-17

Think of it like this

A CSV file is like a paper register with columns. Each line is a row, and a comma separates the boxes on that row.

Where the analogy stops being true. On paper a box can hold anything, but in a CSV file a comma inside a box needs quoting, and by default everything you read back is text.

What to know

csv.reader gives each row as a list of strings. By default it does not turn "1" into a number.
csv.DictReader uses the first row as the field names and gives each row as a dict.
csv.DictWriter needs fieldnames. Call writeheader() once, then writerow or writerows.
A writer accepts any object with a write() method, so io.StringIO works for practice.
The default quoting is QUOTE_MINIMAL: a writer quotes only fields that contain special characters such as the delimiter. In our output the field with a comma was quoted.
Do not use line.split(",") for CSV. It cuts quoted commas apart, as the example output shows.
For a real file, the docs say to open it with newline="", for example with open(path, newline="", encoding="utf-8") as f:.

Example

import csv
import io

rows = [
    {"input": "Sure, Friday works", "label": "meeting_request"},
    {"input": "No thanks", "label": "not_interested"},
]
buf = io.StringIO()
writer = csv.DictWriter(buf, fieldnames=["input", "label"])
writer.writeheader()
writer.writerows(rows)
print(repr(buf.getvalue()))
buf.seek(0)
for row in csv.DictReader(buf):
    print(row["label"], "|", row["input"])
print(next(csv.reader(['a,"b,c",d'])))
print('a,"b,c",d'.split(","))

It prints:

'input,label\r\n"Sure, Friday works",meeting_request\r\nNo thanks,not_interested\r\n'
meeting_request | Sure, Friday works
not_interested | No thanks
['a', 'b,c', 'd']
['a', '"b', 'c"', 'd']

Predict, then check

1. What does this print?

import csv

print(list(csv.reader(['a,"b,c",d'])))

Show the answer

[['a', 'b,c', 'd']]

The quotes keep b,c together as one field, so the row has three fields.

2. What does this print?

import csv

row = next(csv.reader(["1,2"]))
print(row[0] + row[1])

Show the answer

Both fields are strings, so + joins them instead of adding numbers.

3. What does this print?

import csv

print([r["label"] for r in csv.DictReader(["input,label", "hi,x"])])

Show the answer

['x']

The first row names the fields, so the second row becomes {'input': 'hi', 'label': 'x'}.

Try it

Write labels_in(text). The text is CSV with a header row input,label. Return the list of label values, one per data row.

def labels_in(text):
    return []

Show a solution

import csv
import io

def labels_in(text):
    return [row["label"] for row in csv.DictReader(io.StringIO(text))]

Common mistake

If you open a real CSV file without newline="", the docs say newlines embedded inside quoted fields will not be interpreted correctly. They also say it should always be safe to specify it.

In a small-language-model project

Annotation tools and spreadsheets may export labelled replies as CSV. Reading with DictReader lets you rebuild {"input": ..., "label": ...} records, and remember that, by default, every value arrives as text.

Sources

Each quote below appears word for word on the cited page of the official Python documentation (version 3.12).

Each row read from the csv file is returned as a list of strings.
docs.python.org/3.12/library/csv.html — Each reader row is a list of strings, with no type conversion.
If fieldnames is omitted, the values in the first row of file f will be used as the fieldnames and will be omitted from the results.
docs.python.org/3.12/library/csv.html — DictReader takes field names from the first row.
the fieldnames parameter of the DictWriter class is not optional
docs.python.org/3.12/library/csv.html — DictWriter needs fieldnames.
csvfile can be any object with a write() method.
docs.python.org/3.12/library/csv.html — A writer works on any object with write().
Instructs writer objects to only quote those fields which contain special characters such as delimiter , quotechar or any of the characters in lineterminator .
docs.python.org/3.12/library/csv.html — QUOTE_MINIMAL quotes fields with special characters.
If newline='' is not specified, newlines embedded inside quoted fields will not be interpreted correctly
docs.python.org/3.12/library/csv.html — Newlines in quoted fields need newline='' on real files.
If csvfile is a file object, it should be opened with newline='' .
docs.python.org/3.12/library/csv.html — Open real files with newline=''.
It should always be safe to specify newline='' , since the csv module does its own ( universal ) newline handling.
docs.python.org/3.12/library/csv.html — It is always safe to give newline=''.
It can take on any of the QUOTE_* constants and defaults to QUOTE_MINIMAL .
docs.python.org/3.12/library/csv.html — The default quoting is QUOTE_MINIMAL.

Step through this lesson in the Hi! Python desktop