How to Keep Line Breaks Inside CSV Cells Without Creating Extra Rows

A CSV cell can contain a real line break. The line break does not create another record when the entire field is enclosed in straight double quotes. RFC 4180’s commonly used CSV rules say that fields containing line breaks, commas or double quotes should be quoted; a literal double quote inside a quoted field is represented by two double quotes (RFC 4180).

A valid multiline CSV field

Suppose the notes value for Bob contains two lines. The raw CSV can be written as:

id,name,notes
1,Alice,Approved
2,Bob,"First review
Needs follow-up"
3,Cara,Published

This file has:

• 5 physical text lines
• 4 CSV records, including the header
• 3 data records

The opening quote before First review and closing quote after Needs follow-up make the embedded line break part of one field. A quote-aware CSV parser therefore returns one Bob record with three fields, not two records.

That distinction matters when preparing a dataset for publication: a text editor’s line count is not necessarily the dataset’s row count. Python’s CSV documentation makes the distinction explicit: its reader’s line_num counts source lines, which can differ from records because one record may span several lines (Python csv documentation).

Apply all three escaping rules

For a multiline value:

1. Put a straight double quote (") before the field’s first character.
2. Put another straight double quote after its final character, even if that character appears on a later physical line.
3. Replace every literal " inside the value with "".

For example, the two-line cell:

First review
Needs "legal" follow-up

becomes:

2,Bob,"First review
Needs ""legal"" follow-up"

Do not type the two characters \n unless the receiving system explicitly defines that as its own escape convention. In ordinary CSV, preserving a multiline cell means storing an actual newline inside a quoted field.

The same outer-quoting rule handles commas. For more combinations, see how to escape commas and double quotes in CSV.

What creates accidental extra rows

This is malformed for a multiline value because the notes field is not quoted:

id,name,notes
1,Alice,Approved
2,Bob,First review
Needs follow-up
3,Cara,Published

A parser can interpret Needs follow-up as a separate one-field record. The apparent table then has inconsistent field counts, and every later row may appear shifted or broken in software that assumes a fixed schema.

Also avoid these editing mistakes:

• quoting only the second physical line;
• using typographic “curly quotes” instead of the ASCII " character;
• ending the field’s quote before the embedded newline;
• deleting line breaks with a line-oriented cleanup command;
• validating rows by splitting the file on newlines or commas.

Validate parsed records, not physical lines

Use a CSV parser to reopen the finished export and check its structure. Python’s standard library recommends opening CSV files with newline=''; its writer’s minimal-quoting mode treats carriage returns and line feeds as special characters that require quoting (Python csv documentation).

This small check reports both parsed records and physical source lines, then rejects inconsistent field counts:

import csv
from pathlib import Path

path = Path("dataset.csv")

with path.open("r", encoding="utf-8", newline="") as file:
    reader = csv.reader(file)
    rows = list(reader)
    physical_lines = reader.line_num

if not rows:
    raise ValueError("CSV is empty")

expected_fields = len(rows[0])
problems = [
    (record_number, len(row))
    for record_number, row in enumerate(rows[1:], start=2)
    if len(row) != expected_fields
]

print(f"Physical lines read: {physical_lines}")
print(f"CSV records: {len(rows)}")
print(f"Data records: {len(rows) - 1}")

if problems:
    raise ValueError(f"Inconsistent records: {problems}")

record_number here is the parsed CSV record position, not necessarily the text editor’s line number. After this structural check, inspect a few multiline values and confirm that their internal newlines survived the full export-and-import round trip.

Before publishing, also verify the header, expected data-record count, encoding and absence of sensitive information. TablePage currently accepts CSV, TSV, XLSX and XLS uploads and turns them into public dataset pages, so this parser check belongs before upload rather than after sharing (TablePage).