🐧 Tutorial · 6 min read

How to Open a CSV File in Linux (Terminal & GUI)

To open a CSV in Linux, page through it with less data.csv, peek at the top with head -n 5 data.csv, or double-click it into LibreOffice Calc. A CSV is just plain text, so anything can open it — the reason people get stuck is that the tools which split on the comma (awk, cut, a naive column call) misread any field that contains a comma inside quotes. This guide covers the terminal and desktop paths, and the one file that breaks each.

The quick answer

less data.csv                  # page through the raw text
head -n 5 data.csv             # first 5 lines
column -s, -t < data.csv | less -S   # align fields into columns
libreoffice --calc data.csv    # open in the GUI

Use the first two to look, the third to line up fields on a well-formed file, and the fourth when you want a spreadsheet. For anything with quoted commas or embedded line breaks, parse it — see the parser section below.

Look at the file without a parser: less, cat, head, tail

When you only need to see whether the file is comma-separated, tab-separated or full of semicolons, read the raw text:

• less data.csv — scroll a large file without loading it all; q to quit, /term to search.
• head -n 20 data.csv — the header plus the first rows, which is usually enough to spot the delimiter.
• tail -n 20 data.csv — check the end for a footer row or ragged final line.
• cat -A data.csv | head — reveal hidden characters: ^I for a tab, ^M$ for Windows CRLF line endings.

Recent less builds read a CSV as a whole record rather than by physical line, but the classic advice still holds: if rows look wrapped or ragged, you are seeing line breaks that sit inside quoted fields.

Line up the fields — and know where it breaks

The column command (GNU coreutils) turns a delimited file into aligned columns, which makes a small CSV far easier to read than raw text:

column -s, -t < data.csv | less -S

But column splits on the delimiter literally, so it is not a CSV parser. The same trap catches awk -F,, cut -d, and str.split(","). On a file whose second field is "Doe, Jane" and whose third spans two lines, this is what happens:

$ awk -F, '{print NF" fields: "$1" | "$2}' t.csv
3 fields: name | city
4 fields: "Doe |  Jane"
1 fields: and commas" | 
3 fields: Bob | Tokyo

Three records became four fragments, because RFC 4180 lets a field contain a comma when it is wrapped in double quotes. A quote-aware parser reads the same file correctly:

$ python3 -c "import csv;[print(r) for r in csv.reader(open('t.csv',newline='',encoding='utf-8'))]"
['name', 'city', 'note']
['Doe, Jane', 'Paris', 'likes "quotes"\nand commas']
['Bob', 'Tokyo', 'plain']

That is the rule to carry: any time you split a CSV on a single character in the shell, you are trusting that no field holds that character. When your data comes from a CRM, a bank export or a form, do not trust it.

Parse it properly: csvkit and Python

Two ways to stay in the terminal and still parse correctly.

csvkit (a Python tool you install with pipx install csvkit or pip install csvkit) adds CSV-aware commands. csvlook prints an aligned table, csvstat summarises columns, and csvgrep filters rows:

csvlook data.csv
csvstat --count data.csv
csvgrep -c city -m Paris data.csv

The Python standard library needs no install at all — Python ships with every mainstream Linux distribution. One line gives you a correct parse, and you can pipe it through head for a quick look:

python3 -c "import csv;[print(r) for r in csv.reader(open('data.csv',newline='',encoding='utf-8'))]" | head

If you will script the file rather than just look at it, the same module is covered in more detail in how to open a CSV in Python. To query a CSV with SQL instead of filtering by hand, see query a CSV with SQL.

Count rows correctly: wc -l is a trap

wc -l data.csv counts newline characters, not records. Our three-record test file reports four, because one record contains a line break inside a quoted field:

$ wc -l t.csv
4 t.csv

There are two ways this bites: a quoted line break inflates the count, and a file whose last line has no trailing newline makes it one short. For an exact count, parse it — csvstat --count data.csv, or the Python reader above. The row-count problem is the same one described in count rows in a CSV file.

Fix encoding before it garbles your view

If less shows symbols where accents should be, the file is not UTF-8. Identify it, then re-encode:

file -i data.csv
iconv -f LATIN1 -t UTF-8 data.csv > data.utf8.csv

file -i reports the detected type and charset, and iconv converts between them. A byte-order mark at the start of a UTF-8 file can also add a stray character to the first header name; strip it before parsing. The desktop-app version of this symptom is covered in fix garbled CSV.

Desktop and browser options

On a desktop session, the standard choice is LibreOffice Calc, which on open shows a Text Import dialog where you set the separator and encoding explicitly — the reliable path, and the same idea as changing a CSV delimiter. Gnumeric is a lighter alternative. If the file is too large for a spreadsheet, remember the sheet limit is 1,048,576 rows and 16,384 columns per Microsoft's documented Excel limits (Calc is comparable), so either split it or read it in a viewer. For a file you did not create and only need to inspect, the free CSV viewer or CSV analyzer runs in the browser with nothing to install — useful on a headless server where you only have a browser.

Frequently Asked Questions

How do I open a CSV file in Linux?

A CSV is plain text: use less data.csv to page through it, head -n 5 data.csv to see the top, or open it in a desktop app such as LibreOffice Calc. When fields contain commas inside quotes, view it with a real CSV parser rather than splitting on the comma.

Why does awk -F, give the wrong number of columns?

awk -F, splits on every comma, but RFC 4180 allows a comma inside a double-quoted field. A row like "Doe, Jane",Paris is two CSV fields but three comma-separated fragments, so awk reports the wrong count. Use a quote-aware parser such as the Python csv module or csvkit.

How do I count rows in a CSV from the command line?

wc -l counts physical lines, not records. A line break inside a quoted field makes the file have more lines than records, and a missing final newline makes it short. For an exact record count, parse it with csvstat --count or the Python csv module.

How do I fix a garbled CSV in the terminal?

It is an encoding mismatch. file -i data.csv reports the detected type and charset; iconv -f LATIN1 -t UTF-8 in.csv > out.csv re-encodes a legacy file to UTF-8. Check the result with less before feeding it to another tool.