1. Sergey Astanin
  2. python-tabulate




Pretty-print tabular data in Python, a library and a command-line utility.

The main use cases of the library are:

  • printing small tables without hassle: just one function call, formatting is guided by the data itself
  • authoring tabular data for lightweight plain-text markup: multiple output formats suitable for further editing or transformation
  • readable presentation of mixed textual and numeric data: smart column alignment, configurable number formatting, alignment by a decimal point


To install the Python library and the command line utility, run:

pip install tabulate

The command line utility will be installed as tabulate to bin on Linux (e.g. /usr/bin); or as tabulate.exe to Scripts in your Python installation on Windows (e.g. C:\Python27\Scripts\tabulate.exe).

You may consider installing the library only for the current user:

pip install tabulate --user

In this case the command line utility will be installed to ~/.local/bin/tabulate on Linux and to %APPDATA%\Python\Scripts\tabulate.exe on Windows.

To install just the library on Unix-like operating systems:

TABULATE_INSTALL=lib-only pip install tabulate

On Windows:

pip install tabulate

Build status

Build status

Library usage

The module provides just one function, tabulate, which takes a list of lists or another tabular data type as the first argument, and outputs a nicely formatted plain-text table:

>>> from tabulate import tabulate

>>> table = [["Sun",696000,1989100000],["Earth",6371,5973.6],
...          ["Moon",1737,73.5],["Mars",3390,641.85]]
>>> print tabulate(table)
-----  ------  -------------
Sun    696000     1.9891e+09
Earth    6371  5973.6
Moon     1737    73.5
Mars     3390   641.85
-----  ------  -------------

The following tabular data types are supported:

  • list of lists or another iterable of iterables
  • list or another iterable of dicts (keys as columns)
  • dict of iterables (keys as columns)
  • two-dimensional NumPy array
  • NumPy record arrays (names as columns)
  • pandas.DataFrame

Examples in this file use Python2. Tabulate supports Python3 too.


The second optional argument named headers defines a list of column headers to be used:

>>> print tabulate(table, headers=["Planet","R (km)", "mass (x 10^29 kg)"])
Planet      R (km)    mass (x 10^29 kg)
--------  --------  -------------------
Sun         696000           1.9891e+09
Earth         6371        5973.6
Moon          1737          73.5
Mars          3390         641.85

If headers="firstrow", then the first row of data is used:

>>> print tabulate([["Name","Age"],["Alice",24],["Bob",19]],
...                headers="firstrow")
Name      Age
------  -----
Alice      24
Bob        19

If headers="keys", then the keys of a dictionary/dataframe, or column indices are used. It also works for NumPy record arrays and lists of dictionaries or named tuples:

>>> print tabulate({"Name": ["Alice", "Bob"],
...                 "Age": [24, 19]}, headers="keys")
  Age  Name
-----  ------
   24  Alice
   19  Bob

Row Indices

By default, only pandas.DataFrame tables have an additional column called row index. To add a similar column to any other type of table, pass showindex="always" or showindex=True argument to tabulate(). To suppress row indices for all types of data, pass showindex="never" or showindex=False. To add a custom row index column, pass showindex=rowIDs, where rowIDs is some iterable:

>>> print(tabulate([["F",24],["M",19]], showindex="always"))
-  -  --
0  F  24
1  M  19
-  -  --

Table format

There is more than one way to format a table in plain text. The third optional argument named tablefmt defines how the table is formatted.

Supported table formats are:

  • "plain"
  • "simple"
  • "grid"
  • "fancy_grid"
  • "pipe"
  • "orgtbl"
  • "jira"
  • "psql"
  • "rst"
  • "mediawiki"
  • "moinmoin"
  • "youtrack"
  • "html"
  • "latex"
  • "latex_raw"
  • "latex_booktabs"
  • "textile"

plain tables do not use any pseudo-graphics to draw lines:

>>> table = [["spam",42],["eggs",451],["bacon",0]]
>>> headers = ["item", "qty"]
>>> print tabulate(table, headers, tablefmt="plain")
item      qty
spam       42
eggs      451
bacon       0

simple is the default format (the default may change in future versions). It corresponds to simple_tables in Pandoc Markdown extensions:

>>> print tabulate(table, headers, tablefmt="simple")
item      qty
------  -----
spam       42
eggs      451
bacon       0

grid is like tables formatted by Emacs' table.el package. It corresponds to grid_tables in Pandoc Markdown extensions:

>>> print tabulate(table, headers, tablefmt="grid")
| item   |   qty |
| spam   |    42 |
| eggs   |   451 |
| bacon  |     0 |

fancy_grid draws a grid using box-drawing characters:

>>> print tabulate(table, headers, tablefmt="fancy_grid")
│ item   │   qty │
│ spam   │    42 │
│ eggs   │   451 │
│ bacon  │     0 │

psql is like tables formatted by Postgres' psql cli:

>>> print tabulate.tabulate()
| item   |   qty |
| spam   |    42 |
| eggs   |   451 |
| bacon  |     0 |

pipe follows the conventions of PHP Markdown Extra extension. It corresponds to pipe_tables in Pandoc. This format uses colons to indicate column alignment:

>>> print tabulate(table, headers, tablefmt="pipe")
| item   |   qty |
| spam   |    42 |
| eggs   |   451 |
| bacon  |     0 |

orgtbl follows the conventions of Emacs org-mode, and is editable also in the minor orgtbl-mode. Hence its name:

>>> print tabulate(table, headers, tablefmt="orgtbl")
| item   |   qty |
| spam   |    42 |
| eggs   |   451 |
| bacon  |     0 |

jira follows the conventions of Atlassian Jira markup language:

>>> print tabulate(table, headers, tablefmt="jira")
|| item   ||   qty ||
| spam   |    42 |
| eggs   |   451 |
| bacon  |     0 |

rst formats data like a simple table of the reStructuredText format:

>>> print tabulate(table, headers, tablefmt="rst")
======  =====
item      qty
======  =====
spam       42
eggs      451
bacon       0
======  =====

mediawiki format produces a table markup used in Wikipedia and on other MediaWiki-based sites:

>>> print tabulate(table, headers, tablefmt="mediawiki")
{| class="wikitable" style="text-align: left;"
|+ <!-- caption -->
! item   !! align="right"|   qty
| spam   || align="right"|    42
| eggs   || align="right"|   451
| bacon  || align="right"|     0

moinmoin format produces a table markup used in `MoinMoin`_ wikis:

>>> print tabulate(d,headers,tablefmt="moinmoin")
|| ''' item   ''' || ''' quantity   ''' ||
||  spam    ||  41.999      ||
||  eggs    ||  451         ||
||  bacon   ||              ||

youtrack format produces a table markup used in `Youtrack`_ tickets:

>>> print tabulate(d,headers,tablefmt="youtrack")
||  item    ||  quantity   ||
|   spam    |  41.999      |
|   eggs    |  451         |
|   bacon   |              |

textile format produces a table markup used in Textile format:

>>> print tabulate(table, headers, tablefmt='textile')
|_.  item   |_.   qty |
|<. spam    |>.    42 |
|<. eggs    |>.   451 |
|<. bacon   |>.     0 |

html produces standard HTML markup:

>>> print tabulate(table, headers, tablefmt="html")
<tr><th>item  </th><th style="text-align: right;">  qty</th></tr>
<tr><td>spam  </td><td style="text-align: right;">   42</td></tr>
<tr><td>eggs  </td><td style="text-align: right;">  451</td></tr>
<tr><td>bacon </td><td style="text-align: right;">    0</td></tr>

latex format creates a tabular environment for LaTeX markup, replacing special characters like _ or \ to their LaTeX correspondents:

>>> print tabulate(table, headers, tablefmt="latex")
 item   &   qty \\
 spam   &    42 \\
 eggs   &   451 \\
 bacon  &     0 \\

latex_raw behaves like latex but does not escape LaTeX commands and special characters.

latex_booktabs creates a tabular environment for LaTeX markup using spacing and style from the booktabs package.

Column alignment

tabulate is smart about column alignment. It detects columns which contain only numbers, and aligns them by a decimal point (or flushes them to the right if they appear to be integers). Text columns are flushed to the left.

You can override the default alignment with numalign and stralign named arguments. Possible column alignments are: right, center, left, decimal (only for numbers), and None (to disable alignment).

Aligning by a decimal point works best when you need to compare numbers at a glance:

>>> print tabulate([[1.2345],[123.45],[12.345],[12345],[1234.5]])

Compare this with a more common right alignment:

>>> print tabulate([[1.2345],[123.45],[12.345],[12345],[1234.5]], numalign="right")

For tabulate, anything which can be parsed as a number is a number. Even numbers represented as strings are aligned properly. This feature comes in handy when reading a mixed table of text and numbers from a file:

>>> import csv ; from StringIO import StringIO
>>> table = list(csv.reader(StringIO("spam, 42\neggs, 451\n")))
>>> table
[['spam', ' 42'], ['eggs', ' 451']]
>>> print tabulate(table)
----  ----
spam    42
eggs   451
----  ----

Number formatting

tabulate allows to define custom number formatting applied to all columns of decimal numbers. Use floatfmt named argument:

>>> print tabulate([["pi",3.141593],["e",2.718282]], floatfmt=".4f")
--  ------
pi  3.1416
e   2.7183
--  ------

floatfmt argument can be a list or a tuple of format strings, one per column, in which case every column may have different number formatting:

>>> print tabulate([[0.12345, 0.12345, 0.12345]], floatfmt=(".1f", ".3f"))
---  -----  -------
0.1  0.123  0.12345
---  -----  -------

Text formatting

By default, tabulate removes leading and trailing whitespace from text columns. To disable whitespace removal, set the global module-level flag PRESERVE_WHITESPACE:

import tabulate

Wide (fullwidth CJK) symbols

To properly align tables which contain wide characters (typically fullwidth glyphs from Chinese, Japanese or Korean languages), the user should install wcwidth library. To install it together with tabulate:

pip install tabulate[widechars]

Wide character support is enabled automatically if wcwidth library is already installed. To disable wide characters support without uninstalling wcwidth, set the global module-level flag WIDE_CHARS_MODE:

import tabulate
tabulate.WIDE_CHARS_MODE = False

Usage of the command line utility

Usage: tabulate [options] [FILE ...]

FILE                      a filename of the file with tabular data;
                          if "-" or missing, read data from stdin.


-h, --help                show this message
-1, --header              use the first row of data as a table header
-o FILE, --output FILE    print table to FILE (default: stdout)
-s REGEXP, --sep REGEXP   use a custom column separator (default: whitespace)
-F FPFMT, --float FPFMT   floating point number format (default: g)
-f FMT, --format FMT      set output table format; supported formats:
                          plain, simple, grid, fancy_grid, pipe, orgtbl,
                          rst, mediawiki, html, latex, latex_raw,
                          latex_booktabs, tsv
                          (default: simple)

Performance considerations

Such features as decimal point alignment and trying to parse everything as a number imply that tabulate:

  • has to "guess" how to print a particular tabular data type
  • needs to keep the entire table in-memory
  • has to "transpose" the table twice
  • does much more work than it may appear

It may not be suitable for serializing really big tables (but who's going to do that, anyway?) or printing tables in performance sensitive applications. tabulate is about two orders of magnitude slower than simply joining lists of values with a tab, coma or other separator.

In the same time tabulate is comparable to other table pretty-printers. Given a 10x10 table (a list of lists) of mixed text and numeric data, tabulate appears to be slower than asciitable, and faster than PrettyTable and texttable

=================================  ==========  ===========
Table formatter                      time, μs    rel. time
=================================  ==========  ===========
join with tabs and newlines              13.7          1.0
csv to StringIO                          17.1          1.3
asciitable (0.8.0)                      429.9         31.5
tabulate (0.8.0)                        823.8         60.3
tabulate (0.8.0, WIDE_CHARS_MODE)      1112.4         81.5
PrettyTable (0.7.2)                    1667.2        122.1
texttable (0.8.8)                      1919.6        140.6
=================================  ==========  ===========

Version history

  • 0.8.0: latex_raw format, column-specific floating point formatting. Python 3.5 & 3.6 support. Drop support for Python 2.6, 3.2, 3.3.
  • 0.7.7: Identical to 0.7.6, resolving some PyPI issues.
  • 0.7.6: Bug fixes. New table formats (psql, jira, moinmoin, textile). Wide character support. Printing from database cursors. Option to print row indices. Boolean columns. Ragged rows. Option to disable number parsing.
  • 0.7.5: Bug fixes. --float format option for the command line utility.
  • 0.7.4: Bug fixes. fancy_grid and html formats. Command line utility.
  • 0.7.3: Bug fixes. Python 3.4 support. Iterables of dicts. latex_booktabs format.
  • 0.7.2: Python 3.2 support.
  • 0.7.1: Bug fixes. tsv format. Column alignment can be disabled.
  • 0.7: latex tables. Printing lists of named tuples and NumPy record arrays. Fix printing date and time values. Python <= 2.6.4 is supported.
  • 0.6: mediawiki tables, bug fixes.
  • 0.5.1: Fix README.rst formatting. Optimize (performance similar to 0.4.4).
  • 0.5: ANSI color sequences. Printing dicts of iterables and Pandas' dataframes.
  • 0.4.4: Python 2.6 support.
  • 0.4.3: Bug fix, None as a missing value.
  • 0.4.2: Fix manifest file.
  • 0.4.1: Update license and documentation.
  • 0.4: Unicode support, Python3 support, rst tables.
  • 0.3: Initial PyPI release. Table formats: simple, plain, grid, pipe, and orgtbl.

How to contribute

Contributions should include tests and an explanation for the changes they propose. Documentation (examples, docstrings, README.rst) should be updated accordingly.

This project uses nose testing framework and tox to automate testing in different environments. Add tests to one of the files in the test/ folder.

To run tests on all supported Python versions, make sure all Python interpreters, nose and tox are installed, then run tox in the root of the project source tree.

On Linux tox expects to find executables like python2.6, python2.7, python3.4 etc. On Windows it looks for C:\Python26\python.exe, C:\Python27\python.exe and C:\Python34\python.exe respectively.

To test only some Python environements, use -e option. For example, to test only against Python 2.7 and Python 3.4, run:

tox -e py27,py34

in the root of the project source tree.

To enable NumPy and Pandas tests, run:

tox -e py27-extra,py34-extra

(this may take a long time the first time, because NumPy and Pandas will have to be installed in the new virtual environments)

See tox.ini file to learn how to use nosetests directly to test individual Python versions.


Sergey Astanin, Pau Tallada Crespí, Erwin Marsi, Mik Kocikowski, Bill Ryder, Zach Dwiel, Frederik Rietdijk, Philipp Bogensberger, Greg (anonymous), Stefan Tatschner, Emiel van Miltenburg, Brandon Bennett, Amjith Ramanujam, Jan Schulz, Simon Percivall, Javier Santacruz López-Cepero, Sam Denton, Alexey Ziyangirov, acaird, Cesar Sanchez, naught101, John Vandenberg, Zack Dever, Christian Clauss, Benjamin Maier, Andy MacKinlay, Thomas Roten.