What’s New

v0.11.3 (Jul 14, 2017)

  • Bug Fix: When using the xlwings.conf sheet, there was a subscript out of range error (GH708)
  • Enhancement: The add-in is now password protected (pw: xlwings) to declutter the VBA editor (GH710)

You need to update your xlwings add-in to be get the fixes!

v0.11.2 (Jul 6, 2017)

  • Bug Fix: The sql extension was sometimes not correctly assigning the table aliases (GH699)
  • Bug Fix: Permission errors during pip installation should be resolved now (GH693)

v0.11.1 (Jul 5, 2017)

v0.11.0 (Jul 2, 2017)

Big news! This release adds a full blown add-in! We also throw in a great In-Excel SQL Extension and a few bug fixes:



A few highlights:

  • Settings don’t have to be manipulated in VBA code anymore, but can be either set globally via Ribbon/config file or for the workbook via a special worksheet
  • UDF server can be restarted directly from the add-in
  • You can still use a VBA module instead of the add-in, but the recommended way is the add-in
  • Get all the details here: Add-in

In-Excel SQL Extension

The add-in can be extended with own code. We throw in an sql function, that allows you to perform SQL queries on data in your spreadsheets. It’s pretty awesome, get the details here: Extensions.

Bug Fixes

  • [Win]: Running Debug > Compile is not throwing errors anymore (GH678)
  • Pandas deprecation warnings have been fixed (GH675 and GH664)
  • [Mac]: Errors are again shown correctly in a pop up (GH660)
  • [Mac]: Like Windows, Mac now also only shows errors in a popup. Before it was including stdout, too (GH666)

Breaking Changes

  • RunFrozenPython now requires the full path to the executable.
  • The xlwings CLI xlwings template functionality has been removed. Use quickstart instead.

v0.10.4 (Feb 19, 2017)

  • [Win] Bug Fix: v0.10.3 introduced a bug that imported UDFs by default with volatile=True, this has now been fixed. You will need to reimport your functions after upgrading the xlwings package.

v0.10.3 (Jan 28, 2017)

This release adds new features to User Defined Functions (UDFs):

  • categories
  • volatile option
  • suppress calculation in function wizard


import xlwings as xw
@xw.func(category="xlwings", volatile=False, call_in_wizard=True)
def myfunction():
    return ...

For details, check out the (also new) and comprehensive API docs about the decorators: UDF decorators

v0.10.2 (Dec 31, 2016)

  • [Win] Python 3.6 is now supported (GH592)

v0.10.1 (Dec 5, 2016)

  • Writing a Pandas Series with a MultiIndex header was not writing out the header (GH572)
  • [Win] Docstrings for UDF arguments are now working (GH367)
  • [Mac] Range.clear_contents() has been fixed (it was doing clear() instead) (GH576)
  • xw.Book(...) and xw.books.open(...) raise now the same error in case the file doesn’t exist (GH540)

v0.10.0 (Sep 20, 2016)

Dynamic Array Formulas

This release adds an often requested & powerful new feature to User Defined Functions (UDFs): Dynamic expansion for array formulas. While Excel offers array formulas, you need to specify their dimensions up front by selecting the result array first, then entering the formula and finally hitting Ctrl-Shift-Enter. While this makes sense from a data integrity point of view, in practice, it often turns out to be a cumbersome limitation, especially when working with dynamic arrays such as time series data.

This is a simple example that demonstrates the syntax and effect of UDF expansion:

import numpy as np

def dynamic_array(r, c):
    return np.random.randn(int(r), int(c))

Note: Expanding array formulas will overwrite cells without prompting and leave an empty border around them, i.e. they will clear the row to the bottom and the column to the right of the array.

Bug Fixes

  • The int converter works now always as you would expect (e.g.: xw.Range('A1').options(numbers=int).value). Before, it could happen that the number was off by 1 due to floating point issues (GH554).

v0.9.3 (Aug 22, 2016)

  • [Win] App.visible wasn’t behaving correctly (GH551).
  • [Mac] Added support for the new 64bit version of Excel 2016 on Mac (GH549).
  • Unicode book names are again supported (GH546).
  • xlwings.Book.save() now supports relative paths. Also, when saving an existing book under a new name without specifying the full path, it’ll be saved in Python’s current working directory instead of in Excel’s default directory (GH185).

v0.9.2 (Aug 8, 2016)

Another round of bug fixes:

  • [Mac]: Sometimes, a column was referenced instead of a named range (GH545)
  • [Mac]: Python 2.7 was raising a LookupError: unknown encoding: mbcs (GH544)
  • Fixed docs regarding set_mock_caller (GH543)

v0.9.1 (Aug 5, 2016)

This is a bug fix release: As to be expected after a rewrite, there were some rough edges that have now been taken care of:

  • [Win] Opening a file via xw.Book() was causing an additional Book1 to be opened in case Excel was not running yet (GH531)
  • [Win] Some users were getting an ImportError (GH533)
  • [PY 2.7] RunPython was broken with Python 2.7 (GH537)
  • Some corrections in the docs (GH538 and GH536)

v0.9.0 (Aug 2, 2016)

Exciting times! v0.9.0 is a complete rewrite of xlwings with loads of syntax changes (hence the version jump). But more importantly, this release adds a ton of new features and bug fixes that would have otherwise been impossible. Some of the highlights are listed below, but make sure to check out the full migration guide for the syntax changes in details. Note, however, that the syntax for user defined functions (UDFs) did not change. At this point, the API is fairly stable and we’re expecting only smaller changes on our way towards a stable v1.0 release.

  • Active book instead of current book: xw.Range('A1') goes against the active sheet of the active book like you’re used to from VBA. Instantiating an explicit connection to a Book is not necessary anymore:

    >>> import xlwings as xw
    >>> xw.Range('A1').value = 11
    >>> xw.Range('A1').value
  • Excel Instances: Full support of multiple Excel instances (even on Mac!)

    >>> app1 = xw.App()
    >>> app2 = xw.App()
    >>> xw.apps
    Apps([<Excel App 1668>, <Excel App 1644>])
  • New powerful object model based on collections and close to Excel’s original, allowing to fully qualify objects: xw.apps[0].books['MyBook.xlsx'].sheets[0].range('A1:B2').value

    It supports both Python indexing (square brackets) and Excel indexing (round brackets):

    xw.books[0].sheets[0] is the same as xw.books(1).sheets(1)

    It also supports indexing and slicing of range objects:

    >>> rng = xw.Range('A1:E10')
    >>> rng[1]
    <Range [Workbook1]Sheet1!$B$1>
    >>> rng[:2, :2]
    <Range [Workbook1]Sheet1!$A$1:$B$2>

    For more details, see Syntax Overview.

  • UDFs can now also be imported from packages, not just modules (GH437)

  • Named Ranges: Introduction of full object model and proper support for sheet and workbook scope (GH256)

  • Excel doesn’t become the active window anymore so the focus stays on your Python environment (GH414)

  • When writing to ranges while Excel is busy, xlwings is now retrying until Excel is idle again (GH468)

  • xlwings.view() has been enhanced to accept an optional sheet object (GH469)

  • Objects like books, sheets etc. can now be compared (e.g. wb1 == wb2) and are properly hashable

  • Note that support for Python 2.6 has been dropped

Some of the new methods/properties worth mentioning are:

Bug Fixes

  • See here for details about which bugs have been fixed.

v0.7.2 (May 18, 2016)

Bug Fixes

  • [Win] UDFs returning Pandas DataFrames/Series containing nan were failing (GH446).
  • [Win] RunFrozenPython was not finding the executable (GH452).
  • The xlwings VBA module was not finding the Python interpreter if PYTHON_WIN or PYTHON_MAC contained spaces (GH461).

v0.7.1 (April 3, 2016)


  • [Win]: User Defined Functions (UDFs) support now optional/default arguments (GH363)

  • [Win]: User Defined Functions (UDFs) support now multiple source files, see also under API changes below. For example (VBA settings): UDF_MODULES="common;myproject"

  • VBA Subs & Functions are now callable from Python:

    As an example, this VBA function:

    Function MySum(x, y)
        MySum = x + y
    End Function

    can be accessed like this:

    >>> import xlwings as xw
    >>> wb = xw.Workbook.active()
    >>> my_sum = wb.macro('MySum')
    >>> my_sum(1, 2)
  • New xw.view method: This opens a new workbook and displays an object on its first sheet. E.g.:

    >>> import xlwings as xw
    >>> import pandas as pd
    >>> import numpy as np
    >>> df = pd.DataFrame(np.random.rand(10, 4), columns=['a', 'b', 'c', 'd'])
    >>> xw.view(df)
  • New docs about Matplotlib and Custom Converter

  • New method: xlwings.Range.formula_array() (GH411)

API changes

  • VBA settings: PYTHON_WIN and PYTHON_MAC must now include the interpreter if you are not using the default (PYTHON_WIN = "") (GH289). E.g.:

    PYTHON_WIN: "C:\Python35\pythonw.exe"
    PYTHON_MAC: "/usr/local/bin/python3.5"
  • [Win]: VBA settings: UDF_PATH has been replaced with UDF_MODULES. The default behaviour doesn’t change though (i.e. if UDF_MODULES = "", then a Python source file with the same name as the Excel file, but with .py ending will be imported from the same directory as the Excel file).


    UDF_MODULES: "mymodule"
    PYTHONPATH: "C:\path\to"


    UDF_PATH: "C:\path\to\mymodule.py"

Bug Fixes

  • Numpy scalars issues were resolved (GH415)
  • [Win]: xlwings was failing with freezers like cx_Freeze (GH413)
  • [Win]: UDFs were failing if they were returning None or np.nan (GH390)
  • Multiindex Pandas Series have been fixed (GH383)
  • [Mac]: xlwings runpython install was failing (GH424)

v0.7.0 (March 4, 2016)

This version marks an important first step on our path towards a stable release. It introduces converters, a new and powerful concept that brings a consistent experience for how Excel Ranges and their values are treated both when reading and writing but also across xlwings.Range objects and User Defined Functions (UDFs).

As a result, a few highlights of this release include:

  • Pandas DataFrames and Series are now supported for reading and writing, both via Range object and UDFs
  • New Range converter options: transpose, dates, numbers, empty, expand
  • New dictionary converter
  • New UDF debug server
  • No more pyc files when using RunPython

Converters are accessed via the new options method when dealing with xlwings.Range objects or via the @xw.arg and @xw.ret decorators when using UDFs. As an introductory sample, let’s look at how to read and write Pandas DataFrames:


Range object:

>>> import xlwings as xw
>>> import pandas as pd
>>> wb = xw.Workbook()
>>> df = xw.Range('A1:D5').options(pd.DataFrame, header=2).value
>>> df
    a     b
    c  d  e
10  1  2  3
20  4  5  6
30  7  8  9

# Writing back using the defaults:
>>> Range('A1').value = df

# Writing back and changing some of the options, e.g. getting rid of the index:
>>> Range('B7').options(index=False).value = df


This is the same sample as above (starting in Range('A13') on screenshot). If you wanted to return a DataFrame with the defaults, the @xw.ret decorator can be left away.

@xw.arg('x', pd.DataFrame, header=2)
def myfunction(x):
   # x is a DataFrame, do something with it
   return x


  • Dictionary (dict) converter:

    >>> Range('A1:B2').options(dict).value
    {'a': 1.0, 'b': 2.0}
    >>> Range('A4:B5').options(dict, transpose=True).value
    {'a': 1.0, 'b': 2.0}
  • transpose option: This works in both directions and finally allows us to e.g. write a list in column orientation to Excel (GH11):

    Range('A1').options(transpose=True).value = [1, 2, 3]
  • dates option: This allows us to read Excel date-formatted cells in specific formats:

    >>> import datetime as dt
    >>> Range('A1').value
    datetime.datetime(2015, 1, 13, 0, 0)
    >>> Range('A1').options(dates=dt.date).value
    datetime.date(2015, 1, 13)
  • empty option: This allows us to override the default behavior for empty cells:

    >>> Range('A1:B1').value
    [None, None]
    >>> Range('A1:B1').options(empty='NA')
    ['NA', 'NA']
  • numbers option: This transforms all numbers into the indicated type.

    >>> xw.Range('A1').value = 1
    >>> type(xw.Range('A1').value)  # Excel stores all numbers interally as floats
    >>> type(xw.Range('A1').options(numbers=int).value)
  • expand option: This works the same as the Range properties table, vertical and horizontal but is only evaluated when getting the values of a Range:

    >>> import xlwings as xw
    >>> wb = xw.Workbook()
    >>> xw.Range('A1').value = [[1,2], [3,4]]
    >>> rng1 = xw.Range('A1').table
    >>> rng2 = xw.Range('A1').options(expand='table')
    >>> rng1.value
    [[1.0, 2.0], [3.0, 4.0]]
    >>> rng2.value
    [[1.0, 2.0], [3.0, 4.0]]
    >>> xw.Range('A3').value = [5, 6]
    >>> rng1.value
    [[1.0, 2.0], [3.0, 4.0]]
    >>> rng2.value
    [[1.0, 2.0], [3.0, 4.0], [5.0, 6.0]]

All these options work the same with decorators for UDFs, e.g. for transpose:

@xw.arg('x', transpose=True)
def myfunction(x):
    # x will be returned unchanged as transposed both when reading and writing
    return x

Note: These options (dates, empty, numbers) currently apply to the whole Range and can’t be selectively applied to e.g. only certain columns.

  • UDF debug server

    The new UDF debug server allows you to easily debug UDFs: just set UDF_DEBUG_SERVER = True in the VBA Settings, at the top of the xlwings VBA module (make sure to update it to the latest version!). Then add the following lines to your Python source file and run it:

    if __name__ == '__main__':

    When you recalculate the Sheet, the code will stop at breakpoints or print any statements that you may have. For details, see: Debugging.

  • pyc files: The creation of pyc files has been disabled when using RunPython, leaving your directory in an uncluttered state when having the Python source file next to the Excel workbook (GH326).

API changes

  • UDF decorator changes (it is assumed that xlwings is imported as xw and numpy as np):

    New Old
    @xw.func @xw.xlfunc
    @xw.arg @xw.xlarg
    @xw.ret @xw.xlret
    @xw.sub @xw.xlsub

    Pay attention to the following subtle change:

    New Old
    @xw.arg("x", np.array) @xw.xlarg("x", "nparray")
  • Samples of how the new options method replaces the old Range keyword arguments:

    New Old
    Range('A1:A2').options(ndim=2) Range('A1:A2', atleast_2d=True)
    Range('A1:B2').options(np.array) Range('A1:B2', asarray=True)
    Range('A1').options(index=False, header=False).value = df Range('A1', index=False, header=False).value = df
  • Upon writing, Pandas Series are now shown by default with their name and index name, if they exist. This can be changed using the same options as for DataFrames (GH276):

    import pandas as pd
    # unchanged behaviour
    Range('A1').value = pd.Series([1,2,3])
    # Changed behaviour: This will print a header row in Excel
    s = pd.Series([1,2,3], name='myseries', index=pd.Index([0,1,2], name='myindex'))
    Range('A1').value = s
    # Control this behaviour like so (as with DataFrames):
    Range('A1').options(header=False, index=True).value = s
  • NumPy scalar values

    Previously, NumPy scalar values were returned as np.atleast_1d. To keep the same behaviour, this now has to be set explicitly using ndim=1. Otherwise they’re returned as numpy scalar values.

    New Old
    Range('A1').options(np.array, ndim=1).value Range('A1', asarray=True).value

Bug Fixes

A few bugfixes were made: GH352, GH359.

v0.6.4 (January 6, 2016)

API changes



  • Quickstart: It’s now easier than ever to start a new xlwings project, simply use the commmand line client (GH306):

    xlwings quickstart myproject will produce a folder with the following files, ready to be used (see Command Line Client):

  • New documentation about how to use xlwings with other languages like R and Julia, see xlwings with R and Julia.

Bug Fixes

  • [Win]: Importing UDFs with the add-in was throwing an error if the filename was including characters like spaces or dashes (GH331). To fix this, close Excel completely and run xlwings addin update.
  • [Win]: Workbook.caller() is now also accessible within functions that are decorated with @xlfunc. Previously, it was only available with functions that used the @xlsub decorator (GH316).
  • Writing a Pandas DataFrame failed in case the index was named the same as a column (GH334).

v0.6.3 (December 18, 2015)

Bug Fixes

  • [Mac]: This fixes a bug introduced in v0.6.2: When using RunPython from VBA, errors were not shown in a pop-up window (GH330).

v0.6.2 (December 15, 2015)

API changes

  • LOG_FILE: So far, the log file has been placed next to the Excel file per default (VBA settings). This has been changed as it was causing issues for files on SharePoint/OneDrive and Mac Excel 2016: The place where LOG_FILE = "" refers to depends on the OS and the Excel version, see Log File default locations.


  • [Mac]: This version adds support for the VBA module on Mac Excel 2016 (i.e. the RunPython command) and is now feature equivalent with Mac Excel 2011 (GH206).

Bug Fixes

  • [Win]: On certain systems, the xlwings dlls weren’t found (GH323).

v0.6.1 (December 4, 2015)

Bug Fixes

  • [Python 3]: The command line client has been fixed (GH319).
  • [Mac]: It now works correctly with psutil>=3.0.0 (GH315).

v0.6.0 (November 30, 2015)

API changes



  • User Defined Functions (UDFs) - currently Windows only

    The ExcelPython project has been fully merged into xlwings. This means that on Windows, UDF’s are now supported via decorator syntax. A simple example:

    from xlwings import xlfunc
    def double_sum(x, y):
        """Returns twice the sum of the two arguments"""
        return 2 * (x + y)

    For array formulas with or without NumPy, see the docs: VBA: User Defined Functions (UDFs)

  • Command Line Client

    The new xlwings command line client makes it easy to work with the xlwings template and the developer add-in (the add-in is currently Windows-only). E.g. to create a new Excel spreadsheet from the template, run:

    xlwings template open

    For all commands, see the docs: Command Line Client

  • Other enhancements:

v0.5.0 (November 10, 2015)

API changes



This version adds support for Matplotlib! Matplotlib figures can be shown in Excel as pictures in just 2 lines of code:

  1. Get a matplotlib figure object:
  • via PyPlot interface:

    import matplotlib.pyplot as plt
    fig = plt.figure()
    plt.plot([1, 2, 3, 4, 5])
  • via object oriented interface:

    from matplotlib.figure import Figure
    fig = Figure(figsize=(8, 6))
    ax = fig.add_subplot(111)
    ax.plot([1, 2, 3, 4, 5])
  • via Pandas:

    import pandas as pd
    import numpy as np
    df = pd.DataFrame(np.random.rand(10, 4), columns=['a', 'b', 'c', 'd'])
    ax = df.plot(kind='bar')
    fig = ax.get_figure()
  1. Show it in Excel as picture:

    plot = Plot(fig)

See the full API: xlwings.Plot(). There’s also a new example available both on GitHub and as download on the homepage.

Other enhancements:

  • New xlwings.Shape() class
  • New xlwings.Picture() class
  • The PYTHONPATH in the VBA settings now accepts multiple directories, separated by ; (GH258)
  • An explicit exception is raised when Range is called with 0-based indices (GH106)

Bug Fixes

  • Sheet.add was not always acting on the correct workbook (GH287)
  • Iteration over a Range only worked the first time (GH272)
  • [Win]: Sometimes, an error was raised when Excel was not running (GH269)
  • [Win]: Non-default Python interpreters (as specified in the VBA settings under PYTHON_WIN) were not found if the path contained a space (GH257)

v0.4.1 (September 27, 2015)

API changes



This release makes it easier than ever to connect to Excel from Python! In addition to the existing ways, you can now connect to the active Workbook (on Windows across all instances) and if the Workbook is already open, it’s good enough to refer to it by name (instead of having to use the full path). Accordingly, this is how you make a connection to... (GH30 and GH226):

  • a new workbook: wb = Workbook()
  • the active workbook [New!]: wb = Workbook.active()
  • an unsaved workbook: wb = Workbook('Book1')
  • a saved (open) workbook by name (incl. xlsx etc.) [New!]: wb = Workbook('MyWorkbook.xlsx')
  • a saved (open or closed) workbook by path: wb = Workbook(r'C:\\path\\to\\file.xlsx')

Also, there are some new docs:

Bug Fixes

  • The Excel template was updated to the latest VBA code (GH234).
  • Connections to files that are saved on OneDrive/SharePoint are now working correctly (GH215).
  • Various issues with timezone-aware objects were fixed (GH195).
  • [Mac]: A certain range of integers were not written to Excel (GH227).

v0.4.0 (September 13, 2015)

API changes



The most important update with this release was made on Windows: The methodology used to make a connection to Workbooks has been completely replaced. This finally allows xlwings to reliably connect to multiple instances of Excel even if the Workbooks are opened from untrusted locations (network drives or files downloaded from the internet). This gets rid of the dreaded Filename is already open... error message that was sometimes shown in this context. It also allows the VBA hooks (RunPython) to work correctly if the very same file is opened in various instances of Excel.

Note that you will need to update the VBA module and that apart from pywin32 there is now a new dependency for the Windows version: comtypes. It should be installed automatically though when installing/upgrading xlwings with pip.

Other updates:

Bug Fixes

  • The VBA module was not accepting lower case drive letters (GH205).
  • Fixed an error when adding a new Sheet that was already existing (GH211).

v0.3.6 (July 14, 2015)

API changes

Application as attribute of a Workbook has been removed (wb is a Workbook object):

Correct Syntax (as before) Removed
Application(wkb=wb) wb.application


Excel 2016 for Mac Support (GH170)

Excel 2016 for Mac is finally supported (Python side). The VBA hooks (RunPython) are currently not yet supported. In more details:

  • This release allows Excel 2011 and Excel 2016 to be installed in parallel.

  • Workbook() will open the default Excel installation (usually Excel 2016).

  • The new keyword argument app_target allows to connect to a different Excel installation, e.g.:

    Workbook(app_target='/Applications/Microsoft Office 2011/Microsoft Excel')

    Note that app_target is only available on Mac. On Windows, if you want to change the version of Excel that xlwings talks to, go to Control Panel > Programs and Features and Repair the Office version that you want as default.

  • The RunPython calls in VBA are not yet available through Excel 2016 but Excel 2011 doesn’t get confused anymore if Excel 2016 is installed on the same system - make sure to update your VBA module!

Other enhancements

  • New method: xlwings.Application.calculate() (GH207)

Bug Fixes

  • [Win]: When using the OPTIMIZED_CONNECTION on Windows, Excel left an orphaned process running after closing (GH193).

Various improvements regarding unicode file path handling, including:

  • [Mac]: Excel 2011 for Mac now supports unicode characters in the filename when called via VBA’s RunPython (but not in the path - this is a limitation of Excel 2011 that will be resolved in Excel 2016) (GH154).
  • [Win]: Excel on Windows now handles unicode file paths correctly with untrusted documents. (GH154).

v0.3.5 (April 26, 2015)

API changes

Sheet.autofit() and Range.autofit(): The integer argument for the axis has been removed (GH186). Use string arguments rows or r for autofitting rows and columns or c for autofitting columns (as before).


New methods:


>>> rng = Range('A1').table
>>> rng.row, rng.column
(1, 1)
>>> rng.last_cell.row, rng.last_cell.column
(4, 5)

Bug Fixes

  • The unicode bug on Windows/Python3 has been fixed (GH161)

v0.3.4 (March 9, 2015)

Bug Fixes

  • The installation error on Windows has been fixed (GH160)

v0.3.3 (March 8, 2015)

API changes



  • New class Application with quit method and properties screen_updating und calculation (GH101, GH158, GH159). It can be conveniently accessed from within a Workbook (on Windows, Application is instance dependent). A few examples:

    >>> from xlwings import Workbook, Calculation
    >>> wb = Workbook()
    >>> wb.application.screen_updating = False
    >>> wb.application.calculation = Calculation.xlCalculationManual
    >>> wb.application.quit()
  • New headless mode: The Excel application can be hidden either during Workbook instantiation or through the application object:

    >>> wb = Workbook(app_visible=False)
    >>> wb.application.visible
    >>> wb.application.visible = True
  • Newly included Excel template which includes the xlwings VBA module and boilerplate code. This is currently accessible from an interactive interpreter session only:

    >>> from xlwings import Workbook
    >>> Workbook.open_template()

Bug Fixes

  • [Win]: datetime.date objects were causing an error (GH44).
  • Depending on how it was instantiated, Workbook was sometimes missing the fullname attribute (GH76).
  • Range.hyperlink was failing if the hyperlink had been set as formula (GH132).
  • A bug introduced in v0.3.0 caused frozen versions (eg. with cx_Freeze) to fail (GH133).
  • [Mac]: Sometimes, xlwings was causing an error when quitting the Python interpreter (GH136).

v0.3.2 (January 17, 2015)

API changes




Bug Fixes

  • The xlwings.Workbook.save() method has been fixed to show the expected behavior (GH138): Previously, calling save() without a path argument would always create a new file in the current working directory. This is now only happening if the file hasn’t been previously saved.

v0.3.1 (January 16, 2015)

API changes



  • New method xlwings.Workbook.save() (GH110).

  • New method xlwings.Workbook.set_mock_caller() (GH129). This makes calling files from both Excel and Python much easier:

    import os
    from xlwings import Workbook, Range
    def my_macro():
        wb = Workbook.caller()
        Range('A1').value = 1
    if __name__ == '__main__':
        # To run from Python, not needed when called from Excel.
        # Expects the Excel file next to this source file, adjust accordingly.
        path = os.path.abspath(os.path.join(os.path.dirname(__file__), 'myfile.xlsm'))
  • The simulation example on the homepage works now also on Mac.

Bug Fixes

  • [Win]: A long-standing bug that caused the Excel file to close and reopen under certain circumstances has been fixed (GH10): Depending on your security settings (Trust Center) and in connection with files downloaded from the internet or possibly in connection with some add-ins, Excel was either closing the file and reopening it or giving a “file already open” warning. This has now been fixed which means that the examples downloaded from the homepage should work right away after downloading and unzipping.

v0.3.0 (November 26, 2014)

API changes

  • To reference the calling Workbook when running code from VBA, you now have to use Workbook.caller(). This means that wb = Workbook() is now consistently creating a new Workbook, whether the code is called interactively or from VBA.

    New Old
    Workbook.caller() Workbook()


This version adds two exciting but still experimental features from ExcelPython (Windows only!):

  • Optimized connection: Set the OPTIMIZED_CONNECTION = True in the VBA settings. This will use a COM server that will keep the connection to Python alive between different calls and is therefore much more efficient. However, changes in the Python code are not being picked up until the pythonw.exe process is restarted by killing it manually in the Windows Task Manager. The suggested workflow is hence to set OPTIMIZED_CONNECTION = False for development and only set it to True for production - keep in mind though that this feature is still experimental!
  • User Defined Functions (UDFs): Using ExcelPython’s wrapper syntax in VBA, you can expose Python functions as UDFs, see VBA: User Defined Functions (UDFs) for details.

Note: ExcelPython’s developer add-in that autogenerates the VBA wrapper code by simply using Python decorators isn’t available through xlwings yet.

Further enhancements include:

Bug Fixes

  • [Mac]: Environment variables from .bash_profile are now available when called from VBA, e.g. by using: os.environ['USERNAME'] (GH95)

v0.2.3 (October 17, 2014)

API changes



  • New method Sheet.add() (GH71):

    >>> Sheet.add()  # Place at end with default name
    >>> Sheet.add('NewSheet', before='Sheet1')  # Include name and position
    >>> new_sheet = Sheet.add(after=3)
    >>> new_sheet.index
  • New method Sheet.count():

    >>> Sheet.count()
  • autofit() works now also on Sheet objects, not only on Range objects (GH66):

    >>> Sheet(1).autofit()  # autofit columns and rows
    >>> Sheet('Sheet1').autofit('c')  # autofit columns
  • New property number_format for Range objects (GH60):

    >>> Range('A1').number_format
    >>> Range('A1:C3').number_format = '0.00%'
    >>> Range('A1:C3').number_format

    Works also with the Range properties table, vertical, horizontal:

    >>> Range('A1').value = [1,2,3,4,5]
    >>> Range('A1').table.number_format = '0.00%'
  • New method get_address for Range objects (GH7):

    >>> Range((1,1)).get_address()
    >>> Range((1,1)).get_address(False, False)
    >>> Range('Sheet1', (1,1), (3,3)).get_address(True, False, include_sheetname=True)
    >>> Range('Sheet1', (1,1), (3,3)).get_address(True, False, external=True)
  • New method Sheet.all() returning a list with all Sheet objects:

    >>> Sheet.all()
    [<Sheet 'Sheet1' of Workbook 'Book1'>, <Sheet 'Sheet2' of Workbook 'Book1'>]
    >>> [i.name.lower() for i in Sheet.all()]
    ['sheet1', 'sheet2']
    >>> [i.autofit() for i in Sheet.all()]

Bug Fixes

  • xlwings works now also with NumPy < 1.7.0. Before, doing something like Range('A1').value = 'Foo' was causing a NotImplementedError: Not implemented for this type error when NumPy < 1.7.0 was installed (GH73).
  • [Win]: The VBA module caused an error on the 64bit version of Excel (GH72).
  • [Mac]: The error pop-up wasn’t shown on Python 3 (GH85).
  • [Mac]: Autofitting bigger Ranges, e.g. Range('A:D').autofit() was causing a time out (GH74).
  • [Mac]: Sometimes, calling xlwings from Python was causing Excel to show old errors as pop-up alert (GH70).

v0.2.2 (September 23, 2014)

API changes

  • The Workbook qualification changed: It now has to be specified as keyword argument. Assume we have instantiated two Workbooks like so: wb1 = Workbook() and wb2 = Workbook(). Sheet, Range and Chart classes will default to wb2 as it was instantiated last. To target wb1, use the new wkb keyword argument:

    New Old
    Range('A1', wkb=wb1).value wb1.range('A1').value
    Chart('Chart1', wkb=wb1) wb1.chart('Chart1')

    Alternatively, simply set the current Workbook before using the Sheet, Range or Chart classes:

  • Through the introduction of the Sheet class (see Enhancements), a few methods moved from the Workbook to the Sheet class. Assume the current Workbook is: wb = Workbook():

    New Old
    Sheet('Sheet1').activate() wb.activate('Sheet1')
    Sheet('Sheet1').clear() wb.clear('Sheet1')
    Sheet('Sheet1').clear_contents() wb.clear_contents('Sheet1')
    Sheet.active().clear_contents() wb.clear_contents()
  • The syntax to add a new Chart has been slightly changed (it is a class method now):

    New Old
    Chart.add() Chart().add()


  • [Mac]: Python errors are now also shown in a Message Box. This makes the Mac version feature equivalent with the Windows version (GH57):

  • New Sheet class: The new class handles everything directly related to a Sheet. See the section about api_sheet for details (GH62). A few examples:

    >>> Sheet(1).name
    >>> Sheet('Sheet1').clear_contents()
    >>> Sheet.active()
    <Sheet 'Sheet1' of Workbook 'Book1'>
  • The Range class has a new method autofit() that autofits the width/height of either columns, rows or both (GH33).


    axis : string or integer, default None
        - To autofit rows, use one of the following: 'rows' or 'r'
        - To autofit columns, use one of the following: 'columns' or 'c'
        - To autofit rows and columns, provide no arguments


    # Autofit column A
    # Autofit row 1
    # Autofit columns and rows, taking into account Range('A1:E4')
    # AutoFit rows, taking into account Range('A1:E4')
  • The Workbook class has the following additional methods: current() and set_current(). They determine the default Workbook for Sheet, Range or Chart. On Windows, in case there are various Excel instances, when creating new or opening existing Workbooks, they are being created in the same instance as the current Workbook.

    >>> wb1 = Workbook()
    >>> wb2 = Workbook()
    >>> Workbook.current()
    <Workbook 'Book2'>
    >>> wb1.set_current()
    >>> Workbook.current()
    <Workbook 'Book1'>
  • If a Sheet, Range or Chart object is instantiated without an existing Workbook object, a user-friendly error message is raised (GH58).

  • New docs about Debugging and Data Structures Tutorial.

Bug Fixes

  • The atleast_2d keyword had no effect on Ranges consisting of a single cell and was raising an error when used in combination with the asarray keyword. Both have been fixed (GH53):

    >>> Range('A1').value = 1
    >>> Range('A1', atleast_2d=True).value
    >>> Range('A1', atleast_2d=True, asarray=True).value
  • [Mac]: After creating two new unsaved Workbooks with Workbook(), any Sheet, Range or Chart object would always just access the latest one, even if the Workbook had been specified (GH63).

  • [Mac]: When xlwings was imported without ever instantiating a Workbook object, Excel would start upon quitting the Python interpreter (GH51).

  • [Mac]: When installing xlwings, it now requires psutil to be at least version 2.0.0 (GH48).

v0.2.1 (August 7, 2014)

API changes



  • All VBA user settings have been reorganized into a section at the top of the VBA xlwings module:

    PYTHON_WIN = ""
    PYTHON_MAC = GetMacDir("Home") & "/anaconda/bin"
    PYTHON_FROZEN = ThisWorkbook.Path & "\build\exe.win32-2.7"
    PYTHONPATH = ThisWorkbook.Path
    LOG_FILE = ThisWorkbook.Path & "\xlwings_log.txt"
  • Calling Python from within Excel VBA is now also supported on Mac, i.e. Python functions can be called like this: RunPython("import bar; bar.foo()"). Running frozen executables (RunFrozenPython) isn’t available yet on Mac though.

Note that there is a slight difference in the way that this functionality behaves on Windows and Mac:

  • Windows: After calling the Macro (e.g. by pressing a button), Excel waits until Python is done. In case there’s an error in the Python code, a pop-up message is being shown with the traceback.
  • Mac: After calling the Macro, the call returns instantly but Excel’s Status Bar turns into “Running...” during the duration of the Python call. Python errors are currently not shown as a pop-up, but need to be checked in the log file. I.e. if the Status Bar returns to its default (“Ready”) but nothing has happened, check out the log file for the Python traceback.

Bug Fixes


Special thanks go to Georgi Petrov for helping with this release.

v0.2.0 (July 29, 2014)

API changes



  • Cross-platform: xlwings is now additionally supporting Microsoft Excel for Mac. The only functionality that is not yet available is the possibility to call the Python code from within Excel via VBA macros.

  • The clear and clear_contents methods of the Workbook object now default to the active sheet (GH5):

    wb = Workbook()
    wb.clear_contents()  # Clears contents of the entire active sheet

Bug Fixes

  • DataFrames with MultiHeaders were sometimes getting truncated (GH41).

v0.1.1 (June 27, 2014)

API Changes

  • If asarray=True, NumPy arrays are now always at least 1d arrays, even in the case of a single cell (GH14):

    >>> Range('A1', asarray=True).value
  • Similar to NumPy’s logic, 1d Ranges in Excel, i.e. rows or columns, are now being read in as flat lists or 1d arrays. If you want the same behavior as before, you can use the atleast_2d keyword (GH13).


    The table property is also delivering a 1d array/list, if the table Range is really a column or row.

    >>> Range('A1').vertical.value
    [1.0, 2.0, 3.0, 4.0]
    >>> Range('A1', atleast_2d=True).vertical.value
    [[1.0], [2.0], [3.0], [4.0]]
    >>> Range('C1').horizontal.value
    [1.0, 2.0, 3.0, 4.0]
    >>> Range('C1', atleast_2d=True).horizontal.value
    [[1.0, 2.0, 3.0, 4.0]]
    >>> Range('A1', asarray=True).table.value
    array([ 1.,  2.,  3.,  4.])
    >>> Range('A1', asarray=True, atleast_2d=True).table.value
    array([[ 1.],
           [ 2.],
           [ 3.],
           [ 4.]])
  • The single file approach has been dropped. xlwings is now a traditional Python package.


  • xlwings is now officially suppported on Python 2.6-2.7 and 3.1-3.4

  • Support for Pandas Series has been added (GH24):

    >>> import numpy as np
    >>> import pandas as pd
    >>> from xlwings import Workbook, Range
    >>> wb = Workbook()
    >>> s = pd.Series([1.1, 3.3, 5., np.nan, 6., 8.])
    >>> s
    0    1.1
    1    3.3
    2    5.0
    3    NaN
    4    6.0
    5    8.0
    dtype: float64
    >>> Range('A1').value = s
    >>> Range('D1', index=False).value = s
  • Excel constants have been added under their original Excel name, but categorized under their enum (GH18), e.g.:

    # Extra long version
    import xlwings as xl
    # Long version
    from xlwings import constants
    # Short version
    from xlwings import ChartType
  • Slightly enhanced Chart support to control the ChartType (GH1):

    >>> from xlwings import Workbook, Range, Chart, ChartType
    >>> wb = Workbook()
    >>> Range('A1').value = [['one', 'two'],[10, 20]]
    >>> my_chart = Chart().add(chart_type=ChartType.xlLine,
                               name='My Chart',

    alternatively, the properties can also be set like this:

    >>> my_chart = Chart().add()  # Existing Charts: my_chart = Chart('My Chart')
    >>> my_chart.name = 'My Chart'
    >>> my_chart.chart_type = ChartType.xlLine
    >>> my_chart.set_source_data(Range('A1').table)
  • pytz is no longer a dependency as datetime object are now being read in from Excel as time-zone naive (Excel doesn’t know timezones). Before, datetime objects got the UTC timezone attached.

  • The Workbook class has the following additional methods: close()

  • The Range class has the following additional methods: is_cell(), is_column(), is_row(), is_table()

Bug Fixes

  • Writing None or np.nan to Excel works now (GH16 & GH15).
  • The import error on Python 3 has been fixed (GH26).
  • Python 3 now handles Pandas DataFrames with MultiIndex headers correctly (GH39).
  • Sometimes, a Pandas DataFrame was not handling nan correctly in Excel or numbers were being truncated (GH31) & (GH35).
  • Installation is now putting all files in the correct place (GH20).

v0.1.0 (March 19, 2014)

Initial release of xlwings.