IgorWriter

Write Igor Binary Wave (.ibw) or Igor Text (.itx) files from numpy array

Features

• Compatible with multi-dimensional arrays (up to 4 dimensions)
• Supported numpy data types: uint, int, float, complex, bool, str, bytes, datetime64
• Data units (IgorWave.set_datascale)
• Dimension scaling (IgorWave.set_dimscale)
• Dimension labels (IgorWave.set_dimlabel)
• Wave notes (IgorWave.set_note)
• Dependency formula (IgorWave.set_formula)

Installation

From PyPI:

$ pip install igorwriter

Or from conda-forge:

$ conda install conda-forge::igorwriter

Usage

Basic usage

import numpy as np 
from igorwriter import IgorWave 

array = np.array([1,2,3,4,5,6]) 

# make IgorWave objects 
wave = IgorWave(array, name='mywave') 
wave2 = IgorWave(array.astype(np.float32), name='mywave2') 

# save data
wave.save('mywave.ibw') 
wave.save_itx('mywave.itx')

with open('multi_waves.itx', 'w') as fp: 
    # Igor Text files can contain multiples waves per file 
    wave.save_itx(fp)
    wave2.save_itx(fp)

Data units, dimension scaling

wave.set_datascale('DataUnit') 
wave.set_dimscale('x', 0, 0.01, 's')

A two-dimensional array with dimension labels

a2 = np.random.random((10, 3)) 
wave = IgorWave(a2, name='wave2d') 

# you may set optional dimension labels 
wave.set_dimlabel(0, -1, 'points') # label for entire rows 
wave.set_dimlabel(1, -1, 'values') # label for entire columns 
wave.set_dimlabel(1, 0, 'ValueA') # label for column 0 
wave.set_dimlabel(1, 1, 'ValueB') # label for column 1 
wave.set_dimlabel(1, 2, 'ValueC') # label for column 2 
wave.save('my2dwave.ibw')

You can append arbitrary Igor commands to Igor Text files.

wave = IgorWave([1, 4, 9], name='wave0') 
with open('wave0.itx', 'w') as fp: 
    wave.save_itx(fp) 
    fp.write("X Display 'wave0'\n")

Unicode support

From igorwriter 0.5.0, IgorWave stores texts with utf-8 encoding. If you use Igor Pro 6 or older and want to use non-ascii characters, set unicode=False when calling IgorWave(). It will fall back to system text encoding (Windows-1252, Shift JIS, etc.).

Wave Names

There are restrictions on object names in IGOR. From v0.2.0, this package deals with illegal object names.

# This will issue a RenameWarning, and the name will be automatically changed.
wave = IgorWave(array, name='wave:0', on_errors='fix')
print(wave.name)  # wave_0

# This will raise an InvalidNameError.
wave = IgorWave(array, name='wave:0', on_errors='raise')

Igor Pro 8.0+ supports wave names with more than 31 bytes. To support longer wave names, please set long_name_support=True when calling IgorWave(). Note that setting long_name_support=True breaks backward compatibility for older Igor Pro regardless of the actual name length.

Pint integration

If Pint Quantity objects are passed to IgorWave, data units will be set automatically.

import pint 
ureg = pint.UnitRegistry() 
w = IgorWave([3, 4] * ureg.meter / ureg.second, name='wave_with_units')

w.save_itx('wave_with_units.itx')
print(open('wave_with_units.itx', 'r').read())
# ...
# X SetScale d,0,0,"m / s",'wave_with_units' 
# X SetScale /P x,0.0,1.0,"",'wave_with_units'

Pandas integration

You can easily export DataFrame objects with convenience functions.

from igorwriter import utils 
utils.dataframe_to_itx(df, 'df.itx') # all Series are exported in one file 
waves = utils.dataframe_to_ibw(df, prefix='df_bin') # each Series is saved in a separate file, <prefix>_<column>.ibw 
print(waves) # dictionary of generated IgorWaves
# {'col1': <IgorWave 'col1' at 0x...>, 'col2': ...}

IgorWriter tries to convert index info on pandas.Series objects in following manners.

• If the index is evenly-spaced, wave dimension scaling is set accordingly.
• Index names are interpreted as the dimension labels.

Notes on Image Plots

Image Plot in IGOR and imshow in matplotlib use different convention for x and y axes:

• Rows as x, columns as y (IGOR)
• Columns as x, rows as y (Matplotlib)

Thus, image parameter was introduced in save() and save_itx() methods. If you use e.g.

wave.save('path.ibw', image=True)

plt.imshow and Image Plot will give the same results.

The image=True option transposes rows and columns of the underlying array, but does not swap data units, dimension scaling, etc. (You need to change them manually).

Changelog

v0.7.0 (2025-08-26)

• Added support for Long Name Objects (wave names with more than 31 bytes).

v0.6.0 (2024-01-13)

• Wave note support
• Dependency formula support

v0.5.0 (2023-07-08)

• UTF-8 as default encoding. You can instead use system text encoding by setting unicode=False to IgorWave().
• Automatic conversion from pandas index to dimension scaling.
• Exporting 64-bit integer waves (requires Igor Pro 7 or later).
• BUG FIX: Igor Text files created from np.bool_ arrays were broken.

v0.4.1 (2023-07-02)

• Added support for np.str_ and np.bytes_ arrays.
• Automatic type conversion for np.object_ arrays.
• Added support for dimension scaling (IgorWave.set_simlabel).

v0.3.0 (2019-11-16)

• Added utils.dict_to_{ibw, itx}
• Set datascale automatically for pint Quantity object.
• Added support for np.datetime64 array.

v0.2.3 (2019-11-09)

• Added support for 64-bit integers (by automatically casting onto 32-bit integers on save).

v0.2.0 (2019-11-08)

• Added utilities for pandas (utils.dataframe_to_{ibw, itx} ).
• Added unittest scripts.
• Added wave name validator.
• BUG FIX: long (> 3 bytes) units for dimension scaling were ignored in save_itx()
• IgorWriter now uses system locale encoding rather than ASCII (the default behavior of IGOR Pro prior to ver. 7.00)