Skip to main content

SDIF for python

Project description

Needs Cython >= 0.15

It allows to read and write any kind of SDIF file, to define new
kinds of frames and matrices and to read and write metadata.

The matrices read from a sdif file are exposed as numpy arrays.

It exposes both a low level and a high level interface.

The low level interface for reading and writing sdif files mirrors the
sdif library quite transparently so that the example files and
utilities using it can be directly translated with it. In particular
it does not create any intermediate objects, even the data of the matrices
is a numpy array mapped to the c array read from disk, so no allocation takes
place. Whereas this makes for very fast code, one has to take care to copy the
data if it will be used longer, since by the time a new matrix is read this
data is no longer valid.

to read for ex. 1TRC format:

import pysdif

sdif_file = pysdif.SdifFile('filename.sdif')
sig1TRC = pysdif.str2signature("1TRC")
while not sdif_file.eof:
if sdif_file.frame_numerical_signature) == sig1TRC:
print sdif_file.time
for n in range(sdif_file.matrices_in_frame):
if sdif_file.matrix_numerical_signature == sig1TRC:
data = sdif_file.get_matrix_data(copy=True) # default is True
# data is now a numpy array
# here is what happens under the hood: the SDIF library
# reads a whole matrix and puts it in its internal buffer
# This data is memcpy(ed) to create a numpy array
# If you dont intend to keep this data and there are no
# chances that a new matrix or frame will be read before
# you use this data for some calculation, then it is safe
# to call sdif_file.get_matrix_data(copy=False)
# in this case no memory is copied, the numpy array does not
# own its memory and uses the internal buffer of the SDIF runtime
print data

a more natural way:

from pysdif import SdifFile
sdif_file = SdifFile('filename.sdif')
for frame in sdif_file:
if frame.signature == "1TRC":
print frame.time
for matrix in frame:
if matrix.signature == "1TRC":
print matrix.get_data() # data will be copied, use .get_data(copy=False) to avoid this

the frames and the matrices resulting from the iteration
are only guaranteed to be valid as long as no new frames and matrices are read

to write a SdifFile:

f = SdifFile('new_sdif.sdif', 'w')
# these are optional
# add some metadata
'name' : 'my name',
'date' : time.asctime(time.localtime())
# define new frame and matrix types
f.add_frame_type('1NEW', '1ABC NewMatrix, 1FQ0 New1FQ0')
f.add_matrix_type('1ABC', 'Column1, Column2')
# now you can begin adding frames
frame = f.new_frame('1NEW', time_now)
frame.add_matrix('1ABC', array([
[0, 1.2],
[3.5, 8.13],

# say we just want to take the data from an existing
# sdiffile, modify it and write it back
in_sdif = SdifFile("existing-file.sdif")
out_sdif = SdifFile("outfile.sdif", "w")
for in_frame in in_sdif:
if in_frame.signature == "1NEW":
new_frame = out_sdif.new_frame("1NEW", in_frame.time)
# we know there is only one matrix and we dont need to keep the data
in_data = in_frame.get_matrix_data(copy=False)
# multiply it
in_data[:,1] *= 0.5
# add it to the stream
new_frame.add_matrix('1ABC', in_data)
# only one matrix, so write the frame.
# along this operation, only the memory used to allocate the original
# matrix was used, the rest of the operations is performed in place
# This is only safe where only one thread has access to
# the sdif entity at once and data is not kept longer than the time
# read it, transform it and rewrite it.

there are also many utility functions under pysdif.sdiftools

see release notes and changes at

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

pysdif-0.1.4.tar.gz (171.1 kB view hashes)

Uploaded source

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page