Skip to main content

Read SDMX XML files

Project description

Read SDMX XML files. I’ve only added the features I’ve needed, so this is far from being a thorough implementation. Contributions welcome.


pip install sdmx


sdmx.generic_data_message_reader(fileobj, dsd_fileobj=None, lazy=None)

Given a file-like object representing the XML of a generic data message, return a data message reader.

sdmx.compact_data_message_reader(fileobj, dsd_fileobj=None, lazy=None)

Given a file-like object representing the XML of a compact data message, return a data message reader.

Optional arguments for data message readers

  • dsd_fileobj: the file-like object representing the XML of the relevant DSD. Only used if the data message does not contain a URL to the relevant DSD.

  • lazy: set to True to read observations lazily to allow datasets to be read without loading the entire dataset into memory. Use with caution: lazy reading makes some assumptions about the structure of the XML (for instance, that series keys always appear before any observations in that series). These assumptions seem to be safe on files that I’ve tested, but that doesn’t mean they’re universally true.

Data message readers

Each data message reader has the following attributes:

  • datasets(): returns an iterable of DatasetReader instances. Each instance corresponds to a <DataSet> element.


A DatasetReader has the following attributes:

  • key_family(): returns the KeyFamily for the dataset. This corresponds to the <KeyFamilyRef> element.

  • series(): returns an iterable of Series instances. Each instance corresponds to a <Series> element.


A KeyFamily has the following attributes:

  • name(lang): the name of the key family in the language lang.

  • describe_dimensions(lang): for each dimension of the key family, find the referenced concept and use its name in the language lang. Returns a list of strings in the same order as in the source file.


A Series has the following attributes:

  • describe_key(lang): the key of a series is a mapping from each dimension of the dataset to a value. For instance, if the dataset has a dimension named Country, the value for the series might be United Kingdom. Returns an ordered dictionary mapping strings to lists of strings. The items in the dictionary are in the same order as the dimensions returned from describe_dimensions(). For instance, if the dataset has a single dimension called Country, the returned value would be {"Country": ["United Kingdom"]}. All ancestors of a value are also described, with ancestors appearing before descendents. For instance, if the value United Kingdom has the parent value Europe, which has the parent value World, the returned value would be {"Country": ["World", "Europe", "United Kingdom"]}.

  • observations(): returns an iterable of Observation instances. Each instance corresponds to an <Obs> element.


An Observation has the following attributes:

  • time

  • value


The script below can be used to print out the values contained in a generic data message. (If you have a compact data message, then using compact_data_message_reader instead of generic_data_message_reader should also work.) Assuming the script is saved as, it can be used like so:

python path/to/generic-data-message.xml path/to/dsd.xml
import sys

import sdmx

def main():
    dataset_path = sys.argv[1]
    dsd_path = sys.argv[2]

    with open(dataset_path) as dataset_fileobj:
        with open(dsd_path) as dsd_fileobj:
            dataset_reader = sdmx.generic_data_message_reader(

def _print_values(dataset_reader):
    for dataset in dataset_reader.datasets():
        key_family = dataset.key_family()
        name ="en")

        print name

        dimension_names = key_family.describe_dimensions(lang="en") + ["Time", "Value"]

        for series in dataset.series():
            row_template = []
            key = series.describe_key(lang="en")
            for key_name, key_value in key.iteritems():

            for observation in series.observations(lang="en"):
                row = row_template[:]

                print zip(dimension_names, row)


Project details

Download files

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

Source Distribution

sdmx-0.2.10.tar.gz (10.1 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