pydifact

A Python library to parse and serialize UN/EDIFACT interchanges.

Preamble

This is a port of metroplex-systems/edifact to Python. Thanks here at the start to Craig Duncan for this cool piece of software. Porting was like a breeze due to the high code quality there. All credits for the initial code here go to him, I just did the translation to Python(3), some "pythonifications" of the code and little improvements.

Why another EDIFACT library?

Because I did not find a decent UN/EDIFACT library for Python, so I decided to port one of the available good PHP libraries to Python. Here is the result.

ATM this is a Work In Progress, the API is not stable yet. Feel free to help.

Install

As usual, use a virtualenv, and install via pip or pipenv:

pip install pydifact

However, it is not stable yet, so the pypi version, including documentation and code examples, could differ from the latest git code. If in doubt, use the git version:

git clone https://github.com/nerdocs/pydifact.git
cd pydifact
pip install -e .

Usage

To read a full Interchange from a file or string, take the Interchange class and iter over the messages and segments:

from pydifact.segmentcollection import Interchange

interchange = Interchange.from_file("./tests/data/wikipedia.edi")
interchange = Interchange.from_str(
    "UNA:+,? '"
    "UNB+UNOC:1+1234+3333+200102:2212+42'"
    "UNH+42z42+PAORES:93:1:IA'"
    "MSG+1:45'"
    "IFT+3+XYZCOMPANY AVAILABILITY'"
    "ERC+A7V:1:AMD'"
    "UNT+5+42z42'UNZ+2+42'"
)
for message in interchange.get_messages():
    for segment in message.segments:
        print("Segment tag: {}, content: {}".format(segment.tag, segment.elements))

You may also want to iterate directly on segments :

from pydifact.segmentcollection import Interchange

interchange = Interchange.from_str(
    "UNA:+,? '"
    "UNB+UNOC:1+1234+3333+200102:2212+42'"
    "UNH+42z42+PAORES:93:1:IA'"
    "MSG+1:45'"
    "IFT+3+XYZCOMPANY AVAILABILITY'"
    "ERC+A7V:1:AMD'"
    "UNT+5+42z42'UNZ+2+42'"
)

for segment in interchange.segments:
    print("Segment tag: {}, content: {}".format(segment.tag, segment.elements))

Or you can create an EDI interchange on the fly:

from pydifact.segmentcollection import Interchange
from pydifact.segments import Segment

interchange = Interchange(syntax_identifier=("IBMA",1),
                          sender="MeMyselfAndIrene",
                          recipient="TheOtherOne",
                          control_reference="KLuzs7c6")
interchange.add_segment(Segment("QTY", ["12", "3"]))

print(interchange.serialize())

You may also want to parse a « raw » segment bunch which is not an interchange :

from pydifact.segmentcollection import RawSegmentCollection

collection = RawSegmentCollection.from_str("UNH+1+ORDERS:D:96A:UN:EAN008'")

for segment in collection.segments:
    print("Segment tag: {}, content: {}".format(segment.tag, segment.elements))

Limitations

• No support of optional functional groups (UNG→UNE),

Alternatives

In python ecosystem:

• python-edifact - simpler, IMHO less cleaner code, less flexible. may be faster though (not tested). Seems unmaintained.
• bots - huge, with webinterface (bots-monitor), webserver, bots-engine.
• edicat - simple, only for separating lines/segments for CLI-piping.

Development

Setup

To develop pydifact, clone the repository and install the dev requirements:

pip install -e .[dev]

This installs all the python packages needed for development and testing.

Code formatting

Format all python files using black before committing.

Happy coding, PR are more than welcome to make this library better, or to add a feature that matches your needs. Nevertheless, don't forget adding tests for every aspect you add in code.

Testing

pydifact uses pytest for testing. Just exec pytest within the project folder to execute the unit tests. Make sure you installed the library properly before using pip install -e .[dev].

There is one test to check the performance of parsing huge files, named test_huge_message - you can skip that test by calling

pytest --ignore tests/test_huge_message.py

This is recommended for fast testing.

License

This library is licensed under the MIT license, see the LICENSE file.