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

uv pip install pydifact
# or
python -m pip install pydifact

However, it is not marked as 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. You should use uv for the development process:

git clone https://github.com/nerdocs/pydifact.git
cd pydifact
uv sync

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(f"Segment tag: {segment.tag}, content: {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(f"Segment tag: {segment.tags}, content: {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())

To include or override the Service String Advice segment (UNA), just specify it as a regular segment:

interchange.add_segment(Segment("UNA", ":+.? '"))

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(f"Segment tag: {segment.tags}, content: {segment.elements}")

Limitations

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

Alternatives

In the Python ecosystem:

• python-edifact - simpler, IMHO less clean 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:

uv sync

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

Code formatting

Format all python files using black before committing:

uv run black .

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. There is a shortcut in the Makefile for your convenience:

make test

This is recommended for faster testing.

There are some additional tests to check the performance of parsing huge files - you can include that tests by calling

make test-extended

Credits

The official formats for UN/EDIFAT provided by UN Secretariat are raw text files located here and html (.htm) files located here, similarly for service codes as text and html files located at https://www.gefeg.com/jswg/.

There are many other helpful projects and people who made this possible:

• untid-parser by Stefano - thanks for your help.

For a contributors' list, see Contributors.

Sponsor

If you want to support the project, consider becoming a sponsor on GitHub Sponsors. Or buy me a "coffee" at Buy me a coffee.

If you are around Salzburg/Austria, we can meet for a real coffee too. ;-)

License

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