This repository contains the orgxtract package and CLI.
We develop with Pipenv. A guide for setting up a dev environment is on the The Hitchhiker’s Guide to Python.
To run the project Python 3 and a package installer/manager must be installed that can handle pyproject.toml. The following installation steps use Pipenv as example.
Install all dependencies in a virtual environment.
pipenv install -e .
Install the spacy German model.
pipenv run python -m spacy download de_core_news_md
This is a basic example to extract data from the first page of a PDF.
from orgxtract import Document, TextPipeline
import orgxtract.pdf as pdf
# Return the first page of the PDF
drawing = next(pdf.open("examples/orgchart.pdf"))
document = Document.extract(drawing)
# The with statement is only necessary when using threads.
with TextPipeline() as text_pipeline:
texts = document.text_contents.values()
for content in text_pipeline.pipe(texts):
print(content)This package contains a command line tool. It can be executed by running it as script.
The input can be either a PDF or a directory containing multiple PDFs.
pipenv run python -m orgxtract path/to/input -o path/to/output
Use --help to see all parameters
The package does use the Python logging module. It is enabled in the CLI and the level can be configured.
We used a subset from the data from these websites.
- https://www.auswaertiges-amt.de/de/service/terminologie/-/215262
- https://de.wikipedia.org/wiki/Besoldungsordnung_B
- https://www.gesetze-im-internet.de/bbesg/anlage_i.html
- https://www.landesrecht-bw.de/bsbw/document/jlr-AmtsbezZusVBWV14Anlage2
- https://www.verwaltungsvorschriften-im-internet.de/bsvwvbund_13022024_D3302001835.htm
- https://www.verwaltungsvorschriften-im-internet.de/bsvwvbund_14092020_D3302001016.htm
- https://de.wikipedia.org/wiki/Bundesministerium_(Deutschland)
The code is licensed under the MIT license. For distribution, the licenses of the dependencies must be consulted.