docforge-swedish 0.1.0

Swedish NLP utilities for DocForge — zero external dependencies

docforge-swedish

Swedish NLP utilities for the DocForge platform. Zero external dependencies — pure stdlib.

Install

pip install docforge-swedish
# or locally:
pip install -e src/docforge_swedish

Usage

from docforge_swedish import (
    SwedishNormalizer,
    SwedishNER,
    SwedishFormats,
    SwedishStopwords,
    SwedishProcessor,
    Domain,
)

normalizer = SwedishNormalizer()
normalized = normalizer.normalize("socialtjänstlagen § 4 kap")
# → "SoL § 4 kap"

ner = SwedishNER()
entities = ner.extract("JO har granskat Socialnämnden i Stockholm")
print(entities.authorities)  # ['JO', 'Socialnämnden']

stopwords = SwedishStopwords.for_municipal()
filtered = stopwords.filter_text("nämnden beslutade att godkänna rapporten")

Modules

Module	Description
SwedishNormalizer	OCR fixes, law reference normalization, municipality/authority name normalization
SwedishNER	Rule-based named entity recognition (authorities, courts, law refs, persons, dates)
SwedishFormats	Date, SEK amount, personnummer parsing and formatting
SwedishStopwords	Domain-specific stopword sets (general, legal, municipal, social, medical)
SwedishProcessor	High-level pipeline combining all of the above
Domain	Enum for stopword domains