docqa/docxextract/__init__.py

"""Document data extraction service.

Extracts arbitrary user-defined keys from invoices, receipts and scanned
documents using Impira's LayoutLM document-QA models.

    from docxextract import ExtractionService

    svc = ExtractionService()
    result = svc.extract(pdf_bytes, ["Invoice No", "Total Amount", "GSTIN"])
    for field in result.fields:
        print(field.key, field.value, field.status, field.confidence)
"""

from .config import Settings, get_settings
from .engine import Engine, get_engine
from .schema import (DocumentInfo, ExtractedField, ExtractResponse, FieldStatus,
                     SourceType)
from .service import ExtractionService

__version__ = "1.0.0"
__all__ = [
    "Settings", "get_settings", "Engine", "get_engine",
    "ExtractionService", "ExtractResponse", "ExtractedField",
    "DocumentInfo", "FieldStatus", "SourceType", "__version__",
]