universal_parser/__init__.py

from __future__ import annotations

__version__ = "1.0.3"

import universal_parser.extractors.images.scan_extractor
import universal_parser.extractors.mail.mail_extractor
import universal_parser.extractors.office.docx_extractor
import universal_parser.extractors.office.legacy_extractor
import universal_parser.extractors.office.pptx_extractor
import universal_parser.extractors.office.xlsx_extractor

# Import all extractors to trigger their @register decorators.
# Without this, the router registry remains empty until the files are imported.
import universal_parser.extractors.pdf.native
import universal_parser.extractors.structured.csv_extractor
import universal_parser.extractors.structured.json_xml_extractor
import universal_parser.extractors.structured.parquet_extractor
import universal_parser.extractors.web.epub_extractor
import universal_parser.extractors.web.html_extractor  # noqa: F401

# Import the main entry point to expose it at the root of the package
from universal_parser.core.engine import parse
from universal_parser.exports.to_chunks import to_chunks
from universal_parser.exports.to_graph import to_graph
from universal_parser.exports.to_markdown import to_markdown

# Define what is exposed when doing: from universal_parser import *
__all__ = ["__version__", "parse", "to_chunks", "to_graph", "to_markdown"]