Underground-Digital/Workflow-Engine

1import logging2 3from core.rag.extractor.extractor_base import BaseExtractor4from core.rag.models.document import Document5 6logger = logging.getLogger(__name__)7 8 9class UnstructuredMsgExtractor(BaseExtractor):10    """Load msg files.11 12 13    Args:14        file_path: Path to the file to load.15    """16 17    def __init__(self, file_path: str, api_url: str, api_key: str):18        """Initialize with file path."""19        self._file_path = file_path20        self._api_url = api_url21        self._api_key = api_key22 23    def extract(self) -> list[Document]:24        if self._api_url:25            from unstructured.partition.api import partition_via_api26 27            elements = partition_via_api(filename=self._file_path, api_url=self._api_url, api_key=self._api_key)28        else:29            from unstructured.partition.msg import partition_msg30 31            elements = partition_msg(filename=self._file_path)32        from unstructured.chunking.title import chunk_by_title33 34        chunks = chunk_by_title(elements, max_characters=2000, combine_text_under_n_chars=2000)35        documents = []36        for chunk in chunks:37            text = chunk.text.strip()38            documents.append(Document(page_content=text))39 40        return documents41