Function load_document

src/docquery/document.py:238–267

(fpath: str, ocr_reader: Optional[Union[str, OCRReader]] = None, use_embedded_text=True)

@validate_arguments
def load_document(fpath: str, ocr_reader: Optional[Union[str, OCRReader]] = None, use_embedded_text=True):
    base_path = os.path.basename(fpath).split("?")[0].strip()
    doc_type = mimetypes.guess_type(base_path)[0]
    if fpath.startswith("http://") or fpath.startswith("https://"):
        resp = requests.get(fpath, allow_redirects=True, stream=True)
        if not resp.ok:
            raise UnsupportedDocument(f"Failed to download: {resp.content}")

        if "Content-Type" in resp.headers:
            doc_type = resp.headers["Content-Type"].split(";")[0].strip()

        b = resp.raw
    else:
        b = open(fpath, "rb")

    if not ocr_reader or isinstance(ocr_reader, str):
        ocr_reader = get_ocr_reader(ocr_reader)
    elif not isinstance(ocr_reader, OCRReader):
        raise NoOCRReaderFound(f"{ocr_reader} is not a supported OCRReader class")

    if doc_type == "application/pdf":
        return PDFDocument(b.read(), ocr_reader=ocr_reader, use_embedded_text=use_embedded_text)
    elif doc_type == "text/html":
        return WebDocument(fpath)
    else:
        try:
            img = Image.open(b)
        except UnidentifiedImageError as e:
            raise UnsupportedDocument(e)
        return ImageDocument(img, ocr_reader=ocr_reader)