pdf_loader v1.3.3

Suggest edits

Bases: BaseFileLoader

Source Code in griptape/loaders/pdf_loader.py
@define
class PdfLoader(BaseFileLoader):
    def try_parse(
        self,
        data: bytes,
        *,
        password: Optional[str] = None,
    ) -> ListArtifact:
        pypdf = import_optional_dependency("pypdf")
        reader = pypdf.PdfReader(BytesIO(data), strict=True, password=password)
        pages = [TextArtifact(p.extract_text()) for p in reader.pages]

        return ListArtifact(pages)

try_parse(data, *, password=None)

Source Code in griptape/loaders/pdf_loader.py
def try_parse(
    self,
    data: bytes,
    *,
    password: Optional[str] = None,
) -> ListArtifact:
    pypdf = import_optional_dependency("pypdf")
    reader = pypdf.PdfReader(BytesIO(data), strict=True, password=password)
    pages = [TextArtifact(p.extract_text()) for p in reader.pages]

    return ListArtifact(pages)
← Prev

json_loader

↑ Up

loaders

Next →

sql_loader