Tigris Directory

Tigris is a globally distributed object storage service, allowing your files to load fast anywhere in the world.

This covers how to load document objects from an AWS S3 Directory object via Tigris.

%pip install --upgrade --quiet  boto3

from langchain_community.document_loaders import S3DirectoryLoader

loader = S3DirectoryLoader(
    "testing-hwc",  # put your bucket name here
    aws_access_key_id="tid_access_key_id",  # put your access key ID here
    aws_secret_access_key="tsec_secret_access_key",  # put your secret access key here
    region="auto",
    endpoint_url="https://t3.storage.dev",
)

loader.load()

Specifying a prefix

You can also specify a prefix for more fine-grained control over what files to load.

loader = S3DirectoryLoader("testing-hwc", prefix="fake")

loader.load()

[Document(page_content='Lorem ipsum dolor sit amet.', lookup_str='', metadata={'source': 's3://testing-hwc/fake.docx'}, lookup_index=0)]

Related

• Document loader conceptual guide
• Document loader how-to guides