Storage & Extraction¶

Storage Backends¶

fast-agent-stack supports three storage backends for file upload/download.

Local Filesystem¶

pip install "fast-agent-stack[storage-local]"

class Settings(BaseSettings):
    storage_backend: str = "local"
    storage_local_root: str = "./uploads"

Amazon S3¶

pip install "fast-agent-stack[storage-s3]"

class Settings(BaseSettings):
    storage_backend: str = "s3"
    storage_s3_bucket: str = "my-bucket"
    storage_s3_region: str = "us-east-1"

Uses instance role credentials by default. Set AWS_ACCESS_KEY_ID / AWS_SECRET_ACCESS_KEY for explicit credentials.

MinIO¶

pip install "fast-agent-stack[storage-minio]"

class Settings(BaseSettings):
    storage_backend: str = "minio"
    storage_minio_endpoint: str = "http://localhost:9000"
    storage_minio_bucket: str = "my-bucket"
    storage_minio_access_key: str = "minioadmin"
    storage_minio_secret_key: str = "minioadmin"

Using Storage in Routes¶

from fast_agent_stack.core.storage import get_storage_backend

storage = get_storage_backend(settings)

# Upload
await storage.put("uploads/photo.jpg", file_bytes, content_type="image/jpeg")

# Download
data = await storage.get("uploads/photo.jpg")

# Delete
await storage.delete("uploads/photo.jpg")

# URL (for S3/MinIO: presigned; for local: path)
url = await storage.url("uploads/photo.jpg", expires_in=3600)

Document Extraction¶

Extract plain text from common document formats before indexing or analysis.

pip install "fast-agent-stack[extract-pdf,extract-docx,extract-xlsx]"

Extra	Formats	Library
`extract-pdf`	PDF	pdfplumber
`extract-docx`	DOCX	python-docx
`extract-xlsx`	XLSX	openpyxl

from fast_agent_stack.core.extraction import get_extraction_backend

pdf_extractor = get_extraction_backend("pdf")
text = await pdf_extractor.extract(pdf_bytes)

Custom Storage Backend¶

storage_backend: str = "myproject.storage.MyS3CompatibleBackend"

Must implement StorageProtocol (put, get, delete, url).