Errors

PyAIStack raises package-specific exceptions for invalid input, unavailable providers, loaders, and vector stores.

Use format_error() at the application boundary to show a concise Python-style error that points to your code rather than PyAIStack internals.

from pyaistack import RAG, format_error
from pyaistack.exceptions import AIStackError

try:
    rag = RAG()
    rag.add(["A document to index."])
except AIStackError as error:
    print(format_error(error))

Common exceptions:

• ProviderError — Ollama or another model provider failed.
• EmptyIndexError — a search was requested before documents were indexed.
• LoaderError — a file cannot be read or is empty.
• UnsupportedFileTypeError — a requested loader type is not implemented.
• ChunkingError — chunk size, overlap, or input is invalid.
• VectorStoreError — vector validation or persistent-store work failed.
• VectorStoreSchemaError — a PostgreSQL/pgvector schema, dimension, constraint, or required index is incompatible.
• VectorStoreConnectionError — a PostgreSQL connection could not be acquired.
• VectorStoreTimeoutError — PostgreSQL cancelled an operation after its configured statement timeout.
• VectorStoreConstraintError — a PostgreSQL write violated a database constraint.

Database errors preserve their chained driver exception for debugging, while normal PostgreSQL store messages exclude DSNs, credentials, document content, metadata values, and vectors. See PostgreSQL + pgvector.