Vectorstore Module

vectorstore.py handles chunking PDF text and tables, embedding them using Ollama embeddings, and storing in Chroma vectorstore for fast similarity search.

Functions

• build_vectorstore(job_id, pages)

  Parses text and tables from PDF pages, splits into chunks, embeds, and stores in Chroma.

• get_vectorstore(job_id)

  Retrieves a previously saved Chroma vectorstore.

• _table_to_text(table)

  Converts a table (list of lists) into readable text for embedding.