data_processing

You can extract PDF text and query CSV data simultaneously by using the extract_pdf_text function for OCR-capable ingestion and query_csv for deterministic dataframe queries. This combination enables concurrent narrative text extraction and tabular data filtering for operational analysis.

PDF text extraction includes an OCR fallback via PyMuPDF and Tesseract to handle scanned documents. When standard raw text extraction fails, the OCR mechanism activates automatically to ensure all text is captured from your PDF files.

You filter a CSV file by passing a simple expression like 'amount > 100' to a Pandas-powered query function. This executes deterministic dataframe-style filtering against the CSV path and returns usable tabular results for workspace analysis.

The best way is to extract narrative text from a PDF report while simultaneously querying a financials.csv file to summarize results. This dual approach removes manual copying friction by combining OCR-capable PDF ingestion with Pandas-powered dataframe filtering.

You should not use ad-hoc extraction for staging documents because they must be ingested via the rag_ingest pipeline first. Guardrails enforce this rule to ensure staging content is properly processed before querying or extraction occurs.