When improving performance, make sure code neither (a) fetches too much data, nor (b) does the same expensive work repeatedly, nor (c) runs unbounded workloads.

Practical rules

1. Target by primary keys / narrow filters: Prefer get_by_id-style lookups over broad tenant-scoped queries when you already have the identifier.
2. Push filtering to the datastore: If you currently fetch large result sets and filter in Python (or run many queries), move the filter into the ES/DB/search criteria (e.g., include doc_id / source_id in the search condition).
3. Bound user-controlled heavy parameters: Add upper bounds for things like top_k so a bad request can’t trigger massive retrieval.
4. Batch/chunk heavy processing: For page/document loops, use page batches to control memory/CPU and keep progress reporting accurate.
5. Avoid repeated expensive work: Hoist invariant computations out of inner loops, and run file-level expensive steps once (not per sheet/per iteration).
6. Resource-heavy ops must be policy-driven: If you unload/reload models or enable tracing, ensure it’s safe under concurrency (don’t unload while tasks run) and avoid enabling heavy instrumentation by default.

Examples

• Bound retrieval:

  top = int(req.get("top_k", 1024))
  top = min(top, 200)  # cap user input
  results = SearchService.search(..., top_k=top)
  

• Chunk heavy page processing:

  batch_size = max(1, int(os.getenv("PDF_PARSER_PAGE_BATCH_SIZE", "50")))
  for page_from in range(from_page, to_page, batch_size):
    page_to = min(page_from + batch_size, to_page)
    __images__(fnm, page_from=page_from, page_to=page_to)
    chunk_boxes = parse_window_into_boxes()
    all_boxes.extend(to_global(chunk_boxes))
  

• Push doc_id/source_id filtering into search:

  # include source_id/doc_id in the search condition so the store filters
  cond = {"source_id": [doc_id], **other_filters}
  res = SearchService.search(dataset_id=dataset_id, condition=cond, limit=1)