Load the configuration, merging user options with default values.

Read a JSON file, returning None on any error.

Load full document JSON on demand (structure, pages, etc.).

Build a lightweight meta entry from a document dict.

Read and validate _meta.json, returning None on any corruption.

Scan individual doc JSON files and return a meta dict.

Create a clean structure for document description generation, excluding unnecessary fields like 'text'.

Find all direct and indirect children of a parent node

Return document metadata JSON.

Return document tree structure JSON (without text fields).

Return total page count for a PDF document.

For Markdown documents, 'pages' are line numbers. Find nodes whose line_num falls within [min(page_nums), max(page_nums)] and return their te

Extract text for specific PDF pages (1-indexed). Prefer cached pages, fallback to PDF.

Preserve supported Agents SDK prefixes and route other provider paths via LiteLLM.

Parse a pages string like '5-7', '3,8', or '12' into a sorted list of ints.

Return JSON with document metadata: doc_id, doc_name, doc_description, type, status, page_count (PDF) or line_count (Markdown).

Return tree structure JSON with text fields removed (saves tokens).

Retrieve page content for a document. pages format: '5-7', '3,8', or '12' For PDF: pages are physical page numbers (1-indexed). For

Return page content for the given pages string (e.g. '5-7', '3,8', '12').

Helper function to get the parent structure code

Index a document. Returns a document_id.

Run a document QA agent using the OpenAI Agents SDK. Streams text output token-by-token and returns the full answer string. Tool calls are al