tools.prowlarr_search module

Search torrent indexers via a local Prowlarr instance (Docker *arr stack).

async tools.prowlarr_search.run(query, max_results=5, ctx=None)

Search every enabled Prowlarr indexer for a query and return ranked JSON.

Entry point for the search_prowlarr tool. It authorizes the caller, reads the Prowlarr endpoint from config, performs one unfiltered interactive search across all enabled indexers, then ranks the merged releases by seeders and truncates to max_results before returning them as a JSON string.

It gates on _check_web_search_access() (the WEB_SEARCH privilege, resolved against Redis), reads prowlarr_base_url and prowlarr_api_key from ctx.config, normalises the base URL via _normalize_base(), and over-fetches using _fetch_limit_for_sort(). The actual call is an async aiohttp GET to /api/v1/search (120s timeout, X-Api-Key header); the JSON response is ranked and trimmed by _sort_and_limit_releases(). Network and unexpected errors are caught, logged via logger.exception, and folded into an error payload rather than propagated. No module state is mutated.

Registered via the single-tool module contract (TOOL_NAME / run) and dispatched by tool_loader.load_tools; no direct in-repo Python callers.

Parameters:

• query (str) – Search keywords / release title; must be non-empty.

• max_results (int) – Releases to return after seeder-ranking; coerced to an int and capped at 200 (default 5).

• ctx (ToolContext | None) – The tool context supplying user_id, redis, and config (Prowlarr URL and API key).

Returns:

A JSON array of the top releases on success, or a JSON {"error": ...} string when authorization fails, the query is empty, config is missing the URL/key, Prowlarr returns a non-200 or invalid JSON, or a network/unexpected error occurs.

Return type:

str