import { LoadWebOptions, RagLoaderResult } from "./loader.type.mjs";

//#region ../ai/src/rag/loaders/load-web.d.ts
/**
 * Fetch a URL through the SSRF-safe outbound policy and load it into a single
 * {@link RagDocument} of readable text. The fetch ALWAYS goes through
 * `guardedFetch` — never a raw `fetch` — so the scheme allowlist, host
 * allowlist, post-DNS private-IP guard, timeout, and response-size cap from
 * {@link LoadWebOptions.policy} (or the strict defaults) always apply.
 *
 * HTML responses are run through the same tag-strip pass as {@link loadHtml}
 * (scripts/styles dropped, entities decoded, paragraph structure kept);
 * non-HTML text responses (`text/plain`, markdown, …) are used verbatim.
 * The document's `metadata.source` is the resolved URL, `metadata.title` is
 * the page `<title>` (HTML only, unless overridden), `metadata.contentType`
 * is the server-reported type, and `metadata.loader` is `"web"`.
 *
 * The output is the exact shape `index()` consumes, so a load feeds straight
 * in:
 *
 * @example
 * const kb = ai.rag({ embedder, store });
 * await kb.index(await loadWeb("https://example.com/guide"));
 *
 * @example
 * // Tighten the SSRF policy to a single host:
 * await kb.index(await loadWeb(url, {
 *   policy: { hostAllowlist: ["docs.example.com"], maxBytes: 2_000_000 },
 *   tags: ["docs"],
 * }));
 *
 * @param url - The absolute URL to fetch. Validated by the outbound policy.
 * @param options - `policy` (the {@link OutboundPolicy}) plus shared
 *   `id` / `metadata` / `tags` ({@link LoadWebOptions}).
 * @returns A {@link RagLoaderResult} (one document) ready for `rag.index()`.
 * @throws {OutboundPolicyError} when the policy blocks the URL, the request
 *   times out, the body exceeds the cap, or the response is not OK.
 */
declare function loadWeb(url: string, options?: LoadWebOptions): Promise<RagLoaderResult>;
//#endregion
export { loadWeb };
//# sourceMappingURL=load-web.d.mts.map