/**
 * Page-skip detection helpers used by the auditor pipeline (v0.4.1+).
 *
 * Five policy filters live here, each gated by an AuditOption:
 *   - `detectNoindex(page)` honours `<meta name="robots" content="noindex">`
 *     and `X-Robots-Tag: noindex` HTTP headers. The site owner already told
 *     Google not to index these pages — auditing them produces noise the
 *     reader can't act on.
 *   - `detectAuthPage(page)` heuristically classifies pages as
 *     login / signup / password-reset based on three signals (password input
 *     in a thin body, title matches the auth regex, H1 matches the auth
 *     regex). Two signals are required for a positive verdict, which keeps
 *     the false-positive rate low: a marketing landing page with a single
 *     password input or a single auth-shaped heading won't trip it.
 *   - `detectBoilerplatePage(page)` (v0.4.2) flags cookie / legal / consent /
 *     imprint pages. A single signal is enough because the patterns are
 *     anchored + specific (the title or H1 matches the whole-string compliance
 *     regex, or the URL path is one of the well-known compliance slugs).
 *   - `detectSearchResultPage(page)` (v0.4.2) flags pages that look like
 *     internal search-result URLs (query parameter `q` / `query` / `search` /
 *     `s` / `keyword`, or path under `/search`). Per Google's own guidance
 *     these should be noindex'd — auditing them generates noise.
 *   - `detectEmptyBodyPage(page)` (v0.4.2) flags un-hydrated SPA shells:
 *     body text < 100 chars, script tags present, and no substantive
 *     `<noscript>` fallback. These fail every content rule but the underlying
 *     fix is server-side rendering, not content quality.
 *
 * All functions are pure and synchronous so they can be called inside the
 * `parsedPages` map step without disrupting the existing pipeline shape.
 */
import type { ParsedPage } from "./types.js";
/**
 * Returns true when the page is explicitly noindex'd via either the parsed
 * `<meta name="robots">` content or the `X-Robots-Tag` HTTP response header.
 * Match is case-insensitive and substring-based to tolerate combined
 * directives like `"noindex, nofollow"` or `"index, noindex"` (the latter is
 * a common bug — `noindex` wins per spec).
 */
export declare function detectNoindex(page: ParsedPage): boolean;
export interface AuthDetectionResult {
    isAuth: boolean;
    signals: Array<"password-input" | "auth-title" | "auth-h1">;
}
/**
 * Heuristic auth-page detection. Returns `isAuth: true` when 2+ signals
 * fire (high confidence). Single-signal pages return `isAuth: false` but
 * still expose the signal list for diagnostics — useful for tracking
 * borderline cases that might warrant a manual config exclude.
 */
export declare function detectAuthPage(page: ParsedPage): AuthDetectionResult;
export interface BoilerplateDetectionResult {
    isBoilerplate: boolean;
    signals: Array<"boilerplate-title" | "boilerplate-h1" | "boilerplate-url">;
}
/**
 * Heuristic boilerplate-page detection. Single-signal trigger (1+ = positive)
 * because the patterns are anchored and specific — a marketing page that
 * merely mentions "privacy" in its body won't fire any signal. Cookie / legal
 * / consent / imprint pages exist for compliance, never as SEO targets, so
 * auditing them produces routine findings the user already knows about.
 */
export declare function detectBoilerplatePage(page: ParsedPage): BoilerplateDetectionResult;
/**
 * Returns true when the URL has search-result hallmarks: a recognised search
 * query parameter (`?q=`, `?query=`, `?search=`, `?s=`, `?keyword=`) or a
 * pathname of `/search` or starting with `/search/`. Returns false for
 * unparseable URLs (e.g. filesystem paths from local directory audits).
 */
export declare function detectSearchResultPage(page: ParsedPage): boolean;
export interface EmptyBodyDetectionResult {
    isEmpty: boolean;
    reason?: string;
}
/**
 * Heuristic detection of un-hydrated SPA shells. Three conditions must hold:
 *   1. `contentText` (post-render-or-parse body text) trimmed < 100 chars
 *   2. The HTML contains at least one `<script src=...>` tag (script-driven)
 *   3. No substantive `<noscript>` fallback (combined noscript content
 *      <= 200 chars). A page with a long noscript block is doing
 *      progressive enhancement, not failing — don't flag it.
 *
 * Returns `{ isEmpty: true, reason: "spa-shell" }` when all three hold.
 */
export declare function detectEmptyBodyPage(page: ParsedPage): EmptyBodyDetectionResult;
/**
 * Combined skip decision for the auditor pipeline. Returns the reason string
 * to surface in `summary.skippedUrls` diagnostics, or `null` when the page
 * should be audited normally. Order matters — the priority is:
 *   `noindex` > `auth-detected` > `boilerplate` > `search-result` > `spa-shell`.
 *
 *   - `noindex` first: the site-owner-declared signal beats every heuristic.
 *   - `auth-detected` next: a confirmed auth surface wins over generic
 *     boilerplate / search / shell heuristics.
 *   - `boilerplate` next: a clearly compliance-page-shaped URL beats the
 *     generic search/shell signals.
 *   - `search-result` next: explicit search hallmarks beat the empty-body
 *     fallback (a search page MAY be a SPA but the better label is "search").
 *   - `spa-shell` last: the most generic "we couldn't audit this" reason.
 */
export interface PageSkipOptions {
    respectNoindex: boolean;
    skipDetectedAuth: boolean;
    skipBoilerplate: boolean;
    skipSearchPages: boolean;
    skipEmptyBody: boolean;
}
export declare function pageSkipReason(page: ParsedPage, options: PageSkipOptions): "noindex" | "auth-detected" | "boilerplate" | "search-result" | "spa-shell" | null;
//# sourceMappingURL=page-filter.d.ts.map