export interface Sentinels {
    /** Stands for `{…}` / `${…}` template syntax inside a single literal. */
    placeholder: string;
    /** Stands for a value computed by code (a non-constant concatenation operand). */
    hole: string;
}
/**
 * Normalise text before any URL is read out of it.
 *
 *  - Private Use Area characters are removed, so sentinels stay out of band.
 *  - Tab, LF and CR are removed because the WHATWG URL parser removes them
 *    before parsing: `https://example.com\n@vendor.invalid/x` is a request to
 *    vendor.invalid with `example.com` as userinfo, and a scanner that stops at the
 *    newline reads the opposite of what the runtime does.
 */
export declare function sanitizeForScanning(text: string): string;
/**
 * Sanitiser for free text (Markdown, JSON, shell, dotenv).
 *
 * Deliberately does NOT strip newlines. Removing them is right inside a single
 * URL string, where the runtime parser removes them too; applied to a whole
 * file it concatenates unrelated lines, and a verbatim `https://vendor.invalid` at
 * end-of-line became `vendor.invalidUse` — invisible to every host rule.
 */
export declare function sanitizeProseForScanning(text: string): string;
/**
 * Two distinct sentinel characters.
 *
 * They are not searched for: `sanitizeForScanning` deletes the whole Private Use
 * Area from every literal as it is read, so these characters cannot appear in
 * the input and no per-input search is needed.
 */
export declare function pickSentinels(): Sentinels;
export interface UrlReference {
    /** The matched URL, trimmed of trailing punctuation. Sentinels removed. */
    url: string;
    /** Host as far as it is statically known, with sentinels still in place. */
    markedHost: string;
    /** Registrable domain, or undefined when it cannot be determined statically. */
    domain?: string;
    /** When `domain` is undefined, what supplied the undeterminable tail. */
    undeterminedBy?: "placeholder" | "hole";
}
export declare function stripSentinels(text: string, sentinels: Sentinels): string;
/**
 * Every absolute http(s) URL in a blob of text.
 *
 * `text` is expected to be already sentinel-marked when it came from folding.
 * Placeholder syntax inside the text is masked here so that a `{…}` body
 * containing `/`, `?` or `#` cannot cut the authority in half.
 */
export declare function extractUrlReferences(text: string, sentinels?: Sentinels): UrlReference[];
export interface ScanSource {
    /** Package-relative path, used for reporting. */
    file: string;
    content: string;
    /** Set when the bytes could not be decoded as text; `content` is then empty. */
    undecodable?: string;
}
export interface VendorHostFinding {
    file: string;
    url: string;
    host: string;
}
/**
 * Find known vendor domains even when they do not appear in a complete URL.
 */
export declare function findVendorDomainTokens(text: string): {
    domain: string;
    index: number;
}[];
/**
 * WEAK backstop: known vendor domains anywhere in the shipped bytes.
 *
 * A denylist over raw text. It cannot see a vendor domain that is not on
 * `VENDOR_CONTROLLED_DOMAINS`, and it cannot see a host split across string
 * literals. Both gaps are closed for executable code by `findDisallowedCodeUrls`,
 * which folds constants and works from an allowlist. This check exists for prose,
 * which cannot issue a request.
 */
export declare function findVendorHostReferences(sources: Iterable<ScanSource>): VendorHostFinding[];