Function encodePathInSiteContent

packages/gitbook/src/middleware.ts:548–601

* Encode path in a site content. * Special paths are not encoded and passed to be handled by the route handlers.

(rawPathname: string)

 * Special paths are not encoded and passed to be handled by the route handlers.
 */
function encodePathInSiteContent(rawPathname: string): {
    pathname: string;
    routeType?: 'static' | 'dynamic';
} {
    const pathname = removeLeadingSlash(removeTrailingSlash(rawPathname));

    if (pathname.match(/^~gitbook\/ogimage\/\S+$/)) {
        return { pathname };
    }

    // If the pathname is a markdown file, we rewrite it to ~gitbook/markdown/:pathname
    if (pathname.match(/\.md$/)) {
        const pagePathWithoutMD = pathname.slice(0, -3);
        return {
            pathname: `~gitbook/markdown/${encodeURIComponent(pagePathWithoutMD)}`,
            // The markdown content is always static and doesn't depend on the dynamic parameter (customization, theme, etc)
            routeType: 'static',
        };
    }

    // We skip encoding for paginated llms-full.txt pages (i.e. llms-full.txt/100)
    if (pathname.match(/^llms-full\.txt\/\d+$/)) {
        return { pathname, routeType: 'static' };
    }

    // If the pathname is an embedded page
    const embedPage = pathname.match(/^~gitbook\/embed\/page\/(\S+)$/);
    if (embedPage) {
        return {
            pathname: `~gitbook/embed/page/${encodeURIComponent(embedPage[1]!)}`,
        };
    }

    switch (pathname) {
        case '~gitbook/embed/assistant':
        case '~gitbook/icon':
            return { pathname };
        case '~gitbook/mcp':
        case 'llms.txt':
        case 'sitemap.xml':
        case 'sitemap-pages.xml':
        case 'robots.txt':
        case '~gitbook/embed/script.js':
        case '~gitbook/embed/demo':
            // LLMs.txt, sitemap, sitemap-pages and robots.txt are always static
            // as they only depend on the site structure / pages.
            return { pathname, routeType: 'static' };
        case '~gitbook/pdf':
            // PDF routes are always dynamic as they depend on the search params.
            return { pathname, routeType: 'dynamic' };
        default:
            return { pathname: encodeURIComponent(pathname || '/') };
    }
}

/**
 * Append all the query params from a URL to another URL.