@signalwire/docusaurus-plugin-llms-txt/lib/discovery/content-classifier.js

Generate Markdown versions of Docusaurus HTML pages and an llms.txt index file

/**
 * Content classification logic
 * Classify routes as docs/blog/pages and apply include/exclude rules
 */
import { getContentConfig } from '../config';
import { DOCUSAURUS_BLOG_PLUGIN, DOCUSAURUS_PAGES_PLUGIN, CONTENT_TYPES, } from '../constants';
/**
 * Classify a route by its plugin type, with fallback heuristics for routes without plugin info
 * @internal
 */
export function classifyRoute(route) {
    const plugin = route.plugin;
    // If we have plugin info, use it
    if (plugin?.name) {
        switch (plugin.name) {
            case DOCUSAURUS_BLOG_PLUGIN:
                return CONTENT_TYPES.BLOG;
            case DOCUSAURUS_PAGES_PLUGIN:
                return CONTENT_TYPES.PAGES;
            default:
                return CONTENT_TYPES.DOCS;
        }
    }
    // Fallback heuristics for routes without plugin info (e.g., versioned docs)
    return classifyRouteByHeuristics(route);
}
/**
 * Classify routes using reliable component indicators when plugin info is not available
 * Only uses reliable indicators, avoiding fragile path/filename matching
 * @internal
 */
function classifyRouteByHeuristics(route) {
    // Check component type - docs typically use @theme/DocItem
    if (route.component === '@theme/DocItem') {
        return CONTENT_TYPES.DOCS;
    }
    // Check for reliable blog component indicators
    if (route.component === '@theme/BlogListPage' ||
        route.component === '@theme/BlogPostPage') {
        return CONTENT_TYPES.BLOG;
    }
    // Home page is typically a page
    if (route.path === '/') {
        return CONTENT_TYPES.PAGES;
    }
    // If no reliable indicators, mark as unknown rather than guessing
    return CONTENT_TYPES.UNKNOWN;
}
/**
 * Determines if a route should be processed based on plugin configuration
 * @internal
 */
export function shouldProcessRoute(route, options) {
    const contentConfig = getContentConfig(options);
    const routeType = classifyRoute(route);
    // First check if this content type should be included
    let shouldIncludeType = false;
    switch (routeType) {
        case CONTENT_TYPES.BLOG:
            shouldIncludeType = contentConfig.includeBlog;
            break;
        case CONTENT_TYPES.PAGES:
            shouldIncludeType = contentConfig.includePages;
            break;
        case CONTENT_TYPES.DOCS:
        case CONTENT_TYPES.UNKNOWN:
        default:
            shouldIncludeType = contentConfig.includeDocs;
            break;
    }
    if (!shouldIncludeType) {
        return false;
    }
    // For docs routes, check versioned docs filtering
    if ((routeType === CONTENT_TYPES.DOCS || routeType === CONTENT_TYPES.UNKNOWN) &&
        contentConfig.includeVersionedDocs === false) {
        // Check if this is a versioned docs route (not current version)
        const isVersionedRoute = '__docusaurus_isVersioned' in route ?
            route.__docusaurus_isVersioned :
            undefined;
        if (isVersionedRoute === true) {
            return false; // Skip versioned docs when includeVersionedDocs is false
        }
    }
    // Check if this is a generated category index page
    if (contentConfig.includeGeneratedIndex === false) {
        // Generated index pages have a categoryGeneratedIndex prop
        if (route.props?.categoryGeneratedIndex !== undefined) {
            return false; // Skip generated index pages when includeGeneratedIndex is false
        }
    }
    return true;
}