BREAKING CHANGE(api): redesign smartsitemap around builder-based sitemap creation, parsing, validation, and import utilities

ts/smartsitemap.classes.feedimporter.ts

@@ -0,0 +1,159 @@
+import * as plugins from './smartsitemap.plugins.js';
+import type * as interfaces from './interfaces/index.js';
+
+/**
+ * Imports RSS/Atom feeds and converts them to sitemap URL entries.
+ * This is a unique feature of smartsitemap that competitors don't offer.
+ */
+export class FeedImporter {
+  /**
+   * Import from a feed URL, returning standard sitemap URL entries.
+   */
+  static async fromUrl(
+    feedUrl: string,
+    options?: interfaces.IFeedImportOptions,
+  ): Promise<interfaces.ISitemapUrl[]> {
+    const smartfeed = new plugins.smartfeed.Smartfeed();
+    const feed = await smartfeed.parseFeedFromUrl(feedUrl);
+    return FeedImporter.mapItems(feed.items, options);
+  }
+
+  /**
+   * Import from a feed XML string, returning standard sitemap URL entries.
+   */
+  static async fromString(
+    feedXml: string,
+    options?: interfaces.IFeedImportOptions,
+  ): Promise<interfaces.ISitemapUrl[]> {
+    const smartfeed = new plugins.smartfeed.Smartfeed();
+    const feed = await smartfeed.parseFeedFromString(feedXml);
+    return FeedImporter.mapItems(feed.items, options);
+  }
+
+  /**
+   * Import from a feed URL, returning news sitemap URL entries.
+   */
+  static async fromUrlAsNews(
+    feedUrl: string,
+    publicationName: string,
+    publicationLanguage?: string,
+    options?: interfaces.IFeedImportOptions,
+  ): Promise<interfaces.ISitemapUrl[]> {
+    const smartfeed = new plugins.smartfeed.Smartfeed();
+    const feed = await smartfeed.parseFeedFromUrl(feedUrl);
+    return FeedImporter.mapItemsAsNews(feed.items, publicationName, publicationLanguage ?? 'en', options);
+  }
+
+  /**
+   * Import from a feed string, returning news sitemap URL entries.
+   */
+  static async fromStringAsNews(
+    feedXml: string,
+    publicationName: string,
+    publicationLanguage?: string,
+    options?: interfaces.IFeedImportOptions,
+  ): Promise<interfaces.ISitemapUrl[]> {
+    const smartfeed = new plugins.smartfeed.Smartfeed();
+    const feed = await smartfeed.parseFeedFromString(feedXml);
+    return FeedImporter.mapItemsAsNews(feed.items, publicationName, publicationLanguage ?? 'en', options);
+  }
+
+  /**
+   * Map parsed feed items to standard sitemap URLs.
+   */
+  private static mapItems(
+    items: any[],
+    options?: interfaces.IFeedImportOptions,
+  ): interfaces.ISitemapUrl[] {
+    let filtered = FeedImporter.filterItems(items, options);
+
+    if (options?.mapItem) {
+      const results: interfaces.ISitemapUrl[] = [];
+      for (const item of filtered) {
+        const mapped = options.mapItem(item as interfaces.IFeedItem);
+        if (mapped) results.push(mapped);
+      }
+      return results;
+    }
+
+    return filtered
+      .filter((item: any) => item.link)
+      .map((item: any) => {
+        const url: interfaces.ISitemapUrl = {
+          loc: item.link,
+        };
+        if (item.isoDate) {
+          url.lastmod = item.isoDate;
+        }
+        return url;
+      });
+  }
+
+  /**
+   * Map parsed feed items to news sitemap URLs.
+   */
+  private static mapItemsAsNews(
+    items: any[],
+    publicationName: string,
+    publicationLanguage: string,
+    options?: interfaces.IFeedImportOptions,
+  ): interfaces.ISitemapUrl[] {
+    let filtered = FeedImporter.filterItems(items, options);
+
+    if (options?.mapItem) {
+      const results: interfaces.ISitemapUrl[] = [];
+      for (const item of filtered) {
+        const mapped = options.mapItem(item as interfaces.IFeedItem);
+        if (mapped) results.push(mapped);
+      }
+      return results;
+    }
+
+    return filtered
+      .filter((item: any) => item.link)
+      .map((item: any) => {
+        const url: interfaces.ISitemapUrl = {
+          loc: item.link,
+          news: {
+            publication: {
+              name: publicationName,
+              language: publicationLanguage,
+            },
+            publicationDate: item.isoDate || new Date().toISOString(),
+            title: item.title || '',
+            keywords: item.categories,
+          },
+        };
+        if (item.isoDate) {
+          url.lastmod = item.isoDate;
+        }
+        return url;
+      });
+  }
+
+  /**
+   * Apply date and limit filters to feed items.
+   */
+  private static filterItems(items: any[], options?: interfaces.IFeedImportOptions): any[] {
+    let result = [...items];
+
+    // Filter by date
+    if (options?.newerThan != null) {
+      const threshold = options.newerThan instanceof Date
+        ? options.newerThan.getTime()
+        : options.newerThan;
+
+      result = result.filter((item: any) => {
+        if (!item.isoDate) return true; // keep items without dates
+        return new Date(item.isoDate).getTime() >= threshold;
+      });
+    }
+
+    // Apply limit
+    if (options?.limit != null && options.limit > 0) {
+      result = result.slice(0, options.limit);
+    }
+
+    return result;
+  }
+}