@djangocfg/seo 2.1.109 → 2.1.110

package/dist/cli.mjs

@@ -8,7 +8,7 @@ import { searchconsole } from '@googleapis/searchconsole';
 import pLimit from 'p-limit';
 import pRetry from 'p-retry';
 import { JWT } from 'google-auth-library';
-import { load } from 'cheerio';
+import { parseHTML, DOMParser } from 'linkedom';
 import robotsParser from 'robots-parser';
 import { mkdir, writeFile } from 'fs/promises';
 import * as linkinator from 'linkinator';
@@ -753,14 +753,16 @@ var SiteCrawler = class {
    * Parse HTML and extract SEO-relevant data
    */
   parseHtml(html, result, pageUrl, depth) {
-    const $ = load(html);
-    result.title = $("title").first().text().trim() || void 0;
+    const { document } = parseHTML(html);
+    const titleEl = document.querySelector("title");
+    result.title = titleEl?.textContent?.trim() || void 0;
     if (!result.title) {
       result.warnings.push("Missing title tag");
     } else if (result.title.length > 60) {
       result.warnings.push(`Title too long (${result.title.length} chars, recommended: <60)`);
     }
-    result.metaDescription = $('meta[name="description"]').attr("content")?.trim() || void 0;
+    const metaDesc = document.querySelector('meta[name="description"]');
+    result.metaDescription = metaDesc?.getAttribute("content")?.trim() || void 0;
     if (!result.metaDescription) {
       result.warnings.push("Missing meta description");
     } else if (result.metaDescription.length > 160) {
@@ -768,25 +770,28 @@ var SiteCrawler = class {
         `Meta description too long (${result.metaDescription.length} chars, recommended: <160)`
       );
     }
-    result.metaRobots = $('meta[name="robots"]').attr("content")?.trim() || void 0;
-    const xRobots = $('meta[http-equiv="X-Robots-Tag"]').attr("content")?.trim();
-    if (xRobots) {
-      result.metaRobots = result.metaRobots ? `${result.metaRobots}, ${xRobots}` : xRobots;
+    const metaRobots = document.querySelector('meta[name="robots"]');
+    result.metaRobots = metaRobots?.getAttribute("content")?.trim() || void 0;
+    const xRobots = document.querySelector('meta[http-equiv="X-Robots-Tag"]');
+    const xRobotsContent = xRobots?.getAttribute("content")?.trim();
+    if (xRobotsContent) {
+      result.metaRobots = result.metaRobots ? `${result.metaRobots}, ${xRobotsContent}` : xRobotsContent;
     }
-    result.canonicalUrl = $('link[rel="canonical"]').attr("href")?.trim() || void 0;
+    const canonical = document.querySelector('link[rel="canonical"]');
+    result.canonicalUrl = canonical?.getAttribute("href")?.trim() || void 0;
     if (!result.canonicalUrl) {
       result.warnings.push("Missing canonical tag");
     }
-    result.h1 = $("h1").map((_, el) => $(el).text().trim()).get();
-    result.h2 = $("h2").map((_, el) => $(el).text().trim()).get();
+    result.h1 = Array.from(document.querySelectorAll("h1")).map((el) => el.textContent?.trim() || "");
+    result.h2 = Array.from(document.querySelectorAll("h2")).map((el) => el.textContent?.trim() || "");
     if (result.h1.length === 0) {
       result.warnings.push("Missing H1 tag");
     } else if (result.h1.length > 1) {
       result.warnings.push(`Multiple H1 tags (${result.h1.length})`);
     }
-    $("a[href]").each((_, el) => {
-      const href = $(el).attr("href");
-      if (!href) return;
+    for (const el of document.querySelectorAll("a[href]")) {
+      const href = el.getAttribute("href");
+      if (!href) continue;
       try {
         const linkUrl = new URL(href, pageUrl);
         if (linkUrl.hostname === this.baseUrl.hostname) {
@@ -800,18 +805,19 @@ var SiteCrawler = class {
         }
       } catch {
       }
-    });
-    $("img").each((_, el) => {
-      const src = $(el).attr("src");
-      const alt = $(el).attr("alt");
+    }
+    for (const el of document.querySelectorAll("img")) {
+      const src = el.getAttribute("src");
+      const alt = el.getAttribute("alt");
+      const hasAltAttr = alt !== null;
       if (src) {
         result.images.push({
           src,
-          alt,
-          hasAlt: alt !== void 0 && alt.trim().length > 0
+          alt: alt ?? void 0,
+          hasAlt: hasAltAttr && alt.trim().length > 0
         });
       }
-    });
+    }
     const imagesWithoutAlt = result.images.filter((img) => !img.hasAlt);
     if (imagesWithoutAlt.length > 0) {
       result.warnings.push(`${imagesWithoutAlt.length} images without alt text`);
@@ -1145,28 +1151,31 @@ async function analyzeSitemap(sitemapUrl) {
         metadata: { contentType }
       });
     }
-    const $ = load(content, { xmlMode: true });
-    const sitemapIndex = $("sitemapindex");
-    if (sitemapIndex.length > 0) {
+    const parser = new DOMParser();
+    const doc = parser.parseFromString(content, "text/xml");
+    const sitemapIndex = doc.querySelector("sitemapindex");
+    if (sitemapIndex) {
       analysis.type = "sitemap-index";
-      $("sitemap").each((_, el) => {
-        const loc = $("loc", el).text().trim();
+      for (const sitemap of doc.querySelectorAll("sitemap")) {
+        const loc = sitemap.querySelector("loc")?.textContent?.trim();
         if (loc) {
           analysis.childSitemaps.push(loc);
         }
-      });
+      }
       consola3.debug(`Sitemap index contains ${analysis.childSitemaps.length} sitemaps`);
     } else {
       analysis.type = "sitemap";
-      $("url").each((_, el) => {
-        const loc = $("loc", el).text().trim();
+      for (const url of doc.querySelectorAll("url")) {
+        const loc = url.querySelector("loc")?.textContent?.trim();
         if (loc) {
           analysis.urls.push(loc);
         }
-      });
-      const lastmod = $("url lastmod").first().text().trim();
-      if (lastmod) {
-        analysis.lastmod = lastmod;
+        if (!analysis.lastmod) {
+          const lastmod = url.querySelector("lastmod")?.textContent?.trim();
+          if (lastmod) {
+            analysis.lastmod = lastmod;
+          }
+        }
       }
       consola3.debug(`Sitemap contains ${analysis.urls.length} URLs`);
     }