add sanitizing of url-pathname to avoid invalid urls within EDS

tools/importer/import-header.js

@@ -11,7 +11,10 @@
 
 /* global WebImporter */
 
-import { determineEdsBaseUrl } from './import-util.js';
+import {
+  determineEdsBaseUrl,
+  sanitizePathname,
+} from './import-util.js';
 
 /**
  * Exclude generic content from migration
@@ -57,6 +60,9 @@ const handleMenuEntry = (menuEntry, baseUrl) => {
 
   // check if the target of the link is relative: handle EDS-specific url-conversions
   if (href.charAt(0) === '/') {
+    // sanitize url-pathname
+    href = sanitizePathname(href);
+
     // add base-url
     href = baseUrl + href;
 

tools/importer/import-util.js

@@ -150,6 +150,24 @@ export const formatTableData = (table, formats) => {
   });
 };
 
+/**
+ * Sanitize pathname by removing leading and trailing dashes from each section of the url
+ * @param pathname
+ * @returns {string}
+ */
+export const sanitizePathname = (pathname) => {
+  const modifiedPathnameSections = [];
+
+  // handle leading or trailing dashes in url-sections, that are not allowed (anymore) in EDS
+  const pathnameSections = pathname.split('/');
+
+  pathnameSections.forEach((pathnameSection) => {
+    modifiedPathnameSections.push(pathnameSection.replace(/^-+|-+$/g, ''));
+  });
+
+  return modifiedPathnameSections.join('/');
+};
+
 /**
  * Preprocess method that extracts the hreflang=x-default value from the original HTML markup
  * @param document
@@ -306,6 +324,8 @@ export const handleLinks = (main, document, baseUrl) => {
 
       // replace relative urls
       if (href.charAt(0) === '/') {
+        href = sanitizePathname(href);
+
         href = baseUrl + href;
 
         // remove possible parameters from internal links

tools/importer/import.js

@@ -40,6 +40,7 @@ import {
   handleSup,
   handleReferenceRows,
   formatTableData,
+  sanitizePathname
 } from './import-util.js';
 
 const removeGenericContent = (main) => {
@@ -867,7 +868,7 @@ export default {
 
     results.push({
       element: main,
-      path: pathname,
+      path: sanitizePathname(pathname),
     });
 
     return results;