and what is, when I design self, and blend-out some things ?
so, the div ? called xyz will be visible = false per java script ?

does doxygen onready() per jquery ?

I have the following documentation Code:

/**
 * \defgroup keywords_ccpp für C
 * @{
 * \cond german
 * \brief C und C++
 *        eine keiner meiner
 * \endcond
 */
struct auto         {};
struct char         {};
struct register     {};
struct return       {};
struct struct       {};
struct typedef      {};
struct void         {};

/**
 * \htmlonly
 * <script>
 * document.addEventListener("DOMContentLoaded", function() {
 *   document.querySelectorAll("h2.groupheader").forEach(el => {
 *      if (el.textContent.includes("Datenstrukturen")) {
 *          el.textContent = el.textContent.replace("Datenstrukturen", "Schlüsselwörter");
 *      }
 *   })
 * });
 * </script>
 * \endhtmlonly
 */

But \htmlonly does not work - the script Code is not in the output html file.

I use version:
1.13.1 (cb7d0d0)

if you need the sources:
packed.zip

unde doc, you can find a Windows Console Batch file: build.bat

I have re-write the JavaScript Code with older coding style:
And what I get is the following:

I used the following Code:

/**
 * \defgroup keywords_ccpp für C
 * @{
 * \cond german
 * \brief C und C++
 *        eine keiner meiner
 * \endcond
 *
 * \htmlonly
 * <script>
 * document.addEventListener("DOMContentLoaded", function() {
 *   var headers = document.querySelectorAll("h2.groupheader");
 *   for (var i = 0; i < headers.length; i++) {
 *     var el = headers[i];
 *     if (el.textContent.indexOf("Datenstrukturen") !== -1) {
 *       el.textContent = el.textContent.replace("Datenstrukturen", "Schlüsselwörter");
 *     }
 *   }
 * });
 * </script>
 * \endhtmlonly
 */

I see:

 href=""../../d9/d98/structauto.html""

so it looks to me that there are a few double quotes (´"´) too many

How can I pre-process it during the Translation of doxygen ?

What do you mean? What do you want to preprocess`What do you mean with "Translation of doxygen"?

doxygen translate Source Code to HTML.
With pre-process during Translation I mean post processing: if doxygen ready to transform Text after Page Creation ?
So, I can manipulate the HTML File.

I know, I can read-in the HTML Files after all doxygen has finish work.
But then, I have to check all Files of a pattern match.

This is not a realy big Problem for, because I could done this per Python and regex Library.

But I was thinking that it is possible onboard by doxygen, to post process some parts.
Sorry for the pre - should be named post.

OK I understand what you want.
There is no possibility in doxgen to post process the output and hen there would be something like OUTPUT_FILTER this would lead the same as what would have to be done in your description: "I could done this per Python and regex Library", the user still has to provide the "FILTER"