Thank you for your response. Then I will try to manipulate the parsed-subdirectory, as the pre-conversion of the pdf-file has a good resultat and recognizes the paragraphs correctly.
I thought about the following way:
- First conversion (pdf->epub): with options -d DEBUG_PIPELINE and substitute page-break with the marker "*****"
- Processing manually: Then manipulate the generated html-file in the parsed-subdirectory with sed using the introduced marker to still detect the page-break, which is important in my case to detect headings
- Second conversion (opf->epub): Then convert the opf-file in the parsed-subdirectory to epub