Why Calibre struggles with PDFs
A PDF has no paragraphs or chapters, only text placed on pages. Calibre's PDF input mostly keeps that text in order, so page numbers and running heads end up in the middle of the book, paragraphs break where pages ended, and words hyphenated at line ends stay split.
Calibre has no OCR either: a scanned PDF has no text for it to convert, so you get page images or nothing at all.
