Function get_toc

src_classic/utils.py:1025–1078

Create a table of contents. Args: simple: a bool to control output. Returns a list, where each entry consists of outline level, title, page number and link destination (if simple = False). For details see PyMuPDF's documentation.

(
    doc: Document,
    simple: bool = True,
)

def get_toc(
    doc: Document,
    simple: bool = True,
) -> list:
    """Create a table of contents.

    Args:
        simple: a bool to control output. Returns a list, where each entry consists of outline level, title, page number and link destination (if simple = False). For details see PyMuPDF's documentation.
    """

    def recurse(olItem, liste, lvl):
        """Recursively follow the outline item chain and record item information in a list."""
        while olItem:
            if olItem.title:
                title = olItem.title
            else:
                title = " "

            if not olItem.is_external:
                if olItem.uri:
                    if olItem.page == -1:
                        resolve = doc.resolve_link(olItem.uri)
                        page = resolve[0] + 1
                    else:
                        page = olItem.page + 1
                else:
                    page = -1
            else:
                page = -1

            if not simple:
                link = getLinkDict(olItem)
                liste.append([lvl, title, page, link])
            else:
                liste.append([lvl, title, page])

            if olItem.down:
                liste = recurse(olItem.down, liste, lvl + 1)
            olItem = olItem.next
        return liste

    # ensure document is open
    if doc.is_closed:
        raise ValueError("document closed")
    doc.init_doc()
    olItem = doc.outline
    if not olItem:
        return []
    lvl = 1
    liste = []
    toc = recurse(olItem, liste, lvl)
    if doc.is_pdf and simple is False:
        doc._extend_toc_items(toc)
    return toc


def del_toc_item(