Code relying on the method make_bookmark_tree doesn't work since release 53.0

[vicmion]

Hello,

I just noticed that the method make_bookmark_tree of the class Document has been dropped since release 53.0
I was relying on that method to automatically generate an Index table at the beginning of a document.
Is there a way to make it work with the new version of weasyprint?

I took a look at this issue #1420 but couldn't find a solution.

Thanks

[EugenMayer]

@liZe we depend on make_bookmark_tree pretty much due to the way we generate our bookmarks.

You might remember #1121 (comment) and we run this in production.

We cannot upgrade to 53.0 due to this issue. What we do is

   bookmarks = document.make_bookmark_tree()
   toc_inject_page_numbers_for_bookmarks(bookmarks, root)

def toc_inject_page_numbers_for_bookmarks(bookmarks, root: Element):
    """Traverse all given bookmarks and it's children and for each bookmark
    find the referencing toc table entry and update that toc entry with the
    page_number from the bookmark
    """
    for title, (page_number, _, _), children, _ in bookmarks:
        title_ident: str = "toc-%s" % hashlib.md5(title.encode("utf-8")).hexdigest()

        # this should usually only be a single one, it is the toc entry which is
        # referencing the bookmark / header in the content. We identify the toc entry
        # using a md5 hash of the title .. instead of title in the text. this helps
        # us avoiding issues with the special chars/escaping in the xpath expression
        toc_entry = root.find(".//body/div[@class='content content-toctable']//a[@id='%s']//span[@class='toc-entry-page-number']" % title_ident)
        if toc_entry is not None:
            real_page_number = page_number + 1
            toc_entry.text = str(real_page_number)

        toc_inject_page_numbers_for_bookmarks(children, root)

So we really rely on getting all the bookmarks from the PDF. Can we sponsor this (we are a bronze sponsor/supporter). Any chance to get some work on this issue?

Thanks

[vicmion]

I'm posting the solution I found. Hopefully it can help someone with the same problem.

bookmark_tree = document.make_bookmark_tree()

can be replaced with

def make_bookmark_tree(document):
    res = []
    for (i, page) in enumerate(document.pages):
        for bmark in page.bookmarks:
            res.append((i, bmark[1]))
    return res

bookmark_tree = make_bookmark_tree(document)

Consider that, of all the information returned from make_bookmark_tree, all I needed was the label and the page number of the bookmark. Including more informations should be straightforward.

[EugenMayer]

@vicmion thank you for the workarround.

@liZe considering that this custom code operates on a internal data-structure and we already have 3 people requiring the make_bookmark_tree method, I would suggest that it is at least implemented internally, not externally while using internal data-structures.

What are your thoughts?

[EugenMayer]

@liZe thank you for implementing that quickly! Awesome!

[grewn0uille]

Hello!
So we finally re-added the make_bookmark_tree method that works like before :)