Segmentation Fault with Empty Markdown Heading

[AstraBert]

Description:
Hello!

First, thank you for this fantastic tool 🥰.

I encountered a segmentation fault when using my tool, which is built on top of markdown-pdf. The issue arises when converting a Markdown file containing an empty heading (line 59). I’ve attached the file for reference.

While I understand that this edge case is rare, I wanted my tool to handle such situations seamlessly, so I investigated further and identified the root cause in your code.

---

Steps to Reproduce:

1. Use the attached Markdown file: line.md.
2. Attempt to convert it using markdown-pdf.

---

Observed Behavior:
A segmentation fault occurs when processing the file.

---

Analysis:
The issue is triggered at the following line during the second iteration of the while loop in the add_section method:

story.element_positions(self._recorder, {"toc": section.toc, "pdfile": self})

Here's the relevant code snippet for context:

def add_section(self, section: Section, user_css: typing.Optional[str] = None) -> None:
    """Add markdown section to pdf."""
    rect = fitz.paper_rect(section.paper_size)
    print("Rect ok!")
    where = rect + section.borders
    print("Where ok!")
    story = fitz.Story(html=self.m_d.render(section.text), archive=section.root, user_css=user_css)
    print("Story ok!")
    more = 1
    while more:  # loop outputting the story
        self.page += 1
        device = self.writer.begin_page(rect)
        print("Device ok!")
        more, _ = story.place(where)  # layout into allowed rectangle
        print("Place ok!")
        >>>story.element_positions(self._recorder, {"toc": section.toc, "pdfile": self})
        print("Positions ok!")
        story.draw(device)
        print("Device ok!")
        self.writer.end_page()
        print("End page ok!")

(Note: The print statements were added for debugging purposes.)

---

Root Cause:
The segmentation fault seems to be related to the handling of empty headings in Markdown.

---

Proposed Solution:
Upgrading the PyMuPDF dependency resolves the issue. Specifically:

• Current PyMuPDF Issue: The problem appears to relate to [this issue](SegFault 11 when empty H1 H2 H3 H4 etc element is used in insert_htmlbox pymupdf/PyMuPDF#3559), which affects empty headers.
• Resolution: Upgrading PyMuPDF to version 1.24.6 or later resolves the segmentation fault. I verified this by upgrading to version 1.25.1, which generated a clean PDF.

---

Suggestion:
Please consider updating your PyMuPDF dependency to version >= 1.24.6.

---

Thank you for your time, and let me know if I can provide further details! 😊

[vb64]

Thank you for your report. You are an excellent QA :)

[vb64]

Fixed in version 1.3.3. Please, update your installation.

[AstraBert]

Thank you so much!