WP 60.0 regression: UnicodeDecodeError: 'utf-8' codec can't decode byte 0x80 in position 0: invalid start byte

[castedo]

Bad news on 60.0.

STEPS:

Running weasyprint 60.0 with input7.html.txt

weasyprint input7.html out.pdf

I get exception:

  File "/usr/local/lib/python3.11/site-packages/weasyprint/layout/preferred.py", line 312, in inline_line_widths
    child_text[resume_index:].decode(), child.style,
    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
UnicodeDecodeError: 'utf-8' codec can't decode byte 0x80 in position 0: invalid start byte

Here's the full stderr output I get:
stderr.txt

Looking at the HTML and knowing the pipeline I use i don't think there is anything unusual with unicode here.

This is a regression affecting the document pipeline that I run regularly.

[castedo]

Also, I run this within a container so I can share the container image if you have trouble reproducing.

[liZe]

Hi @castedo!

It’s probably a duplicate of #1973. Does the current main branch work for you?

[castedo]

Good news! The main branch does work for me!