Conversion fails with munmap_chunk(): invalid pointer

[marioangst]

Bug

Docling fails in a rare case and aborts fatally when parsing a specific PDF with a C-level error: munmap_chunk(): invalid pointer

If running as part of a larger pipeline, the error I get in my logs is: corrupted size vs. prev_size.

The PDF in question is attached.

Steps to reproduce

from docling.document_converter import DocumentConverter

converter = DocumentConverter()

result_corrupt_pdf = converter.convert(
    "corrupt_pdf.pdf"
)

Docling version

Docling version: 2.26.0 Docling Core version: 2.23.1 Docling IBM Models version: 3.4.1 Docling Parse version: 3.4.0 Python: cpython-312 (3.12.7) Platform: Linux-6.8.0-49-generic-x86_64-with-glibc2.39

Python version

Python 3.12.7

corrupt_pdf.pdf