PdfConverter fails on valid PDF: ValueError not enough values to unpack (expected 4, got 2)

[eliassrm]

Summary

markitdown fails to convert a PDF with PdfConverter threw ValueError: not enough values to unpack (expected 4, got 2).

The ffmpeg / pydub warning appears first, but seems unrelated because this is a PDF conversion.

Command

markitdown chapter11.pdf -o chapter11.md

Result:
markitdown._exceptions.FileConversionException: File conversion failed after 1 attempts:
 - PdfConverter threw ValueError with message: not enough values to unpack (expected 4, got 2)

Traceback:
markitdown\converters\_pdf_converter.py
pdfminer.high_level.extract_text(...)
ValueError: not enough values to unpack (expected 4, got 2)

Environment
OS: Windows
Python: 3.13.13
markitdown: 0.1.6
pdfplumber: 0.11.9
pdfminer.six: 20251230

<img width="1835" height="798" alt="Image" src="https://github.com/user-attachments/assets/3a8681c3-df11-4cc8-a0f1-7c04131ac84b" />

[assinscreedFC]

Update / correction to my comment above. I tracked down the actual root cause, and it's not the MediaBox path (that one is already guarded in 20251230).

The crash comes from a form XObject with a malformed /BBox (e.g. [0 0] instead of 4 numbers). pdfminer.six unpacks it as (x, y, w, h) = bbox in LTFigure.__init__ (via do_Do), so a 2-element BBox raises not enough values to unpack (expected 4, got 2). pdfplumber shares this code path, which is why markitdown's fallback can't recover either.

Fixes are up:

• Root cause, upstream: Tolerate malformed form XObject /BBox instead of crashing pdfminer/pdfminer.six#1264 — validates the /BBox length + numeric values and falls back to a unit rect in non-strict mode (raises in strict).
• markitdown side: Don't abort PDF conversion on a single malformed page #2078 — per-page resilience in _pdf_converter.py so a single malformed page is skipped (with a debug log) instead of failing the whole document.

Together these resolve this issue. No repro PDF needed anymore — apologies for the earlier request.

[Jakeshadow]

This "expected 4, got 2" error is usually a PDF with no real text layer (image-only). Quick check: run pdftotext yourfile.pdf - first — if it returns nothing, you need OCR (Tesseract) before MarkItDown. Full PDF troubleshooting: https://markitdown-pro.com.

[eliassrm]

chapter 11.pdf
Hey guys, this is the file that I was trying to convert it. the file has enough text and images and I want to convert it to use as resource for Claude. anyway, check it out