Pymupdf4llm.to_markdown 1.28 creates unexpected columns that break the text

PyMuPDF
1

Hello pymupdf support!

Here is a small pdf that produces this output when I run pymupdf4llm.to_markdown("repro_layout_split.pdf")

=== Document parser messages ===
Using Tesseract for OCR processing.

|**Sentence**|||**Status**|**Date**|
|---|---|---|---|---|
|The quick brown fox jumped over the lazy d<br>dog again.|og and then the quick bro|wn fox jumped over the lazy|Pending|01/02/2024|
|The five boxing wizards jump quickly|||Done|01/02/2024|
|**Item**|**Category**|**Source**|||
|Widget assembly kit, large size|Hardware|Sheet 2|||
|Sprocket set, metric|Hardware|Sheet 3|||

Is there a way to keep the text `The quick brown fox jumped over the lazy dog and then the quick brown fox jumped over the lazy dog again.` from being broken?

repro_layout_split.pdf (9.0 KB)

2

This is the output under the coming release 1.28.1 - scheduled for next weekend or shortly thereafter:
test.md (394 Bytes)

1 Like