Extracted Table from PDF to Markdown is different from one in JSON

[etetteh]

🔎 Search before asking

• I have searched the PaddleOCR Docs and found no similar bug report.
• I have searched the PaddleOCR Issues and found no similar bug report.
• I have searched the PaddleOCR Discussions and found no similar bug report.

🐛 Bug (问题描述)

I just noticed that the extracted table in the exported markdown file is different from the exported JSON file, which causes data integrity and consistency issue.

The markdown output is accurate, but the JSON output is not.

virology_pg2_0_res.json
virology_pg2_0.md

🏃‍♂️ Environment (运行环境)

OS            macOS-26.2 
Environment   Jupyter
Python        3.13.2
PaddleOCR     3.4.0
Install       uv
RAM           16.00 GB
CPU           Apple M1
CUDA          None

🌰 Minimal Reproducible Example (最小可复现问题的Demo)

from paddlex import create_pipeline

pipeline = create_pipeline(pipeline="PaddleOCR-VL-1.5")

output = pipeline.predict(input="./data/virology_pg2.pdf")

pages_res = list(output)

output = pipeline.restructure_pages(pages_res)

# output = pipeline.restructure_pages(pages_res, merge_table=True) # Merge tables across pages
# output = pipeline.restructure_pages(pages_res, merge_table=True, relevel_titles=True) # Merge tables across pages and reconstruct multi-level titles
# output = pipeline.restructure_pages(pages_res, merge_table=True, relevel_titles=True, concatenate_pages=True) # Merge tables across pages, reconstruct multi-level titles, and merge multiple pages
for res in output:
    res.print() # Print the structured prediction output
    res.save_to_json(save_path="output") # Save the current image's structured result in JSON format
    res.save_to_markdown(save_path="output") # Save the current image's result in Markdown format

virology_pg2.pdf

[scyyh11]

Thanks for the feedback. I checked the JSON and Markdown you provided, after stripping the inline CSS that the Markdown exporter adds for rendering (border=1, style='margin: auto; ...', per-cell text-align: center), the two tables are byte-for-byte identical. In other words, the JSON stores the raw HTML table; the Markdown wraps it with presentation styling.