Document export as markdown missing out some texts

[penquin17]

Bug

When I converted a document and exported it to markdown format, the results showed differences from the two version docling==2.21 and docling==2.20.
I used export labels including DEFAULT_EXPORT_LABELS plus DocItemLabel.FORM, DocItemLabel.KEY_VALUE_REGION

Steps to reproduce

1. Install the version docling==2.21 or docling==2.20
2. Run the following

source = "./test-document.pdf"  # document per local path or URL
print(os.path.exists(source))
IMAGE_RESOLUTION_SCALE = 2.0

# previous `PipelineOptions` is now `PdfPipelineOptions`
pipeline_options = PdfPipelineOptions()
pipeline_options.do_ocr = True
pipeline_options.do_table_structure = True
pipeline_options.images_scale = IMAGE_RESOLUTION_SCALE
pipeline_options.generate_page_images = True
pipeline_options.generate_picture_images = True
ocr_options = TesseractOcrOptions(force_full_page_ocr=False, lang=["auto"])
pipeline_options.ocr_options = ocr_options
accelerator_options = AcceleratorOptions(
    num_threads=4, device=AcceleratorDevice.CPU
)
pipeline_options.accelerator_options = accelerator_options
# ...

# Custom options are now defined per format.
doc_converter = (
    DocumentConverter(  # all of the below is optional, has internal defaults.
        allowed_formats=[
            InputFormat.PDF,
            InputFormat.IMAGE,
            InputFormat.DOCX,
            InputFormat.HTML,
            InputFormat.PPTX,
            InputFormat.XLSX,
        ],  # whitelist formats, non-matching files are ignored.
        format_options={
            InputFormat.PDF: PdfFormatOption(
                pipeline_options=pipeline_options,  # pipeline options go here.
                backend=DoclingParseV2DocumentBackend  # optional: pick an alternative backend
            ),
            InputFormat.DOCX: WordFormatOption(
                pipeline_cls=SimplePipeline  # default for office formats and HTML
            ),
        },
    )
)

# read file
from io import BytesIO
stream = DocumentStream(
    name=source,
    stream=BytesIO(open(source, mode='rb').read())
)
result = doc_converter.convert(stream, raises_on_error=False)

# export to md
print(result.document.export_to_markdown())

Export results

docling==2.20

<!-- image -->

## INTERNATIONAL MEDICAL CENTER

Address: 1234 Fake Street, District 9, City Hotline: (555) 123-4567 | Email: [email protected]

## Patient Medical Record Report

## Title: Well Child Visit (Procedure)

| Aspect                     | Before Treatment              | After Treatment              |
|----------------------------|-------------------------------|------------------------------|
| Immunizations Administered | Hib (PRP-OMP)                 | Hib (PRP-OMP)                |
|                            | Rotavirus, monovalent         | Rotavirus, monovalent        |
|                            | IPV                           | IPV                          |
|                            | DTaP                          | DTaP                         |
|                            | Pneumococcal conjugate PCV 13 | Pneumococcal conjugal PCV 13 |

## Notes:

- ·  The patient received the following immunizations during the well-child visit: Hib (PRP-OMP), rotavirus, monovalent, IPV, DTaP, and Pneumococcal conjugate PCV 13.
- ·  No adverse reactions were noted following the administration of these immunizations.

Confidential Medical Record | All rights reserved Generated on: 2025-02-11

docling==2.21

<!-- image -->

## INTERNATIONAL MEDICAL CENTER

Address: 1234 Fake Street, District 9, City Hotline: (555) 123-4567 | Email: [email protected]

## Patient Medical Record Report

## Title: Well Child Visit (Procedure)

| Aspect                     | Before Treatment              | After Treatment              |
|----------------------------|-------------------------------|------------------------------|
| Immunizations Administered | Hib (PRP-OMP)                 | Hib (PRP-OMP)                |
|                            | Rotavirus, monovalent         | Rotavirus, monovalent        |
|                            | IPV                           | IPV                          |
|                            | DTaP                          | DTaP                         |
|                            | Pneumococcal conjugate PCV 13 | Pneumococcal conjugal PCV 13 |

## Notes:

- ·  The patient received the following immunizations during the well-child visit: Hib (PRP-OMP), rotavirus, monovalent, IPV, DTaP, and Pneumococcal conjugate PCV 13.
- ·  No adverse reactions were noted following the administration of these immunizations.

missing: Confidential Medical Record | All rights reserved Generated on: 2025-02-11

Docling version

docling==2.21 or docling==2.20

Python version

python=3.12

Attachment

test-doc.pdf

[Adolar13]

Same issue in docling 2.22.0.
Markdown is missing footer texts even with explicitly setting label.
Works in docling 2.20.0

[Adolar13]

Possibly tied to this pull request?:
DS4SD/docling-core#148