This project contains iText 9.5 examples and two main entrypoints:

• com.itextpdf.demo.ComplexTaggedPdfMain (creates a complex, tagged PDF 2.0 report)
• com.itextpdf.demo.TaggedPdfToMarkdownMain (converts a PDF to Markdown using the structure tree)

It reads a PDF and tries to build Markdown using the PDF structure tree (tagged content). If tags are missing or empty, it falls back to plain text extraction.

The converter prepends YAML front matter for dataset pipelines:

• source_file
• page_count
• extraction_mode (tagged_structure_tree or plain_text_fallback)

mvn clean compile

mvn -q dependency:build-classpath "-Dmdep.outputFile=target\cp.txt"
$cp = Get-Content "target\cp.txt"
java -cp "target\classes;$cp" com.itextpdf.demo.TaggedPdfToMarkdownMain "input.pdf" "output.md"

If output.md is omitted, the converter writes next to the input PDF using the same file name with .md extension.

mvn -q dependency:build-classpath "-Dmdep.outputFile=target\cp.txt"
$cp = Get-Content "target\cp.txt"
java -cp "target\classes;$cp" com.itextpdf.demo.ComplexTaggedPdfMain

This creates complex-tagged-report.pdf in the project root.

Install Python dependencies:

python -m pip install -r requirements.txt

Run the Docling extractor (defaults to complex-tagged-report.pdf):

python scripts/docling_extract.py

Run with explicit paths:

python scripts/docling_extract.py --input "complex-tagged-report.pdf" --output "complex-tagged-report.docling.md"

By default, the script writes <input>.docling.md next to the input PDF.