PDF Processing Solution

This tool extracts a structured outline (Title, H1, H2, H3) from text-based PDFs using layout and font-based heuristics. It works fully offline, fits within Docker image constraints, and outputs a clean hierarchical JSON for each input document.

• Uses pdfplumber to extract text, layout (font size, position), and structure from each PDF page
• Determines the document title by selecting the largest and most centered text on the first page
• Remaining text is grouped by font size and numerically prefixed patterns (like 1., 1.2, etc.) to infer heading levels
• Heading levels are mapped to:
  • H1: Largest heading size after title
  • H2: Next size down
  • H3: Next size down (optional)
• Outputs a structured JSON outline with heading level, text, and page number

• Python 3.9
• pdfplumber: For PDF parsing and text extraction
• Standard Python libraries: json, re, pathlib

No ML models or training involved — heuristic, layout-aware approach ensures light weight and fast processing.

.
├── input/                  # Place your PDF files here
├── output/                 # JSON outputs appear here
├── main.py                 # Main logic for extraction
├── requirements.txt        # Python dependencies
├── Dockerfile              # Docker build file
├── Commands to run.txt     # Sample terminal commands
└── README.md               # You're reading it!

Recommended for reproducible and dependency-free execution.

1. Place your PDF(s) in the input/ folder.

2. Build the Docker image:

docker build --platform linux/amd64 -t mysolution:latest .

3. Run the container:

docker run --rm \
-v /absolute/path/to/input:/app/input \
-v /absolute/path/to/output:/app/output \
--network none mysolution:latest

Replace /absolute/path/to/input and /absolute/path/to/output with absolute paths on your system.

1. Install requirements:

pip install -r requirements.txt

2. Place your PDF(s) in input/ folder.

3. Run the script:

python main.py

Outputs will be saved as .json in the output/ directory.

Each output JSON will follow this structure:

{
  "title": "Document Title",
  "outline": [
    {"level": "H1", "text": "Introduction", "page": 1},
    {"level": "H2", "text": "1.1 Background", "page": 2},
    {"level": "H3", "text": "1.1.1 Problem Statement", "page": 2}
  ]
}

• Scanned/image-only PDFs are not supported (OCR not implemented)
• Works best with digital/text-based PDFs
• Ensure Docker volumes are mounted correctly and paths are absolute

This project is provided as-is for educational and prototyping purposes. Feel free to modify and extend.