A Model-Driven QTI-Based Pipeline for PDF to Moodle Conversion

A Model-Driven QTI-Based pipeline that converts PDF-based assessments into QTI (Question & Test Interoperability) 3.0 format and then exports them as Moodle-compatible XML quizzes.

Project Overview

This pipeline automates the conversion of PDF documents into structured assessment formats:

1. PDF → QTI XML - Extract questions from PDFs and generate QTI 3.0 XML using GPT API.
2. QTI XML → QTI Model - Parse QTI XML into a QTI-based model.
3. QTI Model → Moodle XML - Generate Moodle quiz XML from the QTI model.

Project Structure

PDF-to-LMS-Converter/
├── src/
│   ├── __init__.py
│   ├── __pycache__/
│   ├── pdf_to_qti/              # PDF processing and QTI generation
│   │   ├── __init__.py
│   │   ├── pdf_to_qti.py        # Main PDF extraction and QTI generation
│   │   ├── config.py            # Configuration settings (paths)
│   │   ├── __pycache__/
│   │   └── llm_assistant/       # Assistant materials for LLM-based QTI generation
│   │
│   └── qti_to_lms/              # QTI conversion and LMS export
│       ├── __init__.py
│       ├── besser_to_moodle.py  # QTI model to Moodle XML conversion
│       ├── qti_to_besser.py     # QTI XML to QTI model conversion
│       ├── qti_to_lms.py        # Main qti_to_lms pipeline orchestrator
│       ├── __pycache__/
│       ├── metamodel/           # Domain models
│       │   └── qti.py           # QTI-based metamodel classes
│       └── templates/           # Jinja2 templates
│           └── moodle_template.py.j2
│
├── evaluation_scripts/          # Evaluation scripts for correctness assessment
│   ├── pdf_to_qti_eval.py       # Evaluates PDF-to-QTI conversion accuracy
│   └── qti_to_moodle_eval.py    # Evaluates QTI-to-Moodle conversion accuracy
│
├── evaluation_results/          # Evaluation outputs
│   ├── Canterbury_Question_Bank/
│   └── QTI3_Examples/
├── extensibility/             # French language resources
├── run_pipeline.py              # Main pipeline module
├── script.py                    # Example usage script
├── requirements.txt             # Python dependencies
├── __pycache__/
└── README.md                   # This file

Installation

Prerequisites

• Python 3.8+
• OpenAI API key (for LLM-based PDF analysis)

Setup

1. Clone or navigate to the project:

   cd PDF-to-LMS-Converter

2. Create a virtual environment (recommended):

   python -m venv venv
   source venv/bin/activate  # On Windows: venv\Scripts\activate

3. Install dependencies:

   pip install -r requirements.txt

4. Configure API key: Obtain an OpenAI API key and set it in your script (see Usage below).

Usage

Quick Start

To run the pipeline, create a Python script (e.g., script.py) that imports and calls the pdf_to_moodle function:

from run_pipeline import pdf_to_moodle

qti_file = pdf_to_moodle(
    api_key="your-openai-api-key-here",
    pdf_path=r"path\to\input\pdf\file.pdf",
    output_folder=r"path\to\output\folder"  # Optional
)

Then run the script:

python script.py

This will:

1. Process the specified PDF file
2. Generate QTI XML files
3. Convert to QTI-based model
4. Export to Moodle XML format in the specified output folder (or output/ by default)

Dependencies

Package	Version	Purpose
requests	≥2.31.0	HTTP requests for OpenAI API
pdfplumber	≥0.11.0	PDF text extraction
retrying	≥1.3.3	Retry logic for API calls
jinja2	≥3.1.2	Template rendering for Moodle XML

Architecture

QTI Metamodel (qti.py)

Represents a QTI-based metamodel with these domain classes:

Enumerations:

• NavigationModeEnum - Assessment navigation (LINEAR, NONLINEAR)
• SubmissionModeEnum - Response submission (INDIVIDUAL, SIMULTANEOUS)
• ShowHideEnum - Conditional visibility (SHOW, HIDE)

Core Classes:

• AssessmentDefinition - Root container for a complete assessment
• AssessmentPart - High-level Assessment division with navigation/submission modes
• AssessmentSection - Mid-level content organizer with visibility control
• Question - Assessment item with body, responses, outcomes, and feedback
• QuestionBody - Question content with selectable blocks
• ResponseDeclaration - Expected response structure and scoring
• ModalFeedback - Conditional feedback based on responses

File Formats

Input

• PDF Files - Assessment documents in input/ directory

Output

• QTI XML (output/qti_output.xml) - QTI 3.0 standard format
• Moodle XML (output/moodle.xml) - Moodle quiz import format

Code Quality

The codebase follows PEP 8 standards with strict pylint compliance.

Troubleshooting

PDF Extraction Issues

• Ensure PDFs are text-based (not scanned images)
• Use pdfplumber to manually Assessment extraction: pdfplumber.open("file.pdf")

QTI Generation Failures

• Check API key validity and quota
• Verify OpenAI API is accessible
• Check PDF file encoding

Moodle Import Issues

• Validate Moodle XML format
• Check compatibility with Moodle version
• Verify character encoding (UTF-8)