Paraxcel Document Toolkit¶

Role Sole Architecture & App Developer

Application Type Local-First Desktop GUI (Windows Executable)

Core Technologies Python, python-docx, Pydantic, Pandas, Tkinter

Accreditation Harvard CS50x Computer Science

Automated ETL Pipeline: Eliminates manual data entry by automatically extracting MCQs, highlighted correct options, and superscripts from DOCX files into validated Excel sheets.

Architecture & Extraction Flow¶

graph LR
    A["Raw DOCX Documents"] --> B["python-docx Run-Level XML Parser"]
    B --> C["Format &amp; Highlight Extraction ('para_utility')"]
    C --> D["Pydantic Schema Validation ('Question' Model)"]
    D --> E["Pandas Tabular Normalization"]
    E --> F["Normalized Excel Workbook (.xlsx)"]

Executive Overview¶

Paraxcel is a modular Python desktop utility built to automate the extraction of multiple-choice questions (MCQs), answers, and option formatting from Microsoft Word (.docx) documents into structured Excel workbooks (.xlsx).

Designed for educators and assessment coordinators, Paraxcel operates entirely offline with zero cloud dependencies. It parses low-level OpenXML document structures to reliably detect marked answers (font color, background highlights) and mathematical notations (superscripts, subscripts).

Technical Challenges & Architectural Solutions¶

1. Granular XML Run-Level Parsing¶

Challenge: Detecting highlighted or color-coded answers embedded within arbitrary paragraph runs across inconsistent Word formatting styles.
Solution: Engineered recursive run inspection routines in para_utility.py that query OpenXML font color, background tint, and strike-through attributes directly at the character run level.

2. Strict Schema Validation & Quality Enforcement¶

Challenge: Preventing corrupted or partially formatted Word documents from outputting malformed Excel rows.
Solution: Implemented declarative Pydantic schemas enforcing strict type bounds (question non-empty, exactly 4 validated options, valid answer index).

3. Dependency-Free Desktop Packaging¶

Challenge: Distributing a Python application to non-technical end-users without requiring a Python runtime environment.
Solution: Configured PyInstaller build pipelines with embedded icon resources (paraxcel.ico), packaging the application into a standalone Windows binary.

Verified Accreditation¶

Harvard CS50x: Introduction to Computer Science • Harvard University (CS50)

Application Screenshots¶

1. Desktop GUI Interface