How I Built a CSV Data Cleaner in 4 Days (Python Beginner Working Project)

Source: Dev.to

Background

After 2+ years in QA (Meta, Microsoft) and RPA consulting, I decided to transition to automation engineering. This is my first Python project, built in 4 days, documented completely.

The Challenge

Build a production‑ready CSV cleaner that:

• Never loses data (even invalid entries)
• Provides detailed error reports
• Handles real‑world messy data
• Uses quality‑first principles

What I Built

A Python script that:

• ✅ Cleans 1000+ contacts in seconds
• ✅ Validates emails, phones, names, ages
• ✅ Separates valid from invalid data
• ✅ Generates detailed error reports

The Journey (Day by Day)

Day 1‑2: Python Fundamentals

• Variables, strings, functions
• Dictionaries and lists
• CSV file handling

Hardest part: Understanding loops and data flow

Day 3: Building the Core

• Wrote 8 cleaning & validation functions
• Implemented error handling

Breakthrough moment: Realizing each function should return errors as a list

Day 4: Integration & Testing

• Combined all functions
• Added file writing
• Tested with messy data

Key learning: Separation of concerns (cleaning vs validation)

Key Code Sections

The Validation Pattern

def validate_email(email):
    """Check email structure"""
    errors = []

    if "@" not in email:
        errors.append("Missing @")

    # More checks...

    return errors

• Returns a list (can collect multiple errors)
• Clear error messages
• Easy to extend

The Main Loop

for row_num, row in enumerate(reader, start=2):
    all_errors = []

    # Clean
    cleaned_name = clean_name(row.get("Name", ""))

    # Validate
    all_errors.extend(validate_name(cleaned_name))

    # Decide
    if all_errors:
        error_contacts.append(...)
    else:
        clean_contacts.append(...)

What I Learned

Technical Skills

• Python fundamentals
• CSV processing
• Error handling patterns
• Function design for reusability

Meta‑Skills

• How to learn efficiently (fundamentals before frameworks)
• How to debug systematically
• How to write readable code
• How to document your work

QA Mindset Applied to Code

• Test edge cases (empty strings, None values)
• Detailed error reporting
• Data integrity (never lose information)
• Clear documentation

Mistakes I Made

• Initially tried to do everything in one function

  • Solution: Split into cleaning and validation

• Forgot error handling on type conversions

  • Solution: Add try/except blocks wherever needed

• Wanted to make it “perfect” before shipping

  • Solution: Ship a working version first, then iterate

The Results

Project Stats

• ~200 lines of code
• 8 functions
• 4 days from start to finish
• 100 % written by myself (with learning resources)

Real‑World Performance

• 1,000 rows:

Feel free to:

• Use it for your projects
• Suggest improvements
• Ask questions in comments