Sample CSV files

This repository contains sample Comma Separated Value (CSV) files. CSV is a generic flat file format used to store structured data. Datasets are split in 3 categories: Customers, Users and Organizations. For each, sample CSV files range from 100 to 2 millions records. Those CSV files can be used for testing purpose. They can be open by any application compatible with CSV files or with a CSV editor.

The datasets are generated using random values. Mosly using Python Faker package.

Customers CSV Sample

• customers-100.csv - Zip version - Customers CSV with 100 records
• customers-1000.csv - Zip version - Customers CSV with 1000 records
• customers-10000.csv - Zip version - Customers CSV with 10000 records
• customers-100000.csv - Zip version - Customers CSV with 100000 records
• customers-500000.csv - Customers CSV with 500000 records
• customers-1000000.csv - Customers CSV with 1000000 records
• customers-2000000.csv - Customers CSV with 2000000 records

Customer Schema

• Index
• Customer Id
• First Name
• Last Name
• Company
• City
• Country
• Phone 1
• Phone 2
• Email
• Subscription Date
• Website

People CSV Samples

• people-100.csv - Zip version - People CSV with 100 records
• people-1000.csv - Zip version - People CSV with 1000 records
• people-10000.csv - Zip version - People CSV with 10000 records
• people-100000.csv - Zip version - People CSV with 100000 records
• people-500000.csv - People CSV with 500000 records
• people-1000000.csv - People CSV with 1000000 records
• people-2000000.csv - People CSV with 2000000 records

People Schema

• Index
• User Id
• First Name
• Last Name
• Sex
• Email
• Phone
• Date of birth
• Job Title

Organizations CSV Samples

• organizations-100.csv - Zip version - Organizations CSV with 100 records
• organizations-1000.csv - Zip version - Organizations CSV with 1000 records
• organizations-10000.csv - Zip version - Organizations CSV with 10000 records
• organizations-100000.csv - Zip version - Organizations CSV with 100000 records
• organizations-500000.csv - Organizations CSV with 500000 records
• organizations-1000000.csv - Organizations CSV with 1000000 records
• organizations-2000000.csv - Organizations CSV with 2000000 records

Organization Schema

• Index
• Organization Id
• Name
• Website
• Country
• Description
• Founded
• Industry
• Number of employees

Leads CSV Samples

• leads-100.csv - Zip version - Leads CSV with 100 records
• leads-1000.csv - Zip version - Leads CSV with 1000 records
• leads-10000.csv - Zip version - Leads CSV with 10000 records
• leads-100000.csv - Zip version - Leads CSV with 100000 records

Lead Schema

• Index
• Account Id
• Lead Owner
• First Name
• Last Name
• Company
• Phone 1
• Phone 2
• Email 1
• Email 2
• Website
• Source
• Deal Stage
• Notes

Products CSV Samples

• products-100.csv - Zip version - Products CSV with 100 records
• products-1000.csv - Zip version - Products CSV with 1000 records
• products-10000.csv - Zip version - Products CSV with 10000 records
• products-100000.csv - Zip version - Products CSV with 100000 records
• products-1000000.csv - Zip version - Products CSV with 1000000 records
• products-2000000.csv - Zip version - Products CSV with 2000000 records

Products Schema

• Index
• Name
• Description
• Brand
• Category
• Price
• Currency
• Stock
• EAN
• Color
• Size
• Availability
• Internal ID

Local Set up to generate files

Python Environments

Create a Python virtual env:

python3 -m venv venv/sample-csv

Activate it

source venv/sample-csv/bin/activate

So you can install dependencies:

pip install -r requirements.txt

Run script

python src/main.py

AI Processing CSV Samples

The generator also creates datasets for testing AI workflows on spreadsheet rows:

• support-tickets: ticket classification, sentiment, and priority routing.
• customer-reviews: sentiment analysis and topic classification.
• messy-company-data: company name cleanup and industry classification.
• product-catalog-ai: ecommerce classification, translation, and attribute extraction.
• product-translation-ai: AI translation testing with realistic product names, product descriptions, feature bullets, glossary terms, and target languages.
• lead-scoring-ai: ICP fit and lead scoring prompts.
• web-page-extraction-ai: structured extraction from page text.
• research-questions-ai: AI agent and web research prompt testing.

AI datasets include synthetic Expected ... columns. Use them to compare prompt output with reference labels during testing.

Broken CSV Fixtures

The broken_csv generator creates intentionally malformed CSV files for parser and repair-tool testing:

• broken encodings (Windows-1252, ISO-8859-1)
• mixed or wrong delimiters
• unescaped and missing quotes
• unquoted multiline fields
• ragged rows
• duplicate headers
• mixed line endings

When the repair is deterministic, a clean version is generated in files/broken_csv/expected-fixed.

Tests

Create and activate a local virtual environment:

python3 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt

Run the test suite:

pytest

Generate all configured CSV files and upload manifests:

python src/main.py
python src/broken_csv.py