CSV

CSV (Comma-Separated Values) is a simple, widely-supported file format for storing tabular data. Despite its simplicity, it remains one of the most common formats for data exchange between systems.

Format

name,email,age,active
Alice,alice@example.com,30,true
Bob,bob@example.com,25,true
Charlie,charlie@example.com,35,false

Key Characteristics

• Plain text: Human-readable, easy to inspect
• Universal support: Every spreadsheet and database can handle CSV
• No types: Everything is a string (parsing required)
• Delimiter variations: TSV (tabs), semicolons, pipes

Common Issues

Issue	Example	Solution
Commas in values	"Smith, John"	Quote the field
Quotes in values	"He said ""hello"""	Escape with double quotes
Newlines in values	Multi-line text	Quote the field
Encoding	Non-ASCII characters	Use UTF-8, specify BOM
Large files	Memory issues	Stream processing

Reading CSV in JavaScript

// Using PapaParse (browser and Node)
import Papa from 'papaparse';

Papa.parse(csvString, {
  header: true,
  dynamicTyping: true,
  complete: (results) => {
    console.log(results.data);
  },
});

// Node.js with csv-parse
import { parse } from 'csv-parse/sync';

const records = parse(csvString, {
  columns: true,
  skip_empty_lines: true,
});

CSV vs Alternatives

Format	Best For
CSV	Simple tabular data, spreadsheet import
JSON	Nested/hierarchical data, APIs
Parquet	Large datasets, analytics
Excel	Rich formatting, formulas

What We Like

• Simplicity: Everyone understands CSV
• Compatibility: Works with every tool
• Streaming: Can process line by line
• Debugging: Easy to inspect and edit

What We Don't Like

• No schema: Type information lost
• No hierarchy: Flat structure only
• Ambiguous spec: Edge cases handled differently
• No compression: Larger than binary formats
• Encoding issues: UTF-8 not always assumed

Best Practices

1. Always use headers: First row should be column names
2. Quote strings: Avoid delimiter issues
3. Specify encoding: UTF-8 with or without BOM
4. Validate on import: Type and format checking
5. Consider alternatives: Parquet for analytics, JSON for nested data