CSV Format Guide

Everything you need to know about CSV files

What is CSV?

CSV (Comma-Separated Values) is a simple, widely-used file format for storing tabular data. Each line represents a row, and values within each row are separated by commas (or other delimiters).

Example CSV:

name,email,age,city
John Doe,john@example.com,30,New York
Jane Smith,jane@example.com,25,Los Angeles
Bob Wilson,bob@example.com,35,Chicago

Advantages

Human-readable plain text
Universal compatibility
Small file size
Easy to create and edit

Limitations

No data type information
No standard specification
Limited to flat data
Encoding ambiguity

CSV Structure

Basic Rules

Header Row: Optional first row containing column names
Data Rows: Each subsequent line is a data record
Fields: Values separated by delimiter (usually comma)
Line Endings: CRLF (\r\n) or LF (\n)

RFC 4180 Standard

While CSV has no official standard, RFC 4180 provides common conventions:

1. Each record is on a separate line
2. Last record may or may not have an ending line break
3. First line may be a header
4. Each line should have the same number of fields
5. Fields containing commas, quotes, or line breaks must be quoted
6. Double-quotes within quoted fields are escaped by doubling them

Delimiters & Variants

Format	Delimiter	Extension	Use Case
CSV	, (comma)	.csv	Most common format
TSV	\t (tab)	.tsv, .txt	Data with commas in values
SSV	; (semicolon)	.csv	European locales (decimal comma)
PSV	\| (pipe)	.txt	Data with commas and semicolons

Quoting Rules

Quoting is essential when your data contains special characters that would otherwise break the CSV structure.

When to Quote

Field contains the delimiter:

"Hello, World",other,values

Field contains line breaks:

"Line 1
Line 2",other,values

Field contains quotes (escape with double quotes):

"She said ""Hello""",other,values

Character Encoding

Character encoding determines how text characters are stored as bytes. Using the wrong encoding can cause garbled text or data loss.

Recommended: UTF-8

Supports all Unicode characters
Backwards compatible with ASCII
Most widely supported encoding
Web standard

Other Common Encodings

UTF-16: Windows default for some apps
ISO-8859-1: Western European
Windows-1252: Legacy Windows
ASCII: Basic English only

Tip: For Excel compatibility, save as UTF-8 with BOM (Byte Order Mark) or use UTF-16 LE encoding.

Best Practices

Creating CSV Files

Always include a header row
Use consistent delimiter throughout
Use UTF-8 encoding
Quote all fields with special characters
Avoid leading/trailing whitespace
Use consistent date/number formats

Reading CSV Files

Detect delimiter automatically when possible
Handle quoted fields properly
Trim whitespace from values
Validate row length consistency
Handle empty values gracefully
Be prepared for encoding issues

Common Issues

1. Garbled Characters

Usually caused by encoding mismatch. Try opening with different encoding (UTF-8, Windows-1252, etc.) or use our encoding converter.

2. Wrong Column Alignment

Happens when delimiter detection fails or fields contain unquoted delimiters. Use our CSV validator to detect issues.

3. Excel Changes Numbers

Excel automatically interprets data (e.g., "1-2" becomes a date). Prefix with apostrophe or import as text to prevent this.

4. Missing Quotes

Fields with commas, quotes, or newlines must be quoted. Use our CSV cleaner to fix formatting issues.

CSV Repair Corpus v1

The CSV Repair Corpus v1 is a versioned set of synthetic edge-case files released under CC0 1.0. It contains no customer or personal data. Its machine-readable manifest records each file's SHA-256 digest, expected findings, allowed automatic and optional repairs, validation statistics, and non-loss invariants.

What it proves

The tested repair engine produces deterministic findings and actions for these exact fixtures. Default structural repair preserves the order and multiplicity of non-empty data cells, while ambiguous malformed quotes remain explicitly flagged. Published hashes let you verify that fixture bytes have not changed.

What it does not prove

The corpus is not exhaustive, a destination-schema guarantee, or a large-file performance benchmark. It does not make every malformed CSV unambiguous; parser errors and optional content changes still require review.

Downloads

Start with the manifest for the complete fixture inventory and integrity contract. The Latin-1 sample uses an ASCII-safe Base64 transport and includes a separate decoded-byte SHA-256.