Ruby has a little-known feature inherited from Perl: the ability to embed arbitrary data at the end of your source file using the __END__ marker. Everything after __END__ becomes accessible through the DATA constant.

Basic Usage

puts DATA.read

__END__
This text appears after the code ends.
You can put anything here:
- Configuration data
- Sample inputs
- ASCII art
- Whatever you want!

The DATA constant is a File object pointing to the content after __END__. You can use all standard file methods on it: read, readline, each, etc.

Why Would You Use This?

Embedding Test Data

Single-file scripts can include their test data inline:

def parse_csv(data)
  data.lines.map { |line| line.split(',').map(&:strip) }
end

if __FILE__ == $0
  result = parse_csv(DATA.read)
  puts result.inspect
end

__END__
name,age,city
Alice,30,Brooklyn
Bob,25,Queens
Charlie,35,Manhattan

Self-Contained Scripts

Distributing a single-file tool becomes easier when configuration or assets are embedded:

require 'yaml'

config = YAML.load(DATA)
puts "Connecting to #{config['host']}"

__END__
host: example.com
port: 5432
timeout: 30

Documentation

You can embed long help text or documentation without cluttering the code:

#!/usr/bin/env ruby

def show_help
  puts DATA.read
  exit
end

show_help if ARGV.include?('--help')

# ... actual program logic ...

__END__
MyScript v1.0

Usage:
  myscript [options] file

Options:
  --help     Show this help
  --verbose  Enable verbose output

Examples:
  myscript input.txt
  myscript --verbose data.csv

Important Details

One Per File

You can only have one __END__ marker per file. Everything after it is DATA—there’s no way to have multiple sections.

Only in the Main File

The DATA constant only exists if __END__ appears in the main executed file. Required libraries with __END__ don’t create a DATA constant:

# main.rb
require './library'
puts DATA.read  # Works if main.rb has __END__

__END__
Main file data

# library.rb
# DATA is not available here even if we have:
__END__
Library data (not accessible)

It’s a Real File Object

Since DATA is a file object, you can seek, rewind, and read multiple times:

section1 = DATA.readline
DATA.rewind
all = DATA.read

__END__
First line
Second line
Third line

Comparison with Heredocs

Why use __END__ instead of heredocs?

# Heredoc
text = <<~TEXT
  Some long text
  Multiple lines
  Embedded in code
TEXT

# vs __END__
text = DATA.read

__END__
Some long text
Multiple lines
Separated from code

Heredocs are better for small snippets. __END__ is better when:

• Data is large enough to distract from code logic
• You want syntax highlighting to stop (some editors dim post-__END__ content)
• The data is conceptually “附件” rather than part of the program flow

Real-World Usage

This feature appears in several contexts:

1. Single-file gems: Embedding gem metadata or documentation
2. Installation scripts: Including configuration templates
3. Code golf: Minimizing file count in challenges
4. Teaching examples: Keeping example data with code

Modern Alternatives

For most use cases, modern Ruby developers prefer:

• External data files (more flexible)
• Embedded resources via constants or methods
• Configuration gems (better structure)

But __END__ remains useful for truly self-contained scripts where distributing a single file is important.

Try It Yourself

The __END__ marker is one of those features you might never need, but when you do need it, nothing else works quite as well. It’s a reminder that Ruby values practicality—even when that means including a quirky Perl-ism that makes certain problems trivial to solve.

Next time you’re writing a throwaway script that needs some test data, give __END__ a try. You might find it’s exactly the right tool for the job.