A Ruby regular expression to parse data URIs based on RFC 2397.

data_uri.rb

require 'base64'

class DataUri
  REGEXP = %r{
    data:
    (?<mediatype>
      (?<mimetype> .+? / .+? )?
      (?<parameters> (?: ; .+? = .+? )* )
    )?
    (?<extension>;base64)?
    ,
    (?<data>.*)
  }x.freeze

  attr_reader :uri, :match

  def initialize(uri)
    match = REGEXP.match(uri)
    raise ArgumentError, 'invalid data URI' unless match

    @uri = uri
    @match = match
  end

  def mediatype
    "#{mimetype}#{parameters}"
  end

  def decoded_data
    return data unless base64?

    Base64.decode64(data)
  end
  
  def base64?
    !String(extension).empty?
  end

  private

  def mimetype
    return 'text/plain' if String(match[:mimetype]).empty?
    
    match[:mimetype]
  end

  def parameters
    return ';charset=US-ASCII' if String(match[:mimetype]).empty? && String(match[:parameters]).empty?
    
    match[:parameters]
  end

  def extension
    match[:extension]
  end

  def data
    match[:data]
  end
end

data_uri_spec.rb

RSpec.describe DataUri do
  it 'raises an Argument Error if not given a valid data URI' do
    expect { described_class.new('not a data URI') }
      .to raise_error(ArgumentError)
  end

  describe '#mediatype' do
    it 'returns the mediatype of the URI' do
      uri = described_class.new('data:text/plain;charset=iso-8859-7,hello')

      expect(uri.mediatype).to eq('text/plain;charset=iso-8859-7')
    end

    it 'returns the default mediatype if none is given' do
      uri = described_class.new('data:,hello')

      expect(uri.mediatype).to eq('text/plain;charset=US-ASCII')
    end

    it 'returns the default mediatype if only a charset is given' do
      uri = described_class.new('data:;charset=UTF-8,hello')

      expect(uri.mediatype).to eq('text/plain;charset=UTF-8')
    end
    
    it 'does not include the default parameters if a MIME type is given' do
      uri = described_class.new('data:text/csv,hello')

      expect(uri.mediatype).to eq('text/csv')
    end
  end

  describe '#base64?' do
    it 'returns true if the base64 extension is present' do
      uri = described_class.new('data:;base64,R0lGODdh')

      expect(uri).to be_base64
    end

    it 'returns false if there is no base64 extension' do
      uri = described_class.new('data:,R0lGODdh')

      expect(uri).not_to be_base64
    end
  end

  describe '#decoded_data' do
    it 'returns the data if it is not base64 encoded' do
      uri = described_class.new('data:,hello')

      expect(uri.decoded_data).to eq('hello')
    end

    it 'decodes data if it is base64 encoded' do
      uri = described_class.new('data:;base64,aGVsbG8=')

      expect(uri.decoded_data).to eq('hello')
    end
  end
end