pg_array_parser.rb

require 'strscan'
require 'bacon'

Bacon.summary_on_exit

class PGArrayParser
  attr_reader :s

  def parse(input)
    return [] if input.empty?
    @s = StringScanner.new(input)

    parse_array
  end

  private

  def parse_string
    out = ""
    s.scan(/"/)
    until s.scan(/"/)
      if s.scan(/\\(.)/)
        out << s[1]
      else
        out << s.scan(/[^"\\"]+/)
      end
    end

    return out
  end

  def parse_number
    if s.scan(/[+-]?\d+\.\d+/)
      Float(s.matched)
    elsif s.scan(/[+-]?\d+/)
      Integer(s.matched)
    end
  end

  def parse_array
    unless s.scan(/\{/)
      raise("not an array")
    end

    array = []

    until s.eos?
      case s.peek(1)
      when '{'
        array << parse_array
      when '"'
        array << parse_string
      when /\d/
        array << parse_number
      when '}'
        @s.scan(/./)
        return array
      end

      @s.scan(/,\s*/)
    end
  end
end

describe 'PostgreSQL Arrays' do
  def parse(input)
    PGArrayParser.new.parse(input)
  end

  it 'may be empty' do
    parse('').should == []
  end

  it 'may have a string' do
    parse('{"hi"}').should == ["hi"]
  end

  it 'may have an int' do
    parse('{123}').should == [123]
  end

  it 'may have more than one int' do
    parse('{123, 456}').should == [123, 456]
  end

  it 'may be a float' do
    parse('{123.456, 789.0}').should == [123.456, 789.0]
  end

  it 'may have more than one string' do
    parse('{"hello", "world"}').should == ["hello", "world"]
  end

  it 'may have more than one array of strings in the array' do
    parse('{{"hello"}, {"world"}}').should == [["hello"], ["world"]]
  end

  it 'may have more than one array of ints in the array' do
    parse('{{123}, {456}}').should == [[123], [456]]
  end

  it 'parses a bunch more' do
    {
      '{"hi","there"}' => ["hi", "there"],
      '{"\"","quote"}' => ['"', 'quote'],
      '{"\\\\","quote"}' => ['\\', 'quote'],
    }.each do |input, output|
      parse(input).should == output
    end
  end
end