Ruby: Parsing a string representation of nested arrays into an Array?

That particular example is being parsed correctly using JSON:

s = "[1,2,[3,4,[5,6]],7]"#=> "[1,2,[3,4,[5,6]],7]"require 'json'#=> trueJSON.parse s#=> [1, 2, [3, 4, [5, 6]], 7]

If that doesn't work, you can try running the string through eval, but you have to ensure that no actual ruby code has been passed, as eval could be used as injection vulnerability.

Edit: Here is a simple recursive, regex based parser, no validation, not tested, not for production use etc:

def my_scan s  res = []  s.scan(/((\d+)|(\[(.+)\]))/) do |match|    if match[1]      res << match[1].to_i    elsif match[3]      res << my_scan(match[3])    end  end  resends = "[1,2,[3,4,[5,6]],7]"p my_scan(s).first #=> [1, 2, [3, 4, [5, 6]], 7]

The same can be done using Ruby standard libaray YAML as below :

require 'yaml's = "[1,2,[3,4,[5,6]],7]"YAML.load(s)# => [1, 2, [3, 4, [5, 6]], 7]

"Obviously" the best solution is to write your own parser. [ If you like writing parsers, have never done it before and want to learn something new, or want control over the exact grammar ]

require 'parslet'class Parser < Parslet::Parser  rule(:space)       { str(' ') }  rule(:space?)      { space.repeat(0) }  rule(:openbrace_)  { str('[').as(:op) >> space? }  rule(:closebrace_) { str(']').as(:cl) >> space? }  rule(:comma_)      { str(',') >> space?  }  rule(:integer)     { match['0-9'].repeat(1).as(:int) }  rule(:value)       { (array | integer) >> space? }  rule(:list)        { value >> ( comma_ >> value ).repeat(0) }  rule(:array)       { (openbrace_ >> list.maybe.as(:list) >> closebrace_ )}  rule(:nest)        { space? >> array.maybe }  root(:nest)endclass Arr  def initialize(args)    @val = args  end  def val    @val.map{|v| v.is_a?(Arr) ? v.val : v}  endendclass MyTransform < Parslet::Transform  rule(:int => simple(:x))      { Integer(x) }  rule(:op => '[', :cl => ']')  { Arr.new([]) }  rule(:op => '[', :list => simple(:x), :cl => ']')   {  Arr.new([x]) }  rule(:op => '[', :list => sequence(:x), :cl => ']')   { Arr.new(x) }enddef parse(s)  MyTransform.new.apply(Parser.new.parse(s)).valendparse " [   1  ,   2  ,  [  3  ,  4  ,  [  5   ,  6  , [ ]]   ]  ,  7  ]  "

Parslet transforms will match a single value as "simple" but if that value returns an array, you soon get arrays of arrays, then you have to start using subtree. returning objects however are fine as they represent a single value when transforming the layer above... so sequence will match fine.

Couple the trouble with returning bare arrays, with the problem that Array([x]) and Array(x) give you the same thing... and you get very confusing results.

To avoid this I made a helper class called Arr which represents an array of items. I could then dictate what I pass into it. Then I can get the parser to keep all the brackets even if you have the example that @MateuszFryc called out :) (thanks @MateuszFryc)