Question

I have an unsorted list of area postcodes as follows:

["E1", "E1C", "E1D", "E10", "E11", "E12", "E2", "E3", "E4", "EC1", "EC1A", "EC1M", "EC1N",
"EC1R", "EC1V", "EC1Y", "EC2", "EC2A", "EC2M", "EC2N", "N1", "N10", "N11", "N12",
"N13", "N2", "NW1", "NW10", "NW2" etc]

I’d like to sort them as follows:

["E1", "E1C", "E1D", "E2", "E3", "E4", "E10", "E11", "E12", "EC1", "EC1A", "EC1M", "EC1N",
"EC1R", "EC1V", "EC1Y", "EC2", "EC2A", "EC2M", "EC2N", "N1", "N2", "N10", "N11", "N12",
"N13", "NW1", "NW2, "NW10" etc]

So to sum up the order of the formats for postcodes beginning with E would be:

• E1
• E1C
• E11
• EC1
• EC1V

Same order for postcodes beginning with N, etc.

What would be the recommended way of sorting such strings? In this case the format of the string is always known, i.e. it will always be 2-4 alphanumberic characters, the first always being a letter.

Should I order the strings by length first and then order within each length group, or is there a more elegant method?

Answer 1

I’d use

array.sort_by do |str|
  /\A(\w)(\d+)\Z/ === str
  [$1, $2.to_i]
end

or, if you have arbitrary sequences of alternating letters and digits,

array.sort_by do |str|
  /\A(\D*)(\d*)(\D*)(\d*)\Z/.match(str)[1..-1].reject(&:blank?).collect do |item|
    /\d/ === item ? item.to_i : item
  end
end