optimized way to search multi-dimensional ruby hash

Question

I'm working with the LinkedIn API to get companies' details. They are sending an XML response, so I simply converted the XML to a hash using the .to_hash method This is a sample hash I'm getting: http://pastebin.com/1bXtHZ2F

in some companies they have more than one locations and contact information, i want to parse this data and get the details like phone number, city, postal_code etc.

The structure of the response is not consistent. Sometimes location field itself is missing or the postal_code is available only at the fourth location.

I tried two ways:

1.

def phone(locations)
  (locations && locations["values"][0]["contactInfo"]["phone1"]) || nil
end

This is not working if the phone number is not available in the first array

2.

def phone(locations)
  if locations["locations"]["total"].to_i == 1 
    locations["locations"]["location"]["contact_info"]["phone1"]
  else
    locations["locations"]["location"].each do |p|
      if (!p["contact_info"]["phone1"].nil? || !p['contact_info'].nil?)
        return p["contact_info"]["phone1"]
        break
      end
    end
  end
end

This is not working if the "location" hash itself is missing from the response. I need a solution where I can search with the keys "city", "phone" and "postal_code" and update if it is present. If it returns an array, parse the array and get the non-empty data.

I've also read this StackOverflow answer.

So do you want to get the data only from locations which have all the elements (e.g. the first location that has city and postal_code and phone1) or are you happy to mix and match (city and postal_code from first location, phone1 from second location) — SteveTurczyn, May 19 '14 at 17:57
no i need city and postal_code from the same location and phone number could be from any other location.This is more meaning full.You can't get city from one location and postal_code from other right? — Rajanand02, May 19 '14 at 18:44
So you're saying a locatio may be present but it may not have an address node, or it may not have a contact node. — SteveTurczyn, May 19 '14 at 19:33
Yes and sometimes locations may be in an array..inside that array phone number can be at any position. — Rajanand02, May 20 '14 at 04:07

pdobb · Answer 1 · 2014-05-19T18:50:18.503

I see this as a question about code confidence. That is, I'm betting you can figure out how to guess your way through all the possible conditions... but that will create a mess of unconfident code. Confident code states what it wants and it gets it and moves on. (Note: I get all of my inspiration on this topic from this wonderful book: http://www.confidentruby.com/ by Avdi Grimm).

That said, I'd recommend the following.

Install the naught gem: https://github.com/avdi/naught
In your code, utilize the Maybe conversion function (read through the gem documetnation for info) to confidently arrive at your values:

At the top of your class or controller:

NullObject = Naught.build

include NullObject::Conversions

In your method:

def phone(locations)
  return {} if locations["location"].blank?
  Maybe(locations["locations"])["location"].to_a.inject({}) do |location, acc|
    contact_info = Maybe(location["contact_info"])
    acc[location][:city] = contact_info["city1"].to_s
    acc[location][:phone] = contact_info["phone1"].to_i
    acc[location][:postal_code] = contact_info["postal_code1"].to_s
    acc
  end
end

I'm not sure exactly what you're trying to accomplish but the above may be a start. It is simply attempting to assume all of the keys exist. Whether they do or they don't they get converted to a object (an array, a string or an integer). And then, ultimately, collected into a hash (call acc -- short for "accumulator" -- internal to the loop above) to be returned.

If any of the above needs clarification let me know and we can chat.

score 1 · Answer 2 · answered May 20 '14 at 11:56

Ok, this code basically works through the hash and isn't concerned about node names (other than the specific nodes it's searching for)

the find_and_get_values method takes two arguments: object to search, and an array of nodes to find. It will only return a result if all nodes in the array are siblings under the same parent node. (so "city" and "postal_code" must be under the same parent otherwise neither is returned)

The data returned is a simple hash.

The get_values method takes one argument (the company hash) and calls find_and_get_values twice, once for %w(city postal_code) and once for %w(phone1) and merges the hash results into one hash.

def get_values(company)
  answer = {}
  answer.merge!(find_and_get_values(company["locations"], %w(city postal_code))
  answer.merge!(find_and_get_values(company["locations"], ["phone1"]))
  answer
end

def find_and_get_values(source, match_keys)
  return {} if source.nil?
  if source.kind_of?(Array)
    source.each do |sub_source|
      result = find_and_get_values(sub_source, match_keys)
      return result unless result.empty?
    end
  else
    result = {}
    if source.kind_of?(Hash)
      match_keys.each do |key|
        result[key] = source[key] unless source[key].nil?
      end
      return result if result.count == match_keys.count
      source.each do |sub_source|
        result = find_and_get_values(sub_source, match_keys)
        return result unless result.empty?
      end
    end
  end
  return {}
end

  p get_values(company)

optimized way to search multi-dimensional ruby hash

2 Answers2