哈希初始化的Ruby哈希

Question

Consider @data is an array of ActiveRecords with fields date , class , name , and grade . Assuming I want to end up with two hashes, one being a unique set of all the dates per name; the other broken down by class, then date, then name to reveal the grade.

dates  = {}
grades = {}

@data.each do |d|
   dates[ d.name ][ d.date ] = ''                              # 1
   grades[ d.class => { d.date => { d.name => d.grade } } ]    # 2 
end

1. results in an error:

   undefined method `[]=' for nil:NilClass

   dates[ d.name => { d.date => '' } ] only stores the last record (does not merge into the older results)

2. doesn't error, but ends with an empty hash {}

After getting back into Perl, I've forgotten how to initialize and add to a hash of hashes in Ruby.

---

Update

For the first one ( dates ), it looks like I can do the following inside the loop:

dates[ d.name ] ||= { d.date => '' }
dates[ d.name ][ d.date ] = ''

Looking at the docs, I don't know if this is the proper way to do it, it also seems inefficient, since it will be doing the assignment twice for the first value. It also would suggest that for each key, I need to declare/initialize the underlying data structure, which would get lengthy the deeper the hash of hashes gets (like the grades hash).

---

Source:

[{ name: 'greg', class:'science', date:'jan', grade:'a' },
 { name: 'greg', class:'math'   , date:'jan', grade:'b' },
 { name: 'barb', class:'history', date:'jan', grade:'a' },
 { name: 'barb', class:'science', date:'feb', grade:'c' }]

dates would look something like:

{ 'greg' => { 'jan' => '' },
  'barb' => { 'jan' => '' ,
              'feb' => '' 
            }
}

grades would look something like:

{ 'science' => { 'jan' => { 'greg' => 'a' } ,
                 'feb' => { 'barb' => 'c' } },
  'math'    => { 'jan' => { 'greg' => 'b' } },
  'history' => { 'jan' => { 'barb' => 'a' } }
}

Answer 1

This will actually work, as when you try to access the key from the hash, it not only returns the default but also sets it:

Hash.new do |h, k|
  h[k] = Hash.new do |hh, kk|
    hh[kk] = {}
  end
end

Update:

This might be a little cleaner albeit potentially more confusing for future developers than just checking if nil.

dates = Hash.new { |hash, key| hash[key] = {} }
grades = Hash.new { |hash, key| hash[key] = dates.clone }

Answer 2

try with:

d={}
c=Client.first
d[c.last_name]=[]
d[c.last_name][c.id]=//your value

d[c.last_name]=[], initialize an array of hash for c.last_name

Answer 3

I was able to build the dates hash like this:

arr = [{ name: 'greg', class:'science', date:'jan', grade:'a' },
       { name: 'greg', class:'math'   , date:'jan', grade:'b' },
       { name: 'barb', class:'history', date:'jan', grade:'a' },
       { name: 'barb', class:'science', date:'feb', grade:'c' }]


dates = arr.group_by { |x| x[:name] }.each_with_object({}) do |(k, v), hash|
  dates = v.map { |i| i[:date] }.uniq
  dates_hash = dates.each_with_object({}) { |date, h| h[date] = "" }
  hash[k] = dates_hash
end

puts dates

{"greg"=>{"jan"=>""}, "barb"=>{"jan"=>"", "feb"=>""}}

Answer 4

Code

To obtain the unique :date values for each :name , you could use the form of Hash.update (same as merge! ) that uses a hash to determine the values of keys are are contained in both hashes being merged:

def dates_by_name(arr)
  arr.each_with_object({}) { |g,h|
    h.update({g[:name]=>[g[:date]]}) {|_,ov,nv|
      ov.include?(nv.first) ? ov : ov + nv}}
end

There are many ways to organize the results for grades by class, date and name. The following employs recursion:

def grade_by_class_date_name(arr, fields)
  fd = fields.dup
  f = fd.shift
  return arr.first[:grade] if f==:grade
  g = arr.each_with_object(Hash.new { |h,k| h[k] = [] }) { |e,g|
    g[e[f]] << e.reject { |k,_| k == f } }
  g.each_key { |k| g[k] = grade_by_class_date_name(g[k], fd) }
  g 
end          

Examples

arr = [{ name: 'greg', class:'science', date:'jan', grade:'a' },
       { name: 'greg', class:'math'   , date:'jan', grade:'b' },
       { name: 'barb', class:'history', date:'jan', grade:'a' },
       { name: 'barb', class:'science', date:'feb', grade:'c' }]

dates_by_name(arr)
  #=> {"greg"=>["jan"], "barb"=>["jan", "feb"]} 

The hash returned by dates_by_name is not in the form you requested, but you could manipulate it easily to suit your requirements.

fields = [:class, :date, :name, :grade] 

grade_by_class_date_name(arr, fields)
  #=> {"science"=>{"jan"=>{"greg"=>"a"},
  #                "feb"=>{"barb"=>"c"}},
  #    "math"   =>{"jan"=>{"greg"=>"b"}},
  #    "history"=>{"jan"=>{"barb"=>"a"}}}