For example, Study123 might have two people acting as the "Billing Coordinator" and thus require two entries in the file.
The logic to parse the file and build the resulting Study object is pretty typical and this morning I thought that it would be nice to have a "rollup" method that would take the boilerplate code off of our hands.
I could envision adding the #rollup method to Enumerable, Array or CSV::Reader to encapsulate this logic and perhaps provide a cleaner interface.
Here is an example I coded dealing with repeating data for People and their addresses:
Pastie
1 module Pattern
2 # roll-up - iterating through an ordered array and rolling up
3 # related entries into a single entry
4 #
5 # require '/Users/chiefwahoo/rails/patterns/patterns.rb'
6 # class Array; include Pattern; end
7 #
8 # class Person
9 # attr_accessor :name, :address
10 # def initialize(name, address)
11 # @name = name; @address = address;
12 # end
13 # end
14 #
15 # # Assumes items are grouped by "key" in this case Person's name
16 # people = []
17 # people << Person.new("Compton, Kevin", "123 Mockingbird Lane")
18 # people << Person.new("Compton, Kevin", "9823 Beach View Drive")
19 # people << Person.new("Compton, Jessica", "823 City Park Avenue")
20 # people << Person.new("Compton, Jessica", "9824 Beach View Drive")
21 # people << Person.new("Compton, Keith", "777 College Avenue")
22 # people << Person.new("Compton, Megan", "123 California")
23 # people << Person.new("Compton, Megan", "876 Mount Vista")
24 #
25 # rollup_items = []
26 #
27 # people.rollup(
28 # Proc.new {|person| person.name},
29 # Proc.new {|person| "User #{person.name} has residences at: "},
30 # Proc.new {|container, person| container << " #{person.address};" },
31 # Proc.new {|container|
32 # rollup_items << container;
33 # puts "Detail is : #{container}" }
34 # )
35 #
36 # ### OUTPUT ###
37 # Detail is : User Compton, Kevin has residences at: 123 Mockingbird Lane; 9823 Beach View Drive;
38 # Detail is : User Compton, Jessica has residences at: 823 City Park Avenue; 9824 Beach View Drive;
39 # Detail is : User Compton, Keith has residences at: 777 College Avenue;
40 # Detail is : User Compton, Megan has residences at: 123 California; 876 Mount Vista;
41 # => nil
42 # irb(main):098:0> rollup_items
43 # => ["User Compton, Kevin has residences at: 123 Mockingbird Lane; 9823 Beach View Drive;", "User Compton, Jessica has residences at: 823 City Park Avenue; 9824 Beach View Drive;", "User Compton, Keith has residences at: 777 College Avenue;", "User Compton, Megan has residences at: 123 California; 876 Mount Vista;"]
44 #
45 def rollup(bookmark_detail_proc, initialize_item_proc, merge_into_item_proc, finalize_item_proc)
46 saved_bookmark = nil
47 current_item = nil
48
49 self.each do |line|
50 current_bookmark = bookmark_detail_proc.call(line)
51 if saved_bookmark != current_bookmark
52 finalize_item_proc.call(current_item) unless current_item.nil?
53
54 saved_bookmark = current_bookmark
55 current_item = initialize_item_proc.call(line)
56 end
57
58 merge_into_item_proc.call(current_item, line)
59 end
60 finalize_item_proc.call(current_item) unless current_item.nil?
61 end
62 end
2 comments:
Using Enumerable methods (esp. Rail's group_by / index_by) is more idiomatic, I think:
people.group_by(&:name).each do |name, people|
puts "Details: User #{name} has residencies at: #{people.map(&:address).join('; ')"
end
I agree that would be a nice implementation for smaller data sets.
However, the initial usage case that led to my implementation involved parsing a large file and I did not want to bring it all into memory but instead parse it and only have a single item in memory at any given point.
Post a Comment