ruby CSV duplicate row parsing

Question

I have some CSV data I need to process, and having trouble figuring out a way to match the duplicates.

data looks a bit like this:

line    id    name   item_1    item_2    item_3    item_4
1      251   john    foo       foo       foo       foo
2      251   john    foo       bar       bar       bar
3      251   john    foo       bar       baz       baz
4      251   john    foo       bar       baz       pat

lines 1-3 are duplicates in this case.

line    id    name   item_1    item_2    item_3    item_4
5      347   bill    foo       foo       foo       foo
6      347   bill    foo       bar       bar       bar

in this case only line 5 is a duplicate

line    id    name   item_1    item_2    item_3    item_4
7      251   mary    foo       foo       foo       foo
8      251   mary    foo       bar       bar       bar
9      251   mary    foo       bar       baz       baz

here lines 7 and 8 are the duplicates

so basically if the pattern adds a new "item" the previous line is a duplicate. I want to end up with a single line for each person, regardless of how many items they have

I am using Ruby 1.9.3 like this:

require 'csv'
puts "loading data"
people = CSV.read('input-file.csv')

CSV.open("output-file", "wb") do |csv|
    #write the first row (header) to the output file
    csv << people[0]
    people.each do |p|
        ... logic to test for dupe ...
        csv << p.unique
    end
end

Answer 1

First, there's a slight bug with your code. Instead of:

csv << people[0]

You would need to do the following if you don't want to change your loop code:

csv << people.shift

Now, the following solution will add only the first occurrence of a person, discarding any subsequent duplicates as determined by id (as I am assuming ids are unique).

require 'csv'
puts "loading data"
people = CSV.read('input-file.csv')
ids = [] # or you could use a Set

CSV.open("output-file", "wb") do |csv|
  #write the first row (header) to the output file
  csv << people.shift
  people.each do |p|
    # If the id of the current records is in the ids array, we've already seen 
    # this person
    next if ids.include?(p[0])

    # Now add the new id to the front of the ids array since the example you gave
    # the duplicate records directly follow the original, this will be slightly
    # faster than if we added the array to the end, but above we still check the
    # entire array to be safe
    ids.unshift p[0]
    csv << p
  end
end

Note that there is a more performant solution if your duplicate records always directly follow the original, you would only need to keep the last original id and check the current record's id rather than inclusion in an entire array. The difference may be negligible if your input file doesn't contain many records.

That would look like this:

require 'csv'
puts "loading data"
people = CSV.read('input-file.csv')
previous_id = nil

CSV.open("output-file", "wb") do |csv|
  #write the first row (header) to the output file
  csv << people.shift
  people.each do |p|
    next if p[0] == previous_id
    previous_id = p[0]
    csv << p
  end
end

Answer 2

It sounds like you're trying to get a list of unique items associated with each person, where a person is identified by an id and a name. If that's right, you can do something like this:

peoplehash = {}
maxitems = 0
people.each do |id, name, *items|
    (peoplehash[[id, name]] ||= []) += items
peoplehash.keys.each do |k|
    peoplehash[k].uniq!
    peoplehash[k].sort!
    maxitems = [maxitems, peoplehash[k].size].max

This'll give you a structure like:

{
    [251, "john"] => ["bar", "bat", "baz", "foo"],
    [347, "bill"] => ["bar", "foo"]
}

and a maxitems that tells you how long the longest items array is, which you can then use for whatever you need.

Answer 3

You can use 'uniq'

irb(main):009:0> row= ['ruby', 'rails', 'gem', 'ruby']
irb(main):010:0> row.uniq
=> ["ruby", "rails", "gem"]
or 

row.uniq!
=> ["ruby", "rails", "gem"]

irb(main):017:0> row
=> ["ruby", "rails", "gem"]

irb(main):018:0> row = [1,      251,   'john',    'foo',       'foo',       'foo',       'foo']
=> [1, 251, "john", "foo", "foo", "foo", "foo"]
irb(main):019:0> row.uniq
=> [1, 251, "john", "foo"]