Typedcsv

Here's your standard untyped CSV:

name,income,created_at,tags,great
Seamus,12301.2,2012-02-21,red;blue,true

Now, you and I know that 12301.2 is a number and 2012-02-21 is a date and red;blue is a list... so let's just write that into the headers:

name,income:number,created_at:date,tags:list,great:boolean
Seamus,12301.2,2012-02-21,red;blue,true

Now let's parse it:

Typedcsv.foreach('file.csv', headers: true) do |row|
  row['income']     # will be a Float
  row['created_at'] # will be a Date
  row['tags']       # will be an Array
  row['great']      # will be TrueClass or FalseClass
end

This gem provides Typedcsv.foreach(), which takes exactly the same arguments as ruby stdlib CSV.foreach.

Types

  • text (default)
  • number
  • list (must be semicolon-separated)
  • date (must be ISO8601)
  • time (must be ISO8601)
  • boolean (recognizes "true" or "false")

Benchmarks

It's about 10x slower than ruby's stdlib CSV.foreach:

cd benchmark && ruby benchmark.rb
[...]
CSV.foreach - array mode
                          2.503  (

Faraday logo

We use typedcsv for B2C customer intelligence at Faraday.

Copyright 2017 Faraday