Rails: Upload CSV file without header - ruby-on-rails

Question

I followed railscasts #396 Importing CSV and implemented CSV upload in my rails project.
This is my view file:
<%= form_tag import_customers_path, multipart: true do %>
<%= file_field_tag :file %>
<%= submit_tag "Import" %>
<% end %>
This is my controller action:
def import
current_user.customers.import(params[:file])
redirect_to customers_path, notice: "Users imported."
end
And these are my model methods:
def self.to_csv(options = {})
CSV.generate(options) do |csv|
csv << column_names
all.each do |customer|
csv << customer.attributes.values_at(*column_names)
end
end
end
def self.import(file)
CSV.foreach(file.path, headers: true) do |row|
Customer.create! row.to_hash
end
end
Here I don't want user to include header in CSV. When I replace headers: true with headers: false, I get error:
NoMethodError in CustomersController#import
undefined method `to_hash' for ["abc#wer.com"]:Array
Can anybody tell how to upload CSV files without need of header line?

Answer 1

As far as upload and handling of the CSV file goes, you're very, very close. You just have an issue with reading the rows of data to populate the database with, via the Customer.create! call
It looks like you've been testing with a CSV file that only has a single line of data. With the headers: true, that single line was converted to headers and subsequently ignored in the CSV.foreach iterator. So, in effect, you had no data in the file, and no iterations occurred. If you had two rows of data in the input file, you'd have encountered the error, anyway.
Now, when you use headers: false, that line of data is treated as data. And that's where the issue lies: handling the data isn't done correctly.
Since there's no schema in your question, I'll assume a little bit of leeway on fields; you should be able to extrapolate pretty easily to make it work in your situation. This code shows how it works:
CSV.parse(csv_data, headers: false) do |row|
hash = {
first_name: row[0],
last_name: row[1],
age: row[2],
phone: row[3],
address: row[4]
}
Customer.create!(hash)
end
If you wanted a CSV version with headers, this would work well in this case, and has the benefit of not allowing arbitrary access to columns that shouldn't be assigned from an outside source:
CSV.parse(csv_data, headers: true, header_converters: :symbol) do |row|
hash = {
first_name: row[:first_name],
surname: row[:last_name],
birth_year: Date.today - row[:age],
phone: row[:phone],
street_address: row[:address]
}
Customer.create!(hash)
end
Note that the Customer#to_csv in your model is not quite correct, either. First, it creates the CSV file with a header, so you wouldn't be able to export and then import again with this implementation. Next, the header fields variable column_names is not actually defined in this code. Finally, the code doesn't control the order of columns written to the CSV, which means that the headers and values could possibly go out of sync. A correct (non-header) version of this is very simple:
csv_data = CSV.generate do |csv|
csv.each do |customer|
csv << [customer.first_name, customer.last_name, customer.age, customer.phone, customer.address]
end
end
The header-based version is this:
csv_data = CSV.generate do |csv|
csv << ["First Name","Last Name","Age","Phone","Address"]
csv.each do |customer|
csv << [customer.first_name, customer.last_name, customer.age, customer.phone, customer.address]
end
end
Personally, I'd use the header-based version, because it's far more robust, and it's easy to understand which columns are which. If you've ever received a headerless CSV file and had to figure out how to make sense of it without any keys, you'd know why the header is important.

Answer 2

You could just load the CSV file into an array of arrays and remove the first row:
data = CSV.read("path/to/file.csv")
data = data[1..-1]
However this will store the data as an array of values only.
When you use headers: true it uses a hash where the keys are the column header names.