csv_to_couch.rb

#!/usr/bin/ruby

# csv_to_couch.rb - Simple script which plonks CSV files into CouchDB
# (C) 2009 James S Urquhart (jamesu at gmail dot com)
# Licensed under MIT

require 'rubygems'
require 'fastercsv'
require 'couchrest'

DOCBUFFER_SIZE = 1024  # number of entries to send per-batch

rows = {}

puts "#{ARGV[0]} -> #{ARGV[1]}..."

@db = CouchRest.database!("http://127.0.0.1:5984/#{ARGV[1]}")
@doc_fields = []
@doc_buffer = []
first = true

# Output
unless @db.nil?
	FasterCSV.foreach(ARGV[0], :col_sep => ',') do |row|
		# Header
		if first
			@doc_fields = []
			row.each {|v| @doc_fields << v}
			first = false
			next
		end
		
		new_doc = {}
		count = 0
		for value in row
			next if count >= @doc_fields.length
			new_doc[@doc_fields[count]] = value
			count += 1
		end
		
		@doc_buffer << new_doc
		
		if @doc_buffer.length >= DOCBUFFER_SIZE
		  @db.bulk_save(@doc_buffer)
		  @doc_buffer = []
		end
	end
end

if @doc_buffer.length > 0
  @db.bulk_save(@doc_buffer)
end

puts "Done."