Mutual Fund Sector Weights without an API

mutual_fund_info.rb

#!/usr/bin/env ruby

require 'mechanize'
require 'csv'

tickers = ["EWX","EZM","FMIJX","IGM","IGV","ITOT","POGRX","PRHSX","SCZ","SPEM","SPVU","SYE","VEA","XSLV","TRBIX","VEMIX","VIIIX","VWILX","VSCIX"]

mechanize = Mechanize.new

puts "Creating new CSV"

csv = CSV.open("/tmp/mutual_fund_data_as_of_#{Date.today.to_s}.csv", 'w',{:col_sep => ",", :quote_char => '\'', :force_quotes => true})

tickers.each do |t|
	puts "Getting data for #{t}"
	page = mechanize.get("http://portfolios.morningstar.com/fund/summary?t=#{t}&region=usa&culture=en-US")

	data = page.xpath('//table[@id="equity_style_tab"]/tbody').text.split(/[\n\t\s*]+/)
	data = data.reject { |c| c.empty? }

	ticker = [t]

# Market Cap section
category = ["Market Capitalization"]
headers = ["Size", "% of Portfolio", "Benchmark", "Category Average"]
data = data.each_slice(4).to_a

# write to the CSV
csv << ticker
csv << category
csv << headers
data.each do |c|
	csv << c
end


# Sector Weightings section
data = page.xpath('//table[@id="sector_we_tab"]/tbody').text.split(/[\n\t]+/)
data.each do |s|
	s.gsub!(/\s+/,'')
end
data = data.reject { |c| c.empty? }
data = data.each_slice(17).to_a

csv << [nil]

category = ["Sector Weightings"]
headers = ["Sector", "% Stocks", "Benchmark", "Category Avg"]

# write to the csv

csv << category
csv << headers

data.each do |d|
	csv << [d[0]]
	d.shift
	new_d = d.each_slice(4).to_a
	new_d.each do |c|
		csv << c
	end
end
csv << [nil]
end

csv.close

puts "Finished!"