added code to output master.csv
parent
a60395dd5b
commit
4e1b862586
42
sorter.rb
42
sorter.rb
|
@ -54,12 +54,13 @@ end
|
||||||
def write_output_csv (output, filename)
|
def write_output_csv (output, filename)
|
||||||
CSV.open(filename, 'wb') do |csv|
|
CSV.open(filename, 'wb') do |csv|
|
||||||
csv << ["bin", "words", "total"]
|
csv << ["bin", "words", "total"]
|
||||||
|
output.delete(:filename)
|
||||||
output.each_key do |key|
|
output.each_key do |key|
|
||||||
line = []
|
line = []
|
||||||
line.push(key)
|
line.push(key)
|
||||||
output[key].each_key do |sub_key|
|
output[key].each_key do |sub_key|
|
||||||
line.push(output[key][sub_key])
|
line.push(output[key][sub_key])
|
||||||
end
|
end
|
||||||
csv << line
|
csv << line
|
||||||
end
|
end
|
||||||
end
|
end
|
||||||
|
@ -127,6 +128,7 @@ def process_file (file_name, binfile, type)
|
||||||
outfile = file_name
|
outfile = file_name
|
||||||
outfile.slice!('.txt')
|
outfile.slice!('.txt')
|
||||||
#puts outfile
|
#puts outfile
|
||||||
|
output[:filename] = outfile
|
||||||
csv.each { |bin| bins[bin[0]] = bin[1..].compact } #turn the csv array into a hash, remove nils
|
csv.each { |bin| bins[bin[0]] = bin[1..].compact } #turn the csv array into a hash, remove nils
|
||||||
bins.each_key do |bin_number|
|
bins.each_key do |bin_number|
|
||||||
key = bin_number.to_sym
|
key = bin_number.to_sym
|
||||||
|
@ -146,11 +148,46 @@ end
|
||||||
#This method will process all .txt files in the supplied directory
|
#This method will process all .txt files in the supplied directory
|
||||||
def process_dir(dir_name, binfile, type)
|
def process_dir(dir_name, binfile, type)
|
||||||
Dir.glob(dir_name + '*.txt') do |file_name|
|
Dir.glob(dir_name + '*.txt') do |file_name|
|
||||||
puts file_name
|
puts "Processing" + file_name
|
||||||
process_file(file_name, binfile, type)
|
process_file(file_name, binfile, type)
|
||||||
end
|
end
|
||||||
end
|
end
|
||||||
|
|
||||||
|
def generate_master_output(dir_name, binfile)
|
||||||
|
puts dir_name
|
||||||
|
file=File.open(binfile,"r")
|
||||||
|
bin_count = file.readlines.size
|
||||||
|
file.close
|
||||||
|
bin_header = []
|
||||||
|
bin_count.times do |num|
|
||||||
|
num += 1
|
||||||
|
words_head = "Bin " + num.to_s + " words"
|
||||||
|
total_head = "Bin " + num.to_s + " total"
|
||||||
|
bin_header.push(words_head)
|
||||||
|
bin_header.push(total_head)
|
||||||
|
end
|
||||||
|
CSV.open('master.csv', 'wb') do |csv|
|
||||||
|
header = ["File", "Total Words" ] + bin_header
|
||||||
|
csv << header
|
||||||
|
Dir.glob(dir_name + '*.json') do |file_name|
|
||||||
|
puts file_name
|
||||||
|
csv_row = []
|
||||||
|
json_file = File.read(file_name)
|
||||||
|
data_hash = JSON.parse(json_file)
|
||||||
|
csv_row.push(data_hash["filename"])
|
||||||
|
data_hash.delete("filename")
|
||||||
|
word_total = 0
|
||||||
|
data_hash.each_key do |key|
|
||||||
|
csv_row.push(data_hash[key]["words"])
|
||||||
|
csv_row.push(data_hash[key]["total"])
|
||||||
|
word_total += data_hash[key]["total"]
|
||||||
|
end
|
||||||
|
csv_row = csv_row.insert(1, word_total)
|
||||||
|
csv << csv_row
|
||||||
|
end
|
||||||
|
end
|
||||||
|
end
|
||||||
|
|
||||||
options = Hash.new
|
options = Hash.new
|
||||||
OptionParser.new do |opts|
|
OptionParser.new do |opts|
|
||||||
opts.banner = 'sorter.rb --options'
|
opts.banner = 'sorter.rb --options'
|
||||||
|
@ -174,5 +211,6 @@ elsif options[:file]
|
||||||
process_file(options[:file], options[:binfile], options[:type])
|
process_file(options[:file], options[:binfile], options[:type])
|
||||||
elsif options[:dir]
|
elsif options[:dir]
|
||||||
process_dir(options[:dir], options[:binfile], options[:type])
|
process_dir(options[:dir], options[:binfile], options[:type])
|
||||||
|
generate_master_output(options[:dir], options[:binfile])
|
||||||
end
|
end
|
||||||
|
|
||||||
|
|
Loading…
Reference in New Issue