parser.rb

#!/usr/bin/env ruby
require 'csv'
require 'prawn'
# require 'pry'
# require 'prawn/table'

Prawn::Font::AFM.hide_m17n_warning = true

NUMERIC_FORMAT = /^(\d*). /
SIMULATION_FORMAT = /(\d+\. .*)/

ParseSimulation = Struct.new(:title, :start, :end)

def putsd(string)
  if $DEBUG
    puts string
  end
end

class QuestionColumn < Struct.new(:title, :responses, :type)
end

class Simulation < Struct.new(:title, :results)
  def upsert_response(question_title, value)
    if res = self.results.select { |r| r.title == question_title }.first
      res.responses.push value
    else
      self.results = self.results.push(QuestionColumn.new(question_title, [value], parse_answer_type(value)))
    end
  end

  def print(pdf)
    free_text = []
    table_responses = []
    time_responses = []

    self.results.each do |r|
      if r[:type] == :numeric || r[:type] == :yesno
        table_responses.push(r)
      elsif r[:type] == :text
        free_text.push(r)
      elsif r[:type] == :minutes
        time_responses.push(r)
      else
        puts r[:responses]
        raise "Unknown column response type for cell in column with title: #{r[:title]}!"
      end
    end

    pdf.text("Results for #{self.title}", size: 24)
    pdf.move_down 20

    table_responses.each do |r|
      pdf.text(r[:title], size: 18)
      pdf.move_down 5
      pdf.text(generate_summary(r), size: 16)
      pdf.move_down 15
    end

    time_responses.each do |tr|
      pdf.text(tr[:title], size: 18)
      pdf.move_down 5
      pdf.text(generate_summary(tr), size: 12)
      pdf.move_down 15
    end

    free_text.each do |ft|
      pdf.text(ft[:title], size: 18)
      pdf.move_down 20

      ft[:responses].each do |resp|
        pdf.text resp
        pdf.move_down 10
      end
    end
  end
end

def generate_summary(response)
  if response[:type] == :numeric
    return decimal_to_percent(summarize_numeric(response[:responses])).to_s
  elsif response[:type] == :yesno
    return decimal_to_percent(summarize_yesno(response[:responses])).to_s
  elsif response[:type] == :minutes
    freq = response[:responses].inject(Hash.new(0)) { |h,v| h[v] += 1; h }
    peak = response[:responses].max_by { |v| freq[v] }
    return "Most common response: " + peak + " (#{freq[peak]} / #{response[:responses].length})"
  else
    raise "Should not be coercing text / time into table format"
  end
end


def decimal_to_percent(d)
  # if this value is more than 1, we're assuming a 1 - 4 scale
  # d * 25.0 is really (d * 100) / 4
  if d > 1.0
    return (d * 25.0).round(0).to_s + "%"
  else
    return (d * 100).round(0).to_s + "%"
  end
end

def summarize_yesno(vals)
  counts = Hash.new(0)
  vals.each { |name| counts[name] += 1 }
  return counts["Yes"] / vals.length.to_f
end
# puts summarize_yesno(["Yes", "Yes", "No"])

def summarize_numeric(vals)
  total = 0

  vals.each do |v|
    c = v.match(NUMERIC_FORMAT)&.captures

    raise "Failed to parse one numeric value from number response" unless c&.length == 1
    total += c[0].to_i
  end

  return total / vals.length.to_f
end

# puts summarize_numeric(["1. Hi", "5. There", "10. Buddy boy 1."]) # puts summarize_numeric(["LOL 1. Hi", "5. There", "10. Buddy boy 1."]) # puts summarize_numeric(["1. Hi"])

def role_type(value)
  if value == "1. Trainee"
    return :trainee
  elsif value == "2. Instructor"
    return :instructor
  elsif value == "3. Volunteer"
    return :volunteer
  else
    puts "Saw '#{value}', expected something like '1. Trainee'"
    raise "Expected to see a role of Trainee, Instructor, or Volunteer. Something is probably badly formatted about the source CSV."
  end
end


def parse_answer_type(value)
  # digits have to be first in the cell to be numeric
  # ugh, this stupid response in the first instance broke our guessing of the column type
  # 1. The creating was good but could be a little more louder and energetic.
  # 2. Asked questions to understand my needs, but wasn't aware about the company's products so could not suggest properly.
  if value =~ /minutes/ && value.length < 23 # "3-5 minutes", "longer than 10 minutes"
    # someone could write minutes in free text oof
    # putsd "MINUTES"
    return :minutes
  elsif value =~ /^\d+\. [\w+ ]{0,}$/
    # digit(s) followed by a period, space, and 0 - 2 'words'
    # putsd "NUMERIC"
    return :numeric
  elsif (value == "Yes") || (value == "No")
    # putsd "BOOLEAN"
    return :yesno
  else
    # putsd "TEXT"
    return :text
  end
end

if ARGV.length != 1
  puts 'need an argument to run!'
  puts 'EXAMPLE: ./parser.rb {target.csv}'
  puts 'Will create a folder "target" and put PDF results in that folder'
  return
end

dest = ARGV[0]
sanitzied = dest.delete_suffix('.csv')
if !Dir.exist?(sanitzied)
  puts "Creating '#{sanitzied}' destination directory"
  Dir.mkdir(sanitzied)
else
  puts "Directory '#{sanitzied}' detected"
end

iter = 0
sims = []
$DEBUG = false
generic_headers = []

CSV.foreach(dest) do |row|
  if iter == 0
    # do nothing
  elsif iter == 1
    row.each_with_index do |cell, idx|
      if cell && cell.strip != ""
        generic_headers.push({ key: idx, value: cell.strip })
      end
    end
  else
    if role_type(row[1]) == :volunteer
      putsd "processing row #{iter}"
      row_sim_title = nil

      row.each_with_index do |cell, i|
        if cell && !cell.empty?
          putsd "#{i} -> #{cell}"
          if i == 12
            # this is the simulation column
            # find the sim for this row...
            cell.match(SIMULATION_FORMAT) do |m|
              sim = sims.select { |s| s.title == m[1] }
              if sim.empty?
                s = Simulation.new(m[1], [])
                row_sim_title = m[1]
                sims = sims.push(s)
              else
                row_sim_title = sim.first.title
              end
            end
          elsif i == 13
            # ignore, this is the volunteer's name
          elsif i > 13
            if cell && !cell.empty?
              # upsert header (because we have values in those columns)
              sim = sims.select { |s| s.title == row_sim_title }.first
              column_header = generic_headers.select { |gh| gh[:key] == i }.first

              # these are responses we care about
              # insert result for headers
              sim.upsert_response(column_header[:value], cell)
            end
          end
        end
      end
    end
  end

  iter = iter + 1
end


simnumber = 1

sims.each do |sim|
  if sim.results.length != 0
    safe_title = (sim.title.delete_suffix('.') + ".pdf").gsub(/\s+/, "").gsub(/:/, "")
    puts "generating #{safe_title} in #{sanitzied}/"

    Prawn::Document.generate(sanitzied + "/" + safe_title) do |pdf|
      pdf.font("./Roboto-Regular.ttf")

      sim.print(pdf)
    end

    simnumber += 1
  end
end