Extrai resultado das votações na Assembleia da República

extracao_votacoes_parlamento.rb

Proposal = Struct.new(:title, :votes_favor, :votes_against, :votes_abstention, :conclusion)

CONCLUSIONS = ['Aprovad', 'Rejeitad']
status = 'FINDING_START'
PARTIES_INDICES = nil
PARTIES_LINE_START = '                      PS'



#                       PS    PSD   BE     PCP     CDS-PP        PAN      PEV   CH   IL     NiJKM      NiCR
def extract_parties_index(line)
  obj = {}
  parties = line.split(/[\s.]+/).reject(&:empty?)
  parties.each do |party|
    obj[line.index(party)] = party
  end
  obj
end

def find_closest_index(i)
  PARTIES_INDICES.keys.min_by{ |x| (i - x).abs }
end

# FAVOR                       X                        X          X             X    X        X            X
def extract_votes(line)
  (0..line.length).find_all { |i| line[i,1] == 'X' }
                  .map { |i| PARTIES_INDICES[find_closest_index(i)] }
                  .compact
end

def contains_conclusion(line)
  CONCLUSIONS.any? { |word| line.include?(word) }
end

cur_proposal = nil
proposals = []

File.readlines(ARGV[0]).each_with_index do |line, index|
  status = 'TITLE_READ' if line.start_with?(PARTIES_LINE_START)

  if line.start_with?('') || status == 'READING_TITLE'
    cur_proposal = Proposal.new('') if line.start_with?('')

    status = 'READING_TITLE'
    cur_proposal.title << line.strip + " "

    status = 'TITLE_READ' if line.strip.end_with?(';')
  end

  if line.start_with?(PARTIES_LINE_START) && !PARTIES_INDICES
    PARTIES_INDICES = extract_parties_index(line)
  end


  if status == 'TITLE_READ'
    status = 'FINDING_VOTES' if line.include?(PARTIES_LINE_START)
    status = 'READING_CONCLUSION' if contains_conclusion(line)
  end

  status = 'READING_VOTES' if line.start_with?('FAVOR')

  if status == 'READING_VOTES'
    if line.start_with?('FAVOR')
      cur_proposal.votes_favor = extract_votes(line)
    elsif line.start_with?('CONTRA')
      cur_proposal.votes_against = extract_votes(line)
    elsif line.start_with?('ABSTEN')
      cur_proposal.votes_abstention = extract_votes(line)
      status = 'READING_CONCLUSION'
    end
  end

  if status == 'READING_CONCLUSION'
    if contains_conclusion(line)
      cur_proposal.conclusion = line.strip
      proposals << cur_proposal

      status = 'FINDING_START'
    end
  end
end

proposals.each do |prop|
  p prop
end

Documento de exemplo: https://app.parlamento.pt/WebUtils/docs/doc.pdf?Path=6148523063446f764c304653546d56304c334e706447567a4c31684a566b786c5a79394e52564e424c30464f52566850553046485255354551564e42636e463161585a764c7a4c43716942545a584e7a77364e764945786c5a326c7a6247463061585a684c31684a566c3879587a4531587a49774d6a41744d5441744d545a664d6a41794d4330784d4330784e6935775a47593d&Fich=XIV_2_15_2020-10-16_2020-10-16.pdf&Inline=true

pdftotext votos.pdf -layout
ruby test.rb votos.txt

Resultado:
#<struct Proposal title="Projeto de Resolução n.º 690/XIV/2.ª (IL) – Portal online de transparência e monitorização do processo de execução dos Fundos Europeus; ", votes_favor=["PSD", "CDS-PP", "PAN", "CH", "IL", "NiJKM", "NiCR"], votes_against=["PS"], votes_abstention=["BE", "PCP", "PEV"], conclusion="Rejeitado">