arquivo 0.1.9 → 0.2.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 76258c87d7c0c3a63f261359001ea4623dd00a2f53f5d4a6469fa121bd361bae
-  data.tar.gz: '02804d07fa8c7db3ba86995568ed76aa5948775e854a5d0a5159a90cf1e72f7d'
+  metadata.gz: 5327acf76aa8ca135c8e6faac8aeb2b5aee4a66a950cf07f3a7498f98da503f5
+  data.tar.gz: d37de979686bfef7182714d1b654da47dae151b3b68b52ebc0b7cf79e25cfa4b
 SHA512:
-  metadata.gz: 75ab73db2816fe15e4f80a8f8bdbeafa349c87dd50bb6ac53f6981e21e5349734d79461ed7a2e7b3b4dafe3b9f8fdf3bf35e5930b34132ee299ff2fdd7d6b021
-  data.tar.gz: a81b2f6d65e75608db0a22b54718e9c8433fedf6702923680568cd08384db924982f6d27b1c2aa1c3009d07a8a071dc8e6034f074d72a8785b77e34423cac3a7
+  metadata.gz: 2d82bbfe746b9dfb2337abcb2c8ef54cb0adfe9466c2b51cde53dfb54eaf1f1cb25828216531350f5c1c5f7e2cf7664f0b80b223b00b17852681826ef714c3fb
+  data.tar.gz: 57dd9546e5d54eb0ad94da2b2fca720e0ac2511e671c1d39f169ff8b03773452f705e25e460ac884d7ac6fb5c4ebd1c558dac341da9d56cfff1a0233ba8560e8

data/Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    arquivo (0.1.9)
+    arquivo (0.2.0)
       fastimage (~> 2.1)
       google-api-client (~> 0.34)
       pdf-reader (~> 2.3)

data/lib/arquivo/dir.rb

@@ -4,11 +4,11 @@ require 'google/apis/sheets_v4'
 require 'googleauth'
 require 'googleauth/stores/file_token_store'
 
+require 'arquivo/noise'
+
 module Arquivo
-  # O1 = '2>/dev/null'
-  # O2 = '1>/dev/null 2>&1'
-  O1 = ''
-  O2 = ''
+  O1 = '2>/dev/null'
+  O2 = '1>/dev/null 2>&1'
 
   # analisar/processar pasta
   class C118dir < Enumerator
@@ -22,7 +22,7 @@ module Arquivo
     # @return [Float] maximo segundos de silencio encontrados
     attr_reader :silence
     # @return [String] noiseprof do silencio encontrado
-    attr_reader :nprof
+    attr_reader :noiseprof
 
     # @return [String] base nome ficheiros finais (pdf, tar.gz)
     attr_reader :base
@@ -30,29 +30,35 @@ module Arquivo
     # @return [C118dir] documentos c118
     def initialize(pasta)
       @items = Dir.glob(File.join(pasta, '*')).each
-      @base = File.basename(pasta, File.extname(pasta)) +
+      @base = File.basename(pasta, File.extname(pasta)) + '-' +
               Date.today.strftime('%Y%m%d')
     end
 
-    def processa_pasta(options)
+    def processa_items(options)
       n = 0
       while next_item
         if File.ftype(item) == 'directory'
-          C118dir.new(item).processa_pasta(options)
+          C118dir.new(item).processa_pasta(item, options)
         else
           processa_file(options, File.extname(item).downcase)
           n += 1
         end
       end
-      processa_fim(n) if n.positive?
+      processa_fim(n)
     end
 
     def processa_fim(num)
-      system "rm -f #{base}.*;" \
-             "pdftk tmp/stamped*.pdf cat output #{base}.pdf;" \
-             "cd tmp/zip;tar cf ../../#{base}.tar *.pdf;" \
-             "cd ../..;gzip --best #{base}.tar;" \
-             'rm -rf ./tmp'
+      return unless num.positive?
+
+      cmd = if /minuta/i.match?(base)
+              "rm -f #{base}.*"
+            else
+              "rm -f #{base}.*;pdftk tmp/stamped*.pdf cat output #{base}.pdf"
+            end
+      # ;rm -rf tmp
+      system cmd + ";cd tmp/zip;tar cf ../../#{base}.tar *" \
+                   ";cd ../..;gzip --best #{base}.tar;rm -rf tmp"
+
       puts "#{base} (#{num})"
     end
 
@@ -61,7 +67,7 @@ module Arquivo
       when '.jpg' then C118jpg.new(item).processa_jpg(options, dados)
       when '.pdf' then C118pdf.new(item).processa_pdf(options, dados)
       when '.mp3', '.m4a', '.wav'
-        C118mp3.new(item).processa_mp3(options, num)
+        C118mp3.new(item).processa_mp3(options, noiseprof)
       else
         puts "erro: #{item} so posso processar mp3, jpg, pdf"
       end
@@ -74,28 +80,20 @@ module Arquivo
       @item = nil
     end
 
-    def prepara(pasta, options)
-      obtem_dados(pasta)
-      obtem_noiseprof(pasta, options)
-      system 'mkdir -p tmp/zip'
-    end
-
-    def obtem_noiseprof(dir, options)
-      return unless /minuta/i.match?(dir) || silence&.zero?
+    def processa_pasta(pasta, options)
+      unless File.ftype(items.peek) == 'directory'
+        @dados = {}
+        obtem_dados(pasta)
 
-      if options[:nred]
         @silence = 0.0
-        silencio(1, duracao(item), options[:som]) while next_item
-        @nprof = noiseprof
+        system 'mkdir -p tmp/zip'
+        obtem_noiseprof(pasta, options)
       end
-    rescue StandardError
-      @silence = 0.0
+      processa_items(options)
     end
 
     def obtem_dados(dir)
-      return unless /fac?tura/i.match?(dir) ||
-                    /recibo/i.match?(dir) ||
-                    dados&.empty?
+      return unless /fac?tura/i.match?(dir) || /recibo/i.match?(dir)
 
       # obtem dados (faturas/recibos) da sheet c118-contas
       id = '1PbiMrtTtqGztZMhe3AiJbDS6NQE9o3hXebnQEFdt954'

data/lib/arquivo/extrato.rb

@@ -5,33 +5,6 @@ require 'pdf-reader'
 module Arquivo
   # analisar/processar pdf
   class C118pdf < String
-    # @return [String] nome do documento
-    attr_reader :file
-    # @return [String] extensao do documento
-    attr_reader :ext
-    # @return [String] base do documento
-    attr_reader :base
-    # @return [String] key do documento ft????/rc????/ex??0??/sc??????
-    attr_reader :key
-    # @return [Integer] tamanho do pdf
-    attr_reader :size
-
-    # @return [Array<Integer>] numeros pagina do extrato final
-    attr_reader :paginas
-    # @return [String] texto pagina pdf
-    attr_reader :pagina
-    # @return [String] nome extrato
-    attr_reader :nome
-
-    # @return [C118pdf] pdf c118
-    def initialize(fpdf)
-      @file = fpdf
-      @ext = File.extname(fpdf).downcase
-      @base = File.basename(fpdf, File.extname(fpdf))
-      @key = @base[/\w+/]
-      @size = File.size(fpdf)
-    end
-
     def c118_gs
       # filtrar images para scq e extratos
       fi = /^[se]/i.match?(key) ? ' -dFILTERIMAGE' : ''

data/lib/arquivo/mp3.rb

@@ -9,61 +9,56 @@ module Arquivo
     attr_reader :ext
     # @return [String] base do ficheiro
     attr_reader :base
-    # @return [String] final do nome do segmento
-    attr_reader :final
     # @return [Float] segundos do mp3
     attr_reader :size
 
+    # @return [String] nome segmento
+    attr_reader :nome
+
     # @return [C118mp3] mp3 c118
     def initialize(fmp3)
       @file = fmp3
       @ext = File.extname(fmp3).downcase
       @base = File.basename(fmp3, File.extname(fmp3))
-      @final = "-#{@base[/\d{8}/]}#{@base[/-\w+/]}#{@ext}"
       @size = `soxi -V0 -D #{fmp3} #{O1}`.to_f
     end
 
-    def forca_mp3
-      o = "tmp/#{base}.mp3"
-      system "sox \"#{file}\" #{o} #{O2}" unless ext == '.mp3'
-      File.size?(o) ? C118mp3.new(o) : self
-    end
-
-    def processa_mp3
-      system processa_segmentos(Dir.glob("tmp/#{base}*#{ext}"),
-                                0, '', options[:amount])
-    end
-
-    def processa_segmentos(ase, pse, cmd, amt)
-      return cmd[1..-1] unless pse < ase.size
-
-      fls = "#{base}/s#{pse}#{final} " \
-        "tmp/s#{pse}-#{final}#{base[/-\w+/]}.mp3"
-      cmd += nprof ? ";sox #{fls} noisered #{nprof} #{ff(amt)}" : ";cp #{fls}"
-
-      processa_segmentos(ase, pse + 1, cmd, amt)
-    end
-
-    def ff(val)
-      format('%<valor>.5f', valor: val)
+    def processa_mp3(options, npr)
+      cmd = if npr
+              "noisered #{npr} #{format('%<v>.5f', v: options[:amount])} "
+            else
+              ''
+            end
+      cmd += "rate -v #{options[:rate]}k"
+      system "sox -G #{file} tmp/zip/#{base}.mp3 #{cmd} #{O2}"
+      # puts base
     end
 
     def segmenta(tps, pse, cmd)
       return cmd[1..-1] unless pse < tps.size
 
-      cmd += ";sox #{file} #{base}/s#{pse}#{final} trim #{tps[pse]}"
+      puts proximo_segmento(pse)
+
+      cmd += ";sox #{file} #{nome} trim #{tps[pse]}"
       pse += 1
       cmd += " =#{tps[pse]}" if pse < tps.size
 
       segmenta(tps, pse, cmd + " #{O2}")
     end
 
+    def proximo_segmento(pse)
+      out = "s#{format('%<v>02d', v: pse)}-#{base[/\d{8}/]}#{base[/-\w+/]}"
+      @nome = "#{base}/#{out}#{ext}"
+      out
+    end
+
     def processa_minuta(options)
       system segmenta(['0'] + options[:tempos], 0, '')
     end
 
     def processa_minuta?
       return true if ['.mp3', '.m4a', '.wav'].include?(ext) &&
+                     size.positive? &&
                      !File.exist?(base)
 
       if File.exist?(base)

data/lib/arquivo/noise.rb

@@ -3,15 +3,23 @@
 module Arquivo
   # analisar/processar pasta
   class C118dir < Enumerator
+    def obtem_noiseprof(dir, options)
+      return unless /minuta/i.match?(dir) && !options[:noise]
+
+      silencio(1, duracao(item), options[:sound]) while next_item
+      items.rewind
+      @noiseprof = processa_noiseprof
+    end
+
     def silencio(thr, tse, som)
-      out = "tmp/silencio-#{File.basename(item)}"
+      o = "tmp/silencio-#{File.basename(item)}"
 
-      system "sox #{item} #{out} " \
+      system "sox #{item} #{o} " \
              "silence 1 #{format('%<valor>.5f', valor: som)}t #{thr}% #{O2}"
 
-      return if silencio?(out, tse) || thr == 3
+      return if silencio?(o, tse) || thr == 3
 
-      silencio(thr + 1, pse, tse, som)
+      silencio(thr + 1, tse, som)
     end
 
     def silencio?(fss, tse)
@@ -19,20 +27,21 @@ module Arquivo
       return false unless s.positive? && (tse - s > silence)
 
       @silence = tse - s
-      @nprof = fss
+      @noiseprof = fss
     end
 
     def duracao(seg)
       `soxi -V0 -D #{seg} #{O1}`.to_f
     end
 
-    def noiseprof
+    def processa_noiseprof
       return unless silence&.positive?
 
-      o = "tmp/prof-#{base}"
+      e = File.extname(noiseprof)
+      o = "tmp/noiseprof-#{File.basename(noiseprof, e)}"
       # obter noiseprof do silencio encontrado
-      system "sox #{nprof} #{o}#{ext} trim 0 #{silence} #{O2};" \
-             "sox #{o}#{ext} -n noiseprof #{o} #{O2}"
+      system "sox #{noiseprof} #{o}#{e} trim 0 #{silence} #{O2};" \
+             "sox #{o}#{e} -n noiseprof #{o} #{O2}"
 
       # so noiseprof validos sao devolvidos
       @silence = 0.0 unless File.size?(o)

data/lib/arquivo/pdf.rb

@@ -1,5 +1,6 @@
 # frozen_string_literal: true
 
+require 'arquivo/extrato'
 require 'i18n'
 
 I18n.config.available_locales = :pt
@@ -7,6 +8,33 @@ I18n.config.available_locales = :pt
 module Arquivo
   # analisar/processar pdf
   class C118pdf < String
+    # @return [String] nome do documento
+    attr_reader :file
+    # @return [String] extensao do documento
+    attr_reader :ext
+    # @return [String] base do documento
+    attr_reader :base
+    # @return [String] key do documento ft????/rc????/ex??0??/sc??????
+    attr_reader :key
+    # @return [Integer] tamanho do pdf
+    attr_reader :size
+
+    # @return [Array<Integer>] numeros pagina do extrato final
+    attr_reader :paginas
+    # @return [String] texto pagina pdf
+    attr_reader :pagina
+    # @return [String] nome extrato
+    attr_reader :nome
+
+    # @return [C118pdf] pdf c118
+    def initialize(fpdf)
+      @file = fpdf
+      @ext = File.extname(fpdf).downcase
+      @base = File.basename(fpdf, File.extname(fpdf))
+      @key = @base[/\w+/]
+      @size = File.size(fpdf)
+    end
+
     def processa_pdf(options, dados)
       # em caso de scanned pdf extract.trim.jpg -> trimed pdf
       tpdf = jpg? ? extract.trim(options).converte(options) : self
@@ -22,18 +50,16 @@ module Arquivo
       system "#{c118_gs} -sOutputFile=tmp/stamp-#{key}.pdf -c \"#{s}\";" \
              "pdftk tmp/zip/#{base}.pdf " \
              "stamp tmp/stamp-#{key}.pdf output #{o} #{O2}"
-      puts key
+      # puts key
     end
 
     def final(kda)
       c118_stamp(kda)
       o = "tmp/zip/#{base}.pdf"
 
-      recibo = key[0] == 'r'
-      # google print has better && smaller pdf then c118_gs
-      system "#{c118_gs} -sOutputFile=#{o} \"#{file}\" #{O2}" unless recibo
+      system "#{c118_gs} -sOutputFile=#{o} \"#{file}\" #{O2}"
       # usar copia do original se processado for maior
-      system "cp \"#{file}\" #{o}" if recibo || File.size(o) > size
+      system "cp \"#{file}\" #{o}" if File.size(o) > size
 
       C118pdf.new(o)
     end

data/lib/arquivo/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Arquivo
-  VERSION = '0.1.9'
+  VERSION = '0.2.0'
 end

data/lib/arquivo.rb

@@ -2,9 +2,7 @@
 
 require 'thor'
 require 'arquivo/version'
-require 'arquivo/extrato'
 require 'arquivo/dir'
-require 'arquivo/noise'
 require 'arquivo/pdf'
 require 'arquivo/jpg'
 require 'arquivo/mp3'
@@ -24,13 +22,14 @@ module Arquivo
       f = C118mp3.new(file)
       return unless f.processa_minuta?
 
-      system "mkdir -p tmp #{f.base}"
+      system "mkdir -p #{f.base}"
       f.processa_minuta(options)
     end
 
-    desc 'pdf PDF', 'processa PDF criando pasta com documentos para arquivo'
+    desc 'pdf EXTRATO', 'processa EXTRATO criando pasta ' \
+                        'com documentos para arquivo'
     def pdf(file)
-      return unless File.ftype(file) == 'file'
+      return unless File.exist?(file) && File.ftype(file) == 'file'
 
       f = C118pdf.new(file)
       return unless f.processa_extrato?
@@ -44,21 +43,26 @@ module Arquivo
       end
     end
 
-    desc 'dir PASTA', 'processa faturas/recibos/extratos/minutas'
+    desc 'dir PASTA', 'processa faturas/recibos/extratos/minutas ' \
+                      ' e cria arquivos c118'
     option :fuzz, type: :numeric, default: 29,
                   desc: 'fuzz trim jpg N-1, escolhe menor -> scanned pdf'
     option :quality, type: :numeric, default: 15,
                      desc: 'compress jpg N% -> scanned pdf (less=low quality)'
-    option :nred, type: :boolean, default: true,
-                  desc: 'fazer reducao do ruido de fundo'
-    option :som, type: :numeric, default: 1.0,
-                 desc: 'minimo som que determina fim do silencio (segundos)'
-    option :amount, type: :numeric, default: 0.00001,
+
+    option :noise, type: :boolean, default: false,
+                   desc: 'ruido de fundo - sim ou nao'
+    option :sound, type: :numeric, default: 1.0,
+                   desc: 'minimo som que determina fim do silencio (segundos)'
+    option :amount, type: :numeric, default: 0.0001,
                     desc: 'qtd ruido a ser removido'
+    option :rate, type: :numeric, default: 16,
+                  desc: 'sample rate - radio-16k, CD-44.1k, PC-48k, pro-96k'
+
     def dir(fdir)
       return unless File.ftype(fdir) == 'directory'
 
-      C118dir.new(fdir).prepara(fdir, options).processa_pasta(options)
+      C118dir.new(fdir).processa_pasta(fdir, options)
     end
   end
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: arquivo
 version: !ruby/object:Gem::Version
-  version: 0.1.9
+  version: 0.2.0
 platform: ruby
 authors:
 - Hernâni Rodrigues Vaz
 autorequire: 
 bindir: exe
 cert_chain: []
-date: 2019-12-19 00:00:00.000000000 Z
+date: 2019-12-20 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: bundler