pdfh 0.1.4 → 0.1.9

data/Rakefile

@@ -1,8 +1,25 @@
 # frozen_string_literal: true
 
-require 'bundler/gem_tasks'
-require 'rspec/core/rake_task'
+require "colorize"
+require "bundler/gem_tasks"
+require "rspec/core/rake_task"
+require "versionomy"
 
 RSpec::Core::RakeTask.new(:spec)
 
 task default: :spec
+
+desc "Bump gem version number (tiny|minor|major)"
+task :bump, :type do |_t, args|
+  args.with_defaults(type: :tiny)
+  version_file = File.join(__dir__, "lib", "pdfh", "version.rb")
+  content = File.read(version_file)
+
+  version_pattern = /(?<major>\d+)\.(?<minor>\d+)\.(?<tiny>\d+)/
+  current_version = content.match(version_pattern)
+  next_version    = Versionomy.parse(current_version.to_s).bump(args.type).to_s
+
+  File.write(version_file, content.gsub(version_pattern, "\\1#{next_version}\\3"))
+
+  puts "Successfully bumped from #{current_version.to_s.red} to #{next_version.green}"
+end

data/bin/console

@@ -1,15 +1,15 @@
 #!/usr/bin/env ruby
 # frozen_string_literal: true
 
-require 'bundler/setup'
-require 'pdfh'
+require "bundler/setup"
+require "pdfh"
 
 # You can add fixtures and/or initialization code here to make experimenting
 # with your gem easier. You can also use a different console, if you like.
 
 # (If you use this, don't forget to add pry to your Gemfile!)
-# require 'pry'
-# Pry.start
+require "pry"
+Pry.start
 
-require 'irb'
-IRB.start(__FILE__)
+# require "irb"
+# IRB.start(__FILE__)

data/exe/pdfh

@@ -1,21 +1,22 @@
 #!/usr/bin/env ruby
 # frozen_string_literal: true
 
-require 'optparse'
-require 'pdfh'
-require 'pdfh/version'
+require "optparse"
+require "pdfh"
+require "pdfh/version"
+require "pdfh/utils"
 
 options = {}
 opt = OptionParser.new do |opts|
   # Process ARGV
   opts.banner = "Usage: #{opts.program_name} [options]"
-  opts.on('-v', '--verbose', 'Show more output, useful for debug') { |o| options[:verbose] = o }
-  opts.on('-d', '--dry', 'Dry run, does not write new pdf') { |o| options[:dry] = o }
-  opts.on_tail('-V', '--version', 'Show version') do
+  opts.on("-v", "--verbose", "Show more output, useful for debug") { |o| options[:verbose] = o }
+  opts.on("-d", "--dry", "Dry run, does not write new pdf") { |o| options[:dry] = o }
+  opts.on_tail("-V", "--version", "Show version") do
     puts "#{opts.program_name} v#{Pdfh::VERSION}"
     exit
   end
-  opts.on_tail('-h', '--help', 'Show this message') do
+  opts.on_tail("-h", "--help", "Show this message") do
     puts opts
     exit
   end
@@ -24,9 +25,19 @@ end
 begin
   opt.parse!
 rescue OptionParser::InvalidOption => e
-  puts "Error, #{e.message}".colorize(:red)
+  Pdfh.print_error e, exit_app: false
   puts opt
   exit 1
 end
 
+def validate_installed(app)
+  require "open3"
+  _stdout, _stderr, status = Open3.capture3("command -v #{app}")
+  puts "Missing #{app} command." unless status.success?
+
+  status.success?
+end
+
+exit(1) unless validate_installed("qpdf") && validate_installed("pdftotext")
+
 Pdfh.main(options)

data/lib/ext/string.rb

@@ -0,0 +1,13 @@
+# frozen_string_literal: true
+
+##
+# Extends String class when required
+module Extensions
+  ##
+  # Adds new functionality to string Class
+  refine String do
+    def titleize
+      split.map(&:capitalize).join(" ")
+    end
+  end
+end

data/lib/pdfh.rb

@@ -1,9 +1,9 @@
 # frozen_string_literal: true
 
-require 'pdfh/version'
-require 'pdfh/settings'
-require 'pdfh/document'
-require 'pdfh/utils'
+require "pdfh/version"
+require "pdfh/settings"
+require "pdfh/document"
+require "pdfh/utils"
 
 ##
 # Gem entry point
@@ -12,68 +12,86 @@ module Pdfh
     Verbose.active = options[:verbose]
     Dry.active = options[:dry]
 
-    settings = Settings.new(search_config_file)
+    @settings = Settings.new(search_config_file)
+    puts "Destination path: #{@settings.base_path.light_blue}"
+    @settings.scrape_dirs.each do |work_directory|
+      process_directory(work_directory)
+    end
+  rescue StandardError => e
+    print_error e
+  end
 
-    settings.scrape_dirs.each do |work_directory|
-      print_separator(work_directory)
-      ignored_files = []
-      Dir["#{work_directory}/*.pdf"].each do |file|
-        base_name = File.basename(file, File.extname(file))
-        type = settings.match_doc_type(file)
-        unless type
-          ignored_files << base_name
-          next
-        end
+  def self.search_config_file
+    name = File.basename($PROGRAM_NAME)
+    names_to_look = ["#{name}.yml", "#{name}.yaml"]
+    dir_order = [Dir.pwd, File.expand_path("~")]
 
-        pad = 12
-        puts "Working on #{base_name.colorize(:light_green)}"
-        print_ident 'Type', type.name, :light_blue, width: pad
-        doc = Document.new(file, type)
-        print_ident 'Sub-Type', doc.sub_type, :light_blue, width: pad
-        print_ident 'Period', doc.period, :light_blue, width: pad
-        print_ident 'New Name', doc.new_name, :light_blue, width: pad
-        print_ident 'Store Path', doc.store_path, :light_blue, width: pad
-        print_ident 'Other files', doc.companion_files(join: true), :light_blue, width: pad
-        doc.write_pdf(settings.base_path)
+    dir_order.each do |dir|
+      names_to_look.each do |file|
+        f = File.join(dir, file)
+        return f if File.file?(f)
       end
+    end
 
-      puts "\nNo account was matched for these PDF files:" unless ignored_files.empty?
-      ignored_files.each.with_index(1) do |file, index|
-        print_ident index, file, :red
+    raise StandardError, "no configuraton file (#{names_to_look.join(" or ")}) was found\n"\
+                         "       within paths: #{dir_order.join(", ")}"
+  end
+
+  ##
+  # @param [String] work_directory
+  def self.process_directory(work_directory)
+    print_separator work_directory
+    ignored_files = []
+    Dir["#{work_directory}/*.pdf"].each do |pdf_file|
+      type = @settings.match_doc_type(pdf_file)
+      if type
+        process_document(pdf_file, type)
+      else
+        ignored_files << basename_without_ext(pdf_file)
       end
     end
+
+    puts "\nNo account was matched for these PDF files:" unless ignored_files.empty?
+    ignored_files.each.with_index(1) { |file, index| print_ident index, file, :magenta }
+  end
+
+  ##
+  # Generate document, and process actions
+  # @param [String] file
+  # @param [Type] type
+  # rubocop:disable Metrics/AbcSize
+  def self.process_document(file, type)
+    puts "Working on #{basename_without_ext(file).colorize(:light_green)}"
+    pad = 12
+    print_ident "Type", type.name, :light_blue, width: pad
+    doc = Document.new(file, type)
+    print_ident "Sub-Type", doc.sub_type, :light_blue, width: pad
+    print_ident "Period", doc.period, :light_blue, width: pad
+    print_ident "New Name", doc.new_name, :light_blue, width: pad
+    print_ident "Store Path", doc.store_path, :light_blue, width: pad
+    print_ident "Other files", doc.companion_files(join: true), :light_blue, width: pad
+    print_ident "Print CMD", doc.print_cmd, :light_blue, width: pad
+    doc.write_pdf(@settings.base_path)
   rescue StandardError => e
-    line = e.backtrace[0].match(/:(\d+)/)[1]
-    puts "Error, #{e.message}. #{line}".colorize(:red)
-    exit 1
+    puts "       Doc Error: #{e.message}".colorize(:red)
   end
+  # rubocop:enable Metrics/AbcSize
 
   def self.print_separator(title)
     _rows, cols = `stty size`.split.map(&:to_i)
-    sep = "\n#{'-' * 40} #{title} "
+    sep = "\n#{"-" * 40} #{title} "
     remaining_cols = cols - sep.size
-    sep += '-' * remaining_cols if remaining_cols.positive?
+    sep += "-" * remaining_cols if remaining_cols.positive?
     puts sep.colorize(:light_yellow)
   end
 
   def self.print_ident(field, value, color = :green, width: 3)
     field_str = field.to_s.rjust(width)
     value_str = value.colorize(color)
-    puts "#{' ' * 4}#{field_str}: #{value_str}"
+    puts "#{" " * 4}#{field_str}: #{value_str}"
   end
 
-  def self.search_config_file
-    name = File.basename($PROGRAM_NAME)
-    names_to_look = ["#{name}.yml", "#{name}.yaml"]
-    dir_order = [Dir.pwd, File.expand_path('~')]
-
-    dir_order.each do |dir|
-      names_to_look.each do |file|
-        f = File.join(dir, file)
-        return f if File.file?(f)
-      end
-    end
-
-    raise StandardError, "no configuraton file (#{names_to_look.join(' or ')}) was found\n       within paths: #{dir_order.join(', ')}"
+  def self.basename_without_ext(file)
+    File.basename(file, File.extname(file))
   end
 end

data/lib/pdfh/document.rb

@@ -1,73 +1,77 @@
 # frozen_string_literal: true
 
-require 'fileutils'
-require 'ext/string'
+require "fileutils"
+require "pdfh/month"
+require "pdfh/pdf_handler"
+require "ext/string"
 
+##
+# main module
 module Pdfh
-  using Pdfh::String
+  using Extensions
+
+  ##
+  # Regular Date Error, when there is not match
+  class ReDateError < StandardError
+    def initialize(message = "No data matched your date regular expression")
+      super(message)
+    end
+  end
 
   ##
   # Handles the PDF detected by the rules
-  # TODO: Replace command utils with this gem
-  #  require 'pdf-reader'
-  #
-  #  reader = PDF::Reader.new(temp)
-  #  reader.pages.each do |page|
-  #      @text << page.text
-  #  end
   class Document
-    attr_accessor :text, :type, :file, :extra
-
-    MONTHS = {
-      enero: 1,
-      febrero: 2,
-      marzo: 3,
-      abril: 4,
-      mayo: 5,
-      junio: 6,
-      julio: 7,
-      agosto: 8,
-      septiembre: 9,
-      octubre: 10,
-      noviembre: 11,
-      diciembre: 12
-    }.freeze
+    attr_reader :text, :type, :file, :extra
 
     def initialize(file, type, _options = {})
+      raise IOError, "File #{file} not found" unless File.exist?(file)
+
       @file = file
       @type = type
       @month_offset = 0
       @year_offset = 0
 
-      raise IOError, "File #{file} not found" unless File.exist?(file)
+      @pdf_doc = PdfHandler.new(file, @type.pwd)
 
       Verbose.print "=== Type: #{type_name} =================="
-      @text = pdf_text
-      Verbose.print "  Text extracted: #{@text}"
+      @text = @pdf_doc.extract_text
       @sub_type = extract_subtype(@type.sub_types)
       Verbose.print "  SubType: #{@sub_type}"
       @month, @year, @extra = match_data
       Verbose.print "==== Assigned: #{@month}, #{@year}, #{@extra} ==( Month, Year, Extra )================"
-      @companion = find_companion_files
+      find_companion_files
+    end
+
+    def write_pdf(base_path)
+      full_path = File.join(base_path, store_path, new_name)
+      dir_path = File.join(base_path, store_path)
+
+      FileUtils.mkdir_p(dir_path) unless File.exist?(dir_path)
+
+      @pdf_doc.write_pdf(dir_path, full_path)
+
+      return if Dry.active?
+
+      make_document_backup
+      copy_companion_files(dir_path)
     end
 
     def period
-      m = month.to_s.rjust(2, '0')
-      y = year
-      "#{y}-#{m}"
+      formated_month = month.to_s.rjust(2, "0")
+      "#{year}-#{formated_month}"
     end
 
     def month
-      m = normalize_month + @month_offset
+      month = Month.normalize(@month) + @month_offset
 
-      case m
-      when 0 then
+      case month
+      when 0
         @year_offset = -1
         12
-      when 13 then
+      when 13
         @year_offset = 1
         1
-      else m
+      else month
       end
     end
 
@@ -89,21 +93,21 @@ module Pdfh
     end
 
     def type_name
-      @type ? @type.name.titleize : 'N/A'
+      @type ? @type.name.titleize : "N/A"
     end
 
     def sub_type
-      @sub_type ? @sub_type.name.titleize : 'N/A'
+      @sub_type ? @sub_type.name.titleize : "N/A"
     end
 
     def new_name
-      template = @type.to_h.key?(:name_template) ? @type.name_template : '{original}'
+      template = @type.to_h.key?(:name_template) ? @type.name_template : "{original}"
       new_name = template
-                 .gsub(/\{original\}/, file_name_only)
-                 .gsub(/\{period\}/, period)
-                 .gsub(/\{type\}/, type_name)
-                 .gsub(/\{subtype\}/, sub_type)
-                 .gsub(/\{extra\}/, extra || '')
+                 .sub("{original}", file_name_only)
+                 .sub("{period}", period)
+                 .sub("{type}", type_name)
+                 .sub("{subtype}", sub_type)
+                 .sub("{extra}", extra || "")
       "#{new_name}.pdf"
     end
 
@@ -111,6 +115,13 @@ module Pdfh
       @type.store_path.gsub(/\{YEAR\}/, year.to_s)
     end
 
+    def print_cmd
+      return "N/A" if type.print_cmd.nil? || type.print_cmd.empty?
+
+      relative_path = File.join(store_path, new_name)
+      "#{type.print_cmd} #{relative_path}"
+    end
+
     def to_s
       <<~STR
         Name:      #{file_name_only}
@@ -122,49 +133,26 @@ module Pdfh
         New Name:  #{new_name}
         StorePath: #{store_path}
         Companion: #{companion_files(join: true)}
+        Print Cmd: #{print_cmd}
       STR
     end
 
     def companion_files(join: false)
       @companion unless join
 
-      @companion.empty? ? 'N/A' : @companion.join(', ')
-    end
-
-    def write_pdf(base_path)
-      Verbose.print '~~~~~~~~~~~~~~~~~~ Writing PDFs'
-      full_path = File.join(base_path, store_path, new_name)
-      dir_path = File.join(base_path, store_path)
-
-      raise IOError, "Path #{dir_path} was not found." unless Dir.exist?(dir_path)
-
-      password_opt = "--password='#{@type.pwd}'" if @type.pwd
-      cmd = %(qpdf #{password_opt} --decrypt '#{@file}' '#{full_path}')
-      Verbose.print "  Write pdf command: #{cmd}"
-
-      return if Dry.active?
-
-      _result = `#{cmd}`
-      raise IOError, "File #{full_path} was not created." unless File.file?(full_path)
-
-      # Making backup of original
-      Dir.chdir(home_dir) do
-        File.rename(file, backup_name)
-      end
-
-      copy_companion_files(dir_path)
+      @companion.empty? ? "N/A" : @companion.join(", ")
     end
 
     private
 
     ##
-    # @param [Array] subtypes
+    # @param sub_types [Array]
     # @return [OpenStruct]
     def extract_subtype(sub_types)
       return nil if sub_types.nil? || sub_types.empty?
 
       sub_types.each do |st|
-        is_matched = Regexp.new(st['name']).match?(@text)
+        is_matched = Regexp.new(st["name"]).match?(@text)
         next unless is_matched
 
         sub = OpenStruct.new(st)
@@ -174,70 +162,56 @@ module Pdfh
       nil
     end
 
-    def normalize_month
-      month_num = @month.to_i
-      return month_num if month_num.positive?
-
-      if @month.size == 3
-        MONTHS.keys.each do |mon|
-          return MONTHS[mon] if mon.to_s[0, 3] == @month
-        end
-      end
-
-      MONTHS[@month.to_sym]
-    end
-
     def home_dir
       File.dirname(@file)
     end
 
-    def pdf_text
-      temp = `mktemp`.chomp
-      Verbose.print "  --> #{temp} temporal file assigned."
-
-      password_opt = "--password='#{@type.pwd}'" if @type.pwd
-      cmd = %(qpdf #{password_opt} --decrypt --stream-data=uncompress '#{@file}' '#{temp}')
-      Verbose.print "  Command: #{cmd}"
-      _result = `#{cmd}`
-      Verbose.print 'Password removed.'
-
-      cmd2 = %(pdftotext -enc UTF-8 '#{temp}' -)
-      Verbose.print "  Command: #{cmd2}"
-      `#{cmd2}`
-    end
-
+    ##
+    # named matches can appear in any order with names 'd', 'm' and 'y'
+    # unamed matches needs to be in order month, year
+    # @return [Array] - format [month, year, day]
     def match_data
-      Verbose.print '~~~~~~~~~~~~~~~~~~ RegEx'
+      Verbose.print "~~~~~~~~~~~~~~~~~~ RegEx"
       Verbose.print "  Using regex: #{@type.re_date}"
       Verbose.print "        named:   #{@type.re_date.named_captures}"
       matched = @type.re_date.match(@text)
+      raise ReDateError unless matched
+
       Verbose.print "     captured: #{matched.captures}"
 
       return matched.captures.map(&:downcase) if @type.re_date.named_captures.empty?
 
-      extra = matched.captures.size > 2 ? matched[3] : nil
+      extra = matched.captures.size > 2 ? matched[:d] : nil
       [matched[:m].downcase, matched[:y], extra]
     end
 
+    ##
+    # Create a backup of original document
+    def make_document_backup
+      Dir.chdir(home_dir) do
+        File.rename(file, backup_name)
+      end
+    end
+
     def find_companion_files
-      Verbose.print '~~~~~~~~~~~~~~~~~~ Searching Companion files'
+      Verbose.print "~~~~~~~~~~~~~~~~~~ Searching Companion files"
       Verbose.print "   Working on dir: #{home_dir}"
       Dir.chdir(home_dir) do
         all_files = Dir["#{file_name_only}.*"]
-        companion = all_files.reject { |f| f.include? 'pdf' }
-        Verbose.print "     - #{companion.join(', ')}"
+        companion = all_files.reject { |file| file.include? "pdf" }
+        Verbose.print "     - #{companion.join(", ")}"
 
-        companion || []
+        @companion = companion || []
       end
     end
 
     def copy_companion_files(destination)
-      Verbose.print '~~~~~~~~~~~~~~~~~~ Writing Companion files'
+      Verbose.print "~~~~~~~~~~~~~~~~~~ Writing Companion files"
       @companion.each do |file|
         Verbose.print "  Working on #{file}..."
         src_name = File.join(home_dir, file)
         src_ext = File.extname(file)
-        dest_name = File.basename(new_name, '.pdf')
+        dest_name = File.basename(new_name, ".pdf")
         dest_full = File.join(destination, "#{dest_name}#{src_ext}")
         Verbose.print "    cp #{src_name} --> #{dest_full}"
         FileUtils.cp(src_name, dest_full)