pdfh 0.1.4 → 0.1.5

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: abfe40299db0b68678beba21f329093fa9ecde2caa3e9df684e16b312c9b8fe2
-  data.tar.gz: 49ee2fc202fbc0cbe279185c4177c27f543c6946d9decdbc4fbf67af9e6b8bf8
+  metadata.gz: 84b58e13bf988d132601b4a22e7d58d2a9fedf0dbaefbe31263d49e441457361
+  data.tar.gz: c5c5faa977695d1e6cb0d42a692f1c951ce23d1fc07fe27e1dc714f16890a9e8
 SHA512:
-  metadata.gz: 60b8883dd62e730ac46306d79d3cd967cd220d8486fe3cf0e21597a7f1c642271528d8a94a37ab21a3cb01bb805cd4f054f46bb918381c39b4325cb3467a2ae0
-  data.tar.gz: 1cd81d3feb7305d5df3119eee85de42587dcae07ddffc9966cdc86e3c87420f0925f105c41fa9d689c2cc68f6b2c75f5ad8260974abd6997a209691595e03c12
+  metadata.gz: c76919c53cc31219265484721f441dc36b5194e859ca8bb0c71528374012e72dab1dde8e3edf47745494501c27d0a0ffcbfb12480a385cef63dfe9dd20e8bffe
+  data.tar.gz: 041110ee2ffa94ce826d020d3e44e7779768358a3e1b8abe89b2743751ee1f1f46319d3ca680d8de9fab04c1dd816e14fa90571d327ab88772d3f53927dabf40

data/.rubocop.yml

@@ -6,3 +6,13 @@ AllCops:
 
 Metrics/LineLength:
   Enabled: false
+
+Metrics/BlockLength:
+  Exclude:
+    - pdfh.gemspec
+
+Metrics/MethodLength:
+  Max: 15
+
+Metrics/AbcSize:
+  Max: 15

data/.ruby-version

@@ -1 +1 @@
-ruby-2.6.0
+ruby-2.6.2

data/CHANGELOG.md

@@ -1,3 +1,11 @@
+## v0.1.5
+* Add print_cmd field in config file for information purposes
+* Settings now validates a no existing directory
+* Refactor for easier maintenance
+
+## v0.1.4
+* Add titleize format when writing new file
+
 ## v0.1.3
 * Fixed copy companion files, which was not copying the files.
 

data/Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    pdfh (0.1.3)
+    pdfh (0.1.5)
       colorize (~> 0.8.1)
 
 GEM
@@ -12,7 +12,7 @@ GEM
     diff-lcs (1.3)
     docile (1.3.1)
     hirb (0.7.3)
-    json (2.1.0)
+    json (2.2.0)
     rake (10.5.0)
     rspec (3.8.0)
       rspec-core (~> 3.8.0)

data/exe/pdfh

@@ -4,6 +4,7 @@
 require 'optparse'
 require 'pdfh'
 require 'pdfh/version'
+require 'pdfh/utils'
 
 options = {}
 opt = OptionParser.new do |opts|
@@ -24,7 +25,7 @@ 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

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

@@ -12,41 +12,69 @@ 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
-  rescue StandardError => e
-    line = e.backtrace[0].match(/:(\d+)/)[1]
-    puts "Error, #{e.message}. #{line}".colorize(:red)
-    exit 1
+
+    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 => err
+    puts "       Doc Error: #{err.message}".colorize(:red)
   end
+  # rubocop:enable Metrics/AbcSize
 
   def self.print_separator(title)
     _rows, cols = `stty size`.split.map(&:to_i)
@@ -62,18 +90,7 @@ module Pdfh
     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 +1,75 @@
 # frozen_string_literal: true
 
 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)
+
+      @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
+      case month
       when 0 then
         @year_offset = -1
         12
       when 13 then
         @year_offset = 1
         1
-      else m
+      else month
       end
     end
 
@@ -99,11 +101,11 @@ module Pdfh
     def new_name
       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 +113,13 @@ 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}
@@ -122,6 +131,7 @@ module Pdfh
         New Name:  #{new_name}
         StorePath: #{store_path}
         Companion: #{companion_files(join: true)}
+        Print Cmd: #{print_cmd}
       STR
     end
 
@@ -131,30 +141,6 @@ module Pdfh
       @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)
-    end
-
     private
 
     ##
@@ -174,60 +160,46 @@ 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 "  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 "   Working on dir: #{home_dir}"
       Dir.chdir(home_dir) do
         all_files = Dir["#{file_name_only}.*"]
-        companion = all_files.reject { |f| f.include? 'pdf' }
+        companion = all_files.reject { |file| file.include? 'pdf' }
         Verbose.print "     - #{companion.join(', ')}"
 
-        companion || []
+        @companion = companion || []
       end
     end
 

data/lib/pdfh/month.rb

@@ -0,0 +1,41 @@
+# frozen_string_literal: true
+
+module Pdfh
+  ##
+  # Handles Month convertions
+  class Month
+    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
+
+    ##
+    # @param [String] month
+    # @return [Integer]
+    def self.normalize(month)
+      # When param is a number
+      month_num = month.to_i
+      return month_num if month_num.between?(1, 12) # (1..12).include?(month_num)
+
+      # When param is a 3 char month: 'mar', 'nov'
+      if month.size == 3
+        MONTHS.keys.each do |mon|
+          return MONTHS[mon] if mon.to_s[0, 3] == month
+        end
+      end
+
+      # When param has a direct match
+      MONTHS[month.to_sym]
+    end
+  end
+end

data/lib/pdfh/pdf_handler.rb

@@ -0,0 +1,54 @@
+# frozen_string_literal: true
+
+module Pdfh
+  ##
+  # Handles the Pdf document text extraction and password removal
+  # 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 PdfHandler
+    attr_reader :file, :password
+
+    def initialize(file, password)
+      @file = file
+      @password = password
+    end
+
+    ##
+    # Gets the text from the pdf in order to execute
+    # the regular expresiom matches
+    def extract_text
+      temp = `mktemp`.chomp
+      Verbose.print "  --> #{temp} temporal file assigned."
+
+      password_opt = "--password='#{@password}'" if @password
+      cmd = %(qpdf #{password_opt} --decrypt --stream-data=uncompress '#{@file}' '#{temp}')
+      Verbose.print "  Command: #{cmd}"
+      _result = `#{cmd}`
+
+      cmd2 = %(pdftotext -enc UTF-8 '#{temp}' -)
+      Verbose.print "  Command: #{cmd2}"
+      text = `#{cmd2}`
+      Verbose.print "  Text extracted: #{text}"
+      text
+    end
+
+    def write_pdf(dir_path, full_path)
+      Verbose.print '~~~~~~~~~~~~~~~~~~ Writing PDFs'
+      raise IOError, "Path #{dir_path} not found." unless Dir.exist?(dir_path)
+
+      password_opt = "--password='#{@password}'" if @password
+      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)
+    end
+  end
+end

data/lib/pdfh/settings.rb

@@ -8,20 +8,18 @@ module Pdfh
   ##
   # Handles the config yaml data mapping, and associates a file name with a doc type
   class Settings
-    attr_accessor :scrape_dirs, :base_path, :document_types
+    attr_reader :scrape_dirs, :base_path, :document_types
 
     def initialize(file)
       file_hash = YAML.load_file(file)
       Verbose.print "Loaded configuration file: #{file}"
 
-      self.scrape_dirs = file_hash['scrape_dirs'].map do |d|
-        File.expand_path(d)
-      end
-      self.base_path = File.expand_path(file_hash['base_path'])
-      self.document_types = process_doc_types(file_hash['document_types'])
+      @scrape_dirs = process_scrape_dirs(file_hash['scrape_dirs'])
+      @base_path = File.expand_path(file_hash['base_path'])
+      @document_types = process_doc_types(file_hash['document_types'])
 
       Verbose.print 'Processing directories:'
-      scrape_dirs.each { |d| Verbose.print "  - #{d}" }
+      scrape_dirs.each { |dir| Verbose.print "  - #{dir}" }
       Verbose.print
     end
 
@@ -38,6 +36,18 @@ module Pdfh
 
     private
 
+    def process_scrape_dirs(scrape_dirs_list)
+      scrape_dirs_list.map do |dir|
+        expanded = File.expand_path(dir)
+        dir_exists = File.directory?(expanded)
+        if dir_exists
+          expanded
+        else
+          Verbose.print "  ** Directory #{dir} does not exists."
+        end
+      end.compact
+    end
+
     def process_doc_types(doc_types)
       doc_types.map do |x|
         object = OpenStruct.new(x)

data/lib/pdfh/utils.rb

@@ -2,9 +2,8 @@
 
 require 'colorize'
 
+# Contains all generic short functionality
 module Pdfh
-  class Error < StandardError; end
-
   ##
   # Keeps Verbose option in whole project
   class Verbose
@@ -34,4 +33,10 @@ module Pdfh
       end
     end
   end
+
+  def self.print_error(exception, exit_app: true)
+    line = exception.backtrace[0].match(/:(?<line>\d+)/)[:line]
+    puts "Error, Line[#{line}]: #{exception.message}.".colorize(:red)
+    exit 1 if exit_app
+  end
 end

data/lib/pdfh/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Pdfh
-  VERSION = '0.1.4'
+  VERSION = '0.1.5'
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: pdfh
 version: !ruby/object:Gem::Version
-  version: 0.1.4
+  version: 0.1.5
 platform: ruby
 authors:
 - Isaias Piña
 autorequire: 
 bindir: exe
 cert_chain: []
-date: 2019-01-28 00:00:00.000000000 Z
+date: 2019-04-01 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: colorize
@@ -119,8 +119,11 @@ files:
 - bin/console
 - bin/setup
 - exe/pdfh
+- lib/ext/string.rb
 - lib/pdfh.rb
 - lib/pdfh/document.rb
+- lib/pdfh/month.rb
+- lib/pdfh/pdf_handler.rb
 - lib/pdfh/settings.rb
 - lib/pdfh/utils.rb
 - lib/pdfh/version.rb
@@ -148,7 +151,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
-rubygems_version: 3.0.1
+rubygems_version: 3.0.3
 signing_key: 
 specification_version: 4
 summary: Organize PDF files