pdf_paradise 0.3.20

data/lib/pdf_paradise/utility_scripts/convert_pdf_to_text.rb

@@ -0,0 +1,179 @@
+#!/usr/bin/ruby -w
+# Encoding: UTF-8
+# frozen_string_literal: true
+# =========================================================================== #
+# === PdfParadise::ConvertPdfToText
+#
+# This file can convert one or several .pdf files to text files.
+#
+# Usage examples:
+#
+#   PdfParadise::ConvertPdfToText.new(ARGV)
+#
+# =========================================================================== #
+# require 'pdf_paradise/utility_scripts/convert_pdf_to_text.rb'
+# PdfParadise::ConvertPdfToText.new(ARGV)
+# =========================================================================== #
+require 'pdf_paradise/base/base.rb'
+
+module PdfParadise
+
+class ConvertPdfToText < ::PdfParadise::Base # === PdfParadise::ConvertPdfToText
+
+  # ========================================================================= #
+  # === NAMESPACE
+  # ========================================================================= #
+  NAMESPACE = inspect
+
+  # ========================================================================= #
+  # === initialize
+  # ========================================================================= #
+  def initialize(
+      i           = ARGV,
+      run_already = true
+    )
+    reset
+    # ======================================================================= #
+    # Next designate which files are to be converted.
+    # ======================================================================= #
+    set_convert_these(i)
+    run if run_already
+  end
+
+  # ========================================================================= #
+  # === reset
+  # ========================================================================= #
+  def reset
+    super()
+  end
+
+  # ========================================================================= #
+  # === get_all_pdf_files_from_current_directory
+  #
+  # This method will obtain all pdf files from the current directory.
+  # ========================================================================= #
+  def get_all_pdf_files_from_current_directory
+    return Dir['*.pdf']
+  end
+
+  # ========================================================================= #
+  # === set_convert_these
+  #
+  # This method will preferentially try to work on only .pdf files.
+  # ========================================================================= #
+  def set_convert_these(
+      i = return_pwd
+    )
+    i = return_pwd if i.nil?
+    if i.is_a? String
+      if i.empty?
+        i = get_all_pdf_files_from_current_directory()
+      elsif File.directory? i
+        i = get_all_pdf_files_from_current_directory()
+      end
+    elsif i.is_a? Array
+      i = get_all_pdf_files_from_current_directory() if i.empty?
+    end
+    i = [i] unless i.is_a? Array
+    @convert_these = i # This must be an Array.
+  end
+
+  # ========================================================================= #
+  # === start_conversion
+  # ========================================================================= #
+  def start_conversion
+    @convert_these.each {|pdf| convert_this_pdf(pdf) }
+  end
+
+  # ========================================================================= #
+  # === opnn
+  # ========================================================================= #
+  def opnn
+    super(NAMESPACE)
+  end
+
+  # ========================================================================= #
+  # === convert_this_pdf
+  #
+  # This will make use of the binary called "pdftotext".
+  # ========================================================================= #
+  def convert_this_pdf(
+      i = @convert_these
+    )
+    if i.is_a? Array
+      i.each {|entry| convert_this_pdf(entry) }
+    else
+      if File.exist? i
+        opnn; e "Now converting `#{sfile(i)}#{rev}` via #{simp('pdftotext')}#{rev}."
+        # =================================================================== #
+        # Delegate towards pdftotext next; since as of January 2024 we will
+        # use -enc UTF-8 by default.
+        # =================================================================== #
+        cmdline_to_use = 'pdftotext -enc UTF-8 "'+i+'"' # We use "". 
+        @output_file = i.gsub(/#{File.extname(i)}/, '')+'.txt'
+        system(cmdline_to_use)
+        opnn; e "Storing into the file `#{sfile(@output_file)}`."
+      else
+        unless i.start_with? '--'
+          opnn; e 'No file called '+sfile(i)+' could be found.'
+        end
+      end
+    end
+  end; alias do_convert convert_this_pdf # === do_convert
+
+  # ========================================================================= #
+  # === output_file?
+  # ========================================================================= #
+  def output_file?
+    @output_file
+  end
+
+  # ========================================================================= #
+  # === menu                                                       (menu tag)
+  # ========================================================================= #
+  def menu(
+      i = @convert_these
+    )
+    if i.is_a? Array
+      i.each {|entry| menu(entry) }
+    else
+      case i
+      # ===================================================================== #
+      # === cpdf --gui
+      # ===================================================================== #
+      when /^-?-?gui/i
+        do_start_the_GUI_interface
+      end
+    end
+  end
+
+  # ========================================================================= #
+  # === do_start_the_GUI_interface
+  # ========================================================================= #
+  def do_start_the_GUI_interface
+    require 'pdf_paradise/gui/gtk3/convert_pdf_to_text/convert_pdf_to_text.rb'
+    ::PdfParadise::GUI::Gtk::ConvertPdfToText.run
+  end
+
+  # ========================================================================= #
+  # === run                                                         (run tag)
+  # ========================================================================= #
+  def run
+    menu
+    start_conversion
+  end
+
+end
+
+# =========================================================================== #
+# === PdfParadise.pdf_to_text
+# =========================================================================== #
+def self.pdf_to_text(i = ARGV)
+  PdfParadise::ConvertPdfToText.new(i)
+end
+
+end
+
+if __FILE__ == $PROGRAM_NAME
+  PdfParadise::ConvertPdfToText.new(ARGV)
+end # cpdf foobar.pdf

data/lib/pdf_paradise/utility_scripts/delete_last_page_of_this_pdf_file.rb

@@ -0,0 +1,180 @@
+#!/usr/bin/ruby -w
+# Encoding: UTF-8
+# frozen_string_literal: true
+# =========================================================================== #
+# === PdfParadise::DeleteLastPageOfThisPdfFile
+#
+# This class will accept one (or several) .pdf files and then proceed
+# to remove the very last .pdf page in that file. So if that pdf-file
+# has had 100 pages, we will end up with 99 pages.
+#
+# By default, the class here will create a new output file and leave
+# the original file untouched.
+#
+# Note that this functionality depends on either "qpdf" or "pdftk".
+# =========================================================================== #
+# require 'pdf_paradise/utility_scripts/delete_last_page_of_this_pdf_file.rb'
+# =========================================================================== #
+require 'pdf_paradise/base/base.rb'
+
+module PdfParadise
+
+class DeleteLastPageOfThisPdfFile < Base 
+
+  require 'pdf_paradise/utility_scripts/pdf_file_n_total_pages.rb'
+
+  # ========================================================================= #
+  # === OVERWRITE_THE_OLD_PDF_FILE
+  #
+  # If the following constant is set to true then the given input
+  # file will be overwritten.
+  # ========================================================================= #
+  OVERWRITE_THE_OLD_PDF_FILE = true
+
+  # ========================================================================= #
+  # === initialize
+  # ========================================================================= #
+  def initialize(
+      input_files = ARGV,
+      run_already = true
+    )
+    reset
+    set_input_files(input_files)
+    run if run_already
+  end
+
+  # ========================================================================= #
+  # === reset
+  # ========================================================================= #
+  def reset
+    super()
+    infer_the_namespace
+    # ======================================================================= #
+    # === @overwrite_the_old_pdf_file
+    # ======================================================================= #
+    @overwrite_the_old_pdf_file = OVERWRITE_THE_OLD_PDF_FILE
+  end
+
+  # ========================================================================= #
+  # === set_input_files
+  # ========================================================================= #
+  def set_input_files(i)
+    i = [i] unless i.is_a? Array
+    if i.empty? and is_on_roebe?
+      # ===================================================================== #
+      # In this case we will use the main-pdf file, if it exists.
+      # ===================================================================== #
+      if ENV.has_key?('MAIN_BOOK') and
+        File.exist?(ENV['MAIN_BOOK'])
+        i << ENV['MAIN_BOOK'].dup
+      end
+    end
+    @input_files = i
+    sanitize_input_files
+  end
+
+  # ========================================================================= #
+  # === sanitize_input_files
+  # ========================================================================= #
+  def sanitize_input_files
+    unless @input_files.empty?
+      # ===================================================================== #
+      # Select only .pdf files next.
+      # ===================================================================== #
+      @input_files.select! {|entry|
+        entry.end_with? '.pdf'
+      }
+    end
+  end
+
+  # ========================================================================= #
+  # === process_each_pdf
+  # ========================================================================= #
+  def process_each_pdf
+    @input_files.each {|this_pdf_file|
+      # ===================================================================== #
+      # First, we must find out how many pdf pages are in the given
+      # pdf file at hand.
+      # ===================================================================== #
+      if File.exist? this_pdf_file
+        has_n_pages = ::PdfParadise.has_n_pages?(this_pdf_file)
+        opne 'Now working on the .pdf file '+sfancy(this_pdf_file)+
+             rev+' ('+steelblue(has_n_pages.to_s+' pages')+rev+')'
+        has_n_pages = PdfParadise.n_pages_in_this_pdf_file?(this_pdf_file) # This is also the last page.
+        if use_pdftk?
+          _ = 'pdftk '+this_pdf_file+' cat 1-'+(has_n_pages.to_i - 1).to_s+' '+
+               has_n_pages.to_s+'-end output '+
+               name_of_the_output_file?(this_pdf_file)
+        else # else we use qpdf
+          _ = 'qpdf'.dup
+          if @overwrite_the_old_pdf_file
+            _ << ' --replace-input'
+          end
+          if @overwrite_the_old_pdf_file
+            _ << ' --pages '+this_pdf_file+' 1-'+(has_n_pages.to_i - 1).to_s+' -- '+
+                 this_pdf_file
+          else
+            _ << ' --pages '+this_pdf_file+' 1-'+(has_n_pages.to_i - 1).to_s+' -- '+
+                 this_pdf_file+' '+name_of_the_output_file?(this_pdf_file)
+          end
+        end
+        begin
+          esystem _ # Need to rescue here, e. g. due to permission errors and what not.
+        rescue Exception => error
+          pp error
+        end
+      else
+        opne 'No file exists at `'+sfile(this_pdf_file)+'`.'
+      end
+    }
+  end
+
+  # ========================================================================= #
+  # === name_of_the_output_file?
+  # ========================================================================= #
+  def name_of_the_output_file?(i)
+    if @overwrite_the_old_pdf_file
+      i.sub(/\.pdf$/,'')+'.pdf'
+    else
+      i.sub(/\.pdf$/,'')+'_output_file.pdf'
+    end
+  end; alias output? name_of_the_output_file? # === output?
+
+  # ========================================================================= #
+  # === use_pdftk?
+  # ========================================================================= #
+  def use_pdftk?
+    false
+  end
+
+  # ========================================================================= #
+  # === input?
+  # ========================================================================= #
+  def input?
+    @input_files
+  end
+
+  # ========================================================================= #
+  # === run
+  # ========================================================================= #
+  def run
+    process_each_pdf
+  end
+
+end
+
+# =========================================================================== #
+# === PdfParadise.delete_last_page_of_this_pdf_file
+#
+# Easier method-way to invoke the above class.
+# =========================================================================== #
+def self.delete_last_page_of_this_pdf_file(i = ARGV)
+  PdfParadise::DeleteLastPageOfThisPdfFile.new(i)
+end; self.instance_eval { alias remove_the_last_page_of_this_pdf_file delete_last_page_of_this_pdf_file } # === PdfParadise.remove_the_last_page_of_this_pdf_file
+     self.instance_eval { alias remove_last_page_of_this_pdf_file     delete_last_page_of_this_pdf_file } # === PdfParadise.remove_last_page_of_this_pdf_file
+
+end
+
+if __FILE__ == $PROGRAM_NAME
+  PdfParadise::DeleteLastPageOfThisPdfFile.new(ARGV)
+end # delete_last_page_of_this_pdf_file