bioroebe 0.10.80

data/lib/bioroebe/shell/reset.rb

@@ -0,0 +1,335 @@
+#!/usr/bin/ruby -w
+# Encoding: UTF-8
+# frozen_string_literal: true
+# =========================================================================== #
+# require 'bioroebe/shell/reset.rb'
+# =========================================================================== #
+module Bioroebe
+
+class Shell < ::Bioroebe::CommandlineApplication
+
+  require 'bioroebe/raw_sequence/raw_sequence.rb'
+  require 'bioroebe/nucleotides/show_nucleotide_sequence.rb'
+
+  # ========================================================================= #
+  # === reset                                                     (reset tag)
+  # ========================================================================= #
+  def reset(
+      be_verbose = true
+    )
+    super()
+    # ======================================================================= #
+    # First we try to enable the configuration.
+    # ======================================================================= #
+    enable_configuration
+    reset_to_initial_state(be_verbose)
+    set_default_length # Set the default length of 1000 here.
+    # ======================================================================= #
+    # Make sure that the base directories exist.
+    # ======================================================================= #
+    ensure_that_the_base_directories_exist
+    if defined? DEFAULT_DNA_INPUT_YAML_FILE # Main DNA string.
+      set_string(DEFAULT_DNA_INPUT_YAML_FILE)
+    end
+  end
+
+  require 'bioroebe/sequence/sequence.rb'
+  # ========================================================================= #
+  # === initialize_main_sequence
+  #
+  # Initialize a Sequence-object on startup. We use the length 150
+  # for now.
+  # ========================================================================= #  
+  def initialize_main_sequence
+    @internal_hash[:array_dna_sequence] << Bioroebe::Sequence.new(250)
+  end; alias reset_string initialize_main_sequence # === reset_string
+
+  # ========================================================================= #
+  # === reset_to_initial_state
+  # ========================================================================= #
+  def reset_to_initial_state(
+      be_verbose = true
+    )
+    # ======================================================================= #
+    # === @internal_hash
+    #
+    # The internal Hash can be used to store internal behaviour variables
+    # rather than spawn so many different instance variables.
+    #
+    # It should be defined very early in this method.
+    # ======================================================================= #
+    @internal_hash = {}
+    # ======================================================================= #
+    # === :first_argument
+    #
+    # This variable refers to the first argument passed into a method,
+    # from the menu.rb file.
+    # ======================================================================= #
+    @internal_hash[:first_argument] = nil
+    # ======================================================================= #
+    # === :show_nucleotide_sequence
+    # ======================================================================= #
+    @internal_hash[:show_nucleotide_sequence] = ::Bioroebe::ShowNucleotideSequence.new(
+      '', :do_not_run_yet
+    )
+    # ======================================================================= #
+    # === :array_aminoacid_sequence
+    #
+    # This variable should be an empty Array, aka [], initially, so that we
+    # can distinguish the case when the user has added data onto that
+    # Array.
+    # ======================================================================= #
+    @internal_hash[:array_aminoacid_sequence] = []
+    # ======================================================================= #
+    # === :array_rna_sequence
+    #
+    # This can be populated with sequence objects such as:
+    #
+    #   Bioroebe::Sequence.new(''.dup, :rna)
+    #
+    # ======================================================================= #
+    @internal_hash[:array_rna_sequence] = []
+    # ======================================================================= #
+    # === :array_dna_sequence
+    # ======================================================================= #
+    @internal_hash[:array_dna_sequence] = []
+    # ======================================================================= #
+    # === @use_working_directory_as_prompt
+    # ======================================================================= #
+    @internal_hash[:use_working_directory_as_prompt] = false
+    # ======================================================================= #
+    # === @debug
+    #
+    # Determine whether we will debug or whether we will not
+    # ======================================================================= #
+    @internal_hash[:debug] = SHALL_WE_DEBUG
+    # ======================================================================= #
+    # === @array_sequences
+    # ======================================================================= #
+    @internal_hash[:array_sequences] = []
+    # ======================================================================= #
+    # === @all_arguments
+    # ======================================================================= #
+    @internal_hash[:all_arguments] = []
+    # ======================================================================= #
+    # === :remaining_arguments
+    # ======================================================================= #
+    @internal_hash[:remaining_arguments] = []
+    # ======================================================================= #
+    # === :registered_actions
+    # ======================================================================= #
+    @internal_hash[:registered_actions] = ARRAY_REGISTERED_ACTIONS
+    # ======================================================================= #
+    # === :the_main_sequence_is_frozen
+    # ======================================================================= #
+    @internal_hash[:the_main_sequence_is_frozen] = false
+    # ======================================================================= #
+    # === :file_dna_string_saved
+    #
+    # Where to store our DNA string by default.
+    # ======================================================================= #
+    @internal_hash[:file_dna_string_saved] = "#{log_dir?}dna_string_saved.yml"
+    # ======================================================================= #
+    # === :array_jumper_directories
+    # ======================================================================= #
+    @internal_hash[:array_jumper_directories] = []
+    # ======================================================================= #
+    # === :array_fasta
+    # ======================================================================= #
+    @internal_hash[:array_fasta] = []
+    # ======================================================================= #
+    # === :raw_sequence
+    # ======================================================================= #
+    @internal_hash[:raw_sequence] = Bioroebe::RawSequence.new
+    # ======================================================================= #
+    # === :use_xsel
+    #
+    # This method will determine whether we will use the external
+    # program "xsel". By default, at the least on my home system,
+    # I prefer to make use of xsel.
+    # ======================================================================= #
+    @internal_hash[:use_xsel] = true
+    # ======================================================================= #
+    # === :silent_startup
+    #
+    # This variable keeps track as to whether we will display a welcome
+    # message on startup or whether we will not.
+    # ======================================================================= #
+    @internal_hash[:silent_startup] = false
+    # ======================================================================= #
+    # === exit_the_shell_how
+    #
+    # The next variable has two valid modes:
+    #
+    #   :instantly
+    #   :exit_gracefully
+    #
+    # The first is the default; the second can be used if the Bioshell
+    # is embedded into another program.
+    # ======================================================================= #
+    @internal_hash[:exit_the_shell_how] = :instantly
+    # ======================================================================= #
+    # === search_for
+    #
+    # This is the sequence we wish to find, in a given nucleotide sequence.
+    # By default this will be nil.
+    # ======================================================================= #
+    @internal_hash[:search_for] = nil
+    # ======================================================================= #
+    # === analyse_the_local_dataset_on_startup
+    #
+    # If this setting is set to true then, on startup, the bioshell
+    # will report some information about the local dataset (FASTA
+    # files and pdb files).
+    # ======================================================================= #
+    @internal_hash[:analyse_the_local_dataset_on_startup] = true
+    # ======================================================================= #
+    # === user_input
+    # ======================================================================= #
+    @internal_hash[:user_input] = nil
+    # ======================================================================= #
+    # === array_history
+    #
+    # This array will keep track of the input-history, aka the commands
+    # that the user has used.
+    # ======================================================================= #
+    @internal_hash[:array_history] = []
+    # ======================================================================= #
+    # === remove_last_character_if_it_is_a_question_mark
+    #
+    # If this variable is set to true then we will remove the last
+    # input character - at the least if it is a '?'.
+    # ======================================================================= #
+    @internal_hash[:remove_last_character_if_it_is_a_question_mark] = false
+    # ======================================================================= #
+    # === prompt_to_use
+    #
+    # This variable keeps track as to which prompt is to be used.
+    # ======================================================================= #
+    @internal_hash[:prompt_to_use] = :default
+    # ======================================================================= #
+    # === use_xsel
+    # ======================================================================= #
+    @internal_hash[:use_xsel] = false
+    # ======================================================================= #
+    # === search_for
+    # ======================================================================= #
+    @internal_hash[:search_for] = nil
+    # ======================================================================= #
+    # === coding_area
+    # ======================================================================= #
+    @internal_hash[:coding_area] = nil # This can keep track of the real coding area.
+    # ======================================================================= #
+    # === misc_sequence
+    # ======================================================================= #
+    @internal_hash[:misc_sequence] = nil # This can store a misc DNA or RNA sequence.
+    # ======================================================================= #
+    # === fasta_file
+    #
+    # This used to be a standalone @hash variable, but during the
+    # rewrite in April 2020 this was integrated into @internal_hash.
+    # ======================================================================= #
+    @internal_hash[:fasta_file] = {}
+    # ======================================================================= #
+    # === @array_all_downloads
+    # ======================================================================= #
+    @internal_hash[:array_all_downloads] = [] # Must be an Array.
+    # ======================================================================= #
+    # === @use_working_directory_as_prompt
+    # ======================================================================= #
+    @internal_hash[:use_working_directory_as_prompt] = true
+    # ======================================================================= #
+    # === @mode
+    #
+    # @mode can be either :dna or :rna or :aminoacid
+    # ======================================================================= #
+    @internal_hash[:mode] = :dna
+    # ======================================================================= #
+    # === @array_these_sequences_were_chopped_away
+    #
+    # Keep an Array that can be used to keep track of which sequences
+    # were chopped away. Whenever we perform a chop-related action we
+    # will store that sequence in an Array.
+    # ======================================================================= #
+    @internal_hash[:array_these_sequences_were_chopped_away] = []
+    # ======================================================================= #
+    # === @save_file
+    # ======================================================================= #
+    @internal_hash[:save_file] = BIOSHELL_SAVE_FILE
+    # ======================================================================= #
+    # === create_directories_on_startup_of_the_shell
+    #
+    # If this variable is set to true then the bio-shell will create
+    # some directories on startup. The user can modify this from
+    # the commandline, though, and specifically disable it.
+    # ======================================================================= #
+    @internal_hash[:create_directories_on_startup_of_the_shell] = true
+    # ======================================================================= #
+    # === show_the_leader
+    # ======================================================================= #
+    @internal_hash[:show_the_leader]  = true # Show the 5' leader.
+    # ======================================================================= #
+    # === show_the_trailer
+    # ======================================================================= #
+    @internal_hash[:show_the_trailer] = true # Show the 3' trailer.
+    # ======================================================================= #
+    # === array_timer_snapshots
+    # ======================================================================= #
+    @internal_hash[:array_timer_snapshots] = [] # This Array keeps track of the time.
+    # ======================================================================= #
+    # === may_we_show_the_startup_information
+    #
+    # This variable determines whether the startup-information may be
+    # shown, on start-up of the bioshell. By default this will be true
+    # typically, but a config-variable may overrule this.
+    # ======================================================================= #
+    @internal_hash[:may_we_show_the_startup_information] = true
+    if @config and @config.respond_to? :may_we_show_the_startup_information
+      @internal_hash[:may_we_show_the_startup_information] = @config.may_we_show_the_startup_information
+    end
+    # ======================================================================= #
+    # === @use_colours
+    #
+    # Let's enable the colours.
+    # ======================================================================= #
+    enable_colours(:be_quiet)
+    initialize_clipboard
+    # ======================================================================= #
+    # === @sequence
+    #
+    # We set @sequence to a new instance of Bioroebe::Sequence next.
+    # ======================================================================= #
+    initialize_main_sequence
+# ======================================================================= #
+    # === Setting towards a default padding
+    #
+    # Next comes left-padding, defaulting to '  '. This should come after
+    # the clipboard has been initialized, and after @internal_hash has
+    # been populated.
+    # ======================================================================= #
+    set_padding :default
+    set_sequence_2
+    set_sequence_3
+    set_sequence_4
+    set_sequence_5
+    set_sequence_6
+    ::Bioroebe.clear_array_colourize_this_aminoacid
+    # ======================================================================= #
+    # === Determine whether silent startup is active
+    #
+    # Next, determine whether we will show a small startup-message or
+    # whether we shall not show it. By default this should be false -
+    # but if a file called 'use_silent_startup.yml' exists then we
+    # will use that file.
+    # ======================================================================= #
+    if File.exist? FILE_USE_SILENT_STARTUP
+      dataset = YAML.load_file(FILE_USE_SILENT_STARTUP)
+      if dataset.is_a?(Hash) and dataset.has_key?('use_silent_startup')
+        @internal_hash[:silent_startup] = dataset['use_silent_startup']
+      end
+    else # else it is 
+      @internal_hash[:silent_startup] = false
+    end
+  end
+
+end; end

data/lib/bioroebe/shell/scan_and_parse.rb

@@ -0,0 +1,135 @@
+#!/usr/bin/ruby -w
+# Encoding: UTF-8
+# frozen_string_literal: true
+# =========================================================================== #
+# require 'bioroebe/shell/scan_and_parse.rb'
+# =========================================================================== #
+module Bioroebe
+
+class Shell < ::Bioroebe::CommandlineApplication
+
+  # ========================================================================= #
+  # === parse                                                     (parse tag)
+  # ========================================================================= #
+  def parse(i)
+    if i.is_a? Array
+      i = i.join(' ').strip
+    end
+    i = i.dup if i.frozen?
+    i.chop! if i.end_with?('.')
+    # ======================================================================= #
+    # We will parse here based on the file-suffix.
+    # ======================================================================= #
+    case i
+    # ======================================================================= #
+    # === fasta file
+    # ======================================================================= #
+    when /\.fasta$/i,
+         /\.fa$/i
+      parse_fasta_format(i)
+      unless array_fasta?.empty?
+        # =================================================================== #
+        # Assign this dataset next to become the new main sequence.
+        # =================================================================== #
+        opnn; erev "We will now assign this data to #{sfancy('@_')}#{rev}."
+        set_mode :protein if array_fasta?.last.is_protein?
+        assign(array_fasta?.last.sequence)
+      end
+    # ======================================================================= #
+    # === pdb file
+    # ======================================================================= #
+    when /\.pdb$/i
+      parse_this_pdb_file(i)
+    else
+      erev "Not yet handled: #{steelblue(i)}"
+    end
+  end
+
+  # ========================================================================= #
+  # === scan_for_leucine_zippers
+  #
+  # What is a leucine zipper?
+  #
+  #   https://en.wikipedia.org/wiki/Leucine_zipper
+  #
+  # ========================================================================= #
+  def scan_for_leucine_zippers(
+      i = amino_acid_sequence?
+    )
+    if i.nil?
+      i = amino_acid_sequence?
+    end
+    chars = i.chars
+    # ======================================================================= #
+    # We start to count at the first leucine zipper.
+    # ======================================================================= #
+    index = chars.index('L')
+    erev 'The string '+simp(i)+rev+' has these at every 7th step.'
+    print '  ' # Leading indent.
+    (index..chars.size).step(7) {|token|
+      print simp(chars[token]),' '
+    }; e # And a final newline.
+  end
+
+  # ========================================================================= #
+  # === parse_this_gff_file
+  #
+  # Use this method if you wish to parse a .gff or .gff3 file.
+  # ========================================================================= #
+  def parse_this_gff_file(i)
+    if i.is_a? Array
+      i.each {|entry| parse_this_gff_file(entry) }
+    else
+      ::Bioroebe::Parser::GFF.new(i)
+    end
+  end
+
+  # ========================================================================= #
+  # === scan_for_gff_files
+  # ========================================================================= #
+  def scan_for_gff_files
+    results_for_gff_files  = Dir['*.gff']
+    results_for_gff3_files = Dir['*.gff3']
+    unless results_for_gff_files.empty?
+      erev 'There are '+sfancy(results_for_gff_files.size.to_s)+rev+
+           ' .gff files in the current directory.'
+    end
+    unless results_for_gff3_files.empty?
+      erev 'There are '+sfancy(results_for_gff3_files.size.to_s)+
+           ' .gff3 files in the current directory.'
+      erev 'We will simply pass the first entry there into '\
+           'class Bioroebe::Parser::GFF.'
+      parse_this_gff_file(results_for_gff3_files.first)
+    end
+  end
+
+  # ========================================================================= #
+  # === scan_or_parse_for_this_gff_file_or_any_gff_file
+  # ========================================================================= #
+  def scan_or_parse_for_this_gff_file_or_any_gff_file(i)
+    if i.is_a? Array
+      i.each {|entry|
+        if entry.nil?
+          scan_for_gff_files
+        else
+          scan_or_parse_for_this_gff_file_or_any_gff_file(entry)
+        end
+      }
+    else
+      if File.exist? i.to_s
+        parse_this_gff_file(i)
+      else # else scan for any .gff or .gff3 file.
+        scan_for_gff_files
+      end
+    end
+  end
+
+  require 'bioroebe/pdb/parse_pdb_file.rb'
+  # ========================================================================= #
+  # === parse_this_pdb_file
+  # ========================================================================= #
+  def parse_this_pdb_file(i)
+    ParsePdbFile.new(i)
+  end
+
+end; end