fixnames 0.3.0

data/fixnames.gemspec

@@ -0,0 +1,92 @@
+# Generated by jeweler
+# DO NOT EDIT THIS FILE DIRECTLY
+# Instead, edit Jeweler::Tasks in Rakefile, and run 'rake gemspec'
+# -*- encoding: utf-8 -*-
+
+Gem::Specification.new do |s|
+  s.name = %q{fixnames}
+  s.version = "0.3.0"
+
+  s.required_rubygems_version = Gem::Requirement.new(">= 0") if s.respond_to? :required_rubygems_version=
+  s.authors = [%q{Brent Sanders}]
+  s.date = %q{2011-10-04}
+  s.description = %q{Cleans up filenames so they can easily be used
+in scripts, without annoyances such as spaces or other bad characters}
+  s.email = %q{git@thoughtnoise.net}
+  s.executables = [%q{fixnames}, %q{fixdirs}]
+  s.extra_rdoc_files = [
+    "LICENSE.txt",
+    "README.md",
+    "README.rdoc"
+  ]
+  s.files = [
+    ".document",
+    ".rspec",
+    "COPYING",
+    "LICENSE.txt",
+    "README.md",
+    "README.rdoc",
+    "Rakefile",
+    "VERSION",
+    "bin/fixdirs",
+    "bin/fixnames",
+    "fixnames.gemspec",
+    "lib/fixnames.rb",
+    "lib/fixnames/debug.rb",
+    "lib/fixnames/engine.rb",
+    "lib/fixnames/engine/scan_dir.rb",
+    "lib/fixnames/filters.rb",
+    "lib/fixnames/helpers.rb",
+    "lib/fixnames/interface.rb",
+    "lib/fixnames/option.rb",
+    "lib/fixnames/version.rb",
+    "spec/fixnames/banners_spec.rb",
+    "spec/fixnames/brackets_spec.rb",
+    "spec/fixnames/camelcase_spec.rb",
+    "spec/fixnames/charstrip_spec.rb",
+    "spec/fixnames/checksums_spec.rb",
+    "spec/fixnames/fixdots_spec.rb",
+    "spec/fixnames/hack_and_spec.rb",
+    "spec/fixnames/lowercase_spec.rb",
+    "spec/fixnames/semicolon_spec.rb",
+    "spec/fixnames/whitespace_spec.rb",
+    "spec/spec_helper.rb",
+    "spec/support/should_fix_helpers.rb",
+    "test/helper.rb",
+    "test/test_charstrip.rb",
+    "test/test_hack_and.rb",
+    "test/test_lowercase.rb",
+    "test/test_semicolon.rb",
+    "test/test_whitespace.rb"
+  ]
+  s.homepage = %q{http://github.com/pdkl95/fixnames}
+  s.licenses = [%q{MIT}]
+  s.require_paths = [%q{lib}]
+  s.rubygems_version = %q{1.8.5}
+  s.summary = %q{Filename cleanup for script compatability}
+
+  if s.respond_to? :specification_version then
+    s.specification_version = 3
+
+    if Gem::Version.new(Gem::VERSION) >= Gem::Version.new('1.2.0') then
+      s.add_runtime_dependency(%q<term-ansicolor>, [">= 1.0.6"])
+      s.add_development_dependency(%q<yard>, [">= 0.6.0"])
+      s.add_development_dependency(%q<rspec>, [">= 2.3.0"])
+      s.add_development_dependency(%q<jeweler>, [">= 1.6.4"])
+      s.add_development_dependency(%q<simplecov>, [">= 0"])
+    else
+      s.add_dependency(%q<term-ansicolor>, [">= 1.0.6"])
+      s.add_dependency(%q<yard>, [">= 0.6.0"])
+      s.add_dependency(%q<rspec>, [">= 2.3.0"])
+      s.add_dependency(%q<jeweler>, [">= 1.6.4"])
+      s.add_dependency(%q<simplecov>, [">= 0"])
+    end
+  else
+    s.add_dependency(%q<term-ansicolor>, [">= 1.0.6"])
+    s.add_dependency(%q<yard>, [">= 0.6.0"])
+    s.add_dependency(%q<rspec>, [">= 2.3.0"])
+    s.add_dependency(%q<jeweler>, [">= 1.6.4"])
+    s.add_dependency(%q<simplecov>, [">= 0"])
+  end
+end
+

data/lib/fixnames.rb

@@ -0,0 +1,12 @@
+module Fixnames
+end
+
+require 'fixnames/option'
+require 'fixnames/debug'
+require 'fixnames/interface'
+
+module Fixnames
+  include Debug
+end
+
+

data/lib/fixnames/debug.rb

@@ -0,0 +1,43 @@
+require 'term/ansicolor'
+
+module Fixnames
+  module Debug
+    class Color
+      extend Term::ANSIColor
+
+      def self.prefix(chr, cname)
+        #raise "#{cname.inspect}, #{send(cname).inspect}"
+        [Color.send(cname), Color.bold, "#{chr}#{chr}>", Color.clear].join
+      end
+
+      def self.puts_msg(str, chr, cname)
+        puts "#{prefix(chr, cname)} #{str}"
+      end
+    end
+
+    def bold(str)
+      [ Color.bold,
+        Color.yellow,
+        Color.on_blue,
+        str,
+        Color.clear
+      ].join
+    end
+
+    def warn(msg)
+      Color.puts_msg(msg, '*', :red)    if @option.verbose > 0
+    end
+
+    def note(msg)
+      Color.puts_msg(msg, '!', :yellow) if @option.verbose > 0
+    end
+
+    def info(msg)
+      Color.puts_msg(msg, '-', :green)  if @option.verbose > 1
+    end
+
+    def debug(msg)
+      Color.puts_msg(msg, '>', :cyan)   if @option.verbose > 2
+    end
+  end
+end

data/lib/fixnames/engine.rb

@@ -0,0 +1,86 @@
+require 'fixnames/debug'
+require 'fixnames/helpers'
+require 'fixnames/filters'
+require 'fixnames/engine/scan_dir'
+
+module Fixnames
+  # the main filtering-engine that fixes
+  # a single filename
+  class Engine
+    include Debug
+    include Helpers
+    include Filters
+
+    attr_reader :orig, :fixed, :dir, :option
+    alias_method :to_s, :fixed
+
+    # Creates an engine to fix a single filename.
+    #
+    # @param name [String] The filename to be fixed
+    # @param options [{Symbol => Object}] An options hash
+    def initialize(name, opts=Option.new)
+      @option = opts
+
+      @dir  = File.dirname(name)
+      @orig = File.basename(name)
+
+      if option.recursive && File.directory?(@orig)
+        @scandir = ScanDir.new(@orig, option)
+      end
+
+      @fixed = @orig.dup
+
+      option.filter_order.each do |optname|
+        if option.send(optname) and respond_to?(optname)
+          debug "FILTER[:#{optname}]"
+          old = fixed.dup
+          case method(optname).arity
+          when 1 then send optname, option.send(optname)
+          when 0 then send optname
+          else raise "Unsupported arity in ##{optname}"
+          end
+          if old != fixed
+            debug "\t    old -- #{old.inspect}"
+            debug "\t    new -- #{fixed.inspect}"
+          end
+        end
+      end
+    end
+
+    def orig_path
+      "#{dir}/#{orig}"
+    end
+
+    def fixed_path
+      "#{dir}/#{fixed}"
+    end
+
+    def scandir_changed?
+      @scandir ? @scandir.changed? : false
+    end
+
+    def changed?
+      fixed != orig
+    end
+
+    def collision?
+      File.exists? fixed_path
+    end
+
+    def fix!
+      @scandir.fix! if @scandir
+
+      if changed?
+        if collision?
+          warn "NAME COLLISION: #{fixed_path.inspect}"
+        else
+          note "mv #{orig_path.inspect} #{fixed_path.inspect}"
+          File.rename orig_path, fixed_path unless option.pretend
+        end
+      else
+        info "no change: #{orig_path.inspect}"
+      end
+      self
+    end
+  end
+end

data/lib/fixnames/engine/scan_dir.rb

@@ -0,0 +1,34 @@
+module Fixnames
+  class Engine
+    class ScanDir
+      attr_reader :option, :name, :base, :prefix
+      def initialize(dirname, opts)
+        raise "Not a directory: #{dirname}" unless File.directory?(dirname)
+        @name = File.realpath(dirname)
+        raise "Not a directory: #{name}" unless File.directory?(name)
+
+        @option = opts
+      end
+
+      def glob_str
+        "#{name}/#{option.dir_glob}"
+      end
+
+      def files
+        @files ||= Dir.glob(glob_str)
+      end
+
+      def engines
+        @engies ||= files.map do |name|
+          Engine.new(name, option)
+        end
+      end
+
+      def fix!
+        engines.map do |en|
+          en.fix!
+        end
+      end
+    end
+  end
+end

data/lib/fixnames/filters.rb

@@ -0,0 +1,79 @@
+module Fixnames
+  module Filters
+    def expunge(re)
+      replace re, option.mendstr
+    end
+
+    def hack_and
+      replace '&', '_and_'
+    end
+
+    def semicolon
+      translate ';', '-'
+      fixed.squeeze! '-'
+    end
+
+    def banners
+      option.banner_types.each do |x|
+        remove_bracket_ranges(x)
+      end
+    end
+
+    def brackets
+      remove wrap_brackets('.+?')
+    end
+
+    def checksums
+      remove wrap_brackets('[0-9a-fA-F]{8}')
+    end
+
+    def lowercase
+      translate 'A-Z', 'a-z'
+    end
+
+    def fix_dots
+      last = fixed.rindex('.')
+      translate '.', '_'
+      replace '(.*)\.(.*\.)', '\1_\2'
+      fixed[last] = '.' if last
+    end
+
+    def fix_dashes
+      fixed.squeeze! '-'
+      remove  '^-' while fixed =~ /^-/
+      remove  '-$' while fixed =~ /-$/
+    end
+
+    def camelcase
+      replace '([a-z])([A-Z])', '\1_\2'
+      fixed.downcase!
+    end
+
+    def junkwords(wordlist)
+      wordlist.each do |word|
+        replace "[_-]#{word}[_-]", '_'
+        remove  "[_-]#{word}$"
+        remove     "^#{word}[_-]"
+      end
+    end
+
+    def whitespace(chrlist)
+      replace "[#{Regexp.escape chrlist}]", '_'
+      replace '[_-]\.', '.' while fixed =~ /[_-]\./
+      replace '_-',     '-' while fixed =~ /_-/
+      replace '-_',     '-' while fixed =~ /-_/
+      remove  '^_'          while fixed =~ /^_/
+      remove  '_$'          while fixed =~ /_$/
+      fixed.squeeze! '_'
+    end
+
+    def charstrip(chrlist)
+      re = Regexp.escape( if option.charstrip_allow_brackets
+                            remove_bracket_characters_from(chrlist)
+                          else
+                            chrlist
+                          end )
+      remove "[#{re}]"
+    end
+  end
+end

data/lib/fixnames/helpers.rb

@@ -0,0 +1,41 @@
+module Fixnames
+  module Helpers
+    def replace(re, replacement)
+      re_str = bold "/#{re}/"
+      replacement_str = bold "\"#{replacement}\""
+      debug "\t<replace>  #{re_str}  ->  #{replacement_str}"
+      fixed.gsub! Regexp.new(re), replacement
+    end
+
+    def remove(re)
+      re_str = bold "/#{re}/"
+      debug "\t<expunge>  #{re_str}"
+      fixed.gsub! Regexp.new(re), ''
+    end
+
+    def translate(src, dst)
+      debug "\t<translate>  #{bold src.inspect}  ->  #{bold dst.inspect}"
+      fixed.tr! src, dst
+    end
+
+    def match_bracket_open
+      "[#{Regexp.escape(option.bracket_characters_open)}]"
+    end
+
+    def match_bracket_close
+      "[#{Regexp.escape(option.bracket_characters_close)}]"
+    end
+
+    def wrap_brackets(re)
+      "#{match_bracket_open}#{re}#{match_bracket_close}"
+    end
+
+    def remove_bracket_ranges(re)
+      remove wrap_brackets(".*?#{re}.*?")
+    end
+
+    def remove_bracket_characters_from(str)
+      str.gsub /(#{match_bracket_open}|#{match_bracket_close})/, ''
+    end
+  end
+end

data/lib/fixnames/interface.rb

@@ -0,0 +1,41 @@
+require 'fixnames/engine'
+
+module Fixnames
+  module FixFile
+    def self.parse(name, opts)
+      Engine.new(name, opts)
+    end
+
+    # Returns the fixed version of a filename, without
+    # actually changing anything on the filesystem.
+    def self.fix_name(*args)
+      parse(*args).fixed
+    end
+
+    def self.fix!(*args)
+      parse(*args).fix!
+    end
+
+    def self.fix_list!(list, *args)
+      list.map do |x|
+        fix! x, *args
+      end
+    end
+  end
+
+  module FixDir
+    def self.parse(name, optsw)
+      Engine::ScanDir.new(name, opts)
+    end
+
+    def self.fix!(*args)
+      parse(*args).fix!
+    end
+
+    def self.fix_list!(list, *args)
+      list.map do |x|
+        fix! x, *args
+      end
+    end
+  end
+end

data/lib/fixnames/option.rb

@@ -0,0 +1,163 @@
+module Fixnames
+  class Option
+    # filters that MUST run early
+    SETUP_FILTERS = [ :expunge ]
+
+    # filters that only accept a simple boolean on/off
+    FLAG_FILTERS = [ :hack_and, :checksums, :banners,
+                     :brackets, :semicolon,
+                     :camelcase, :lowercase,
+                     :fix_dots, :fix_dashes ]
+
+    # filters that accept character ranges
+    CHAR_FILTERS = [ :junkwords, :charstrip, :whitespace]
+
+    # standard order to apply the filters
+    DEFAULT_FILTER_ORDER = [ SETUP_FILTERS,
+                             FLAG_FILTERS,
+                             CHAR_FILTERS
+                           ].flatten
+
+    DEFAULT_DIR_GLOB   = '*'
+    DEFAULT_MENDSTR    = ''
+    DEFAULT_WHITESPACE = " \t_"
+    DEFAULT_BRACKET_CHARACTERS_OPEN  = '[({<'
+    DEFAULT_BRACKET_CHARACTERS_CLOSE = '])}>'
+    DEFAULT_CHARSTRIP = "[]{}'\",()+!~@#/\\"
+    DEFAULT_JUNKWORDS = [ 'x264', 'hdtv', '2hd', '720p', 'dvdrip']
+    DEFAULT_BANNER_TYPES = [ 'xxx', 'dvdrip', 'dual_audio',
+                             'xvid', 'h264', 'divx' ]
+
+    # Creates an option
+    #
+    # @param [String] name the name of the option to create
+    # @param [Array] types a list of classes that are valid
+    # @param [Object] default_val the value to set initially
+    def self.mkopt(name, types, default_val)
+      types = [types] unless types.is_a?(Array)
+      var = "@#{name}"
+
+      define_method "valid_for_#{name}?" do |value|
+        types.map do |type|
+          value.is_a?(type)
+        end.reduce(false) do |t,x|
+          t || x
+        end
+      end
+
+      define_method name do |*args|
+        unless instance_variable_defined?(var)
+          instance_variable_set(var, default_val)
+        end
+        if args.length == 1
+          unless send("valid_for_#{name}?", args[0])
+            raise "bad type for option"
+          end
+          instance_variable_set(var, args[0])
+        end
+        instance_variable_get(var)
+      end
+
+      define_method "#{name}=" do |value|
+        unless send("valid_for_#{name}?", value)
+          raise "bad type for option"
+        end
+        instance_variable_set(var, value)
+      end
+    end
+
+    # set to turn of the ANSI-color output
+    # @macro [attach] mkopt
+    #   default: `$3`
+    #
+    #   @overload $1()
+    #   @overload $1(new_value)
+    #   @overload $1=(new_value)
+    #   @param [$2] new_value
+    #   @return [$2]
+    mkopt :nocolor, [TrueClass, FalseClass], false
+
+    # Verbosity levels
+    #
+    # * `verbose=0` ; no output
+    # * `verbose=1` ; only names that change are output
+    # * `verbose=2` ; all names are output with their change-status
+    # * `verbose=3` ; all *filters* are output as they run for debugging. Very noisy.
+    mkopt :verbose, Integer, 0
+
+    # When {#recursive} is set, use this pattern to glob each
+    # directory for files.
+    mkopt :dir_glob, String, DEFAULT_DIR_GLOB
+
+    # Recursively descend into directories if true.
+    mkopt :recursive, [TrueClass, FalseClass], false
+
+    # A generic pattern to remove from all filenames.
+    # @note Enables {Fixnames::Filters#expunge}
+    mkopt :expunge, String, nil
+
+    # After we {#expunge} a pattern, it is replaced with this string.
+    mkopt :mendstr, String, DEFAULT_MENDSTR
+
+    # @note Enables {Fixnames::Filters#hack_and}
+    mkopt :hack_and,  [TrueClass, FalseClass], false
+
+    # @note Enables {Fixnames::Filters#checksums}
+    mkopt :checksums, [TrueClass, FalseClass], false
+
+    # @note Enables {Fixnames::Filters#banners}
+    mkopt :banners,   [TrueClass, FalseClass], false
+
+    # @note Enables {Fixnames::Filters#brackets}
+    mkopt :brackets, [TrueClass, FalseClass], false
+
+    # @note Enables {Fixnames::Filters#semicolon}
+    mkopt :semicolon, [TrueClass, FalseClass], false
+
+    # @note Enables {Fixnames::Filters#fix_dots}
+    mkopt :fix_dots, [TrueClass, FalseClass], false
+
+    # @note Enables {Fixnames::Filters#fix_dashes}
+    mkopt :fix_dashes, [TrueClass, FalseClass], false
+
+    # @note Enables {Fixnames::Filters#camelcase}
+    mkopt :camelcase, [TrueClass, FalseClass], false
+
+    # @note Enables {Fixnames::Filters#lowercase}
+    mkopt :lowercase, [TrueClass, FalseClass], false
+
+    # @note Enables {Fixnames::Filters#junkwords} if non-nil
+    mkopt :junkwords, Array, DEFAULT_JUNKWORDS
+
+    # @note Enables {Fixnames::Filters#charstrip} if non-nil
+    mkopt :charstrip, String, DEFAULT_CHARSTRIP
+
+    # @note Enables {Fixnames::Filters#whitespace} if non-nil
+    mkopt :whitespace, String, DEFAULT_WHITESPACE
+
+    # The list of strings to find for removal in {Fixnames::Filters#banners}
+    mkopt :banner_types, Array, DEFAULT_BANNER_TYPES
+
+    # Set to true to have {Fixnames::Filters#charstrip} ignore its
+    # default behavior and allow the brackets through. This is
+    # potentially ignored if you change {#charstrip}.
+    mkopt :charstrip_allow_brackets, [TrueClass, FalseClass], false
+
+    # What is considered an *open* bracket in things
+    # like {#checksums} or {#brackets}
+    mkopt :bracket_characters_open,  String, DEFAULT_BRACKET_CHARACTERS_OPEN
+
+    # What is considered a *close* bracket in things
+    # like {#checksums} or {#brackets}
+    mkopt :bracket_characters_close, String, DEFAULT_BRACKET_CHARACTERS_CLOSE
+
+    # The order we should apply the filter to the filename.
+    # This order is significant, and can dramatically affect
+    # the output.
+    mkopt :filter_order, Array, DEFAULT_FILTER_ORDER
+
+    # if set, we just pretend to work, and skip the final
+    # move command, so the filesystem is never altered
+    mkopt :pretend, [TrueClass, FalseClass], false
+  end
+end