Refactor config, add time tracking facility, optimize cleaner

In particular, I was curious what caused a performance problem on
a large document I'm working on. Turned out, it was a
remove_inner_whitespace procedure in Cleaner. With a simple fix I
managed to make it finish in 1 second, instead of 170s.

All the rest of the processing combined takes 10s, so we will be
able to progress much faster on next issues.

exe/reverse_adoc

@@ -28,6 +28,10 @@ OptionParser.new do |opts|
     Coradoc::ReverseAdoc.config.unknown_tags = v
   end
 
+  opts.on("--track-time", "Track time spent on each step") do
+    Coradoc::ReverseAdoc.config.track_time = true
+  end
+
   opts.on("-v", "--version", "Version information") do |_v|
     puts "reverse_adoc: v#{Coradoc::ReverseAdoc::VERSION}"
     exit

lib/coradoc/reverse_adoc/cleaner.rb

@@ -1,11 +1,21 @@
 module Coradoc::ReverseAdoc
   class Cleaner
     def tidy(string)
-      result = remove_inner_whitespaces(String.new(string))
-      result = remove_newlines(result)
-      result = remove_leading_newlines(result)
-      result = clean_tag_borders(result)
-      clean_punctuation_characters(result)
+      result = HtmlConverter.track_time "Removing inner whitespace" do
+        remove_inner_whitespaces(String.new(string))
+      end
+      result = HtmlConverter.track_time "Removing newlines" do
+        remove_newlines(result)
+      end
+      result = HtmlConverter.track_time "Removing leading newlines" do
+        remove_leading_newlines(result)
+      end
+      result = HtmlConverter.track_time "Cleaning tag borders" do
+        clean_tag_borders(result)
+      end
+      result = HtmlConverter.track_time "Cleaning punctuation characters" do
+        clean_punctuation_characters(result)
+      end
     end
 
     def remove_newlines(string)
@@ -22,11 +32,13 @@ def remove_inner_whitespaces(string)
         string.gsub!(/(stem:\[([^\]]|\\\])*\])\n(?=\S)/, "\\1 ")
         string.gsub!(/(stem:\[([^\]]|\\\])*\])\s+(?=[\^-])/, "\\1")
       end
-      string.each_line.inject("") do |memo, line|
-        memo + preserve_border_whitespaces(line) do
+      result = +""
+      string.each_line do |line|
+        result << preserve_border_whitespaces(line) do
           line.strip.gsub(/[ \t]{2,}/, " ")
         end
       end
+      result
     end
 
     # Find non-asterisk content that is enclosed by two or

lib/coradoc/reverse_adoc/config.rb

@@ -2,9 +2,6 @@
 
 module Coradoc::ReverseAdoc
   class Config
-    attr_accessor :unknown_tags, :tag_border, :mathml2asciimath, :external_images,
-                  :destination, :sourcedir, :image_counter, :image_counter_pattern, :input_format
-
     def initialize
       @unknown_tags     = :pass_through
       @input_format     = :html
@@ -26,29 +23,36 @@ def initialize
       @strong_delimiter = "*".freeze
       @inline_options   = {}
       @tag_border       = " ".freeze
+
+      # Debugging options
+      @track_time       = false
     end
 
     def with(options = {})
+      old_options = @inline_options
       @inline_options = options
       result = yield
-      @inline_options = {}
+      @inline_options = old_options
       result
     end
 
-    def unknown_tags
-      @inline_options[:unknown_tags] || @unknown_tags
-    end
-
-    def mathml2asciimath
-      @inline_options[:mathml2asciimath] || @mathml2asciimath
-    end
+    def self.declare_option(option)
+      define_method(option) do
+        @inline_options[option] || instance_variable_get(:"@#{option}")
+      end
 
-    def external_images
-      @inline_options[:external_images] || @external_images
+      attr_writer option
     end
 
-    def tag_border
-      @inline_options[:tag_border] || @tag_border
-    end
+    declare_option :unknown_tags
+    declare_option :tag_border
+    declare_option :mathml2asciimath
+    declare_option :external_images
+    declare_option :destination
+    declare_option :sourcedir
+    declare_option :image_counter
+    declare_option :image_counter_pattern
+    declare_option :input_format
+    declare_option :track_time
   end
 end

lib/coradoc/reverse_adoc/html_converter.rb

@@ -36,28 +36,59 @@
 require_relative "converters/video"
 require_relative "converters/math"
 
-module Coradoc::ReverseAdoc
-  class HtmlConverter
-    def self.to_coradoc(input, options = {})
-      root = case input
-             when String
-               Nokogiri::HTML(input).root
-             when Nokogiri::XML::Document
-               input.root
-             when Nokogiri::XML::Node
-               input
-             end
+module Coradoc
+  module ReverseAdoc
+    class HtmlConverter
+      def self.to_coradoc(input, options = {})
+        ReverseAdoc.config.with(options) do
+          root = track_time "Loading input HTML document" do
+            case input
+            when String
+              Nokogiri::HTML(input).root
+            when Nokogiri::XML::Document
+              input.root
+            when Nokogiri::XML::Node
+              input
+            end
+          end
 
-      return "" unless root
+          return "" unless root
 
-      Coradoc::ReverseAdoc.config.with(options) do
-        Coradoc::ReverseAdoc::Converters.lookup(root.name).to_coradoc(root)
+          track_time "Converting input document tree to Coradoc tree" do
+            Converters.lookup(root.name).to_coradoc(root)
+          end
+        end
+      end
+
+      def self.convert(input, options = {})
+        ReverseAdoc.config.with(options) do
+          coradoc = to_coradoc(input)
+          result = track_time "Converting Coradoc tree into Asciidoc" do
+            Coradoc::Generator.gen_adoc(coradoc)
+          end
+          track_time "Cleaning up the result" do
+            ReverseAdoc.cleaner.tidy(result)
+          end
+        end
       end
-    end
 
-    def self.convert(input, options = {})
-      result = Coradoc::Generator.gen_adoc(to_coradoc(input, options))
-      Coradoc::ReverseAdoc.cleaner.tidy(result)
+      @track_time_indentation = 0
+      def self.track_time(task)
+        if ReverseAdoc.config.track_time
+          warn "  " * @track_time_indentation +
+            "* #{task} is starting..."
+          @track_time_indentation += 1
+          t0 = Time.now
+          ret = yield
+          time_elapsed = Time.now - t0
+          @track_time_indentation -= 1
+          warn "  " * @track_time_indentation +
+            "* #{task} took #{time_elapsed.round(3)} seconds"
+          ret
+        else
+          yield
+        end
+      end
     end
   end
 end