reverse_adoc 0.2.5 → 0.2.10

data/lib/reverse_adoc.rb

@@ -0,0 +1,27 @@
+# frozen_string_literal: true
+
+require 'digest'
+require 'nokogiri'
+require 'reverse_adoc/version'
+require 'reverse_adoc/errors'
+require 'reverse_adoc/cleaner'
+require 'reverse_adoc/config'
+require 'reverse_adoc/converters'
+require 'reverse_adoc/converters/base'
+require "reverse_adoc/html_converter"
+
+module ReverseAdoc
+  def self.convert(input, options = {})
+    ReverseAdoc::HtmlConverter.convert(input, options)
+  end
+
+  def self.config
+    @config ||= Config.new
+    yield @config if block_given?
+    @config
+  end
+
+  def self.cleaner
+    @cleaner ||= Cleaner.new
+  end
+end

data/lib/{reverse_asciidoctor → reverse_adoc}/cleaner.rb

@@ -1,4 +1,4 @@
-module ReverseAsciidoctor
+module ReverseAdoc
   class Cleaner
 
     def tidy(string)
@@ -36,19 +36,19 @@ module ReverseAsciidoctor
     # Same for underscores and brackets.
     def clean_tag_borders(string)
       result = string.gsub(/\s?\*{2,}.*?\*{2,}\s?/) do |match|
-        preserve_border_whitespaces(match, default_border: ReverseAsciidoctor.config.tag_border) do
+        preserve_border_whitespaces(match, default_border: ReverseAdoc.config.tag_border) do
           match.strip.sub('** ', '**').sub(' **', '**')
         end
       end
 
       result = result.gsub(/\s?\_{2,}.*?\_{2,}\s?/) do |match|
-        preserve_border_whitespaces(match, default_border: ReverseAsciidoctor.config.tag_border) do
+        preserve_border_whitespaces(match, default_border: ReverseAdoc.config.tag_border) do
           match.strip.sub('__ ', '__').sub(' __', '__')
         end
       end
 
       result = result.gsub(/\s?~{2,}.*?~{2,}\s?/) do |match|
-        preserve_border_whitespaces(match, default_border: ReverseAsciidoctor.config.tag_border) do
+        preserve_border_whitespaces(match, default_border: ReverseAdoc.config.tag_border) do
           match.strip.sub('~~ ', '~~').sub(' ~~', '~~')
         end
       end
@@ -64,6 +64,29 @@ module ReverseAsciidoctor
       string.gsub(/(\*\*|~~|__)\s([\.!\?'"])/, "\\1".strip + "\\2")
     end
 
+     # preprocesses HTML, rather than postprocessing it
+    def preprocess_word_html(string)
+      clean_headings(scrub_whitespace(string.dup))
+    end
+
+    def scrub_whitespace(string)
+      string.gsub!(/ |\ |\u00a0/i, ' ')       # HTML encoded spaces
+      string.sub!(/^\A[[:space:]]+/m, '') # document leading whitespace
+      string.sub!(/[[:space:]]+\z$/m, '') # document trailing whitespace
+      string.gsub!(/([ ]+)$/, ' ')       # line trailing whitespace
+      string.gsub!(/\n\n\n\n/, "\n\n")  # Quadruple line breaks
+      #string.delete!('?| ')               # Unicode non-breaking spaces, injected as tabs
+      string
+    end
+
+    # following added by me
+    def clean_headings(string)
+      string.gsub!(%r{<h([1-9])[^>]*></h\1>}, " ") # I don't know why Libre Office is inserting them, but they need to go
+      string.gsub!(%r{<h([1-9])[^>]* style="vertical-align: super;[^>]*>(.+?)</h\1>},
+                   "<sup>\\2</sup>")         # I absolutely don't know why Libre Office is rendering superscripts as h1
+      string
+    end
+
     private
 
     def preserve_border_whitespaces(string, options = {}, &block)

data/lib/{reverse_asciidoctor → reverse_adoc}/config.rb

@@ -1,12 +1,13 @@
 require 'tmpdir'
 
-module ReverseAsciidoctor
+module ReverseAdoc
   class Config
     attr_accessor :unknown_tags, :tag_border, :mathml2asciimath, :external_images,
-      :destination, :sourcedir, :image_counter, :image_counter_pattern
+      :destination, :sourcedir, :image_counter, :image_counter_pattern, :input_format
 
     def initialize
       @unknown_tags     = :pass_through
+      @input_format     = :html
       @mathml2asciimath = false
       @external_images  = false
 

data/lib/{reverse_asciidoctor → reverse_adoc}/converters.rb

@@ -1,4 +1,4 @@
-module ReverseAsciidoctor
+module ReverseAdoc
   module Converters
     def self.register(tag_name, converter)
       @@converters ||= {}
@@ -16,17 +16,17 @@ module ReverseAsciidoctor
     private
 
     def self.default_converter(tag_name)
-      case ReverseAsciidoctor.config.unknown_tags.to_sym
+      case ReverseAdoc.config.unknown_tags.to_sym
       when :pass_through
-        ReverseAsciidoctor::Converters::PassThrough.new
+        ReverseAdoc::Converters::PassThrough.new
       when :drop
-        ReverseAsciidoctor::Converters::Drop.new
+        ReverseAdoc::Converters::Drop.new
       when :bypass
-        ReverseAsciidoctor::Converters::Bypass.new
+        ReverseAdoc::Converters::Bypass.new
       when :raise
         raise UnknownTagError, "unknown tag: #{tag_name}"
       else
-        raise InvalidConfigurationError, "unknown value #{ReverseAsciidoctor.config.unknown_tags.inspect} for ReverseAsciidoctor.config.unknown_tags"
+        raise InvalidConfigurationError, "unknown value #{ReverseAdoc.config.unknown_tags.inspect} for ReverseAdoc.config.unknown_tags"
       end
     end
   end

data/lib/{reverse_asciidoctor → reverse_adoc}/converters/a.rb

@@ -1,6 +1,6 @@
 require "uri"
 
-module ReverseAsciidoctor
+module ReverseAdoc
   module Converters
     class A < Base
       def convert(node, state = {})

data/lib/{reverse_asciidoctor → reverse_adoc}/converters/aside.rb

@@ -1,4 +1,4 @@
-module ReverseAsciidoctor
+module ReverseAdoc
   module Converters
     class Aside < Base
       def convert(node, state = {})

data/lib/{reverse_asciidoctor → reverse_adoc}/converters/audio.rb

@@ -1,4 +1,4 @@
-module ReverseAsciidoctor
+module ReverseAdoc
   module Converters
     class Audio < Base
       def convert(node, state = {})

data/lib/{reverse_asciidoctor → reverse_adoc}/converters/base.rb

@@ -1,4 +1,4 @@
-module ReverseAsciidoctor
+module ReverseAdoc
   module Converters
     class Base
       def treat_children(node, state)
@@ -8,7 +8,7 @@ module ReverseAsciidoctor
       end
 
       def treat(node, state)
-        ReverseAsciidoctor::Converters.lookup(node.name).convert(node, state)
+        ReverseAdoc::Converters.lookup(node.name).convert(node, state)
       end
 
       def escape_keychars(string)

data/lib/{reverse_asciidoctor → reverse_adoc}/converters/blockquote.rb

@@ -1,4 +1,4 @@
-module ReverseAsciidoctor
+module ReverseAdoc
   module Converters
     class Blockquote < Base
       def convert(node, state = {})
@@ -7,7 +7,7 @@ module ReverseAsciidoctor
         cite = node['cite']
         attrs = cite ? "[quote, #{cite}]\n" : ""
         content = treat_children(node, state).strip
-        content = ReverseAsciidoctor.cleaner.remove_newlines(content)
+        content = ReverseAdoc.cleaner.remove_newlines(content)
         #"\n\n> " << content.lines.to_a.join('> ') << "\n\n"
         "\n\n#{attrs}____\n" << content.lines.to_a.join('') << "\n____\n\n"
       end

data/lib/{reverse_asciidoctor → reverse_adoc}/converters/br.rb

@@ -1,4 +1,4 @@
-module ReverseAsciidoctor
+module ReverseAdoc
   module Converters
     class Br < Base
       def convert(node, state = {})

data/lib/{reverse_asciidoctor → reverse_adoc}/converters/bypass.rb

@@ -1,4 +1,4 @@
-module ReverseAsciidoctor
+module ReverseAdoc
   module Converters
     class Bypass < Base
       def convert(node, state = {})

data/lib/{reverse_asciidoctor → reverse_adoc}/converters/code.rb

@@ -1,4 +1,4 @@
-module ReverseAsciidoctor
+module ReverseAdoc
   module Converters
     class Code < Base
       def convert(node, state = {})

data/lib/{reverse_asciidoctor → reverse_adoc}/converters/div.rb

@@ -1,4 +1,4 @@
-module ReverseAsciidoctor
+module ReverseAdoc
   module Converters
     class Div < Base
       def convert(node, state = {})

data/lib/{reverse_asciidoctor → reverse_adoc}/converters/drop.rb

@@ -1,4 +1,4 @@
-module ReverseAsciidoctor
+module ReverseAdoc
   module Converters
     class Drop < Base
       def convert(node, state = {})

data/lib/{reverse_asciidoctor → reverse_adoc}/converters/em.rb

@@ -1,4 +1,4 @@
-module ReverseAsciidoctor
+module ReverseAdoc
   module Converters
     class Em < Base
       def convert(node, state = {})

data/lib/{reverse_asciidoctor → reverse_adoc}/converters/figure.rb

@@ -1,4 +1,4 @@
-module ReverseAsciidoctor
+module ReverseAdoc
   module Converters
     class Figure < Base
       def convert(node, state = {})

data/lib/reverse_adoc/converters/h.rb

@@ -0,0 +1,34 @@
+module ReverseAdoc
+  module Converters
+    class H < Base
+      def convert(node, state = {})
+        id = node['id']
+        anchor = id ? "[[#{id}]]" : ""
+        internal_anchor = treat_children_anchors(node, state) || ""
+        anchor.empty? and anchor = internal_anchor
+        anchor.empty? or anchor += "\n"
+        prefix = '=' * (node.name[/\d/].to_i + 1)
+        ["\n", anchor, prefix, ' ', treat_children_no_anchors(node, state), "\n"].join
+      end
+
+      def treat_children_no_anchors(node, state)
+        node.children.reject { |a| a.name == "a" }.inject('') do |memo, child|
+          memo << treat(child, state)
+        end
+      end
+
+      def treat_children_anchors(node, state)
+        node.children.select { |a| a.name == "a" }.inject('') do |memo, child|
+          memo << treat(child, state)
+        end
+      end
+    end
+
+    register :h1, H.new
+    register :h2, H.new
+    register :h3, H.new
+    register :h4, H.new
+    register :h5, H.new
+    register :h6, H.new
+  end
+end

data/lib/{reverse_asciidoctor → reverse_adoc}/converters/head.rb

@@ -1,4 +1,4 @@
-module ReverseAsciidoctor
+module ReverseAdoc
   module Converters
     class Head < Base
       def convert(node, state = {})

data/lib/{reverse_asciidoctor → reverse_adoc}/converters/hr.rb

@@ -1,4 +1,4 @@
-module ReverseAsciidoctor
+module ReverseAdoc
   module Converters
     class Hr < Base
       def convert(node, state = {})

data/lib/{reverse_asciidoctor → reverse_adoc}/converters/ignore.rb

@@ -1,4 +1,4 @@
-module ReverseAsciidoctor
+module ReverseAdoc
   module Converters
     class Ignore < Base
       def convert(node, state = {})

data/lib/{reverse_asciidoctor → reverse_adoc}/converters/img.rb

@@ -1,51 +1,32 @@
 require "fileutils"
 require "pathname"
+require "tempfile"
+require "base64"
+require "mimemagic"
 
-module ReverseAsciidoctor
+module ReverseAdoc
   module Converters
     class Img < Base
 
       def image_number
         sprintf(
-          ReverseAsciidoctor.config.image_counter_pattern,
-          ReverseAsciidoctor.config.image_counter
+          ReverseAdoc.config.image_counter_pattern,
+          ReverseAdoc.config.image_counter
         )
       end
 
       def image_number_increment
-        ReverseAsciidoctor.config.image_counter += 1
+        ReverseAdoc.config.image_counter += 1
       end
 
       def datauri2file(src)
         %r{^data:image/(?<imgtype>[^;]+);base64,(?<imgdata>.+)$} =~ src
 
-        dest_dir = Pathname.new(ReverseAsciidoctor.config.destination).dirname
+        dest_dir = Pathname.new(ReverseAdoc.config.destination).dirname
         images_dir = dest_dir + 'images'
         FileUtils.mkdir_p(images_dir)
 
-        ext = ""
-
-        if imgdata
-          file = Tempfile.open(["radoc", ".jpg"]) do |f|
-            begin
-              f.binmode
-              f.write(Base64.strict_decode64(imgdata))
-              f.rewind
-              ext = MimeMagic.by_magic(f)
-            ensure
-              f.close!
-            end
-          end
-
-          image_src_path = file.path
-          # puts "tempfile: #{file}"
-
-        else
-          ext = File.extname(src).strip.downcase[1..-1]
-          image_src_path = Pathname.new(ReverseAsciidoctor.config.sourcedir) + src
-
-        end
-
+        ext, image_src_path = determine_image_src_path(src, imgdata)
         image_dest_path = images_dir + "#{image_number}.#{ext}"
 
         # puts "image_dest_path: #{image_dest_path.to_s}"
@@ -57,6 +38,23 @@ module ReverseAsciidoctor
         image_dest_path.relative_path_from(dest_dir)
       end
 
+      def determine_image_src_path(src, imgdata)
+        return copy_temp_file(imgdata) if imgdata
+
+        ext = File.extname(src).strip.downcase[1..-1]
+        [ext, Pathname.new(ReverseAdoc.config.sourcedir) + src]
+      end
+
+      def copy_temp_file(imgdata)
+        Tempfile.open(['radoc', '.jpg']) do |f|
+          f.binmode
+          f.write(Base64.strict_decode64(imgdata))
+          f.rewind
+          ext = MimeMagic.by_magic(f).subtype
+          [ext, f.path]
+        end
+      end
+
       def convert(node, state = {})
         alt   = node['alt']
         src   = node['src']
@@ -66,7 +64,7 @@ module ReverseAsciidoctor
         anchor = id ? "[[#{id}]]\n" : ""
         title = extract_title(node)
 
-        if ReverseAsciidoctor.config.external_images
+        if ReverseAdoc.config.external_images
           # puts "external image conversion #{id}, #{src}"
           src = datauri2file(src)
         end

data/lib/{reverse_asciidoctor → reverse_adoc}/converters/li.rb

@@ -1,4 +1,4 @@
-module ReverseAsciidoctor
+module ReverseAdoc
   module Converters
     class Li < Base
       def convert(node, state = {})

data/lib/{reverse_asciidoctor → reverse_adoc}/converters/mark.rb

@@ -1,4 +1,4 @@
-module ReverseAsciidoctor
+module ReverseAdoc
   module Converters
     class Mark < Base
       def convert(node, state = {})

data/lib/{reverse_asciidoctor → reverse_adoc}/converters/math.rb

@@ -1,15 +1,15 @@
-# Unless run with ReverseAsciidoctor.config.mathml2asciimath,
+# Unless run with ReverseAdoc.config.mathml2asciimath,
 # this is cheating: we're injecting MathML into Asciidoctor, but
 # Asciidoctor only understands AsciiMath or LaTeX
 
 require "mathml2asciimath"
 
-module ReverseAsciidoctor
+module ReverseAdoc
   module Converters
     class Math < Base
       def convert(node, state = {})
         stem = node.to_s.gsub(/\n/, " ")
-        stem = MathML2AsciiMath.m2a(stem) if ReverseAsciidoctor.config.mathml2asciimath
+        stem = MathML2AsciiMath.m2a(stem) if ReverseAdoc.config.mathml2asciimath
         stem = stem.gsub(/\[/, "\\[").gsub(/\]/, "\\]").gsub(/\(\(([^\)]+)\)\)/, "(\\1)") unless stem.nil?
         " stem:[" << stem << "] "
       end

data/lib/{reverse_asciidoctor → reverse_adoc}/converters/ol.rb

@@ -1,4 +1,4 @@
-module ReverseAsciidoctor
+module ReverseAdoc
   module Converters
     class Ol < Base
       def convert(node, state = {})

data/lib/{reverse_asciidoctor → reverse_adoc}/converters/p.rb

@@ -1,4 +1,4 @@
-module ReverseAsciidoctor
+module ReverseAdoc
   module Converters
     class P < Base
       def convert(node, state = {})

data/lib/{reverse_asciidoctor → reverse_adoc}/converters/pass_through.rb

@@ -1,4 +1,4 @@
-module ReverseAsciidoctor
+module ReverseAdoc
   module Converters
     class PassThrough < Base
       def convert(node, state = {})

data/lib/{reverse_asciidoctor → reverse_adoc}/converters/pre.rb

@@ -1,4 +1,4 @@
-module ReverseAsciidoctor
+module ReverseAdoc
   module Converters
     class Pre < Base
       def convert(node, state = {})

data/lib/{reverse_asciidoctor → reverse_adoc}/converters/q.rb

@@ -1,4 +1,4 @@
-module ReverseAsciidoctor
+module ReverseAdoc
   module Converters
     class Q < Base
       def convert(node, state = {})