coradoc 1.0.0 → 1.1.1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 71bfd2ac723d58ca75b591bceabcdf834abd9cbbf32e54b96352da21b8f15095
-  data.tar.gz: 1a7ea6bf80d64ab4c1349c948aa2aee9ed704dc36ddd07fb7cce78293dca86cd
+  metadata.gz: '0187fda1ca8860383e973840ca92de0d24b7de0f4b548eb92364b83dc76773c2'
+  data.tar.gz: bcccecebdf83ef11138182fc66ad27f5e24bb01d0b2824fa8b470d3c32ba413a
 SHA512:
-  metadata.gz: 555800621a06ffafc07e03a5c2cdca1210de6369d524a606de51df78b1ee9b19aaba30d32d8a448347237e488362a59677781a02502a1d8891fb89aa1cc267f5
-  data.tar.gz: 22707b18c6ed99fb4c6127885d871875ed1fed2906d2d10dcdd5f4ed9bdd1b1bfe3f19fd092768ccc4d488bdb7e705aba9ba7ab948ceace9cc679f1ee733eddc
+  metadata.gz: 37f09a7b2ba8fc861c4cf88f8c5638e6c654cc19a65b1b6c68314c6d9ff1b45f046559d573a849ee972b4dfa0422384271c1951ce5baa02491dc1c03000c2e77
+  data.tar.gz: c6844f3ccad625662b838cba656fa6e1f773675548d4fe664278f93569f7cf0ae86edcd70e26de2e60cb487090d3873720bc0c6e0f6892f8559e77a78129af4d

data/Makefile

@@ -0,0 +1,35 @@
+export SPEC ?= spec
+SPEC_FILE = $(subst ../,, $(SPEC))
+export RUBY_IMAGE ?= ruby:3.1.2-slim
+
+.PHONY: up
+up:
+	docker-compose up
+
+.PHONY: down
+down:
+	docker-compose down
+
+.PHONY: test
+test: rspec
+
+.PHONY: ssh
+ssh:
+	docker-compose run lib bash
+
+.PHONY: install
+install:
+	docker-compose run lib bin/setup
+
+.PHONY: console
+console:
+	docker-compose run lib bin/console
+
+.PHONY: rspec
+rspec:
+	docker-compose run lib bin/rspec ${SPEC_FILE}
+
+.PHONY: setup
+setup:
+	docker-compose build --build-arg RUBY_IMAGE=${RUBY_IMAGE}
+	docker-compose run lib bin/setup

data/{.docker/readme.md → README.Docker.adoc}

@@ -1,61 +1,57 @@
-## Docker
+= Docker
 
-This directory is only meant to be used for development, and contains some
+This Docker image is only meant to be used for development, and contains some
 necessary setup to spin up docker containers with multiple ruby environment.
 
-### Setup
+== Setup
 
-Before doing anything, you might want to create a symlink to the docker file and
-Makefile. This would allow you to avoid some of the unnecessary work related to
-the file paths To do that run the following from the root of the project.
-
-```
-ln -sf .docker/Makefile .
-ln -sf .docker/docker-compose.yml .
-```
-
-By default it usages the most recent ruby version for docker environment, but if
+By default it uses the most recent ruby version for docker environment, but if
 you want to run it in any specific version then you can set it up by exporting
 `RUBY_IMAGE` environment variable in your shell:
 
-```sh
+[source,sh]
+----
 export RUBY_IMAGE=ruby:3.0-buster
-```
+----
 
 Once everything is set then you would need to build the development images for
 the first time and you can do that using:
 
-```sh
+[source,sh]
+----
 make setup
-```
+----
 
 The setup process will install all dependencies and it will also setup a volume
 to speed up the repeated gem installation.
 
-### Playground
+== Playground
 
 The `Makefile` contains two target for tests, and you can run the tests using
 any of the following commands:
 
-```sh
+[source,sh]
+----
 make test
 
 # or
 make rspec
-```
+----
 
 If you need more control, and you want to do some development on the go then you
 can get into the container using:
 
-```sh
+[source,sh]
+----
 make ssh
-```
+----
 
-### Cleanup
+== Cleanup
 
 Once you are done with your experiment then you can cleanup the docker
 environment using the following command.
 
-```sh
+[source,sh]
+----
 make down
-```
+----

data/README.adoc

@@ -0,0 +1,121 @@
+:sandi-metz: footnote:[http://robots.thoughtbot.com/post/50655960596/sandi-metz-rules-for-developers]
+
+= Coradoc
+
+Coradoc is a modern Parser for Asciidoc document. It defines a grammar for
+AsciiDoc, and then build the Parser for that grammar.
+
+Once the document is parsed, it provides a pure ruby object `Coradoc::Document`,
+which can used to customize the document in easiest way.
+
+== Installation
+
+Add this line to your application's Gemfile:
+
+[source,ruby]
+----
+gem "coradoc"
+----
+
+And then execute:
+
+[source,sh]
+----
+bundle install
+----
+
+Or install it yourself as:
+
+[source,sh]
+----
+gem install coradoc
+----
+
+== Development
+
+We are following Sandi Metz's Rules for this gem, you can read the
+description of the rules here.{sandi-metz} All new code should follow these
+rules. If you make changes in a pre-existing file that violates these rules you
+should fix the violations as part of your contribution.
+
+== Setup
+
+Clone the repository.
+
+[source,sh]
+----
+git clone https://github.com/metanorma/coradoc.git
+----
+
+Setup your environment in docker
+
+[source,sh]
+----
+make setup
+----
+
+Run the test suite
+
+[source,sh]
+----
+make test
+----
+
+
+== Usage from command line
+
+=== Converting a document
+
+[source,sh]
+----
+$ coradoc help convert
+$ coradoc convert file.html -o file.adoc
+----
+
+== Usage from Ruby
+
+=== Parsing a document
+
+To parse any AsciiDoc, we can use the following:
+
+[source,ruby]
+----
+Coradoc::Parser.parse(sample_asciidoc)
+----
+
+This interface will return the abstract syntax tree.
+
+=== Converting a document
+
+To convert any document of a supported format (right now: `.html`, `.adoc`, `.docx`) to any supported
+format (right now: `.adoc`), you can execute:
+
+[source,ruby]
+----
+Coradoc::Converter.("input.html", "output.adoc")
+----
+
+The converters are chosen based on file extension, but you can select a converter manually like so:
+
+[source,ruby]
+----
+Coradoc::Converter.("input", "output", input_processor: :html, output_processor: :adoc)
+----
+
+Some converters may support additional options, which can likewise be passed as keyword arguments:
+
+[source,ruby]
+----
+Coradoc::Converter.(
+  "input.html", "output.adoc",
+  input_options: { external_images: true, split_sections: 2 }
+)
+----
+
+It is also possible to pass IO objects instead of filenames. By default, if an argument is not
+provided, it defaults to STDIN/STDOUT. Note that not all combinations of formats and converter
+options are supported in this mode.
+
+=== Legacy README for converting from HTML to AsciiDoc (formerly reverse_adoc)
+
+See: https://github.com/metanorma/coradoc/blob/main/lib/coradoc/input/html/README.adoc[Coradoc::Input::HTML README]

data/coradoc.gemspec

@@ -35,6 +35,8 @@ Gem::Specification.new do |spec|
   spec.add_dependency "parslet"
   spec.add_dependency "premailer", "~> 1.11.0"
   spec.add_dependency "word-to-markdown"
+  spec.add_dependency "base64"
+  spec.add_dependency "thor"
   spec.add_development_dependency "codeclimate-test-reporter"
   spec.add_development_dependency "pry"
   spec.add_development_dependency "rake"
@@ -43,5 +45,7 @@ Gem::Specification.new do |spec|
   spec.add_development_dependency "rubocop"
   spec.add_development_dependency "rubocop-performance"
   spec.add_development_dependency "simplecov"
+  # spec.add_development_dependency "parallel_tests"
+  # spec.add_development_dependency "stackprof"
   # spec.add_runtime_dependency "thor"
 end

data/docker-compose.yml

@@ -0,0 +1,14 @@
+version: "3"
+
+services:
+  lib:
+    build:
+      context: .
+      dockerfile: ./Dockerfile
+
+    volumes:
+      - .:/workspace:z
+      - bundle:/bundle:z
+
+volumes:
+  bundle:

data/exe/coradoc

@@ -0,0 +1,5 @@
+#!/usr/bin/env ruby
+
+require "coradoc/cli"
+
+Coradoc::CLI.start

data/exe/reverse_adoc

@@ -4,28 +4,37 @@
 require "rubygems"
 require "bundler/setup"
 
-require "coradoc/reverse_adoc"
+require "coradoc/input/html"
+require "coradoc/converter"
 require "optparse"
 require "fileutils"
 
+config = {
+  input_options: input_options = {},
+  input_processor: :html,
+  output_options: output_options = {},
+  output_processor: :adoc,
+}
+destination = nil
+
 OptionParser.new do |opts|
   opts.banner = "Usage: reverse_adoc [options] <file>"
   opts.on("-m", "--mathml2asciimath", "Convert MathML to AsciiMath") do |_v|
-    Coradoc::ReverseAdoc.config.mathml2asciimath = true
+    input_options[:mathml2asciimath] = true
   end
 
   opts.on("-oFILENAME", "--output=FILENAME", "Output file to write to") do |v|
-    Coradoc::ReverseAdoc.config.destination = File.expand_path(v)
-    # puts "output goes to #{Coradoc::ReverseAdoc.config.destination}"
+    destination = File.expand_path(v)
+    # puts "output goes to #{Coradoc::Input::HTML.config.destination}"
   end
 
   opts.on("-e", "--external-images", "Export images if data URI") do |_v|
-    Coradoc::ReverseAdoc.config.external_images = true
+    input_options[:external_images] = true
   end
 
   opts.on("-u", "--unknown_tags [pass_through, drop, bypass, raise]",
           "Unknown tag handling (default: pass_through)") do |v|
-    Coradoc::ReverseAdoc.config.unknown_tags = v
+    input_options[:unknown_tags] = v
   end
 
   opts.on("-r", "--require RUBYMODULE", "Require additional Ruby file") do |v|
@@ -33,15 +42,15 @@ OptionParser.new do |opts|
   end
 
   opts.on("--track-time", "Track time spent on each step") do
-    Coradoc::ReverseAdoc.config.track_time = true
+    input_options[:track_time] = true
   end
 
   opts.on("--split-sections LEVEL", "Split sections up to LEVEL") do |i|
-    Coradoc::ReverseAdoc.config.split_sections = i.to_i
+    input_options[:split_sections] = i.to_i
   end
 
   opts.on("-v", "--version", "Version information") do |_v|
-    puts "reverse_adoc: v#{Coradoc::ReverseAdoc::VERSION}"
+    puts "Coradoc: v#{Coradoc::VERSION}"
     exit
   end
 
@@ -52,40 +61,21 @@ OptionParser.new do |opts|
 end.parse!
 
 if filename = ARGV.pop
-  input_content = IO.read(filename)
-  Coradoc::ReverseAdoc.config.sourcedir = File.dirname(File.expand_path(filename))
+  input_content = filename
 else
-  if Coradoc::ReverseAdoc.config.external_images
+  if input_options[:external_images]
     raise "The -e | --external-images feature cannot be used with STDIN input. Exiting."
   end
 
-  input_content = ARGF.read
+  input_content = ARGF
 end
 
-if Coradoc::ReverseAdoc.config.external_images && Coradoc::ReverseAdoc.config.destination.nil?
+if input_options[:external_images] && destination.nil?
   raise "The -e | --external-images feature must be used with -o | --output. Exiting."
 end
 
-if Coradoc::ReverseAdoc.config.split_sections && Coradoc::ReverseAdoc.config.destination.nil?
+if input_options[:split_sections] && destination.nil?
   raise "The --split_sections feature must be used with -o | --output. Exiting."
 end
 
-# Read from STDIN
-adoc_content = Coradoc::ReverseAdoc.convert(input_content)
-
-# Print to STDOUT
-unless Coradoc::ReverseAdoc.config.destination
-  puts adoc_content
-  exit
-end
-
-# Write output to Coradoc::ReverseAdoc.config.destination
-adoc_content = {nil => adoc_content} unless adoc_content.is_a? Hash
-
-adoc_content.each do |file, content|
-  destination = Coradoc::ReverseAdoc.config.destination
-  destdir = File.dirname(destination)
-  filename = file ? "#{destdir}/#{file}" : destination
-  FileUtils.mkdir_p(File.dirname(filename))
-  File.write(filename, content)
-end
+Coradoc::Converter.(input_content, destination, **config)

data/exe/w2a

@@ -6,27 +6,35 @@ require "bundler/setup"
 
 require "word-to-markdown"
 require "optparse"
-require "coradoc/reverse_adoc"
+require "coradoc"
 
 ARGV.push("-h") if ARGV.empty?
 
+config = {
+  input_options: input_options = {},
+  input_processor: :docx,
+  output_options: output_options = {},
+  output_processor: :adoc,
+}
+destination = nil
+
 OptionParser.new do |opts|
   opts.banner = "Usage: w2a [options] <file>"
   opts.on("-m", "--mathml2asciimath", "Convert MathML to AsciiMath") do |_v|
-    Coradoc::ReverseAdoc.config.mathml2asciimath = true
+    input_options[:mathml2asciimath] = true
   end
 
   opts.on("-oFILENAME", "--output=FILENAME", "Output file to write to") do |v|
-    Coradoc::ReverseAdoc.config.destination = File.expand_path(v)
-    # puts "output goes to #{Coradoc::ReverseAdoc.config.destination}"
+    destination = File.expand_path(v)
+    # puts "output goes to #{Coradoc::Input::HTML.config.destination}"
   end
 
   opts.on("-e", "--external-images", "Export images if data URI") do |_v|
-    Coradoc::ReverseAdoc.config.external_images = true
+    input_options[:external_images] = true
   end
 
   opts.on("-v", "--version", "Version information") do |_v|
-    puts "reverse_adoc: v#{Coradoc::ReverseAdoc::VERSION}"
+    puts "Coradoc: v#{Coradoc::VERSION}"
     puts "[dependency] WordToMarkdown: v#{WordToMarkdown::VERSION}"
     if Gem.win_platform?
       puts "[dependency] LibreOffice: version not available on Windows"
@@ -45,28 +53,8 @@ end.parse!
 filename = ARGV.pop
 raise "Please provide an input file to process. Exiting." unless filename
 
-if Coradoc::ReverseAdoc.config.external_images && Coradoc::ReverseAdoc.config.destination.nil?
+if input_options[:external_images] && destination.nil?
   raise "The -e | --external-images feature must be used with -o | --output. Exiting."
 end
 
-Coradoc::ReverseAdoc.config.sourcedir = Dir.mktmpdir
-
-doc = WordToMarkdown.new(filename, Coradoc::ReverseAdoc.config.sourcedir)
-# File.open("test.html", "w:UTF-8") { |f| f.write doc.document.html }
-adoc_content = Coradoc::ReverseAdoc.convert(
-  Coradoc::ReverseAdoc.cleaner.preprocess_word_html(doc.document.html),
-  WordToMarkdown::REVERSE_MARKDOWN_OPTIONS,
-)
-# puts scrub_whitespace(doc.document.html)
-
-# Print to STDOUT
-unless Coradoc::ReverseAdoc.config.destination
-  puts adoc_content
-  exit
-end
-
-# Write output to Coradoc::ReverseAdoc.config.destination
-FileUtils.mkdir_p(File.dirname(Coradoc::ReverseAdoc.config.destination))
-File.open(Coradoc::ReverseAdoc.config.destination, "w") do |file|
-  file.write(adoc_content)
-end
+Coradoc::Converter.(filename, destination, **config)

data/lib/coradoc/cli.rb

@@ -0,0 +1,105 @@
+require "coradoc"
+require "thor"
+
+module Coradoc
+  class CLI < Thor
+    package_name "coradoc"
+
+    desc "convert [FILE]", "Convert document to another format"
+
+    option :output,
+           type: :string, aliases: "-o",
+           desc: "Output file to write"
+
+    option :input_format,
+           type: :string, aliases: "-I",
+           enum: Input.keys.map(&:to_s), default: nil,
+           desc: "Define input format (defaults to input file extension)"
+
+    option :output_format,
+           type: :string, aliases: "-O",
+           enum: Output.keys.map(&:to_s), default: nil,
+           desc: "Define output format (defaults to output file extension)"
+
+    at_least_one :output, :output_format
+
+    option :require,
+           type: :string, aliases: "-r",
+           repeatable: true,
+           desc: "Require additional Ruby file (eg. to load a plugin)"
+
+    option :external_images,
+           type: :boolean, aliases: "-e",
+           desc: "Extract images from input document"
+
+    option :unknown_tags,
+           type: :string, aliases: "-u",
+           enum: %w[pass_through drop bypass raise],
+           default: "pass_through",
+           desc: "Unknown tag handling"
+
+    option :mathml2asciimath,
+           type: :boolean, aliases: "-m",
+           desc: "Convert MathML to AsciiMath"
+
+    option :track_time,
+           type: :boolean,
+           desc: "Track time spent on each step"
+
+    option :split_sections,
+           type: :numeric,
+           default: 0, banner: "LEVEL",
+           desc: "Split sections into separate files up to a provided level"
+
+    def convert(input = nil)
+      options[:require]&.each { |r| Kernel.require r }
+
+      config = {
+        input_options: input_options = {},
+        input_processor: nil,
+        output_options: output_options = {},
+        output_processor: nil,
+      }
+
+      config[:input_processor] = options[:input_format]&.to_sym
+      config[:output_processor] = options[:output_format]&.to_sym
+
+      %i[
+        external_images
+        unknown_tags
+        mathml2asciimath
+        track_time
+        split_sections
+      ].each do |i|
+        input_options[i] = options[i]
+      end
+
+      output = options[:output]
+
+      begin
+        Coradoc::Converter.(input, output, **config)
+      rescue Converter::NoInputPathError => e
+        warn "You must provide INPUT file as a file for this optionset."
+        warn "Detail: #{e.message}"
+      rescue Converter::NoOutputPathError => e
+        warn "You must provide OUTPUT file as a file for this optionset."
+        warn "Detail: #{e.message}"
+      rescue Converter::NoProcessorError => e
+        warn "No processor found for given input/output."
+        warn "Hint: set -I/--input-format or -O/--output-format option."
+        warn "Detail: #{e.message}"
+      end
+    end
+
+    desc "version", "display version information"
+    def version
+      puts "Coradoc: v#{Coradoc::VERSION}"
+      puts "[dependency] WordToMarkdown: v#{WordToMarkdown::VERSION}"
+      if Gem.win_platform?
+        puts "[dependency] LibreOffice: version not available on Windows"
+      else
+        puts "[dependency] LibreOffice: v#{WordToMarkdown.soffice.version}"
+      end
+    end
+  end
+end