hryk/feed_task.rb

## feed_task.rb
#!/usr/bin/env ruby

require "nokogiri"
require "excon"
require "thor"

# Tasks related to maintain my subscriptions.
class FeedTask < Thor
  OPML = <<-EOF
<?xml version="1.0" encoding="UTF-8" ?>
<opml version="1.0">
<head><title>Subscriptions list from jottit.</title></head>
<body></body>
</opml>
EOF

  desc "opml_to_md FILE", "Output OPML content as markdown"
  def opml_to_md(opml_file)
    opml = File.open(opml_file)
    doc = Nokogiri::XML::Document.parse(opml)
    doc.xpath("//outline[@xmlUrl]").each do |outline|
      puts "* [#{outline.attr("text")}](#{outline.attr("xmlUrl")})"
    end
  end

  desc "export_jottit URL", "Create OPML from jottit page."
  option :tag, type: :string, default: :all
  option :output, type: :string
  def export_jottit(url)
    Excon.defaults[:ssl_verify_peer] = false
    subscriptions = get_feeds_from_jottit(url)
    opml = generate_opml(feeds: subscriptions, tag: options[:tag])
    output = if options[:output]
               open(options[:output], "w")
             else
               $stdout
             end
    output.puts opml
  end

  protected

  def add_outline_element(doc, feed)
    outline = Nokogiri::XML::Node.new("outline", doc)
    outline.set_attribute("text", feed[:title])
    outline.set_attribute("title", feed[:title])
    outline.set_attribute("type", "rss")
    outline.set_attribute("xmlUrl", feed[:xml_url])
    doc.css("body").first.add_child(outline)
  end

  def generate_opml(feeds: {}, tag: :all)
    export_feeds = if tag == :all
                     feeds.values.flatten
                   elsif feeds.key? tag
                     feeds[tag]
                   else
                     fail "[generate_opml] tag `#{tag}` not found."
                   end
    doc = Nokogiri::XML.parse(OPML)
    export_feeds.each { |feed| add_outline_element(doc, feed) }
    doc
  end

  def get_feeds_from_jottit(url)
    feeds, feed_stub = {}, {}
    response = Excon.new(url).request(method: :get, expects: [200])
    html = Nokogiri::HTML.parse(response.body)
    html.css("#content").children.each do |node|
      if node.name == "h1"
        if feed_stub.key?(:feeds) && feed_stub[:feeds].size > 0
          feeds[feed_stub[:label]] = feed_stub[:feeds]
        end
        feed_stub = { label: node.text, feeds: [] }
      elsif node.name == "ul"
        feed_stub[:feeds] = node.css("li>a")
          .map { |a| { title: a.text, xml_url: a.attr(:href) } }
      end
    end
    feeds
  rescue => e
    raise "[get_feeds_from_jottit] Failed to fetch jottit page #{e}"
  end
end

FeedTask.start(ARGV)
	#!/usr/bin/env ruby

	require "nokogiri"
	require "excon"
	require "thor"

	# Tasks related to maintain my subscriptions.
	class FeedTask < Thor
	OPML = <<-EOF
	<?xml version="1.0" encoding="UTF-8" ?>
	<opml version="1.0">
	<head><title>Subscriptions list from jottit.</title></head>
	<body></body>
	</opml>
	EOF

	desc "opml_to_md FILE", "Output OPML content as markdown"
	def opml_to_md(opml_file)
	opml = File.open(opml_file)
	doc = Nokogiri::XML::Document.parse(opml)
	doc.xpath("//outline[@xmlUrl]").each do \|outline\|
	puts "* [#{outline.attr("text")}](#{outline.attr("xmlUrl")})"
	end
	end

	desc "export_jottit URL", "Create OPML from jottit page."
	option :tag, type: :string, default: :all
	option :output, type: :string
	def export_jottit(url)
	Excon.defaults[:ssl_verify_peer] = false
	subscriptions = get_feeds_from_jottit(url)
	opml = generate_opml(feeds: subscriptions, tag: options[:tag])
	output = if options[:output]
	open(options[:output], "w")
	else
	$stdout
	end
	output.puts opml
	end

	protected

	def add_outline_element(doc, feed)
	outline = Nokogiri::XML::Node.new("outline", doc)
	outline.set_attribute("text", feed[:title])
	outline.set_attribute("title", feed[:title])
	outline.set_attribute("type", "rss")
	outline.set_attribute("xmlUrl", feed[:xml_url])
	doc.css("body").first.add_child(outline)
	end

	def generate_opml(feeds: {}, tag: :all)
	export_feeds = if tag == :all
	feeds.values.flatten
	elsif feeds.key? tag
	feeds[tag]
	else
	fail "[generate_opml] tag `#{tag}` not found."
	end
	doc = Nokogiri::XML.parse(OPML)
	export_feeds.each { \|feed\| add_outline_element(doc, feed) }
	doc
	end

	def get_feeds_from_jottit(url)
	feeds, feed_stub = {}, {}
	response = Excon.new(url).request(method: :get, expects: [200])
	html = Nokogiri::HTML.parse(response.body)
	html.css("#content").children.each do \|node\|
	if node.name == "h1"
	if feed_stub.key?(:feeds) && feed_stub[:feeds].size > 0
	feeds[feed_stub[:label]] = feed_stub[:feeds]
	end
	feed_stub = { label: node.text, feeds: [] }
	elsif node.name == "ul"
	feed_stub[:feeds] = node.css("li>a")
	.map { \|a\| { title: a.text, xml_url: a.attr(:href) } }
	end
	end
	feeds
	rescue => e
	raise "[get_feeds_from_jottit] Failed to fetch jottit page #{e}"
	end
	end

	FeedTask.start(ARGV)