Podcast RSS generator in Ruby

podcast.rb

#!/usr/bin/env ruby
require 'date'
require 'xml'

def htmlit(infile)
  lines = File.readlines(infile)
  /<!--\s+(.+)\s+-->/.match lines.shift
  lines.unshift('<h2>%s</h2>' % $1)
  html = ''
  lines.each do |line|
    html << line.strip.gsub('="/', '="https://sivers.org/')
  end
  html
end

def subtitle(infile)
  txt = File.readlines(infile)[2].strip
  txt.gsub(/<[^>]+>/, '')
end

def duration(mp3)
  seconds = %x(soxi -D /w/m/#{mp3}).to_i
  Time.at(seconds).utc.strftime('%H:%M:%S')
end

doc = XML::Document.new()
doc.root = XML::Node.new('rss')
rss = doc.root
rss['version'] = '2.0'
rss['xmlns:itunes'] = 'http://www.itunes.com/dtds/podcast-1.0.dtd'
rss['xmlns:atom'] = 'http://www.w3.org/2005/Atom'
rss['xmlns:content'] = 'http://purl.org/rss/1.0/modules/content/'
rss['xmlns:googleplay'] = 'https://www.google.com/schemas/play-podcasts/1.0/'
rss << channel = XML::Node.new('channel')
channel << title = XML::Node.new('title')
title << 'Derek Sivers'
channel << description = XML::Node.new('description')
description << 'sivers.org'
channel << managingEditor = XML::Node.new('managingEditor')
managingEditor << 'derek@sivers.org'
channel << copyright = XML::Node.new('copyright')
copyright << '© 2019 Derek Sivers'
channel << generator = XML::Node.new('generator')
generator << 'Derek Sivers'
channel << atomlink = XML::Node.new('atom:link')
atomlink['href'] = 'https://sivers.org/podcast.rss'
atomlink['rel'] = 'self'
atomlink['type'] = 'application/rss+xml'
channel << link = XML::Node.new('link')
link << 'https://sivers.org/'
channel << itunesnewfeed = XML::Node.new('itunes:new-feed-url')
itunesnewfeed << 'https://sivers.org/podcast.rss'
channel << itunesowner = XML::Node.new('itunes:owner')
itunesowner << itunesemail = XML::Node.new('itunes:email')
itunesemail << 'derek@sivers.org'
itunesowner << itunesname = XML::Node.new('itunes:name')
itunesname << 'Derek Sivers'
channel << itunesauthor = XML::Node.new('itunes:author')
itunesauthor << 'Derek Sivers'
channel << itunessummary = XML::Node.new('itunes:summary')
itunessummary << 'Derek Sivers posts from sivers.org'
channel << itunessubtitle = XML::Node.new('itunes:subtitle')
itunessubtitle << 'Derek Sivers posts from sivers.org'
channel << language = XML::Node.new('language')
language << 'en'
channel << itunesexplicit = XML::Node.new('itunes:explicit')
itunesexplicit << 'no'
channel << cat1 = XML::Node.new('itunes:category')
cat1['text'] = 'Education'
cat1 << subcat1 = XML::Node.new('itunes:category')
subcat1['text'] = 'Self-Improvement'
channel << cat2 = XML::Node.new('itunes:category')
cat2['text'] = 'Society & Culture'
cat2 << subcat2 = XML::Node.new('itunes:category')
subcat2['text'] = 'Philosophy'
channel << cat3 = XML::Node.new('itunes:category')
cat3['text'] = 'Arts'
cat3 << subcat3 = XML::Node.new('itunes:category')
subcat3['text'] = 'Books'
channel << ituneskeywords = XML::Node.new('itunes:keywords')
ituneskeywords << 'Derek Sivers,sivers,sivers.org'
channel << itunestype = XML::Node.new('itunes:type')
itunestype << 'episodic'
channel << itunesimage = XML::Node.new('itunes:image')
itunesimage['href'] = 'https://sivers.org/images/DerekSivers-20141209a-1400.jpg'
channel << image = XML::Node.new('image')
image << image_url = XML::Node.new('url')
image_url << 'https://sivers.org/images/DerekSivers-20141209a-1400.jpg'
image << image_link = XML::Node.new('link')
image_link << 'https://sivers.org/'
image << image_title = XML::Node.new('title')
image_title << 'Derek Sivers'

d = %w(trd exan trav1 pg2 masch exex vls1 ment wtg polut d22 gofear daydream antic autom negz sauna erra hooky ldn isms metac ddown uncomf hf dq pe abio ww tp3 tp2 tp1 pinit)
d.each do |x|
  infile = Dir['/w/sivers.org/content/blog/*-' + x][0]
  m = /\A([0-9]{4}-[0-9]{2}-[0-9]{2})-(\S+)\Z/.match(File.basename(infile))
  date = m[1]
  contents = htmlit(infile)
  lines = File.readlines(infile)
  /<!--\s+(.+)\s+-->/.match lines.shift
  blogtitle = $1
  mp3 = 'sivers.org.%s.mp3' % x
  item_url = "https://m.sivers.org/#{mp3}"
  channel << item = XML::Node.new('item')
  item << item_link = XML::Node.new('link')
  item_link << "https://sivers.org/#{x}"
  item << title = XML::Node.new('title')
  title << blogtitle
  item << itunestitle = XML::Node.new('itunes:title')
  itunestitle << blogtitle
  item << itunessubtitle = XML::Node.new('itunes:subtitle')
  itunessubtitle << subtitle(infile)
  item << itunesauthor = XML::Node.new('itunes:author')
  itunesauthor << 'Derek Sivers'
  item << description = XML::Node.new('description')
  description << XML::Node.new_cdata(contents)
  item << summary = XML::Node.new('itunes:summary')
  summary << XML::Node.new_cdata(contents)
  item << contente = XML::Node.new('content:encoded')
  contente << XML::Node.new_cdata(contents)
  item << season = XML::Node.new('itunes:season')
  season << '1'
  item << guid = XML::Node.new('guid')
  guid << 'https://sivers.org/' + x
  item << pubDate = XML::Node.new('pubDate')
  pubDate << Date.parse(date).strftime('%a, %d %b %Y %H:%M:%S %z')
  item << itunesepisodeType = XML::Node.new('itunes:episodeType')
  itunesepisodeType << 'full'
  item << itunesexplicit = XML::Node.new('itunes:explicit')
  itunesexplicit << 'no'
  item << itunesimage = XML::Node.new('itunes:image')
  itunesimage['href'] = 'https://sivers.org/images/DerekSivers-20141209a-1400.jpg'
  item << itunesduration = XML::Node.new('itunes:duration')
  itunesduration << duration(mp3)
  item << enclosure = XML::Node.new('enclosure')
  enclosure['url'] = 'https://m.sivers.org/%s' % mp3
  enclosure['type'] = 'audio/mpeg'
  enclosure['length'] = File.size("/w/m/#{mp3}").to_s
end

doc.save('site/podcast.rss', indent: true, encoding: XML::Encoding::UTF_8)