Import overcast.fm full data export OPML as Hugo posts

grab-overcast-thumbnails.rb

#!/usr/bin/env ruby

# USAGE: $0 path/to/overcast-detailed-export.xml

require 'fileutils'
require 'json'
require 'nokogiri'
require 'open-uri'

puts ARGV[0]
doc = Nokogiri::XML(File.open(ARGV[0]))
cards_cache = JSON.load(File.open('data/cards_cache.json'))

# don't make posts for listens from these feeds
skip_feeds = [
  # secret feed URLs here!
]
doc.xpath('//outline[@type="rss"]').each do |ep|
  next if skip_feeds.include? ep.attr('xmlUrl')
  if cards_cache.include? ep.attr('htmlUrl')
    puts "#{ep.attr('htmlUrl')} already in cache."
    next
  end
  puts "Fetching card data for #{ep.attr('htmlUrl')} from #{ep.attr('xmlUrl')}."
  begin
    feed_doc = Nokogiri::XML(open(ep.attr('xmlUrl')))
    feed_doc.xpath('//channel/itunes:image').each do |img|
      puts "Found: #{img.attr('href')}"
      cards_cache[ep.attr('htmlUrl')] = {
        'photo' => img.attr('href')
      }
    end
  rescue Exception => e
    puts e.inspect
  end
  sleep 1
end

puts cards_cache.to_json

import-overcast-listens.rb

#!/usr/bin/env ruby

# USAGE: $0 path/to/overcast-detailed-export.xml

require 'date'
require 'fileutils'
require 'nokogiri'
require 'json'
require 'yaml'

puts ARGV[0]
doc = Nokogiri::XML(File.open(ARGV[0]))

# don't make posts for listens from these feeds
skip_feeds = [
  # secret feed URLs here!
]
doc.xpath('//outline[@type="podcast-episode" and @played="1"]').each do |ep|
  # ep.parent is the podcast (aka outline type="rss")
  next if skip_feeds.include? ep.parent.attr('xmlUrl')
  urls = [ep_url = ep.attr('overcastUrl')]
  # use the included episode URL if it "makes sense".
  puts ep.attr('url')
  if (!!ep.attr('url')) && (! ep.attr('url').end_with?('mp3')) && (ep.attr('url') != ep.parent.attr('htmlUrl'))
    urls.unshift(ep.attr('url'))
  end
  slug = (('Listened to ') + ep.attr('title')).downcase.gsub(/[\W[:punct:]]+/,'-').gsub(/-$/,'')
  post = {
    'h' => 'entry',
    'date' => ep.attr('userUpdatedDate'),
    'listen-of' => urls.first,
    'slug' => slug,
    'refs' => {
      "#{urls.first}" => {
        'url' => urls.first,
        'data' => {
          'type' => 'entry',
          'published' => ep.attr('pubDate'),
          'url' => urls.first,
          'audio' => [ep.attr('enclosureUrl')],
          'name' => ep.attr('title'),
          'author' => {
            'type' => 'h-card',
            'name' => ep.parent.attr('title'),
            'url' => ep.parent.attr('htmlUrl')
          }
        }
      }
    }
  }
  post_date = DateTime.parse(post['date'])
  file = "content/listens/#{post_date.strftime('%Y/%m/%d')}/#{slug}.md"
  dir = File.dirname(file)
  FileUtils.mkdir_p dir
  open(file, 'w') do |f|
    f.write post.to_yaml
    f.write "---\n"
  end
end