Ruby script that converts a JSON export of Pinboard.in bookmarks into a properly tagged Markdown file to import into Roam

pinboard_roam_converter.rb

#!/usr/bin/env ruby -w
# frozen_string_literal: true

#
# usage
# chmod +x pinboard_converter.rb
# ./pinboard_converter.rb -t [pinboard_export_file] > tag_map.yml
# edit tag_map.yml
# ./pinboard_converter.rb [pinboard_export_file] tag_map.yml > [roam_import_file]
# import [roam_import_file] into Roam
#
# optionally pass -d and a file with a list of URL filters (one per line) for
# links that should not pass through into the markdown file
#

require 'optparse'
require 'json'
require 'yaml'

options = {}

optparse = OptionParser.new do |opts|
  opts.banner = 'Usage: pinboard_converter [options] pinboard_export_file tag_map_file'
  options[:tag_map] = false
  opts.on('-t', '--tags', 'Output a tag map file') do
    options[:tag_map] = true
  end

  opts.on('-d', '--domain-filter FILTER', 'Filter link domains using FILTER') do |filter_file|
    options[:domain_filter] = filter_file
  end
end

optparse.parse!

filters = if options[:domain_filter]
            File.readlines(options[:domain_filter]).map { |filter| Regexp.new(filter.strip) }
          else
            []
end

def filter_link(filters, href)
  filters.find { |filter| filter =~ href }
end

def non_empty_str(s)
  s == '' ? nil : s
end

def convert_tags(tag_map, tags)
  tags.strip.split(/\s+/).map { |tag| non_empty_str(tag_map[tag]) }.compact.uniq.map do |tag|
    if tag =~ /\s+/
      "#[[#{tag}]]"
    else
      "##{tag}"
    end
  end
end

pins = JSON.parse(File.read(ARGV[0]))

if options[:tag_map]
  tags = pins.each_with_object({}) do |pin, tag_acc|
    pin['tags'].split(/\s+/).map { |tag| tag }.each do |tag|
      tag_acc[tag] = tag
    end
  end.sort.to_h

  puts YAML.dump(tags)
else
  tag_map = YAML.safe_load(File.read(ARGV[1]))
  items = pins.map do |pin|
    href = pin['href'].strip

    if filter_link(filters, href)
      nil
    else
      desc = pin['description'].strip
      link = "- [#{desc}](#{href})"
      tags = convert_tags( tag_map, pin['tags'] )

      if tags.empty?
        link
      else
        "#{link} #{tags.join(' ')}"
      end
    end
  end

  puts items.compact.join("\n")
end