Compare/Diff bookmark export HTML files in netscape formats

compare-bookmarks.rb

#!/usr/bin/env ruby

# Take 2 bookmark HTML files in netscape format, and look for differences.
# Doesn't compare hierarchy.

# Requirements: Ruby 2.x; oga ruby gem (gem install oga)
# License: MPL 2.0
HELP_MSG = "Usage: compare-bookmarks.rb [Bookmarks HTML 1] [Bookmarks HTML 2]".freeze

# HTML parser
require "oga"

file_1 = ARGV[0]
file_2 = ARGV[1]

if file_1.nil? || file_2.nil?
  puts HELP_MSG
  exit(1)
end

file_1_data = File.read(file_1)
file_2_data = File.read(file_2)
html_1 = Oga.parse_html(file_1_data)
html_2 = Oga.parse_html(file_2_data)

# @param nodes array of Oga nodes
# @returns array of strings
def nodes_to_strings(nodes)
  strings = []
  nodes.each do |node|
    href = node.attributes.find {|n| n.name == "HREF"}&.value&.strip
    string = href || node.text.strip
    strings.push(string)
  end
  strings
end

folders_1 = nodes_to_strings(html_1.css("h3"))
folders_2 = nodes_to_strings(html_2.css("h3"))
bookmarks_1 = nodes_to_strings(html_1.css("a"))
bookmarks_2 = nodes_to_strings(html_2.css("a"))

folders_in_1_not_in_2 = folders_1 - folders_2
folders_in_2_not_in_1 = folders_2 - folders_1
bookmarks_in_1_not_in_2 = bookmarks_1 - bookmarks_2
bookmarks_in_2_not_in_1 = bookmarks_2 - bookmarks_1

def log_strings(strings, prefix = "  ")
  strings.each do |s|
    puts "#{prefix}#{s}"
  end
end

puts "Folders in #{file_1} not in #{file_2} (#{folders_in_1_not_in_2.size} of #{folders_2.size}):"
log_strings(folders_in_1_not_in_2)
puts "Bookmarks in #{file_1} not in #{file_2} (#{bookmarks_in_1_not_in_2.size} of #{bookmarks_2.size}):"
log_strings(bookmarks_in_1_not_in_2)
puts "Folders in #{file_2} not in #{file_1} (#{folders_in_2_not_in_1.size} of #{folders_1.size}):"
log_strings(folders_in_2_not_in_1)
puts "Bookmarks in #{file_1} not in #{file_2} (#{bookmarks_in_2_not_in_1.size} of #{bookmarks_1.size}):"
log_strings(bookmarks_in_2_not_in_1)