lseq-top

#!/usr/bin/env ruby
# lseq-top — list top-level page hierarchies and their subpage count

require 'set'

def parse_path(path)
  if path.include?('%2F')
    path.split('%2F')[0]
  else
    path.split('.')[0]
  end
end

def count_subpages(path, hierarchy_counts)
  file_name = File.basename(path, ".*")
  top_level = parse_path(file_name)
  hierarchy_counts[top_level] += 1
end

def print_hierarchy_counts(hierarchy_counts)
  hierarchy_counts.each do |hierarchy, count|
    puts "#{hierarchy}/ - #{count} subpages"
  end
end

def main
  unless ENV['LOGSEQ_GRAPH']
    puts "Error: LOGSEQ_GRAPH environment variable must be defined."
    exit 1
  end

  # Get the pages directory from the environment variable
  pages_dir = "#{ENV['LOGSEQ_GRAPH']}/pages"

  unless Dir.exist?(pages_dir)
    puts "#{pages_dir} is not a valid directory"
    exit 1
  end

  hierarchy_counts = Hash.new(0)

  Dir.foreach(pages_dir) do |filename|
    file_path = File.join(pages_dir, filename)
    if File.file?(file_path)
      count_subpages(filename, hierarchy_counts)
    end
  end

  print_hierarchy_counts(hierarchy_counts)
end

main