lseq

#!/usr/bin/env ruby
# Search for and rank pages in logseq graph by number of chars or lines
#
# Usage: lseq [-h] [-m] [-l] [-i] [<page-name-substring>]

require 'optparse'

def count_entities(filepath, mode)
  content = File.read(filepath)
  mode == 'lines' ? content.lines.size : content.size
end

def print_help
  puts <<~HELP
    Usage: lseq [-h] [-m] [-l] [-i] [<page-name-substring>]
    Options:
      -h, --help        Show this help message and exit
      -m                Count characters (default)
      -l                Count lines
      -i                Ignore files starting with 'ARCHIVE'
  HELP
end

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

  directory = "#{ENV['LOGSEQ_GRAPH']}/pages"
  mode = 'chars'
  ignore_archive = false
  substring = nil

  options = {}
  OptionParser.new do |opts|
    opts.banner = "Usage: lseq [options] [<page-name-substring>]"

    opts.on('-h', '--help', 'Show this help message and exit') do
      print_help
      exit
    end

    opts.on('-m', 'Count characters (default)') do
      mode = 'chars'
    end

    opts.on('-l', 'Count lines') do
      mode = 'lines'
    end

    opts.on('-i', 'Ignore files starting with ARCHIVE') do
      ignore_archive = true
    end
  end.parse!

  substring = ARGV.pop if ARGV.any?

  matching_files = Dir.entries(directory).select do |filename|
    next if filename == '.' || filename == '..'
    next if ignore_archive && filename.start_with?('ARCHIVE')
    if substring
      substring_with_percent2f = substring.gsub('/', '%2F')
      substring_with_dot = substring_with_percent2f.gsub('%2F', '.')
      filename.include?(substring_with_percent2f) || filename.include?(substring_with_dot)
    else
      true
    end
  end.compact.map do |filename|
    filepath = File.join(directory, filename)
    entity_count = count_entities(filepath, mode)
    [filename, entity_count]
  end

  matching_files.sort_by! { |_, count| count }

  entity_name = mode == 'lines' ? 'lines' : 'characters'
  matching_files.each do |filename, entity_count|
    base = File.basename(filename, '.*').gsub('%2F', ' / ').gsub('.', ' / ')
    puts "#{entity_count}:\t #{base}"
  end
end

if __FILE__ == $0
  main
end