Only list the latest SRPM files on an online repo

README.md

Exemple:

$ ruby latest-srpm.rb <REPO_URL> | xargs wget -N

It will only download the latest SRPM

Gemfile

source 'https://rubygems.org'

gem 'nokogiri'
gem 'colorize'

latest-srpm.rb

#!/usr/bin/env ruby
require 'nokogiri'
require 'open-uri'
require 'colorize'

# Read packages
base_url = ARGV[0]

abort "Usage: latest-srpm.rb <URL>" unless base_url

page = Nokogiri::HTML(open(base_url))
links = []
page.css('a').each do |a|
  url = a['href']

  # Only filters src.rpm files
  is_srpm = a.content.match(/(?<name>[a-zA-Z0-9_\-\+]*)-(?<version>[0-9\.]*)-(?<rev>[-_a-zA-Z0-9\.]*)[^\-]*\.src\.rpm/)
  next unless is_srpm

  # parse name and version from filename
  name, version, rev = is_srpm.captures
  rev ||= 0
  rev = rev.gsub(/[el]/, '.').gsub(/_/, '-').gsub(/\.\.*/, '.')

  link = {
    url: base_url + url,
    name: name,

    # transform version into a semver object
    version: Gem::Version.new("#{version}-#{rev}")
  }
  links << link
end

# Only keep the latest version
packages = {}
links.each do |link|
  name=link[:name]
  if packages[name] == nil 
    packages[name] = link
  elsif packages[name][:version] < link[:version]
    packages[name] = link
  elsif packages[name][:version] == link[:version]
    $stderr.puts "Same verion for #{packages[name]} and #{link}".red
  end
end

# Print the latest SRPM
packages.each_value do |link|
  puts "#{link[:url]}"
end