aws-cf-log

#!/usr/bin/env ruby

require 'date'
require 'optparse'
require 'pathname'
require 'json'
require 'fileutils'

# ---
# 1. Option Parsing and Help
# ---

options = {}
OptionParser.new do |opts|
  opts.banner = "Usage: #{$0} --start <time> [--end <time>] [--raw] [--flush]"

  opts.on("--start TIME", "Start time of the log range (e.g., '7 days ago').") do |time|
    options[:start_time] = time
  end

  opts.on("--end TIME", "End time of the log range (e.g., '1 day ago').") do |time|
    options[:end_time] = time
  end

  opts.on("--raw", "Outputs the unzipped log content to stdout, suppressing all other logs.") do
    options[:raw_mode] = true
  end

  opts.on("--flush", "Force a re-download and remove existing log files.") do
    options[:flush] = true
  end

  opts.on("-h", "--help", "Prints this help") do
    puts opts
    exit
  end
end.parse!

# ---
# 2. Environment Variable Checks
# ---

required_vars = ['S3_PREFIX', 'DISTRIBUTION_ID']
missing_vars = required_vars.select { |var| ENV[var].nil? || ENV[var].empty? }

unless missing_vars.empty?
  unless options[:raw_mode]
    STDERR.puts "Error: The following environment variables are not set: #{missing_vars.join(', ')}"
    STDERR.puts "Please set them before running the script."
  end
  exit 1
end

# ---
# 3. Date Parsing Logic
# ---

def parse_time(time_str)
  case time_str
  when 'now'
    Date.today
  when /^([0-9]+)\s+(days|day|d)s? ago$/
    Date.today - $1.to_i
  when /^([0-9]+)\s+(weeks|week|w)s? ago$/
    Date.today - $1.to_i * 7
  when /^([0-9]+)\s+(months|month|m)s? ago$/
    Date.today << $1.to_i
  when /^([0-9]+)\s+(years|year|y)s? ago$/
    Date.today << $1.to_i * 12
  when /^[0-9]{4}-[0-9]{2}-[0-9]{2}/
    Date.parse(time_str)
  else
    raise "Unsupported time format: '#{time_str}'"
  end
rescue => e
  STDERR.puts "Error parsing time: #{e.message}"
  exit 1
end

begin
  start_date = parse_time(options[:start_time])
  end_date   = parse_time(options[:end_time] || 'now')
rescue
  STDERR.puts "Error: Invalid date format provided."
  exit 1
end

# ---
# 4. AWS CLI Command Execution with Caching
# ---

s3_prefix = ENV['S3_PREFIX']
distribution_id = ENV['DISTRIBUTION_ID']
CACHE_FILE = '.cloudfront_log_cache.json'

cache_data = {}
if File.exist?(CACHE_FILE)
  cache_data = JSON.parse(File.read(CACHE_FILE), symbolize_names: true)
end

current_options = {
  start_time: options[:start_time],
  end_time: options[:end_time] || 'now',
  s3_prefix: s3_prefix,
  distribution_id: distribution_id,
  raw_mode: options[:raw_mode]
}

cache_match = false
if cache_data[:options] == current_options
  last_run_time = DateTime.parse(cache_data[:timestamp])
  if (DateTime.now - last_run_time) * 24 * 60 < 15 # 15 minutes
    cache_match = true
  end
end

unless options[:raw_mode]
  puts "---"
  puts "Starting CloudFront log download process."
  puts "S3 Prefix: #{s3_prefix}"
  puts "Distribution ID: #{distribution_id}"
  puts "Requested time range: start='#{options[:start_time]}', end='#{options[:end_time] || 'now'}'"
  puts "Converted absolute date range: #{start_date.strftime('%Y-%m-%d')} to #{end_date.strftime('%Y-%m-%d')}"
end

if cache_match && !options[:flush]
  unless options[:raw_mode]
    puts "Cache hit! Skipping download. Last run was less than 15 minutes ago with the same parameters."
    puts "Use --flush to force a re-download."
    puts "---"
  end
else
  unless options[:raw_mode]
    puts "---"
    puts "Downloading logs..."
    puts "---"
  end

  # Cleanup command
  if options[:flush]
    unless options[:raw_mode]
      puts "Flushing existing log files (--flush option is active)."
      puts "---"
    end
    system("rm *.gz")
  end

  current_date = start_date
  while current_date <= end_date
    date_str = current_date.strftime('%Y-%m-%d')
    unless options[:raw_mode]
      puts "Syncing logs for #{date_str}..."
    end
    unless system "aws s3 sync #{s3_prefix}/ . --exclude \"*\" --include \"#{distribution_id}.#{date_str}-*\""
      unless options[:raw_mode]
        STDERR.puts "Error: Failed to sync logs for date #{date_str}."
      end
      exit 1
    end
    current_date = current_date.next_day
  end

  # Update cache file
  cache_data = {
    options: current_options,
    timestamp: DateTime.now.to_s
  }
  File.write(CACHE_FILE, JSON.pretty_generate(cache_data))
end

# ---
# 5. Output and Post-Processing
# ---

if options[:raw_mode]
  system "gunzip -c #{distribution_id}.*.gz"
else
  puts "---"
  system "ls -la"
  puts "The log files are now available in the current directory."
  puts "---"
end