#!/usr/local/bin/ruby
# 
#  backup_svn_repositories.rb
#  
#  Created by Benoit Gagnon on 2008-01-03.
#  Copyright 2008. All rights reserved.
# 

require 'yaml'

# Basic parameters
REPOS_PARENT = "/home/svn"
DUMP_DIR     = "/var/svn_dumps"
SYNC_FILE    = "#{DUMP_DIR}/syncfile.yml"

# Paths to executables
SVNADMIN     = '/usr/local/bin/svnadmin'
SVNLOOK      = '/usr/local/bin/svnlook'
RSYNC        = '/usr/local/bin/rsync'
MD5          = '/sbin/md5'

# Rsync options
USE_RSYNC    = true
RSYNC_FLAGS  = '--ignore-times -avcz'
REMOTE_HOST  = 'www.yourhostname.com'
REMOTE_USER  = 'username'
REMOTE_KEY   = '/path/to/your_ssh_key'
REMOTE_DIR   = "/var/svn_dumps"

# Log formatter
def log( m )
  STDOUT.puts "#{Time.now.strftime('%Y-%m-%d %H:%M:%S')} | #{m}"
end

log "-- Session started --"

# Get a list of all repositories
repos_parent_dir = Dir.new( REPOS_PARENT )
repos_list = repos_parent_dir.reject { |p| [ '.', '..' ].include?(p) }

# Get a fresh hash with the HEAD of each repo
fresh_hash = Hash.new
log "Scanning for repositories..."
repos_list.each do |r|
  log "  found repo: #{r}"
  uuid = `#{SVNLOOK} uuid #{REPOS_PARENT}/#{r}`.chomp
  head = `#{SVNLOOK} youngest #{REPOS_PARENT}/#{r}`.chomp.to_i
  fresh_hash[uuid] = { :name => r, :rev => head }
end

# Load previous hash
prev_hash = Hash.new
if File.exist?( SYNC_FILE )
  prev_hash = YAML.load_file( SYNC_FILE )
  log "Loaded sync file: #{SYNC_FILE}"
else
  log "Sync file not found: #{SYNC_FILE}"
end

# Compute operations
ops = Array.new
log "Finding differences..."
fresh_hash.each do |uuid, h|
  if prev_hash[uuid].nil?
    log "  new: #{h[:name]}"
    ops << { :uuid  => uuid,
             :name  => h[:name],
             :range => "0:#{h[:rev]}",
             :mode  => :full }
  elsif prev_hash[uuid][:rev] < h[:rev]
    log "  changed: #{h[:name]}"
    ops << { :uuid  => uuid,
             :name  => h[:name],
             :range => "#{prev_hash[uuid][:rev]+1}:#{h[:rev]}",
             :mode  => :incremental }
  else
    log "  up to date: #{h[:name]}"
  end
end

# Perform operations
ops.each do |op|
  repo_path = "#{REPOS_PARENT}/#{op[:name]}"
  dump_path = "#{DUMP_DIR}/#{op[:uuid]}.svndump"

  # Use '>' if we're writing to a new file, '>>' if we're appending
  inc, stream = '', '>'
  if op[:mode] == :incremental
    inc = '--incremental'
    stream = '>>'
    
    # Integrity check
    if `#{MD5} -q #{dump_path}`.chomp != prev_hash[op[:uuid]][:md5]
      log "ERROR: Integrity check failed for #{op[:name]}"
      exit
    end
  end
  
  cmd = "#{SVNADMIN} dump -q #{inc} #{repo_path} -r #{op[:range]} #{stream} #{dump_path}"
  
  log "start: #{cmd}"
  if not system(cmd)
    log "ERROR: Command failed."
    exit
  end
  log "done."
  
  # Update hash
  prev_hash[op[:uuid]] = fresh_hash[op[:uuid]]
  prev_hash[op[:uuid]][:md5] = `#{MD5} -q #{dump_path}`.chomp

  # Save updated hash after each op
  File.open( SYNC_FILE, 'w' ) { |f| YAML.dump( prev_hash, f ) }
  log "Saved sync file to #{SYNC_FILE}"
end

if USE_RSYNC
  if ops.empty?
    log "No operations performed; will not rsync."
  else
    # Rsync the dumps to some exotic location
    cmd = "#{RSYNC} --rsh=\"ssh -i #{REMOTE_KEY}\" #{RSYNC_FLAGS} #{DUMP_DIR} #{REMOTE_USER}@#{REMOTE_HOST}:#{REMOTE_DIR}"
    log "Starting rsync to #{REMOTE_USER}@#{REMOTE_HOST}:#{REMOTE_DIR}"
    if not system(cmd)
      log "ERROR: Command failed."
      exit
    end
    log "rsync complete."
  end
end

# At this point, prev_hash should be equivalent to fresh_hash
log "-- Session complete --"