bin/mog

   1 #!/usr/bin/env ruby
   2 require 'mogilefs'
   3 require 'optparse'
   4
   5 trap('INT') { exit 130 }
   6 trap('PIPE') { exit 0 }
   7
   8 # this is to be compatible with config files used by the Perl tools
   9 def parse_config_file!(path, overwrite = false)
  10   dest = {}
  11   File.open(path).each_line do |line|
  12     line.strip!
  13     if /^(domain|class)\s*=\s*(\S+)/.match(line)
  14       dest[$1.to_sym] = $2
  15     elsif m = /^(?:trackers|hosts)\s*=\s*(.*)/.match(line)
  16       dest[:hosts] = $1.split(/\s*,\s*/)
  17     elsif m = /^timeout\s*=\s*(.*)/.match(line)
  18       dest[:timeout] = m[1].to_f
  19     else
  20       STDERR.puts "Ignored configuration line: #{line}" unless /^#/.match(line)
  21     end
  22   end
  23   dest
  24 end
  25
  26 # parse the default config file if one exists
  27 def_file = File.expand_path("~/.mogilefs-client.conf")
  28 def_cfg = File.exist?(def_file) ? parse_config_file!(def_file) : {}
  29
  30 # parse the command-line first, these options take precedence over all else
  31 cli_cfg = {}
  32 config_file = nil
  33 ls_l = false
  34 ls_h = false
  35 test = {}
  36 cat = { :raw => false }
  37
  38 ARGV.options do |x|
  39   x.banner = "Usage: #{$0} [options] <command> [<arguments>]"
  40   x.separator ''
  41
  42   x.on('-c', '--config=/path/to/config',
  43        'config file to load') { |file| config_file = file }
  44
  45   x.on('-t', '--trackers=host1[,host2]', '--hosts=host1[,host2]', Array,
  46        'hostnames/IP addresses of trackers') do |trackers|
  47     cli_cfg[:hosts] = trackers
  48   end
  49
  50   x.on('-e', 'True if key exists') { test[:e] = true }
  51   x.on('-r', '--raw', 'show raw big_info file information') { cat[:raw] = true }
  52
  53   x.on('-C', '--class=s', 'class') { |klass| cli_cfg[:class] = klass }
  54   x.on('-d', '--domain=s', 'domain') { |domain| cli_cfg[:domain] = domain }
  55   x.on('-l', "long listing format (`ls' command)") { ls_l = true }
  56   x.on('-h', '--human-readable',
  57        "print sizes in human-readable format (`ls' command)") { ls_h = true }
  58
  59   x.separator ''
  60   x.on('--help', 'Show this help message.') { puts x; exit }
  61   x.parse!
  62 end
  63
  64 # parse the config file specified at the command-line
  65 file_cfg = config_file ? parse_config_file!(config_file, true) : {}
  66
  67 # read environment variables, too.  This Ruby API favors the term
  68 # "hosts", however upstream MogileFS teminology favors "trackers" instead.
  69 # Favor the term more consistent with what the MogileFS inventors used.
  70 env_cfg = {}
  71 if ENV["MOG_TRACKERS"]
  72   env_cfg[:hosts] = ENV["MOG_TRACKERS"].split(/\s*,\s*/)
  73 end
  74 if ENV["MOG_HOSTS"] && (env_cfg[:hosts] || []).empty?
  75   env_cfg[:hosts] = ENV["MOG_HOSTS"].split(/\s*,\s*/)
  76 end
  77 env_cfg[:domain] = ENV["MOG_DOMAIN"] if ENV["MOG_DOMAIN"]
  78 env_cfg[:class] = ENV["MOG_CLASS"] if ENV["MOG_CLASS"]
  79
  80 # merge the configs, favoring them in order specified:
  81 cfg = {}.merge(def_cfg).merge(env_cfg).merge(file_cfg).merge(cli_cfg)
  82
  83 # error-checking
  84 err = []
  85 err << "trackers must be specified" if cfg[:hosts].nil? || cfg[:hosts].empty?
  86 err << "domain must be specified" unless cfg[:domain]
  87 if err.any?
  88   STDERR.puts "Errors:\n  #{err.join("\n  ")}"
  89   STDERR.puts ARGV.options
  90   exit 1
  91 end
  92
  93 unless cmd = ARGV.shift
  94   STDERR.puts ARGV.options
  95   exit 1
  96 end
  97
  98 cfg[:timeout] ||= 30 # longer timeout for interactive use
  99 include MogileFS::Util
 100 mg = MogileFS::MogileFS.new(cfg)
 101
 102 def store_file_retry(mg, key, storage_class, filepath)
 103   tries = 0
 104   begin
 105     mg.store_file(key, storage_class, filepath)
 106   rescue MogileFS::UnreadableSocketError,
 107          MogileFS::Backend::NoDevicesError => err
 108     if ((tries += 1) < 10)
 109       STDERR.puts "Retrying on error: #{err}: #{err.message} tries: #{tries}"
 110       retry
 111     else
 112       STDERR.puts "FATAL: #{err}: #{err.message} tries: #{tries}"
 113     end
 114     exit 1
 115   end
 116 end
 117
 118 begin
 119   case cmd
 120   when 'cp'
 121     filename = ARGV.shift or raise ArgumentError, '<filename> <key>'
 122     key = ARGV.shift or raise ArgumentError, '<filename> <key>'
 123     ARGV.shift and raise ArgumentError, '<filename> <key>'
 124     store_file_retry(mg, key, cfg[:class], filename)
 125   when 'cat'
 126     ARGV.empty? and raise ArgumentError, '<key1> [<key2> ...]'
 127     ARGV.each do |key|
 128       if (!cat[:raw] && key =~ /^_big_info:/)
 129         mg.bigfile_write(key, STDOUT, {:verify => true})
 130       else
 131         mg.get_file_data(key) { |fp| sysrwloop(fp, STDOUT) }
 132       end
 133     end
 134   when 'ls'
 135     prefixes = ARGV.empty? ? [ nil ] : ARGV
 136     prefixes.each do |prefix|
 137       mg.each_key(prefix) do |key|
 138         if ls_l
 139           path_nr = "% 2d" % mg.get_paths(key).size
 140           size = mg.size(key)
 141           if ls_h && size > 1024
 142             suff = ''
 143             %w(K M G).each do |s|
 144               size /= 1024.0
 145               suff = s
 146               break if size <= 1024
 147             end
 148             size = sprintf("%.1f%s", size, suff)
 149           else
 150             size = size.to_s
 151           end
 152           size = (' ' * (12 - size.length)) << size # right justify
 153           puts [ path_nr, size, key ].pack("A4 A16 A32")
 154         else
 155           puts key
 156         end
 157       end
 158     end
 159   when 'rm'
 160     ARGV.empty? and raise ArgumentError, '<key1> [<key2>]'
 161     ARGV.each { |key| mg.delete(key) }
 162   when 'mv'
 163     from = ARGV.shift or raise ArgumentError, '<from> <to>'
 164     to = ARGV.shift or raise ArgumentError, '<from> <to>'
 165     ARGV.shift and raise ArgumentError, '<from> <to>'
 166     mg.rename(from, to)
 167   when 'stat' # this outputs a RFC822-like format
 168     ARGV.empty? and raise ArgumentError, '<key1> [<key2>]'
 169     ARGV.each_with_index do |key, i|
 170       if size = mg.size(key)
 171         puts "Key: #{key}"
 172         puts "Size: #{size}"
 173         mg.get_paths(key).each_with_index do |path,i|
 174           puts "URL-#{i}: #{path}"
 175         end
 176         puts ""
 177       else
 178         STDERR.puts "No such key: #{key}"
 179       end
 180     end
 181   when 'tee'
 182     require 'tempfile'
 183     key = ARGV.shift or raise ArgumentError, '<key>'
 184     ARGV.shift and raise ArgumentError, '<key>'
 185     cfg[:class] or raise ArgumentError, 'E: --class must be specified'
 186     buf = ''
 187     tmp = Tempfile.new('mog-tee') # TODO: explore Transfer-Encoding:chunked :)
 188     at_exit { tmp.unlink }
 189
 190     # if stdout is pointing to /dev/null, don't bother installing the filter.
 191     STDOUT.sync = true
 192     tee_filter = File.stat('/dev/null') == STDOUT.stat ?
 193                  nil : Proc.new { |buf| STDOUT.write(buf); buf }
 194     begin
 195       sysrwloop(STDIN, tmp, tee_filter)
 196       store_file_retry(mg, key, cfg[:class], tmp.path)
 197     ensure
 198       tmp.close
 199     end
 200   when 'test'
 201     truth, ok = true, nil
 202     raise ArgumentError, "-e must be specified" unless (test.size == 1)
 203
 204     truth, key = case ARGV.size
 205     when 1
 206       [ true, ARGV[0] ]
 207     when 2
 208       if ARGV[0] != "!"
 209         raise ArgumentError, "#{ARGV[0]}: binary operator expected"
 210       end
 211       [ false, ARGV[1] ]
 212     else
 213       raise ArgumentError, "Too many arguments"
 214     end
 215
 216     paths = mg.get_paths(key)
 217     if test[:e]
 218       ok = !!(paths && paths.size > 0)
 219     else
 220       raise ArgumentError, "Unknown flag: -#{test.keys.first}"
 221     end
 222
 223     truth or ok = ! ok
 224     exit ok ? 0 : 1
 225   else
 226     raise ArgumentError, "Unknown command: #{cmd}"
 227   end
 228 rescue ArgumentError => err
 229   STDERR.puts "Usage: #{$0} #{cmd} #{err.message}"
 230   exit 1
 231 end
 232 exit 0