shbackup

   1 #!/usr/bin/python3
   2 """A simple backup solution."""
   3
   4 __version__ = "0.1"
   5 __author__ = "Stefan Huber"
   6
   7 import datetime
   8 import os, shutil, sys
   9 import configparser
  10 import hashlib
  11 import subprocess, fcntl, select
  12 import random, re
  13 import logging
  14
  15
  16 Modes = ["full", "incr", "diff"]
  17
  18 class Epoch:
  19
  20     units = {
  21                 "hour" : datetime.timedelta(0, 3600),
  22                 "day" : datetime.timedelta(1),
  23                 "week" : datetime.timedelta(7),
  24                 "month" : datetime.timedelta(31),
  25                 "year" : datetime.timedelta(365) }
  26
  27     def __init__(self, unit=None, mult=1, mode="full", numkeeps=None):
  28         self.unit = unit
  29         self.mult = mult
  30         self.mode = mode
  31         self.numkeeps = numkeeps
  32         self.excludes = []
  33
  34     def __repr__(self):
  35         return "[unit: " + repr(self.unit) + \
  36                 ", mult:" + repr(self.mult) + \
  37                 ", mode: " + repr(self.mode) + \
  38                 ", numkeeps: " + repr(self.numkeeps) + \
  39                 ", excludes: " + repr(self.excludes) + "]"
  40
  41     def getTimeDelta(self):
  42         if self.unit == None:
  43             return None
  44         return self.mult*Epoch.units[self.unit]
  45
  46     def isRipe(self, oldest, now):
  47
  48         if self.unit==None:
  49             return True
  50
  51         delta = now-oldest
  52         mult = self.mult
  53
  54         if delta >= self.getTimeDelta():
  55             return True
  56
  57         if self.unit == "hour":
  58             return abs(now.hour - oldest.hour) >= mult
  59         elif self.unit == "day":
  60             return abs(now.day - oldest.day) >= mult
  61         elif self.unit == "week":
  62             return abs(now.isocalendar()[1] - oldest.isocalendar()[1]) >= mult
  63         elif self.unit == "month":
  64             return abs(now.month - oldest.month) >= mult
  65         elif self.unit == "year":
  66             return abs(now.year - oldest.year) >= mult
  67
  68         return None
  69
  70
  71     @staticmethod
  72     def parseTimedelta( deltastr ):
  73         tokens = [ s.strip() for s in deltastr.split("*") ]
  74         unit = None
  75         mult = 1
  76         if len(tokens) == 1:
  77             unit = tokens[0]
  78         elif len(tokens) == 2:
  79             mult = int(tokens[0])
  80             unit = tokens[1]
  81         else:
  82             raise ValueError("Invalid format: '{0}'".format(deltastr))
  83
  84         if not unit in Epoch.units:
  85             raise ValueError("Unknown unit '{0}'".format(unit))
  86
  87         if mult <= 0:
  88             raise ValueError("Non-positive factor '{0}' given.".format(mult))
  89
  90         return mult, unit
  91
  92
  93
  94 class FileSet:
  95     """A fileset has a name and a list of directories."""
  96     def __init__(self, name, dirs, excludes):
  97         self.name = name
  98         self.dirs = dirs
  99         self.excludes = excludes
 100
 101     def __repr__(self):
 102         return "[name: " + self.name + \
 103                 ", dirs: " + str(self.dirs) + \
 104                 ", excludes: " + str(self.excludes) + "]"
 105
 106
 107 class Backup:
 108     """A single backup has a date, an epoch and a mode."""
 109
 110     def __init__(self, date, epoch, mode):
 111         self.date = date
 112         self.epoch = epoch
 113         self.mode = mode
 114         self.excludes = []
 115
 116     @staticmethod
 117     def fromDirName(dirname):
 118             [strdate, strtime, epoch, mode] = dirname.split("-")
 119
 120             if not mode in Modes:
 121                 raise ValueError("Invalid mode: " + mode)
 122
 123             date = datetime.datetime(int(strdate[0:4]),
 124                     int(strdate[4:6]), int(strdate[6:8]),\
 125                     int(strtime[0:2]), int(strtime[2:4]))
 126
 127             return Backup(date, epoch, mode)
 128
 129     def __repr__(self):
 130         return "[date: " + self.date.ctime() + \
 131                 ", epoch: " + self.epoch + \
 132                 ", mode: " + self.mode + "]"
 133
 134     def colAlignedString(self):
 135         age = datetime.datetime.now() - self.date
 136         total_hours = age.total_seconds()/3600
 137         if total_hours <= 48:
 138             agestr = "(%s h)" % int(total_hours)
 139         else:
 140             agestr = "(%s d)" % age.days
 141         return "%16s  %7s  %10s  %4s" % (
 142                 self.date.strftime("%Y-%m-%d %H:%M"), agestr,
 143                 self.epoch, self.mode)
 144
 145     @staticmethod
 146     def getDirName(date, epoch, mode):
 147         """Get directory name of backup by given properties."""
 148         return date.strftime("%Y%m%d-%H%M") + "-" + epoch + "-" + mode
 149
 150     @staticmethod
 151     def isBackupDir(dirname):
 152         """Is directory a backup directory?"""
 153         p = re.compile(r'^\d\d\d\d\d\d\d\d-\d\d\d\d-\w+-\w+$')
 154         return p.match(dirname)
 155
 156
 157
 158 class Config:
 159     """Encapsules the configuration for the backup program."""
 160
 161     class ReadError(RuntimeError):
 162         """An exception raised when reading configurations."""
 163         def __init__(self, value):
 164             self.value = value
 165             self.message = value
 166
 167
 168     formats = ["tar", "tar.gz", "tar.bz2", "tar.xz" ]
 169
 170     # Filename where checksum of config is saved
 171     checksumfn = "checksum"
 172
 173     def __init__(self):
 174         self.backupdir = None
 175         self.format = self.formats[1]
 176         self.tarbin = "/bin/tar"
 177         self.excludes = []
 178         self.sets = []
 179         self.checksum = None
 180         self.lastchecksum = None
 181         self.epochs = Epochs = { "sporadic" : Epoch() }
 182
 183
 184     def __repr__(self):
 185         return "[backupdir: " + self.backupdir + \
 186                                   ", format: " + self.format + \
 187                                   ", tarbin: " + self.tarbin + \
 188                                   ", excludes: " + repr(self.excludes) + \
 189                   ", epochs: " + repr(self.epochs) + \
 190                                   ", sets: " + repr(self.sets) + "]"
 191
 192     def getRealEpochsSorted(self):
 193         """Return all epochs with have a non-None unit, sorted by
 194         Epoch.getTimeDelta(), starting with the longest dela."""
 195         epochs = self.epochs
 196         realepochs = [ e for e in epochs.keys() if epochs[e].unit != None ]
 197         deltakey = lambda e: epochs[e].getTimeDelta()
 198         realepochs.sort(key=deltakey, reverse=True)
 199         return realepochs
 200
 201
 202     def _read_global(self, config, sec):
 203         for opt in config.options(sec):
 204             if opt=="backupdir":
 205                 self.backupdir = config.get(sec, opt)
 206                 if not os.path.isdir(self.backupdir):
 207                     raise Config.ReadError("Backupdir '{0}' does not exist.".format(self.backupdir))
 208             elif opt=="format":
 209                 self.format = config.get(sec, opt)
 210                 if not self.format in Config.formats:
 211                     raise Config.ReadError("Invalid 'format' given.")
 212             elif opt=="tarbin":
 213                 self.tarbin = config.get(sec, opt)
 214                 if not os.path.isfile(self.tarbin):
 215                     raise Config.ReadError("Tar binary '{0}' does not exist.".format(self.tarbin))
 216             elif opt.startswith("exclude"):
 217                 self.excludes += [ config.get(sec, opt) ]
 218             else:
 219                 raise Config.ReadError("Unknown option '{0}'.".format(opt))
 220
 221
 222     def _read_epoch(self, config, sec):
 223         name = sec[6:].strip()
 224         e = Epoch()
 225         if name in self.epochs:
 226             raise Config.ReadError("Epoch '{0}' already defined.".format(name))
 227         if name in Epoch.units:
 228             e.unit = name
 229
 230         for opt in config.options(sec):
 231             if opt=="numkeeps":
 232                 try:
 233                     e.numkeeps = int(config.getint(sec, opt))
 234                 except ValueError:
 235                     raise Config.ReadError("Invalid integer given for '{0}'.".format(opt))
 236                 if e.numkeeps <= 0:
 237                     raise Config.ReadError("Non-positive numkeeps '{0}' given.".format(e.numkeeps))
 238
 239             elif opt=="mode":
 240                 e.mode = config.get(sec, opt)
 241                 if not e.mode in Modes:
 242                     raise Config.ReadError("Invalid mode '{0}'.".format(e.mode))
 243
 244             elif opt=="timespan":
 245                 if name in Epoch.units:
 246                     raise Config.ReadError("The time delta of a standard epoch " + \
 247                             "is not supposed to be redefined. ")
 248                 td = config.get(sec,opt)
 249                 try:
 250                     mult, unit = Epoch.parseTimedelta(td)
 251                     e.unit = unit
 252                     e.mult = mult
 253                 except ValueError as e:
 254                     raise Config.ReadError("Invalid timespan '{0}': {1}".format(td, str(e)))
 255
 256             elif opt.startswith("exclude"):
 257                 e.excludes += [config.get(sec, opt)]
 258
 259             else:
 260                 raise Config.ReadError("Unknown option '" + opt + "'.")
 261
 262         if e.numkeeps == None:
 263             raise Config.ReadError("No numkeeps set for epoch '{0}'.".format(name))
 264
 265         self.epochs[name] = e
 266
 267
 268     def _read_set(self, config, sec):
 269         name = sec[4:].strip()
 270         dirs = []
 271         excludes = []
 272
 273         for opt in config.options(sec):
 274             if opt.startswith("dir"):
 275                 dirs += [config.get(sec, opt)]
 276             elif opt.startswith("exclude"):
 277                 excludes += [config.get(sec,opt)]
 278             else:
 279                 raise Config.ReadError("Unknown option '" + opt + "'.")
 280
 281         self.sets += [FileSet(name, dirs, excludes)]
 282
 283
 284     def read(self, filename):
 285         """Read configuration from file"""
 286
 287         if not os.path.isfile(filename):
 288             raise Config.ReadError("Cannot read config file '" + filename + "'.")
 289
 290         config = configparser.RawConfigParser()
 291         config.read(filename)
 292
 293         for reqsec in ["global"]:
 294             if not config.has_section(reqsec):
 295                 raise Config.ReadError("Mandatory section '" + reqsec + "' is missing.")
 296
 297         for sec in config.sections():
 298
 299             if sec=="global":
 300                 self._read_global(config, sec)
 301
 302             elif sec.startswith("epoch "):
 303                 self._read_epoch(config, sec)
 304
 305             elif sec.startswith("set "):
 306                 self._read_set(config, sec)
 307
 308             else:
 309                 raise Config.ReadError("Unknown section '" + sec + "'.")
 310
 311         if self.backupdir == None:
 312             raise Config.ReadError("No backup directory set.")
 313
 314
 315         # Compute checksum of config file
 316         m = hashlib.sha1()
 317         f = open(filename, 'rb')
 318         try:
 319             m.update(f.read())
 320             self.checksum = m.hexdigest()
 321         finally:
 322             f.close()
 323
 324         try:
 325             f = open(os.path.join(self.backupdir, self.checksumfn), 'r')
 326             self.lastchecksum = f.read().strip()
 327             f.close()
 328         except IOError:
 329             self.lastchecksum = None
 330
 331
 332 class BackupManager:
 333     """List and create backups"""
 334
 335     def __init__(self, conffn):
 336         self.conf = Config()
 337         self.conf.read(conffn)
 338
 339
 340     def listAllDirs(self):
 341         """List all dirs in backupdir"""
 342
 343         # Get all entries
 344         basedir = self.conf.backupdir
 345         dirs = os.listdir(basedir)
 346         # Filter directories
 347         return [ d for d in dirs if os.path.isdir(os.path.join(basedir, d)) ]
 348
 349
 350     def listOldBackups(self):
 351         """Returns a list of old backups."""
 352
 353         backups = []
 354
 355         for entry in [ b for b in self.listAllDirs() if Backup.isBackupDir(b) ]:
 356             backups += [ Backup.fromDirName(entry) ]
 357
 358         return backups
 359
 360
 361     def getDesiredEpochs(self, backups, now):
 362         """Get desired epoch based on self.configuration and list of old backups"""
 363
 364         # Find the longest epoch for which we would like the make a backup
 365         latest = datetime.datetime(1900, 1, 1)
 366         for e in self.conf.getRealEpochsSorted():
 367             epoch = self.conf.epochs[e]
 368             if epoch.numkeeps <= 0:
 369                 continue
 370
 371             # Get backups of that epoch
 372             byepoch = list(sorted( [ b for b in backups if b.epoch==e], \
 373                 key=lambda b: b.date))
 374
 375             # If there are any, determine the latest
 376             if len(byepoch) > 0:
 377                 latest = max(latest, byepoch[-1].date )
 378
 379             if epoch.isRipe(latest, now):
 380                 return e
 381
 382         # No backup is to be made
 383         return None
 384
 385
 386
 387     def backupFileSet(self, fileset, targetdir, excludes, since=None):
 388         """Create an archive for given fileset at given target directory."""
 389
 390         logfile = logging.getLogger('backuplog')
 391         logfile.info("Running file set: " + fileset.name)
 392
 393         fsfn = os.path.join(targetdir, fileset.name) + "." + self.conf.format
 394         taropts = []
 395
 396         # Add the since date, if given
 397         if since != None:
 398             taropts += ["-N", since.strftime("%Y-%m-%d %H:%M:%S")]
 399
 400         # Add the exclude patterns
 401         for pat in excludes:
 402             taropts += ["--exclude", pat]
 403
 404         #Add exclude patterns from fileset
 405         for pat in fileset.excludes:
 406             taropts += ["--exclude", pat]
 407
 408         # Adding directories to backup
 409         taropts += ["-C", "/"] + [ "./" + d.lstrip("/") for d in fileset.dirs]
 410
 411         # Launch the tar process
 412         tarargs = [self.conf.tarbin] + ["-cpvaf", fsfn] + taropts
 413         logfile.debug("tar call: " + " ".join(tarargs))
 414         tarp = subprocess.Popen( tarargs, bufsize=-1, \
 415                 stdout=subprocess.PIPE, stderr=subprocess.PIPE )
 416
 417         # Change tarp's stdout and stderr to non-blocking
 418         for s in [tarp.stdout, tarp.stderr]:
 419             fd = s.fileno()
 420             fl = fcntl.fcntl(fd, fcntl.F_GETFL)
 421             fcntl.fcntl(fd, fcntl.F_SETFL, fl | os.O_NONBLOCK)
 422
 423         # Read stdout and stderr of tarp
 424         errmsg = b""
 425         while tarp.poll() == None:
 426             rd,wr,ex = select.select([tarp.stdout, tarp.stderr], [], [], 0.05)
 427             if tarp.stdout in rd:
 428                 logging.debug( tarp.stdout.readline()[:-1].decode() )
 429             if tarp.stderr in rd:
 430                 errmsg += tarp.stderr.read()
 431
 432         # Get the remainging output of tarp
 433         for l in tarp.stdout.readlines():
 434             logging.debug(l.decode().rstrip())
 435         errmsg += tarp.stderr.read()
 436
 437         # Get return code of tarp
 438         rett = tarp.wait()
 439         if rett != 0:
 440             for l in errmsg.decode().split("\n"):
 441                 logfile.error(l)
 442             logfile.error(self.conf.tarbin + " returned with exit status " + \
 443                    str(rett) + ".")
 444
 445
 446     def backup(self, epoch=None, mode=None):
 447         """Make a new backup, if necessary. If epoch is None then determine
 448         desired epoch automatically. Use given epoch otherwise. If mode is None
 449         then use mode for given epoch. Use given mode otherwise."""
 450
 451         now = datetime.datetime.now()
 452         oldbackups = self.listOldBackups()
 453
 454         # Get epoch of backup
 455         if epoch == None:
 456             epoch = self.getDesiredEpochs(oldbackups, now)
 457         if epoch == None:
 458             logging.info("No backup planned.")
 459             return
 460
 461         # Get mode of backup
 462         if mode == None:
 463             mode = self.conf.epochs[epoch].mode
 464         logging.info("Making a backup. Epochs: " + epoch + ", mode: " + mode)
 465
 466         oldfullbackups = [ b for b in oldbackups if b.mode == "full" ]
 467
 468         # No old full backups existing
 469         if mode != "full" and len(oldfullbackups)==0:
 470             logging.info("No full backups existing. Making a full backup.")
 471
 472         # Checksum changed -> self.config file changed
 473         if self.conf.checksum != self.conf.lastchecksum and mode != "full":
 474             logging.warning("Full backup recommended as config file has changed.")
 475
 476
 477         # If we have a full backup, we backup everything
 478         since = None
 479         if mode == "diff":
 480             since = sorted(oldfullbackups, key=lambda b: b.date)[-1].date
 481         elif mode == "incr":
 482             since = sorted(oldbackups, key=lambda b: b.date)[-1].date
 483
 484         if since != None:
 485             logging.debug("Making backup relative to " + since.ctime())
 486
 487         yesno = self.ask_user_yesno("Proceed? [Y, n] ")
 488         if yesno == "n":
 489             return
 490
 491         # Create new backup directory
 492         basedir = self.conf.backupdir
 493         dirname = Backup.getDirName(now, epoch, mode)
 494         tmpdirname = dirname + ("-%x" % (random.random()*2e16) )
 495         targetdir = os.path.join(basedir, tmpdirname)
 496         os.mkdir( targetdir )
 497
 498
 499         # Add file logger
 500         logfile = logging.getLogger("backuplog")
 501         fil = logging.FileHandler( os.path.join(targetdir, "log") )
 502         fil.setLevel(logging.DEBUG)
 503         logfile.addHandler(fil)
 504
 505         logfile.info("Started: " + now.ctime())
 506
 507         # Backup all file sets
 508         for s in self.conf.sets:
 509             excludes = self.conf.excludes + self.conf.epochs[epoch].excludes
 510             self.backupFileSet(s, targetdir, excludes, since)
 511
 512         logfile.info("Stopped: " + datetime.datetime.now().ctime())
 513
 514         # Rename backup directory to final name
 515         os.rename( targetdir, os.path.join(basedir, dirname) )
 516
 517         # We made a full backup -- recall checksum of config
 518         if mode == "full":
 519             f = open( os.path.join(basedir, self.conf.checksumfn), "w")
 520             f.write( self.conf.checksum )
 521             f.close()
 522
 523
 524
 525     def prune(self):
 526         """Prune old backup files"""
 527
 528         allDirs = sorted(self.listAllDirs())
 529         # Collect all directories not matching backup name
 530         removeDirs = [ d for d in allDirs if not Backup.isBackupDir(d) ]
 531
 532         # Get all directories which are kept
 533         backups = self.listOldBackups()
 534         keepdirs = []
 535         byepoch = { e : list(sorted( [ b for b in backups if b.epoch == e ], \
 536                 key=lambda b : b.date, reverse=True)) for e in self.conf.getRealEpochsSorted() }
 537         for e in byepoch:
 538             epoch = self.conf.epochs[e]
 539             old = byepoch[e][epoch.numkeeps:]
 540             removeDirs += [ Backup.getDirName(b.date, b.epoch, b.mode) for b in old]
 541
 542
 543         logging.info("List of stale/outdated entries:")
 544         for d in allDirs:
 545             msg = ""
 546             if d in removeDirs:
 547                 msg = "[*]  "
 548             else:
 549                 msg = "[ ]  "
 550
 551             if Backup.isBackupDir(d):
 552                 msg += Backup.fromDirName(d).colAlignedString()
 553             else:
 554                 msg += d
 555
 556             logging.info(msg)
 557
 558         # Check that dirs to be removed is in list of all dirs
 559         for d in removeDirs:
 560             assert( d in allDirs )
 561
 562         if len(removeDirs) == 0:
 563             logging.info("No stale/outdated entries to remove.")
 564             return
 565
 566         basedir = self.conf.backupdir
 567         yesno = self.ask_user_yesno("Remove entries marked by '*'? [y, N] ")
 568         if yesno == "y":
 569             for d in removeDirs:
 570                 try:
 571                     shutil.rmtree(os.path.join(basedir, d))
 572                 except OSError as e:
 573                     logging.error("Error when removing '%s': %s" % (d,e.strerror) )
 574
 575
 576     def ask_user_yesno(self, question):
 577         if LogConf.con.level <= logging.INFO:
 578             return input(question)
 579         else:
 580             return "y"
 581
 582
 583 def printUsage():
 584     """Print --help text"""
 585
 586     print("shbackup - a simple backup solution.")
 587     print("")
 588     print("Usage:")
 589     print("  " + sys.argv[0] + " {options} [cmd]")
 590     print("  " + sys.argv[0] + " --help")
 591     print("")
 592     print("Commands:")
 593     print("  backup                     make a new backup, if necessary")
 594     print("  list                       list all backups (default)")
 595     print("  prune                      prune outdated/old backups")
 596     print("")
 597     print("Options:")
 598     print("  -h, --help                 print this usage text")
 599     print("  -c, --conf <configfile>    use given configuration file")
 600     print("                             default: /etc/shbackup.conf")
 601     print("  -e, --epoch <epoch>        force to create backup for given epoch, which")
 602     print("                             can be 'sporadic' or one of the configured epochs")
 603     print("  -m, --mode <mode>          override mode: full, diff, or incr")
 604     print("  -v, --verbose              be more verbose and interact with user")
 605     print("  --verbosity LEVEL          set verbosity to LEVEL, which can be")
 606     print("                             error, warning, info, debug")
 607     print("  -V, --version              print version info")
 608
 609
 610
 611 class LogConf:
 612     """Encapsulates logging configuration"""
 613
 614     con = logging.StreamHandler(sys.stderr)
 615
 616     @classmethod
 617     def setup(cls):
 618         """Setup logging system"""
 619         conlog = logging.getLogger()
 620         conlog.setLevel(logging.DEBUG)
 621
 622         cls.con.setLevel(logging.WARNING)
 623         conlog.addHandler(cls.con)
 624
 625         fillog = logging.getLogger("backuplog")
 626         fillog.setLevel(logging.DEBUG)
 627
 628
 629 if __name__ == "__main__":
 630
 631     LogConf.setup()
 632
 633     conffn = "/etc/shbackup.conf"
 634     cmd = "list"
 635     mode = None
 636     epoch = None
 637
 638     i = 0
 639     while i < len(sys.argv)-1:
 640         i += 1
 641         opt = sys.argv[i]
 642
 643         if opt in ["-h", "--help"]:
 644             printUsage()
 645             exit(0)
 646
 647         elif opt in ["-c", "--conf"]:
 648             i += 1
 649             conffn = sys.argv[i]
 650
 651         elif opt in ["-V", "--version"]:
 652             print("shbackup " + __version__)
 653             exit(0)
 654
 655         elif opt in ["-v", "--verbose"]:
 656             LogConf.con.setLevel(logging.INFO)
 657
 658         elif opt in ["--verbosity"]:
 659             i += 1
 660             level = sys.argv[i]
 661             numlevel = getattr(logging, level.upper(), None)
 662             if not isinstance(numlevel, int):
 663                 raise ValueError('Invalid verbosity level: %s' % level)
 664             LogConf.con.setLevel(numlevel)
 665
 666         elif opt in ["-m", "--mode"]:
 667             i += 1
 668             mode = sys.argv[i]
 669             if not mode in Modes:
 670                 logging.error("Unknown mode '" + mode + "'.")
 671                 exit(1)
 672
 673         elif opt in ["-e", "--epoch"]:
 674             i += 1
 675             epoch = sys.argv[i]
 676
 677         elif opt in ["backup", "list", "prune"]:
 678             cmd = opt
 679
 680         else:
 681             logging.error("Unknown option: " + opt)
 682             exit(1)
 683
 684     try:
 685         man = BackupManager(conffn)
 686
 687         logging.debug("Config: " + str(man.conf))
 688
 689         if epoch!=None and not epoch in man.conf.epochs.keys():
 690             logging.error("Unknown epoch '" + epoch + "'.")
 691             exit(1)
 692
 693         if cmd == "backup":
 694             man.backup(epoch, mode)
 695
 696         if cmd == "list":
 697             for b in sorted(man.listOldBackups(), key=lambda b: b.date):
 698                 print(b.colAlignedString())
 699
 700         if cmd == "prune":
 701             man.prune()
 702
 703     except (Config.ReadError, configparser.Error) as e:
 704         logging.error("Error: " + e.message)
 705
 706
 707
 708