tools/dev/scramble-tree.py

   1 #!/usr/bin/env python
   2 #
   3 # scramble-tree.py:  (See scramble-tree.py --help.)
   4 #
   5 # Makes multiple random file changes to a directory tree, for testing.
   6 #
   7 # This script will add some new files, remove some existing files, add
   8 # text to some existing files, and delete text from some existing
   9 # files.  It will also leave some files completely untouched.
  10 #
  11 # The exact set of changes made is always the same for identical trees,
  12 # where "identical" means the names of files and directories are the
  13 # same, and they are arranged in the same tree structure (the actual
  14 # contents of files may differ).  If two are not identical, the sets of
  15 # changes scramble-tree.py will make may differ arbitrarily.
  16 #
  17 # Directories named .svn/ and CVS/ are ignored.
  18 #
  19 # Example scenario, starting with a pristine Subversion working copy:
  20 #
  21 #   $ ls
  22 #   foo/
  23 #   $ svn st foo
  24 #   $ cp -r foo bar
  25 #   $ svn st bar
  26 #   $ scramble-tree.py foo
  27 #   $ svn st foo
  28 #   [... see lots of scary status output ...]
  29 #   $ scramble-tree.py bar
  30 #   [... see the exact same scary status output ...]
  31 #   $ scramble-tree.py foo
  32 #   [... see a new bunch of scary status output ...]
  33 #   $
  34
  35 import os
  36 import sys
  37 import getopt
  38 try:
  39   my_getopt = getopt.gnu_getopt
  40 except AttributeError:
  41   my_getopt = getopt.getopt
  42 import random
  43 import md5
  44 import base64
  45
  46
  47 class VCActions:
  48   def __init__(self):
  49     pass
  50   def add_file(self, path):
  51     """Add an existing file to version control."""
  52     pass
  53   def remove_file(self, path):
  54     """Remove an existing file from version control, and delete it."""
  55     pass
  56
  57
  58 class NoVCActions(VCActions):
  59   def remove_file(self, path):
  60     os.unlink(path)
  61
  62
  63 class CVSActions(VCActions):
  64   def add_file(self, path):
  65     cwd = os.getcwd()
  66     try:
  67       dirname, basename = os.path.split(path)
  68       os.chdir(os.path.join(cwd, dirname))
  69       os.system('cvs -Q add -m "Adding file to repository" "%s"' % (basename))
  70     finally:
  71       os.chdir(cwd)
  72   def remove_file(self, path):
  73     cwd = os.getcwd()
  74     try:
  75       dirname, basename = os.path.split(path)
  76       os.chdir(os.path.join(cwd, dirname))
  77       os.system('cvs -Q rm -f "%s"' % (basename))
  78     finally:
  79       os.chdir(cwd)
  80
  81
  82 class SVNActions(VCActions):
  83   def add_file(self, path):
  84     os.system('svn add --quiet "%s"' % (path))
  85   def remove_file(self, path):
  86     os.remove(path)
  87     os.system('svn rm --quiet --force "%s"' % (path))
  88
  89
  90 class hashDir:
  91   """Given a directory, creates a string containing all directories
  92   and files under that directory (sorted alphanumerically) and makes a
  93   base64-encoded md5 hash of the resulting string.  Call
  94   hashDir.gen_seed() to generate a seed value for this tree."""
  95
  96   def __init__(self, rootdir):
  97     self.allfiles = []
  98     os.path.walk(rootdir, self.walker_callback, len(rootdir))
  99
 100   def gen_seed(self):
 101     # Return a base64-encoded (kinda ... strip the '==\n' from the
 102     # end) MD5 hash of sorted tree listing.
 103     self.allfiles.sort()
 104     return base64.encodestring(md5.md5(''.join(self.allfiles)).digest())[:-3]
 105
 106   def walker_callback(self, baselen, dirname, fnames):
 107     if ((dirname == '.svn') or (dirname == 'CVS')):
 108       return
 109     self.allfiles.append(dirname[baselen:])
 110     for filename in fnames:
 111       path = os.path.join(dirname, filename)
 112       if not os.path.isdir(path):
 113         self.allfiles.append(path[baselen:])
 114
 115
 116 class Scrambler:
 117   def __init__(self, seed, vc_actions, dry_run, quiet):
 118     if not quiet:
 119       print 'SEED: ' + seed
 120
 121     self.rand = random.Random(seed)
 122     self.vc_actions = vc_actions
 123     self.dry_run = dry_run
 124     self.quiet = quiet
 125     self.ops = []  ### ["add" | "munge", path]
 126     self.greeking = """
 127 ======================================================================
 128 This is some text that was inserted into this file by the lovely and
 129 talented scramble-tree.py script.
 130 ======================================================================
 131 """
 132
 133   ### Helpers
 134   def shrink_list(self, list, remove_count):
 135     if len(list) <= remove_count:
 136       return []
 137     for i in range(remove_count):
 138       j = self.rand.randrange(len(list) - 1)
 139       del list[j]
 140     return list
 141
 142   def _make_new_file(self, dir):
 143     i = 0
 144     path = None
 145     for i in range(99999):
 146       path = os.path.join(dir, "newfile.%05d.txt" % i)
 147       if not os.path.exists(path):
 148         open(path, 'w').write(self.greeking)
 149         return path
 150     raise Exception("Ran out of unique new filenames in directory '%s'" % dir)
 151
 152   ### File Mungers
 153   def _mod_append_to_file(self, path):
 154     if not self.quiet:
 155       print 'append_to_file:', path
 156     if self.dry_run:
 157       return
 158     fh = open(path, "a")
 159     fh.write(self.greeking)
 160     fh.close()
 161
 162   def _mod_remove_from_file(self, path):
 163     if not self.quiet:
 164       print 'remove_from_file:', path
 165     if self.dry_run:
 166       return
 167     lines = self.shrink_list(open(path, "r").readlines(), 5)
 168     open(path, "w").writelines(lines)
 169
 170   def _mod_delete_file(self, path):
 171     if not self.quiet:
 172       print 'delete_file:', path
 173     if self.dry_run:
 174       return
 175     self.vc_actions.remove_file(path)
 176
 177   ### Public Interfaces
 178   def get_randomizer(self):
 179     return self.rand
 180
 181   def schedule_munge(self, path):
 182     self.ops.append(tuple(["munge", path]))
 183
 184   def schedule_addition(self, dir):
 185     self.ops.append(tuple(["add", dir]))
 186
 187   def enact(self, limit):
 188     num_ops = len(self.ops)
 189     if limit == 0:
 190       return
 191     elif limit > 0 and limit <= num_ops:
 192       self.ops = self.shrink_list(self.ops, num_ops - limit)
 193     for op, path in self.ops:
 194       if op == "add":
 195         path = self._make_new_file(path)
 196         if not self.quiet:
 197           print "add_file:", path
 198         if self.dry_run:
 199           return
 200         self.vc_actions.add_file(path)
 201       elif op == "munge":
 202         file_mungers = [self._mod_append_to_file,
 203                         self._mod_append_to_file,
 204                         self._mod_append_to_file,
 205                         self._mod_remove_from_file,
 206                         self._mod_remove_from_file,
 207                         self._mod_remove_from_file,
 208                         self._mod_delete_file,
 209                         ]
 210         self.rand.choice(file_mungers)(path)
 211
 212
 213 def usage(retcode=255):
 214   print 'Usage: %s [OPTIONS] DIRECTORY' % (sys.argv[0])
 215   print ''
 216   print 'Options:'
 217   print '    --help, -h  : Show this usage message.'
 218   print '    --seed ARG  : Use seed ARG to scramble the tree.'
 219   print '    --use-svn   : Use Subversion (as "svn") to perform file additions'
 220   print '                  and removals.'
 221   print '    --use-cvs   : Use CVS (as "cvs") to perform file additions'
 222   print '                  and removals.'
 223   print '    --dry-run   : Don\'t actually change the disk.'
 224   print '    --limit N   : Limit the scrambling to a maximum of N operations.'
 225   print '    --quiet, -q : Run in stealth mode!'
 226   sys.exit(retcode)
 227
 228
 229 def walker_callback(scrambler, dirname, fnames):
 230   if ((dirname.find('.svn') != -1) or dirname.find('CVS') != -1):
 231     return
 232   rand = scrambler.get_randomizer()
 233   if rand.randrange(5) == 1:
 234     scrambler.schedule_addition(dirname)
 235   for filename in fnames:
 236     path = os.path.join(dirname, filename)
 237     if not os.path.isdir(path) and rand.randrange(3) == 1:
 238       scrambler.schedule_munge(path)
 239
 240
 241 def main():
 242   seed = None
 243   vc_actions = NoVCActions()
 244   dry_run = 0
 245   quiet = 0
 246   limit = None
 247
 248   # Mm... option parsing.
 249   optlist, args = my_getopt(sys.argv[1:], "hq",
 250                             ['seed=', 'use-svn', 'use-cvs',
 251                              'help', 'quiet', 'dry-run', 'limit='])
 252   for opt, arg in optlist:
 253     if opt == '--help' or opt == '-h':
 254       usage(0)
 255     if opt == '--seed':
 256       seed = arg
 257     if opt == '--use-svn':
 258       vc_actions = SVNActions()
 259     if opt == '--use-cvs':
 260       vc_actions = CVSActions()
 261     if opt == '--dry-run':
 262       dry_run = 1
 263     if opt == '--limit':
 264       limit = int(arg)
 265     if opt == '--quiet' or opt == '-q':
 266       quiet = 1
 267
 268   # We need at least a path to work with, here.
 269   argc = len(args)
 270   if argc < 1 or argc > 1:
 271     usage()
 272   rootdir = args[0]
 273
 274   # If a seed wasn't provide, calculate one.
 275   if seed is None:
 276     seed = hashDir(rootdir).gen_seed()
 277   scrambler = Scrambler(seed, vc_actions, dry_run, quiet)
 278   os.path.walk(rootdir, walker_callback, scrambler)
 279   scrambler.enact(limit)
 280
 281 if __name__ == '__main__':
 282   main()