asure.py

   1 #! /usr/bin/env python
   2
   3 # Directory integrity scanner.
   4
   5 from stat import *
   6 import os
   7 import sys
   8 from os.path import join
   9
  10 from cPickle import dump, load
  11 import gzip
  12
  13 def walk(top):
  14     """Root of directory generator"""
  15     topstat = os.lstat(top)
  16     for x in walker(top, '.', topstat):
  17         yield x
  18
  19 def walker(path, name, topstat):
  20     """Directory tree generator.
  21
  22     At one point, this started as a copy of os.walk from Python's
  23     library.  Even the arguments are different now.
  24     """
  25
  26     try:
  27         names = os.listdir(path)
  28     except OSError:
  29         sys.stderr.write("Warning, can't read dir: %s\n" % path)
  30         return
  31
  32     # The verification algorithm requires the names to be sorted.
  33     names.sort()
  34
  35     # Stat each name found, and put the result in one of two lists.
  36     dirs, nondirs = [], []
  37     for onename in names:
  38         st = os.lstat(join(path, onename))
  39         if S_ISDIR(st.st_mode):
  40             dirs.append((onename, st))
  41         else:
  42             nondirs.append((onename, st))
  43
  44     # Indicate "entering" the directory.
  45     yield 'd', name
  46
  47     # Then recursively walk into all of the subdirectories.
  48     for (onename, st) in dirs:
  49         subpath = join(path, onename)
  50         if st.st_dev == topstat.st_dev:
  51             for x in walker(subpath, onename, topstat):
  52                 yield x
  53
  54     # Then yield each entry that is not a subdirectory.
  55     for (onename, st) in nondirs:
  56         yield '-', onename
  57
  58     # Last, yield the leaving.
  59     yield ('u',)
  60
  61 def compare_trees(prior, cur):
  62     """Compare two scanned trees."""
  63     a = prior.next()
  64     if a[0] != 'd':
  65         raise "Scan doesn't start with a directory"
  66     b = cur.next()
  67     if b[0] != 'd':
  68         raise "Tree walk doesn't start with a directory"
  69
  70     # We don't concern ourselves with whether the names are the same
  71     # at this point.
  72     comp_walk(prior, cur)
  73
  74 def comp_walk(prior, cur, depth=1):
  75     """Inside directory.
  76
  77     Recursively walks both the "prior" and "cur" directories,
  78     comparing the trees found inside.  Returns when each has left this
  79     directory."""
  80
  81     a = prior.next()
  82     b = cur.next()
  83     while True:
  84
  85         # print "Comparing (%d) %s and %s" % (depth, a, b)
  86
  87         # Both are 'leave' nodes.
  88         if a[0] == 'u' and b[0] == 'u':
  89             # print "...leave"
  90             return
  91
  92         if a[0] == 'd' and b[0] == 'd':
  93             # Both are looking at a child subdirectory.
  94
  95             if a[1] == b[1]:
  96                 # Same name, just walk this tree.
  97                 # print "...same dir, enter"
  98                 comp_walk(prior, cur, depth+1)
  99                 a = prior.next()
 100                 b = cur.next()
 101                 continue
 102
 103             elif a[1] < b[1]:
 104                 # A directory has been deleted.
 105                 print "Delete dir: %s" % a[1]
 106                 consume_dir(prior)
 107                 a = prior.next()
 108                 continue
 109
 110             else:
 111                 # A directory has been added.
 112                 print "Add dir: %s" % b[1]
 113                 consume_dir(cur)
 114                 b = cur.next()
 115                 continue
 116
 117         elif a[0] == '-' and b[0] == '-':
 118             # Both are looking at a non-dir.
 119             if a[1] == b[1]:
 120                 # Same name, all is well.
 121                 # print "...same file"
 122                 a = prior.next()
 123                 b = cur.next()
 124                 continue
 125
 126             elif a[1] < b[1]:
 127                 print "Delete nondir: %s" % a[1]
 128                 a = prior.next()
 129                 continue
 130
 131             else:
 132                 print "Add nondir: %s" % b[1]
 133                 b = cur.next()
 134                 continue
 135
 136         elif a[0] == '-' and b[0] == 'u':
 137             print "Delete nondir: %s" % a[1]
 138             a = prior.next()
 139             continue
 140
 141         elif a[0] == 'u' and b[0] == '-':
 142             print "New nondir: %s" % b[1]
 143             b = cur.next()
 144             continue
 145
 146         elif a[0] == 'd' and b[0] == '-':
 147             print "Delete dir: %s" % a[1]
 148             consume_dir(prior)
 149             a = prior.next()
 150             continue
 151
 152         elif (a[0] == '-' or a[0] == 'u') and b[0] == 'd':
 153             print "Add dir: %s" % b[1]
 154             consume_dir(cur)
 155             b = cur.next()
 156             continue
 157
 158         else:
 159             print "Unhandled case: prior: %s, cur: %s" % (a[0], b[0])
 160             sys.exit(2)
 161
 162 def consume_dir(iter):
 163     """Consume entries until this directory has been exhausted"""
 164     while True:
 165         a = iter.next()
 166         if a[0] == 'u':
 167             return
 168         elif a[0] == 'd':
 169             consume_dir(iter)
 170
 171 version = 'Asure scan version 1.0'
 172
 173 def reader(path):
 174     """Iterate over a previously written dump"""
 175     fd = gzip.open(path, 'rb')
 176     vers = load(fd)
 177     if version != vers:
 178         raise "incompatible version of asure file"
 179     try:
 180         while True:
 181             yield load(fd)
 182     except EOFError:
 183         return
 184
 185 def writer(path, tmppath, iter):
 186     """Write the given item (probably assembled iterator)"""
 187     fd = gzip.open(tmppath, 'wb')
 188     dump(version, fd, -1)
 189     for item in iter:
 190         dump(item, fd, -1)
 191     fd.close
 192     os.rename(tmppath, path)
 193
 194 def fresh_scan():
 195     """Perform a fresh scan of the filesystem"""
 196     writer('asure.dat.gz', 'asure.0.gz', walk('.'))
 197
 198 def check_scan():
 199     """Perform a scan of the filesystem, and compare it with the scan
 200     file.  reports differences."""
 201     prior = reader('asure.dat.gz')
 202     cur = walk('.')
 203     compare_trees(prior, cur)
 204
 205 def main(argv):
 206     if len(argv) != 1:
 207         usage()
 208     if argv[0] == 'scan':
 209         fresh_scan()
 210     elif argv[0] == 'update':
 211         print "Update"
 212     elif argv[0] == 'check':
 213         check_scan()
 214     elif argv[0] == 'show':
 215         for i in reader('asure.dat.gz'):
 216             print i
 217
 218 def usage():
 219     print "Usage: asure {scan|update|check}"
 220     sys.exit(1)
 221
 222 if __name__ == '__main__':
 223     "Test this"
 224     main(sys.argv[1:])
 225     #for info in walk('/home/davidb/wd/asure'):
 226     #print info