Lib/lib-old/cmpcache.py

   1 """Efficiently compare files, boolean outcome only (equal / not equal).
   2
   3 Tricks (used in this order):
   4     - Use the statcache module to avoid statting files more than once
   5     - Files with identical type, size & mtime are assumed to be clones
   6     - Files with different type or size cannot be identical
   7     - We keep a cache of outcomes of earlier comparisons
   8     - We don't fork a process to run 'cmp' but read the files ourselves
   9 """
  10
  11 import os
  12 from stat import *
  13 import statcache
  14
  15
  16 # The cache.
  17 #
  18 cache = {}
  19
  20
  21 def cmp(f1, f2, shallow=1):
  22     """Compare two files, use the cache if possible.
  23     May raise os.error if a stat or open of either fails.
  24     Return 1 for identical files, 0 for different.
  25     Raise exceptions if either file could not be statted, read, etc."""
  26     s1, s2 = sig(statcache.stat(f1)), sig(statcache.stat(f2))
  27     if not S_ISREG(s1[0]) or not S_ISREG(s2[0]):
  28         # Either is a not a plain file -- always report as different
  29         return 0
  30     if shallow and s1 == s2:
  31         # type, size & mtime match -- report same
  32         return 1
  33     if s1[:2] != s2[:2]: # Types or sizes differ, don't bother
  34         # types or sizes differ -- report different
  35         return 0
  36     # same type and size -- look in the cache
  37     key = f1 + ' ' + f2
  38     if cache.has_key(key):
  39         cs1, cs2, outcome = cache[key]
  40         # cache hit
  41         if s1 == cs1 and s2 == cs2:
  42             # cached signatures match
  43             return outcome
  44         # stale cached signature(s)
  45     # really compare
  46     outcome = do_cmp(f1, f2)
  47     cache[key] = s1, s2, outcome
  48     return outcome
  49
  50 def sig(st):
  51     """Return signature (i.e., type, size, mtime) from raw stat data."""
  52     return S_IFMT(st[ST_MODE]), st[ST_SIZE], st[ST_MTIME]
  53
  54 def do_cmp(f1, f2):
  55     """Compare two files, really."""
  56     #print '    cmp', f1, f2 # XXX remove when debugged
  57     bufsize = 8*1024 # Could be tuned
  58     fp1 = open(f1, 'rb')
  59     fp2 = open(f2, 'rb')
  60     while 1:
  61         b1 = fp1.read(bufsize)
  62         b2 = fp2.read(bufsize)
  63         if b1 != b2: return 0
  64         if not b1: return 1