Home | History | Annotate | Download | only in Lib 
      1 """Utilities for comparing files and directories.
      2 
      3 Classes:
      4     dircmp
      5 
      6 Functions:
      7     cmp(f1, f2, shallow=1) -> int
      8     cmpfiles(a, b, common) -> ([], [], [])
      9 
     10 """
     11 
     12 import os
     13 import stat
     14 from itertools import ifilter, ifilterfalse, imap, izip
     15 
     16 __all__ = ["cmp","dircmp","cmpfiles"]
     17 
     18 _cache = {}
     19 BUFSIZE=8*1024
     20 
     21 def cmp(f1, f2, shallow=1):
     22     """Compare two files.
     23 
     24     Arguments:
     25 
     26     f1 -- First file name
     27 
     28     f2 -- Second file name
     29 
     30     shallow -- Just check stat signature (do not read the files).
     31                defaults to 1.
     32 
     33     Return value:
     34 
     35     True if the files are the same, False otherwise.
     36 
     37     This function uses a cache for past comparisons and the results,
     38     with a cache invalidation mechanism relying on stale signatures.
     39 
     40     """
     41 
     42     s1 = _sig(os.stat(f1))
     43     s2 = _sig(os.stat(f2))
     44     if s1[0] != stat.S_IFREG or s2[0] != stat.S_IFREG:
     45         return False
     46     if shallow and s1 == s2:
     47         return True
     48     if s1[1] != s2[1]:
     49         return False
     50 
     51     outcome = _cache.get((f1, f2, s1, s2))
     52     if outcome is None:
     53         outcome = _do_cmp(f1, f2)
     54         if len(_cache) > 100:      # limit the maximum size of the cache
     55             _cache.clear()
     56         _cache[f1, f2, s1, s2] = outcome
     57     return outcome
     58 
     59 def _sig(st):
     60     return (stat.S_IFMT(st.st_mode),
     61             st.st_size,
     62             st.st_mtime)
     63 
     64 def _do_cmp(f1, f2):
     65     bufsize = BUFSIZE
     66     with open(f1, 'rb') as fp1, open(f2, 'rb') as fp2:
     67         while True:
     68             b1 = fp1.read(bufsize)
     69             b2 = fp2.read(bufsize)
     70             if b1 != b2:
     71                 return False
     72             if not b1:
     73                 return True
     74 
     75 # Directory comparison class.
     76 #
     77 class dircmp:
     78     """A class that manages the comparison of 2 directories.
     79 
     80     dircmp(a,b,ignore=None,hide=None)
     81       A and B are directories.
     82       IGNORE is a list of names to ignore,
     83         defaults to ['RCS', 'CVS', 'tags'].
     84       HIDE is a list of names to hide,
     85         defaults to [os.curdir, os.pardir].
     86 
     87     High level usage:
     88       x = dircmp(dir1, dir2)
     89       x.report() -> prints a report on the differences between dir1 and dir2
     90        or
     91       x.report_partial_closure() -> prints report on differences between dir1
     92             and dir2, and reports on common immediate subdirectories.
     93       x.report_full_closure() -> like report_partial_closure,
     94             but fully recursive.
     95 
     96     Attributes:
     97      left_list, right_list: The files in dir1 and dir2,
     98         filtered by hide and ignore.
     99      common: a list of names in both dir1 and dir2.
    100      left_only, right_only: names only in dir1, dir2.
    101      common_dirs: subdirectories in both dir1 and dir2.
    102      common_files: files in both dir1 and dir2.
    103      common_funny: names in both dir1 and dir2 where the type differs between
    104         dir1 and dir2, or the name is not stat-able.
    105      same_files: list of identical files.
    106      diff_files: list of filenames which differ.
    107      funny_files: list of files which could not be compared.
    108      subdirs: a dictionary of dircmp objects, keyed by names in common_dirs.
    109      """
    110 
    111     def __init__(self, a, b, ignore=None, hide=None): # Initialize
    112         self.left = a
    113         self.right = b
    114         if hide is None:
    115             self.hide = [os.curdir, os.pardir] # Names never to be shown
    116         else:
    117             self.hide = hide
    118         if ignore is None:
    119             self.ignore = ['RCS', 'CVS', 'tags'] # Names ignored in comparison
    120         else:
    121             self.ignore = ignore
    122 
    123     def phase0(self): # Compare everything except common subdirectories
    124         self.left_list = _filter(os.listdir(self.left),
    125                                  self.hide+self.ignore)
    126         self.right_list = _filter(os.listdir(self.right),
    127                                   self.hide+self.ignore)
    128         self.left_list.sort()
    129         self.right_list.sort()
    130 
    131     def phase1(self): # Compute common names
    132         a = dict(izip(imap(os.path.normcase, self.left_list), self.left_list))
    133         b = dict(izip(imap(os.path.normcase, self.right_list), self.right_list))
    134         self.common = map(a.__getitem__, ifilter(b.__contains__, a))
    135         self.left_only = map(a.__getitem__, ifilterfalse(b.__contains__, a))
    136         self.right_only = map(b.__getitem__, ifilterfalse(a.__contains__, b))
    137 
    138     def phase2(self): # Distinguish files, directories, funnies
    139         self.common_dirs = []
    140         self.common_files = []
    141         self.common_funny = []
    142 
    143         for x in self.common:
    144             a_path = os.path.join(self.left, x)
    145             b_path = os.path.join(self.right, x)
    146 
    147             ok = 1
    148             try:
    149                 a_stat = os.stat(a_path)
    150             except os.error, why:
    151                 # print 'Can\'t stat', a_path, ':', why[1]
    152                 ok = 0
    153             try:
    154                 b_stat = os.stat(b_path)
    155             except os.error, why:
    156                 # print 'Can\'t stat', b_path, ':', why[1]
    157                 ok = 0
    158 
    159             if ok:
    160                 a_type = stat.S_IFMT(a_stat.st_mode)
    161                 b_type = stat.S_IFMT(b_stat.st_mode)
    162                 if a_type != b_type:
    163                     self.common_funny.append(x)
    164                 elif stat.S_ISDIR(a_type):
    165                     self.common_dirs.append(x)
    166                 elif stat.S_ISREG(a_type):
    167                     self.common_files.append(x)
    168                 else:
    169                     self.common_funny.append(x)
    170             else:
    171                 self.common_funny.append(x)
    172 
    173     def phase3(self): # Find out differences between common files
    174         xx = cmpfiles(self.left, self.right, self.common_files)
    175         self.same_files, self.diff_files, self.funny_files = xx
    176 
    177     def phase4(self): # Find out differences between common subdirectories
    178         # A new dircmp object is created for each common subdirectory,
    179         # these are stored in a dictionary indexed by filename.
    180         # The hide and ignore properties are inherited from the parent
    181         self.subdirs = {}
    182         for x in self.common_dirs:
    183             a_x = os.path.join(self.left, x)
    184             b_x = os.path.join(self.right, x)
    185             self.subdirs[x]  = dircmp(a_x, b_x, self.ignore, self.hide)
    186 
    187     def phase4_closure(self): # Recursively call phase4() on subdirectories
    188         self.phase4()
    189         for sd in self.subdirs.itervalues():
    190             sd.phase4_closure()
    191 
    192     def report(self): # Print a report on the differences between a and b
    193         # Output format is purposely lousy
    194         print 'diff', self.left, self.right
    195         if self.left_only:
    196             self.left_only.sort()
    197             print 'Only in', self.left, ':', self.left_only
    198         if self.right_only:
    199             self.right_only.sort()
    200             print 'Only in', self.right, ':', self.right_only
    201         if self.same_files:
    202             self.same_files.sort()
    203             print 'Identical files :', self.same_files
    204         if self.diff_files:
    205             self.diff_files.sort()
    206             print 'Differing files :', self.diff_files
    207         if self.funny_files:
    208             self.funny_files.sort()
    209             print 'Trouble with common files :', self.funny_files
    210         if self.common_dirs:
    211             self.common_dirs.sort()
    212             print 'Common subdirectories :', self.common_dirs
    213         if self.common_funny:
    214             self.common_funny.sort()
    215             print 'Common funny cases :', self.common_funny
    216 
    217     def report_partial_closure(self): # Print reports on self and on subdirs
    218         self.report()
    219         for sd in self.subdirs.itervalues():
    220             print
    221             sd.report()
    222 
    223     def report_full_closure(self): # Report on self and subdirs recursively
    224         self.report()
    225         for sd in self.subdirs.itervalues():
    226             print
    227             sd.report_full_closure()
    228 
    229     methodmap = dict(subdirs=phase4,
    230                      same_files=phase3, diff_files=phase3, funny_files=phase3,
    231                      common_dirs = phase2, common_files=phase2, common_funny=phase2,
    232                      common=phase1, left_only=phase1, right_only=phase1,
    233                      left_list=phase0, right_list=phase0)
    234 
    235     def __getattr__(self, attr):
    236         if attr not in self.methodmap:
    237             raise AttributeError, attr
    238         self.methodmap[attr](self)
    239         return getattr(self, attr)
    240 
    241 def cmpfiles(a, b, common, shallow=1):
    242     """Compare common files in two directories.
    243 
    244     a, b -- directory names
    245     common -- list of file names found in both directories
    246     shallow -- if true, do comparison based solely on stat() information
    247 
    248     Returns a tuple of three lists:
    249       files that compare equal
    250       files that are different
    251       filenames that aren't regular files.
    252 
    253     """
    254     res = ([], [], [])
    255     for x in common:
    256         ax = os.path.join(a, x)
    257         bx = os.path.join(b, x)
    258         res[_cmp(ax, bx, shallow)].append(x)
    259     return res
    260 
    261 
    262 # Compare two files.
    263 # Return:
    264 #       0 for equal
    265 #       1 for different
    266 #       2 for funny cases (can't stat, etc.)
    267 #
    268 def _cmp(a, b, sh, abs=abs, cmp=cmp):
    269     try:
    270         return not abs(cmp(a, b, sh))
    271     except (os.error, IOError):
    272         return 2
    273 
    274 
    275 # Return a copy with items that occur in skip removed.
    276 #
    277 def _filter(flist, skip):
    278     return list(ifilterfalse(skip.__contains__, flist))
    279 
    280 
    281 # Demonstration and testing.
    282 #
    283 def demo():
    284     import sys
    285     import getopt
    286     options, args = getopt.getopt(sys.argv[1:], 'r')
    287     if len(args) != 2:
    288         raise getopt.GetoptError('need exactly two args', None)
    289     dd = dircmp(args[0], args[1])
    290     if ('-r', '') in options:
    291         dd.report_full_closure()
    292     else:
    293         dd.report()
    294 
    295 if __name__ == '__main__':
    296     demo()
    297