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     result = _cache.get((f1, f2))
     52     if result and (s1, s2) == result[:2]:
     53         return result[2]
     54     outcome = _do_cmp(f1, f2)
     55     _cache[f1, f2] = s1, s2, outcome
     56     return outcome
     57 
     58 def _sig(st):
     59     return (stat.S_IFMT(st.st_mode),
     60             st.st_size,
     61             st.st_mtime)
     62 
     63 def _do_cmp(f1, f2):
     64     bufsize = BUFSIZE
     65     with open(f1, 'rb') as fp1, open(f2, 'rb') as fp2:
     66         while True:
     67             b1 = fp1.read(bufsize)
     68             b2 = fp2.read(bufsize)
     69             if b1 != b2:
     70                 return False
     71             if not b1:
     72                 return True
     73 
     74 # Directory comparison class.

     75 #

     76 class dircmp:
     77     """A class that manages the comparison of 2 directories.
     78 
     79     dircmp(a,b,ignore=None,hide=None)
     80       A and B are directories.
     81       IGNORE is a list of names to ignore,
     82         defaults to ['RCS', 'CVS', 'tags'].
     83       HIDE is a list of names to hide,
     84         defaults to [os.curdir, os.pardir].
     85 
     86     High level usage:
     87       x = dircmp(dir1, dir2)
     88       x.report() -> prints a report on the differences between dir1 and dir2
     89        or
     90       x.report_partial_closure() -> prints report on differences between dir1
     91             and dir2, and reports on common immediate subdirectories.
     92       x.report_full_closure() -> like report_partial_closure,
     93             but fully recursive.
     94 
     95     Attributes:
     96      left_list, right_list: The files in dir1 and dir2,
     97         filtered by hide and ignore.
     98      common: a list of names in both dir1 and dir2.
     99      left_only, right_only: names only in dir1, dir2.
    100      common_dirs: subdirectories in both dir1 and dir2.
    101      common_files: files in both dir1 and dir2.
    102      common_funny: names in both dir1 and dir2 where the type differs between
    103         dir1 and dir2, or the name is not stat-able.
    104      same_files: list of identical files.
    105      diff_files: list of filenames which differ.
    106      funny_files: list of files which could not be compared.
    107      subdirs: a dictionary of dircmp objects, keyed by names in common_dirs.
    108      """
    109 
    110     def __init__(self, a, b, ignore=None, hide=None): # Initialize

    111         self.left = a
    112         self.right = b
    113         if hide is None:
    114             self.hide = [os.curdir, os.pardir] # Names never to be shown

    115         else:
    116             self.hide = hide
    117         if ignore is None:
    118             self.ignore = ['RCS', 'CVS', 'tags'] # Names ignored in comparison

    119         else:
    120             self.ignore = ignore
    121 
    122     def phase0(self): # Compare everything except common subdirectories

    123         self.left_list = _filter(os.listdir(self.left),
    124                                  self.hide+self.ignore)
    125         self.right_list = _filter(os.listdir(self.right),
    126                                   self.hide+self.ignore)
    127         self.left_list.sort()
    128         self.right_list.sort()
    129 
    130     def phase1(self): # Compute common names

    131         a = dict(izip(imap(os.path.normcase, self.left_list), self.left_list))
    132         b = dict(izip(imap(os.path.normcase, self.right_list), self.right_list))
    133         self.common = map(a.__getitem__, ifilter(b.__contains__, a))
    134         self.left_only = map(a.__getitem__, ifilterfalse(b.__contains__, a))
    135         self.right_only = map(b.__getitem__, ifilterfalse(a.__contains__, b))
    136 
    137     def phase2(self): # Distinguish files, directories, funnies

    138         self.common_dirs = []
    139         self.common_files = []
    140         self.common_funny = []
    141 
    142         for x in self.common:
    143             a_path = os.path.join(self.left, x)
    144             b_path = os.path.join(self.right, x)
    145 
    146             ok = 1
    147             try:
    148                 a_stat = os.stat(a_path)
    149             except os.error, why:
    150                 # print 'Can\'t stat', a_path, ':', why[1]

    151                 ok = 0
    152             try:
    153                 b_stat = os.stat(b_path)
    154             except os.error, why:
    155                 # print 'Can\'t stat', b_path, ':', why[1]

    156                 ok = 0
    157 
    158             if ok:
    159                 a_type = stat.S_IFMT(a_stat.st_mode)
    160                 b_type = stat.S_IFMT(b_stat.st_mode)
    161                 if a_type != b_type:
    162                     self.common_funny.append(x)
    163                 elif stat.S_ISDIR(a_type):
    164                     self.common_dirs.append(x)
    165                 elif stat.S_ISREG(a_type):
    166                     self.common_files.append(x)
    167                 else:
    168                     self.common_funny.append(x)
    169             else:
    170                 self.common_funny.append(x)
    171 
    172     def phase3(self): # Find out differences between common files

    173         xx = cmpfiles(self.left, self.right, self.common_files)
    174         self.same_files, self.diff_files, self.funny_files = xx
    175 
    176     def phase4(self): # Find out differences between common subdirectories

    177         # A new dircmp object is created for each common subdirectory,

    178         # these are stored in a dictionary indexed by filename.

    179         # The hide and ignore properties are inherited from the parent

    180         self.subdirs = {}
    181         for x in self.common_dirs:
    182             a_x = os.path.join(self.left, x)
    183             b_x = os.path.join(self.right, x)
    184             self.subdirs[x]  = dircmp(a_x, b_x, self.ignore, self.hide)
    185 
    186     def phase4_closure(self): # Recursively call phase4() on subdirectories

    187         self.phase4()
    188         for sd in self.subdirs.itervalues():
    189             sd.phase4_closure()
    190 
    191     def report(self): # Print a report on the differences between a and b

    192         # Output format is purposely lousy

    193         print 'diff', self.left, self.right
    194         if self.left_only:
    195             self.left_only.sort()
    196             print 'Only in', self.left, ':', self.left_only
    197         if self.right_only:
    198             self.right_only.sort()
    199             print 'Only in', self.right, ':', self.right_only
    200         if self.same_files:
    201             self.same_files.sort()
    202             print 'Identical files :', self.same_files
    203         if self.diff_files:
    204             self.diff_files.sort()
    205             print 'Differing files :', self.diff_files
    206         if self.funny_files:
    207             self.funny_files.sort()
    208             print 'Trouble with common files :', self.funny_files
    209         if self.common_dirs:
    210             self.common_dirs.sort()
    211             print 'Common subdirectories :', self.common_dirs
    212         if self.common_funny:
    213             self.common_funny.sort()
    214             print 'Common funny cases :', self.common_funny
    215 
    216     def report_partial_closure(self): # Print reports on self and on subdirs

    217         self.report()
    218         for sd in self.subdirs.itervalues():
    219             print
    220             sd.report()
    221 
    222     def report_full_closure(self): # Report on self and subdirs recursively

    223         self.report()
    224         for sd in self.subdirs.itervalues():
    225             print
    226             sd.report_full_closure()
    227 
    228     methodmap = dict(subdirs=phase4,
    229                      same_files=phase3, diff_files=phase3, funny_files=phase3,
    230                      common_dirs = phase2, common_files=phase2, common_funny=phase2,
    231                      common=phase1, left_only=phase1, right_only=phase1,
    232                      left_list=phase0, right_list=phase0)
    233 
    234     def __getattr__(self, attr):
    235         if attr not in self.methodmap:
    236             raise AttributeError, attr
    237         self.methodmap[attr](self)
    238         return getattr(self, attr)
    239 
    240 def cmpfiles(a, b, common, shallow=1):
    241     """Compare common files in two directories.
    242 
    243     a, b -- directory names
    244     common -- list of file names found in both directories
    245     shallow -- if true, do comparison based solely on stat() information
    246 
    247     Returns a tuple of three lists:
    248       files that compare equal
    249       files that are different
    250       filenames that aren't regular files.
    251 
    252     """
    253     res = ([], [], [])
    254     for x in common:
    255         ax = os.path.join(a, x)
    256         bx = os.path.join(b, x)
    257         res[_cmp(ax, bx, shallow)].append(x)
    258     return res
    259 
    260 
    261 # Compare two files.

    262 # Return:

    263 #       0 for equal

    264 #       1 for different

    265 #       2 for funny cases (can't stat, etc.)

    266 #

    267 def _cmp(a, b, sh, abs=abs, cmp=cmp):
    268     try:
    269         return not abs(cmp(a, b, sh))
    270     except os.error:
    271         return 2
    272 
    273 
    274 # Return a copy with items that occur in skip removed.

    275 #

    276 def _filter(flist, skip):
    277     return list(ifilterfalse(skip.__contains__, flist))
    278 
    279 
    280 # Demonstration and testing.

    281 #

    282 def demo():
    283     import sys
    284     import getopt
    285     options, args = getopt.getopt(sys.argv[1:], 'r')
    286     if len(args) != 2:
    287         raise getopt.GetoptError('need exactly two args', None)
    288     dd = dircmp(args[0], args[1])
    289     if ('-r', '') in options:
    290         dd.report_full_closure()
    291     else:
    292         dd.report()
    293 
    294 if __name__ == '__main__':
    295     demo()
    296