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