"""Utilities for comparing files and directories.

[0] Fix | Delete

[1] Fix | Delete

Classes:

[2] Fix | Delete

dircmp

[3] Fix | Delete

[4] Fix | Delete

Functions:

[5] Fix | Delete

cmp(f1, f2, shallow=1) -> int

[6] Fix | Delete

cmpfiles(a, b, common) -> ([], [], [])

[7] Fix | Delete

[8] Fix | Delete

"""

[9] Fix | Delete

[10] Fix | Delete

import os

[11] Fix | Delete

import stat

[12] Fix | Delete

from itertools import ifilter, ifilterfalse, imap, izip

[13] Fix | Delete

[14] Fix | Delete

__all__ = ["cmp","dircmp","cmpfiles"]

[15] Fix | Delete

[16] Fix | Delete

_cache = {}

[17] Fix | Delete

BUFSIZE=8*1024

[18] Fix | Delete

[19] Fix | Delete

def cmp(f1, f2, shallow=1):

[20] Fix | Delete

"""Compare two files.

[21] Fix | Delete

[22] Fix | Delete

Arguments:

[23] Fix | Delete

[24] Fix | Delete

f1 -- First file name

[25] Fix | Delete

[26] Fix | Delete

f2 -- Second file name

[27] Fix | Delete

[28] Fix | Delete

shallow -- Just check stat signature (do not read the files).

[29] Fix | Delete

defaults to 1.

[30] Fix | Delete

[31] Fix | Delete

Return value:

[32] Fix | Delete

[33] Fix | Delete

True if the files are the same, False otherwise.

[34] Fix | Delete

[35] Fix | Delete

This function uses a cache for past comparisons and the results,

[36] Fix | Delete

with a cache invalidation mechanism relying on stale signatures.

[37] Fix | Delete

[38] Fix | Delete

"""

[39] Fix | Delete

[40] Fix | Delete

s1 = _sig(os.stat(f1))

[41] Fix | Delete

s2 = _sig(os.stat(f2))

[42] Fix | Delete

if s1[0] != stat.S_IFREG or s2[0] != stat.S_IFREG:

[43] Fix | Delete

return False

[44] Fix | Delete

if shallow and s1 == s2:

[45] Fix | Delete

return True

[46] Fix | Delete

if s1[1] != s2[1]:

[47] Fix | Delete

return False

[48] Fix | Delete

[49] Fix | Delete

outcome = _cache.get((f1, f2, s1, s2))

[50] Fix | Delete

if outcome is None:

[51] Fix | Delete

outcome = _do_cmp(f1, f2)

[52] Fix | Delete

if len(_cache) > 100: # limit the maximum size of the cache

[53] Fix | Delete

_cache.clear()

[54] Fix | Delete

_cache[f1, f2, s1, s2] = outcome

[55] Fix | Delete

return outcome

[56] Fix | Delete

[57] Fix | Delete

def _sig(st):

[58] Fix | Delete

return (stat.S_IFMT(st.st_mode),

[59] Fix | Delete

st.st_size,

[60] Fix | Delete

st.st_mtime)

[61] Fix | Delete

[62] Fix | Delete

def _do_cmp(f1, f2):

[63] Fix | Delete

bufsize = BUFSIZE

[64] Fix | Delete

with open(f1, 'rb') as fp1, open(f2, 'rb') as fp2:

[65] Fix | Delete

while True:

[66] Fix | Delete

b1 = fp1.read(bufsize)

[67] Fix | Delete

b2 = fp2.read(bufsize)

[68] Fix | Delete

if b1 != b2:

[69] Fix | Delete

return False

[70] Fix | Delete

if not b1:

[71] Fix | Delete

return True

[72] Fix | Delete

[73] Fix | Delete

# Directory comparison class.

[74] Fix | Delete

#

[75] Fix | Delete

class dircmp:

[76] Fix | Delete

"""A class that manages the comparison of 2 directories.

[77] Fix | Delete

[78] Fix | Delete

dircmp(a,b,ignore=None,hide=None)

[79] Fix | Delete

A and B are directories.

[80] Fix | Delete

IGNORE is a list of names to ignore,

[81] Fix | Delete

defaults to ['RCS', 'CVS', 'tags'].

[82] Fix | Delete

HIDE is a list of names to hide,

[83] Fix | Delete

defaults to [os.curdir, os.pardir].

[84] Fix | Delete

[85] Fix | Delete

High level usage:

[86] Fix | Delete

x = dircmp(dir1, dir2)

[87] Fix | Delete

x.report() -> prints a report on the differences between dir1 and dir2

[88] Fix | Delete

or

[89] Fix | Delete

x.report_partial_closure() -> prints report on differences between dir1

[90] Fix | Delete

and dir2, and reports on common immediate subdirectories.

[91] Fix | Delete

x.report_full_closure() -> like report_partial_closure,

[92] Fix | Delete

but fully recursive.

[93] Fix | Delete

[94] Fix | Delete

Attributes:

[95] Fix | Delete

left_list, right_list: The files in dir1 and dir2,

[96] Fix | Delete

filtered by hide and ignore.

[97] Fix | Delete

common: a list of names in both dir1 and dir2.

[98] Fix | Delete

left_only, right_only: names only in dir1, dir2.

[99] Fix | Delete

common_dirs: subdirectories in both dir1 and dir2.

[100] Fix | Delete

common_files: files in both dir1 and dir2.

[101] Fix | Delete

common_funny: names in both dir1 and dir2 where the type differs between

[102] Fix | Delete

dir1 and dir2, or the name is not stat-able.

[103] Fix | Delete

same_files: list of identical files.

[104] Fix | Delete

diff_files: list of filenames which differ.

[105] Fix | Delete

funny_files: list of files which could not be compared.

[106] Fix | Delete

subdirs: a dictionary of dircmp objects, keyed by names in common_dirs.

[107] Fix | Delete

"""

[108] Fix | Delete

[109] Fix | Delete

def __init__(self, a, b, ignore=None, hide=None): # Initialize

[110] Fix | Delete

self.left = a

[111] Fix | Delete

self.right = b

[112] Fix | Delete

if hide is None:

[113] Fix | Delete

self.hide = [os.curdir, os.pardir] # Names never to be shown

[114] Fix | Delete

else:

[115] Fix | Delete

self.hide = hide

[116] Fix | Delete

if ignore is None:

[117] Fix | Delete

self.ignore = ['RCS', 'CVS', 'tags'] # Names ignored in comparison

[118] Fix | Delete

else:

[119] Fix | Delete

self.ignore = ignore

[120] Fix | Delete

[121] Fix | Delete

def phase0(self): # Compare everything except common subdirectories

[122] Fix | Delete

self.left_list = _filter(os.listdir(self.left),

[123] Fix | Delete

self.hide+self.ignore)

[124] Fix | Delete

self.right_list = _filter(os.listdir(self.right),

[125] Fix | Delete

self.hide+self.ignore)

[126] Fix | Delete

self.left_list.sort()

[127] Fix | Delete

self.right_list.sort()

[128] Fix | Delete

[129] Fix | Delete

def phase1(self): # Compute common names

[130] Fix | Delete

a = dict(izip(imap(os.path.normcase, self.left_list), self.left_list))

[131] Fix | Delete

b = dict(izip(imap(os.path.normcase, self.right_list), self.right_list))

[132] Fix | Delete

self.common = map(a.__getitem__, ifilter(b.__contains__, a))

[133] Fix | Delete

self.left_only = map(a.__getitem__, ifilterfalse(b.__contains__, a))

[134] Fix | Delete

self.right_only = map(b.__getitem__, ifilterfalse(a.__contains__, b))

[135] Fix | Delete

[136] Fix | Delete

def phase2(self): # Distinguish files, directories, funnies

[137] Fix | Delete

self.common_dirs = []

[138] Fix | Delete

self.common_files = []

[139] Fix | Delete

self.common_funny = []

[140] Fix | Delete

[141] Fix | Delete

for x in self.common:

[142] Fix | Delete

a_path = os.path.join(self.left, x)

[143] Fix | Delete

b_path = os.path.join(self.right, x)

[144] Fix | Delete

[145] Fix | Delete

ok = 1

[146] Fix | Delete

try:

[147] Fix | Delete

a_stat = os.stat(a_path)

[148] Fix | Delete

except os.error, why:

[149] Fix | Delete

# print 'Can\'t stat', a_path, ':', why[1]

[150] Fix | Delete

ok = 0

[151] Fix | Delete

try:

[152] Fix | Delete

b_stat = os.stat(b_path)

[153] Fix | Delete

except os.error, why:

[154] Fix | Delete

# print 'Can\'t stat', b_path, ':', why[1]

[155] Fix | Delete

ok = 0

[156] Fix | Delete

[157] Fix | Delete

if ok:

[158] Fix | Delete

a_type = stat.S_IFMT(a_stat.st_mode)

[159] Fix | Delete

b_type = stat.S_IFMT(b_stat.st_mode)

[160] Fix | Delete

if a_type != b_type:

[161] Fix | Delete

self.common_funny.append(x)

[162] Fix | Delete

elif stat.S_ISDIR(a_type):

[163] Fix | Delete

self.common_dirs.append(x)

[164] Fix | Delete

elif stat.S_ISREG(a_type):

[165] Fix | Delete

self.common_files.append(x)

[166] Fix | Delete

else:

[167] Fix | Delete

self.common_funny.append(x)

[168] Fix | Delete

else:

[169] Fix | Delete

self.common_funny.append(x)

[170] Fix | Delete

[171] Fix | Delete

def phase3(self): # Find out differences between common files

[172] Fix | Delete

xx = cmpfiles(self.left, self.right, self.common_files)

[173] Fix | Delete

self.same_files, self.diff_files, self.funny_files = xx

[174] Fix | Delete

[175] Fix | Delete

def phase4(self): # Find out differences between common subdirectories

[176] Fix | Delete

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

[177] Fix | Delete

# these are stored in a dictionary indexed by filename.

[178] Fix | Delete

# The hide and ignore properties are inherited from the parent

[179] Fix | Delete

self.subdirs = {}

[180] Fix | Delete

for x in self.common_dirs:

[181] Fix | Delete

a_x = os.path.join(self.left, x)

[182] Fix | Delete

b_x = os.path.join(self.right, x)

[183] Fix | Delete

self.subdirs[x] = dircmp(a_x, b_x, self.ignore, self.hide)

[184] Fix | Delete

[185] Fix | Delete

def phase4_closure(self): # Recursively call phase4() on subdirectories

[186] Fix | Delete

self.phase4()

[187] Fix | Delete

for sd in self.subdirs.itervalues():

[188] Fix | Delete

sd.phase4_closure()

[189] Fix | Delete

[190] Fix | Delete

def report(self): # Print a report on the differences between a and b

[191] Fix | Delete

# Output format is purposely lousy

[192] Fix | Delete

print 'diff', self.left, self.right

[193] Fix | Delete

if self.left_only:

[194] Fix | Delete

self.left_only.sort()

[195] Fix | Delete

print 'Only in', self.left, ':', self.left_only

[196] Fix | Delete

if self.right_only:

[197] Fix | Delete

self.right_only.sort()

[198] Fix | Delete

print 'Only in', self.right, ':', self.right_only

[199] Fix | Delete

if self.same_files:

[200] Fix | Delete

self.same_files.sort()

[201] Fix | Delete

print 'Identical files :', self.same_files

[202] Fix | Delete

if self.diff_files:

[203] Fix | Delete

self.diff_files.sort()

[204] Fix | Delete

print 'Differing files :', self.diff_files

[205] Fix | Delete

if self.funny_files:

[206] Fix | Delete

self.funny_files.sort()

[207] Fix | Delete

print 'Trouble with common files :', self.funny_files

[208] Fix | Delete

if self.common_dirs:

[209] Fix | Delete

self.common_dirs.sort()

[210] Fix | Delete

print 'Common subdirectories :', self.common_dirs

[211] Fix | Delete

if self.common_funny:

[212] Fix | Delete

self.common_funny.sort()

[213] Fix | Delete

print 'Common funny cases :', self.common_funny

[214] Fix | Delete

[215] Fix | Delete

def report_partial_closure(self): # Print reports on self and on subdirs

[216] Fix | Delete

self.report()

[217] Fix | Delete

for sd in self.subdirs.itervalues():

[218] Fix | Delete

print

[219] Fix | Delete

sd.report()

[220] Fix | Delete

[221] Fix | Delete

def report_full_closure(self): # Report on self and subdirs recursively

[222] Fix | Delete

self.report()

[223] Fix | Delete

for sd in self.subdirs.itervalues():

[224] Fix | Delete

print

[225] Fix | Delete

sd.report_full_closure()

[226] Fix | Delete

[227] Fix | Delete

methodmap = dict(subdirs=phase4,

[228] Fix | Delete

same_files=phase3, diff_files=phase3, funny_files=phase3,

[229] Fix | Delete

common_dirs = phase2, common_files=phase2, common_funny=phase2,

[230] Fix | Delete

common=phase1, left_only=phase1, right_only=phase1,

[231] Fix | Delete

left_list=phase0, right_list=phase0)

[232] Fix | Delete

[233] Fix | Delete

def __getattr__(self, attr):

[234] Fix | Delete

if attr not in self.methodmap:

[235] Fix | Delete

raise AttributeError, attr

[236] Fix | Delete

self.methodmap[attr](self)

[237] Fix | Delete

return getattr(self, attr)

[238] Fix | Delete

[239] Fix | Delete

def cmpfiles(a, b, common, shallow=1):

[240] Fix | Delete

"""Compare common files in two directories.

[241] Fix | Delete

[242] Fix | Delete

a, b -- directory names

[243] Fix | Delete

common -- list of file names found in both directories

[244] Fix | Delete

shallow -- if true, do comparison based solely on stat() information

[245] Fix | Delete

[246] Fix | Delete

Returns a tuple of three lists:

[247] Fix | Delete

files that compare equal

[248] Fix | Delete

files that are different

[249] Fix | Delete

filenames that aren't regular files.

[250] Fix | Delete

[251] Fix | Delete

"""

[252] Fix | Delete

res = ([], [], [])

[253] Fix | Delete

for x in common:

[254] Fix | Delete

ax = os.path.join(a, x)

[255] Fix | Delete

bx = os.path.join(b, x)

[256] Fix | Delete

res[_cmp(ax, bx, shallow)].append(x)

[257] Fix | Delete

return res

[258] Fix | Delete

[259] Fix | Delete

[260] Fix | Delete

# Compare two files.

[261] Fix | Delete

# Return:

[262] Fix | Delete

# 0 for equal

[263] Fix | Delete

# 1 for different

[264] Fix | Delete

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

[265] Fix | Delete

#

[266] Fix | Delete

def _cmp(a, b, sh, abs=abs, cmp=cmp):

[267] Fix | Delete

try:

[268] Fix | Delete

return not abs(cmp(a, b, sh))

[269] Fix | Delete

except (os.error, IOError):

[270] Fix | Delete

return 2

[271] Fix | Delete

[272] Fix | Delete

[273] Fix | Delete

# Return a copy with items that occur in skip removed.

[274] Fix | Delete

#

[275] Fix | Delete

def _filter(flist, skip):

[276] Fix | Delete

return list(ifilterfalse(skip.__contains__, flist))

[277] Fix | Delete

[278] Fix | Delete

[279] Fix | Delete

# Demonstration and testing.

[280] Fix | Delete

#

[281] Fix | Delete

def demo():

[282] Fix | Delete

import sys

[283] Fix | Delete

import getopt

[284] Fix | Delete

options, args = getopt.getopt(sys.argv[1:], 'r')

[285] Fix | Delete

if len(args) != 2:

[286] Fix | Delete

raise getopt.GetoptError('need exactly two args', None)

[287] Fix | Delete

dd = dircmp(args[0], args[1])

[288] Fix | Delete

if ('-r', '') in options:

[289] Fix | Delete

dd.report_full_closure()

[290] Fix | Delete

else:

[291] Fix | Delete

dd.report()

[292] Fix | Delete

[293] Fix | Delete

if __name__ == '__main__':

[294] Fix | Delete

demo()

[295] Fix | Delete

[296] Fix | Delete