Lib/lib-old/dircmp.py

   1 """A class to build directory diff tools on."""
   2
   3 import os
   4
   5 import dircache
   6 import cmpcache
   7 import statcache
   8 from stat import *
   9
  10 class dircmp:
  11     """Directory comparison class."""
  12
  13     def new(self, a, b):
  14         """Initialize."""
  15         self.a = a
  16         self.b = b
  17         # Properties that caller may change before calling self.run():
  18         self.hide = [os.curdir, os.pardir] # Names never to be shown
  19         self.ignore = ['RCS', 'tags'] # Names ignored in comparison
  20
  21         return self
  22
  23     def run(self):
  24         """Compare everything except common subdirectories."""
  25         self.a_list = filter(dircache.listdir(self.a), self.hide)
  26         self.b_list = filter(dircache.listdir(self.b), self.hide)
  27         self.a_list.sort()
  28         self.b_list.sort()
  29         self.phase1()
  30         self.phase2()
  31         self.phase3()
  32
  33     def phase1(self):
  34         """Compute common names."""
  35         self.a_only = []
  36         self.common = []
  37         for x in self.a_list:
  38             if x in self.b_list:
  39                 self.common.append(x)
  40             else:
  41                 self.a_only.append(x)
  42
  43         self.b_only = []
  44         for x in self.b_list:
  45             if x not in self.common:
  46                 self.b_only.append(x)
  47
  48     def phase2(self):
  49         """Distinguish files, directories, funnies."""
  50         self.common_dirs = []
  51         self.common_files = []
  52         self.common_funny = []
  53
  54         for x in self.common:
  55             a_path = os.path.join(self.a, x)
  56             b_path = os.path.join(self.b, x)
  57
  58             ok = 1
  59             try:
  60                 a_stat = statcache.stat(a_path)
  61             except os.error, why:
  62                 # print 'Can\'t stat', a_path, ':', why[1]
  63                 ok = 0
  64             try:
  65                 b_stat = statcache.stat(b_path)
  66             except os.error, why:
  67                 # print 'Can\'t stat', b_path, ':', why[1]
  68                 ok = 0
  69
  70             if ok:
  71                 a_type = S_IFMT(a_stat[ST_MODE])
  72                 b_type = S_IFMT(b_stat[ST_MODE])
  73                 if a_type != b_type:
  74                     self.common_funny.append(x)
  75                 elif S_ISDIR(a_type):
  76                     self.common_dirs.append(x)
  77                 elif S_ISREG(a_type):
  78                     self.common_files.append(x)
  79                 else:
  80                     self.common_funny.append(x)
  81             else:
  82                 self.common_funny.append(x)
  83
  84     def phase3(self):
  85         """Find out differences between common files."""
  86         xx = cmpfiles(self.a, self.b, self.common_files)
  87         self.same_files, self.diff_files, self.funny_files = xx
  88
  89     def phase4(self):
  90         """Find out differences between common subdirectories.
  91         A new dircmp object is created for each common subdirectory,
  92         these are stored in a dictionary indexed by filename.
  93         The hide and ignore properties are inherited from the parent."""
  94         self.subdirs = {}
  95         for x in self.common_dirs:
  96             a_x = os.path.join(self.a, x)
  97             b_x = os.path.join(self.b, x)
  98             self.subdirs[x] = newdd = dircmp().new(a_x, b_x)
  99             newdd.hide = self.hide
 100             newdd.ignore = self.ignore
 101             newdd.run()
 102
 103     def phase4_closure(self):
 104         """Recursively call phase4() on subdirectories."""
 105         self.phase4()
 106         for x in self.subdirs.keys():
 107             self.subdirs[x].phase4_closure()
 108
 109     def report(self):
 110         """Print a report on the differences between a and b."""
 111         # Assume that phases 1 to 3 have been executed
 112         # Output format is purposely lousy
 113         print 'diff', self.a, self.b
 114         if self.a_only:
 115             print 'Only in', self.a, ':', self.a_only
 116         if self.b_only:
 117             print 'Only in', self.b, ':', self.b_only
 118         if self.same_files:
 119             print 'Identical files :', self.same_files
 120         if self.diff_files:
 121             print 'Differing files :', self.diff_files
 122         if self.funny_files:
 123             print 'Trouble with common files :', self.funny_files
 124         if self.common_dirs:
 125             print 'Common subdirectories :', self.common_dirs
 126         if self.common_funny:
 127             print 'Common funny cases :', self.common_funny
 128
 129     def report_closure(self):
 130         """Print reports on self and on subdirs.
 131         If phase 4 hasn't been done, no subdir reports are printed."""
 132         self.report()
 133         try:
 134             x = self.subdirs
 135         except AttributeError:
 136             return # No subdirectories computed
 137         for x in self.subdirs.keys():
 138             print
 139             self.subdirs[x].report_closure()
 140
 141     def report_phase4_closure(self):
 142         """Report and do phase 4 recursively."""
 143         self.report()
 144         self.phase4()
 145         for x in self.subdirs.keys():
 146             print
 147             self.subdirs[x].report_phase4_closure()
 148
 149
 150 def cmpfiles(a, b, common):
 151     """Compare common files in two directories.
 152     Return:
 153         - files that compare equal
 154         - files that compare different
 155         - funny cases (can't stat etc.)"""
 156
 157     res = ([], [], [])
 158     for x in common:
 159         res[cmp(os.path.join(a, x), os.path.join(b, x))].append(x)
 160     return res
 161
 162
 163 def cmp(a, b):
 164     """Compare two files.
 165     Return:
 166         0 for equal
 167         1 for different
 168         2 for funny cases (can't stat, etc.)"""
 169
 170     try:
 171         if cmpcache.cmp(a, b): return 0
 172         return 1
 173     except os.error:
 174         return 2
 175
 176
 177 def filter(list, skip):
 178     """Return a copy with items that occur in skip removed."""
 179
 180     result = []
 181     for item in list:
 182         if item not in skip: result.append(item)
 183     return result
 184
 185
 186 def demo():
 187     """Demonstration and testing."""
 188
 189     import sys
 190     import getopt
 191     options, args = getopt.getopt(sys.argv[1:], 'r')
 192     if len(args) != 2:
 193         raise getopt.error, 'need exactly two args'
 194     dd = dircmp().new(args[0], args[1])
 195     dd.run()
 196     if ('-r', '') in options:
 197         dd.report_phase4_closure()
 198     else:
 199         dd.report()
 200
 201 if __name__ == "__main__":
 202     demo()