/Tools/scripts/classfix.py

  1#! /usr/bin/env python
  2
  3# This script is obsolete -- it is kept for historical purposes only.
  4#
  5# Fix Python source files to use the new class definition syntax, i.e.,
  6# the syntax used in Python versions before 0.9.8:
  7#       class C() = base(), base(), ...: ...
  8# is changed to the current syntax:
  9#       class C(base, base, ...): ...
 10#
 11# The script uses heuristics to find class definitions that usually
 12# work but occasionally can fail; carefully check the output!
 13#
 14# Command line arguments are files or directories to be processed.
 15# Directories are searched recursively for files whose name looks
 16# like a python module.
 17# Symbolic links are always ignored (except as explicit directory
 18# arguments).  Of course, the original file is kept as a back-up
 19# (with a "~" attached to its name).
 20#
 21# Changes made are reported to stdout in a diff-like format.
 22#
 23# Undoubtedly you can do this using find and sed or perl, but this is
 24# a nice example of Python code that recurses down a directory tree
 25# and uses regular expressions.  Also note several subtleties like
 26# preserving the file's mode and avoiding to even write a temp file
 27# when no changes are needed for a file.
 28#
 29# NB: by changing only the function fixline() you can turn this
 30# into a program for a different change to Python programs...
 31
 32import sys
 33import re
 34import os
 35from stat import *
 36
 37err = sys.stderr.write
 38dbg = err
 39rep = sys.stdout.write
 40
 41def main():
 42    bad = 0
 43    if not sys.argv[1:]: # No arguments
 44        err('usage: ' + sys.argv[0] + ' file-or-directory ...\n')
 45        sys.exit(2)
 46    for arg in sys.argv[1:]:
 47        if os.path.isdir(arg):
 48            if recursedown(arg): bad = 1
 49        elif os.path.islink(arg):
 50            err(arg + ': will not process symbolic links\n')
 51            bad = 1
 52        else:
 53            if fix(arg): bad = 1
 54    sys.exit(bad)
 55
 56ispythonprog = re.compile('^[a-zA-Z0-9_]+\.py$')
 57def ispython(name):
 58    return ispythonprog.match(name) >= 0
 59
 60def recursedown(dirname):
 61    dbg('recursedown(%r)\n' % (dirname,))
 62    bad = 0
 63    try:
 64        names = os.listdir(dirname)
 65    except os.error, msg:
 66        err('%s: cannot list directory: %r\n' % (dirname, msg))
 67        return 1
 68    names.sort()
 69    subdirs = []
 70    for name in names:
 71        if name in (os.curdir, os.pardir): continue
 72        fullname = os.path.join(dirname, name)
 73        if os.path.islink(fullname): pass
 74        elif os.path.isdir(fullname):
 75            subdirs.append(fullname)
 76        elif ispython(name):
 77            if fix(fullname): bad = 1
 78    for fullname in subdirs:
 79        if recursedown(fullname): bad = 1
 80    return bad
 81
 82def fix(filename):
 83##  dbg('fix(%r)\n' % (filename,))
 84    try:
 85        f = open(filename, 'r')
 86    except IOError, msg:
 87        err('%s: cannot open: %r\n' % (filename, msg))
 88        return 1
 89    head, tail = os.path.split(filename)
 90    tempname = os.path.join(head, '@' + tail)
 91    g = None
 92    # If we find a match, we rewind the file and start over but
 93    # now copy everything to a temp file.
 94    lineno = 0
 95    while 1:
 96        line = f.readline()
 97        if not line: break
 98        lineno = lineno + 1
 99        while line[-2:] == '\\\n':
100            nextline = f.readline()
101            if not nextline: break
102            line = line + nextline
103            lineno = lineno + 1
104        newline = fixline(line)
105        if newline != line:
106            if g is None:
107                try:
108                    g = open(tempname, 'w')
109                except IOError, msg:
110                    f.close()
111                    err('%s: cannot create: %r\n' % (tempname, msg))
112                    return 1
113                f.seek(0)
114                lineno = 0
115                rep(filename + ':\n')
116                continue # restart from the beginning
117            rep(repr(lineno) + '\n')
118            rep('< ' + line)
119            rep('> ' + newline)
120        if g is not None:
121            g.write(newline)
122
123    # End of file
124    f.close()
125    if not g: return 0 # No changes
126
127    # Finishing touch -- move files
128
129    # First copy the file's mode to the temp file
130    try:
131        statbuf = os.stat(filename)
132        os.chmod(tempname, statbuf[ST_MODE] & 07777)
133    except os.error, msg:
134        err('%s: warning: chmod failed (%r)\n' % (tempname, msg))
135    # Then make a backup of the original file as filename~
136    try:
137        os.rename(filename, filename + '~')
138    except os.error, msg:
139        err('%s: warning: backup failed (%r)\n' % (filename, msg))
140    # Now move the temp file to the original file
141    try:
142        os.rename(tempname, filename)
143    except os.error, msg:
144        err('%s: rename failed (%r)\n' % (filename, msg))
145        return 1
146    # Return succes
147    return 0
148
149# This expression doesn't catch *all* class definition headers,
150# but it's pretty darn close.
151classexpr = '^([ \t]*class +[a-zA-Z0-9_]+) *( *) *((=.*)?):'
152classprog = re.compile(classexpr)
153
154# Expressions for finding base class expressions.
155baseexpr = '^ *(.*) *( *) *$'
156baseprog = re.compile(baseexpr)
157
158def fixline(line):
159    if classprog.match(line) < 0: # No 'class' keyword -- no change
160        return line
161
162    (a0, b0), (a1, b1), (a2, b2) = classprog.regs[:3]
163    # a0, b0 = Whole match (up to ':')
164    # a1, b1 = First subexpression (up to classname)
165    # a2, b2 = Second subexpression (=.*)
166    head = line[:b1]
167    tail = line[b0:] # Unmatched rest of line
168
169    if a2 == b2: # No base classes -- easy case
170        return head + ':' + tail
171
172    # Get rid of leading '='
173    basepart = line[a2+1:b2]
174
175    # Extract list of base expressions
176    bases = basepart.split(',')
177
178    # Strip trailing '()' from each base expression
179    for i in range(len(bases)):
180        if baseprog.match(bases[i]) >= 0:
181            x1, y1 = baseprog.regs[1]
182            bases[i] = bases[i][x1:y1]
183
184    # Join the bases back again and build the new line
185    basepart = ', '.join(bases)
186
187    return head + '(' + basepart + '):' + tail
188
189if __name__ == '__main__':
190    main()