bin/filter-bash-environment.py

   1 #!/usr/bin/env python
   2 # Copyright 1999-2007 Gentoo Foundation
   3 # Distributed under the terms of the GNU General Public License v2
   4 # $Id$
   5
   6 import os, re, sys
   7
   8 egrep_compat_map = {
   9         "[:alnum:]" : r'\w',
  10         "[:digit:]" : r'\d',
  11         "[:space:]" : r'\s',
  12 }
  13
  14 here_doc_re = re.compile(r'.*\s<<[-]?(\w+)$')
  15 func_start_re = re.compile(r'^[-\w]+\s*\(\)\s*$')
  16 func_end_re = re.compile(r'^\}$')
  17
  18 var_assign_re = re.compile(r'(^|^declare\s+-\S+\s+|^export\s+)([^=\s]+)=("|\')?.*$')
  19 close_quote_re = re.compile(r'(\\"|"|\')\s*$')
  20
  21 def compile_egrep_pattern(s):
  22         for k, v in egrep_compat_map.iteritems():
  23                 s = s.replace(k, v)
  24         return re.compile(s)
  25
  26 def have_end_quote(quote, line):
  27         """
  28         Check if the line has an end quote (useful for handling multi-line
  29         quotes). This handles escaped double quotes that may occur at the
  30         end of a line. The posix spec does not allow escaping of single
  31         quotes inside of single quotes, so that case is not handled.
  32         """
  33         close_quote_match = close_quote_re.search(line)
  34         return close_quote_match is not None and \
  35                 close_quote_match.group(1) == quote
  36
  37 def filter_bash_environment(pattern, file_in, file_out):
  38         here_doc_delim = None
  39         in_func = None
  40         multi_line_quote = None
  41         multi_line_quote_filter = None
  42         for line in file_in:
  43                 if multi_line_quote is not None:
  44                         if not multi_line_quote_filter:
  45                                 file_out.write(line)
  46                         if have_end_quote(multi_line_quote, line):
  47                                 multi_line_quote = None
  48                                 multi_line_quote_filter = None
  49                         continue
  50                 if here_doc_delim is None and in_func is None:
  51                         var_assign_match = var_assign_re.match(line)
  52                         if var_assign_match is not None:
  53                                 quote = var_assign_match.group(3)
  54                                 filter_this = pattern.match(var_assign_match.group(2)) \
  55                                         is not None
  56                                 if quote is not None and not have_end_quote(quote, line):
  57                                         multi_line_quote = quote
  58                                         multi_line_quote_filter = filter_this
  59                                 if not filter_this:
  60                                         file_out.write(line)
  61                                 continue
  62                 if here_doc_delim is not None:
  63                         if here_doc_delim.match(line):
  64                                 here_doc_delim = None
  65                         file_out.write(line)
  66                         continue
  67                 here_doc = here_doc_re.match(line)
  68                 if here_doc is not None:
  69                         here_doc_delim = re.compile("^%s$" % here_doc.group(1))
  70                         file_out.write(line)
  71                         continue
  72                 # Note: here-documents are handled before functions since otherwise
  73                 # it would be possible for the content of a here-document to be
  74                 # mistaken as the end of a function.
  75                 if in_func:
  76                         if func_end_re.match(line) is not None:
  77                                 in_func = None
  78                         file_out.write(line)
  79                         continue
  80                 in_func = func_start_re.match(line)
  81                 if in_func is not None:
  82                         file_out.write(line)
  83                         continue
  84                 # This line is not recognized as part of a variable assignment,
  85                 # function definition, or here document, so just allow it to
  86                 # pass through.
  87                 file_out.write(line)
  88
  89 if __name__ == "__main__":
  90         description = "Filter out variable assignments for varable " + \
  91                 "names matching a given PATTERN " + \
  92                 "while leaving bash function definitions and here-documents " + \
  93                 "intact. The PATTERN is a space separated list of variable names" + \
  94                 " and it supports python regular expression syntax in addition to" + \
  95                 " [:alnum:], [:digit:], and [:space:] " + \
  96                 "character classes which will be automatically translated " + \
  97                 "for compatibility with egrep syntax."
  98         usage = "usage: %s PATTERN" % os.path.basename(sys.argv[0])
  99         from optparse import OptionParser
 100         parser = OptionParser(description=description, usage=usage)
 101         options, args = parser.parse_args(sys.argv[1:])
 102         if len(args) != 1:
 103                 parser.error("Missing required PATTERN argument.")
 104         file_in = sys.stdin
 105         file_out = sys.stdout
 106         var_pattern = args[0].split()
 107
 108         # Filter invalid variable names that are not supported by bash.
 109         var_pattern.append(r'\d.*')
 110         var_pattern.append(r'.*\W.*')
 111
 112         var_pattern = "^(%s)$" % "|".join(var_pattern)
 113         filter_bash_environment(
 114                 compile_egrep_pattern(var_pattern), file_in, file_out)
 115         file_out.flush()