tools/check_val_to_str.py

   1 #!/usr/bin/env python3
   2 # Wireshark - Network traffic analyzer
   3 # By Gerald Combs <gerald@wireshark.org>
   4 # Copyright 1998 Gerald Combs
   5 #
   6 # SPDX-License-Identifier: GPL-2.0-or-later
   7
   8 # Scan dissectors for calls to val_to_str() and friends,
   9 # checking for appropriate format specifier strings in
  10 # 'unknown' arg.
  11 # TODO:
  12 # - more detailed format specifier checking (check letter, that there is only 1)
  13 # - scan conformance (.cnf) files for ASN1 dissectors?
  14
  15 import os
  16 import re
  17 import subprocess
  18 import argparse
  19 import signal
  20
  21
  22 # Try to exit soon after Ctrl-C is pressed.
  23 should_exit = False
  24
  25 def signal_handler(sig, frame):
  26     global should_exit
  27     should_exit = True
  28     print('You pressed Ctrl+C - exiting')
  29
  30 signal.signal(signal.SIGINT, signal_handler)
  31
  32
  33 # Test for whether the given file was automatically generated.
  34 def isGeneratedFile(filename):
  35     # Check file exists - e.g. may have been deleted in a recent commit.
  36     if not os.path.exists(filename):
  37         return False
  38
  39     # Open file
  40     f_read = open(os.path.join(filename), 'r', encoding="utf8")
  41     lines_tested = 0
  42     for line in f_read:
  43         # The comment to say that its generated is near the top, so give up once
  44         # get a few lines down.
  45         if lines_tested > 10:
  46             f_read.close()
  47             return False
  48         if (line.find('Generated automatically') != -1 or
  49             line.find('Generated Automatically') != -1 or
  50             line.find('Autogenerated from') != -1 or
  51             line.find('is autogenerated') != -1 or
  52             line.find('automatically generated by Pidl') != -1 or
  53             line.find('Created by: The Qt Meta Object Compiler') != -1 or
  54             line.find('This file was generated') != -1 or
  55             line.find('This filter was automatically generated') != -1 or
  56             line.find('This file is auto generated, do not edit!') != -1 or
  57             line.find('This file is auto generated') != -1):
  58
  59             f_read.close()
  60             return True
  61         lines_tested = lines_tested + 1
  62
  63     # OK, looks like a hand-written file!
  64     f_read.close()
  65     return False
  66
  67
  68
  69 def removeComments(code_string):
  70     code_string = re.sub(re.compile(r"/\*.*?\*/",re.DOTALL ) ,"" ,code_string) # C-style comment
  71     code_string = re.sub(re.compile(r"//.*?\n" ) ,"" ,code_string)             # C++-style comment
  72     return code_string
  73
  74
  75 def is_dissector_file(filename):
  76     p = re.compile(r'.*(packet|file)-.*\.c')
  77     return p.match(filename)
  78
  79 def findDissectorFilesInFolder(folder, recursive=False):
  80     dissector_files = []
  81
  82     if recursive:
  83         for root, subfolders, files in os.walk(folder):
  84             for f in files:
  85                 if should_exit:
  86                     return
  87                 f = os.path.join(root, f)
  88                 dissector_files.append(f)
  89     else:
  90         for f in sorted(os.listdir(folder)):
  91             if should_exit:
  92                 return
  93             filename = os.path.join(folder, f)
  94             dissector_files.append(filename)
  95
  96     return [x for x in filter(is_dissector_file, dissector_files)]
  97
  98
  99
 100 warnings_found = 0
 101 errors_found = 0
 102
 103 # Check the given dissector file.
 104 def checkFile(filename, generated):
 105     global warnings_found
 106     global errors_found
 107
 108     # Check file exists - e.g. may have been deleted in a recent commit.
 109     if not os.path.exists(filename):
 110         print(filename, 'does not exist!')
 111         return
 112
 113     with open(filename, 'r', encoding="utf8") as f:
 114         contents = f.read()
 115
 116         # Remove comments so as not to trip up RE.
 117         contents = removeComments(contents)
 118
 119         matches =   re.finditer(r'(?<!try_)(?<!char_)(?<!bytes)(r?val_to_str(?:_ext|)(?:_const|))\(.*?,.*?,\s*(".*?\")\s*\)', contents)
 120         for m in matches:
 121             function = m.group(1)
 122             format_string = m.group(2)
 123
 124             # Ignore what appears to be a macro.
 125             if format_string.find('#') != -1:
 126                 continue
 127
 128             if function.endswith('_const'):
 129                 # These ones shouldn't have a specifier - its an error if they do.
 130                 # TODO: I suppose it could be escaped, but haven't seen this...
 131                 if format_string.find('%') != -1:
 132                     # This is an error as format specifier would show in app
 133                     print('Error:', filename, "  ", m.group(0),
 134                           '   - should not have specifiers in unknown string',
 135                           '(GENERATED)' if generated else '')
 136                     errors_found += 1
 137             else:
 138                 # These ones need to have a specifier, and it should be suitable for an int
 139                 count = format_string.count('%')
 140                 if count == 0:
 141                     print('Warning:', filename, "  ", m.group(0),
 142                           '   - should have suitable format specifier in unknown string (or use _const()?)',
 143                           '(GENERATED)' if generated else '')
 144                     warnings_found += 1
 145                 elif count > 1:
 146                     print('Warning:', filename, "  ", m.group(0),
 147                           '   - has more than one specifier?',
 148                           '(GENERATED)' if generated else '')
 149                 # TODO: check allowed specifiers (d, u, x, ?) and modifiers (0-9*) in re ?
 150                 if format_string.find('%s') != -1:
 151                     # This is an error as this likely causes a crash
 152                     print('Error:', filename, "  ", m.group(0),
 153                           '    - inappropriate format specifier in unknown string',
 154                           '(GENERATED)' if generated else '')
 155                     errors_found += 1
 156
 157
 158
 159 #################################################################
 160 # Main logic.
 161
 162 # command-line args.  Controls which dissector files should be checked.
 163 # If no args given, will scan all dissectors.
 164 parser = argparse.ArgumentParser(description='Check calls in dissectors')
 165 parser.add_argument('--file', action='append',
 166                     help='specify individual dissector file to test')
 167 parser.add_argument('--commits', action='store',
 168                     help='last N commits to check')
 169 parser.add_argument('--open', action='store_true',
 170                     help='check open files')
 171 parser.add_argument('--generated', action='store_true',
 172                     help='check generated files')
 173
 174 args = parser.parse_args()
 175
 176
 177 # Get files from wherever command-line args indicate.
 178 files = []
 179 if args.file:
 180     # Add specified file(s)
 181     for f in args.file:
 182         if not os.path.isfile(f) and not f.startswith('epan'):
 183             f = os.path.join('epan', 'dissectors', f)
 184         if not os.path.isfile(f):
 185             print('Chosen file', f, 'does not exist.')
 186             exit(1)
 187         else:
 188             files.append(f)
 189 elif args.commits:
 190     # Get files affected by specified number of commits.
 191     command = ['git', 'diff', '--name-only', 'HEAD~' + args.commits]
 192     files = [f.decode('utf-8')
 193              for f in subprocess.check_output(command).splitlines()]
 194     # Will examine dissector files only
 195     files = list(filter(lambda f : is_dissector_file(f), files))
 196 elif args.open:
 197     # Unstaged changes.
 198     command = ['git', 'diff', '--name-only']
 199     files = [f.decode('utf-8')
 200              for f in subprocess.check_output(command).splitlines()]
 201     # Only interested in dissector files.
 202     files = list(filter(lambda f : is_dissector_file(f), files))
 203     # Staged changes.
 204     command = ['git', 'diff', '--staged', '--name-only']
 205     files_staged = [f.decode('utf-8')
 206                     for f in subprocess.check_output(command).splitlines()]
 207     # Only interested in dissector files.
 208     files_staged = list(filter(lambda f : is_dissector_file(f), files_staged))
 209     for f in files_staged:
 210         if f not in files:
 211             files.append(f)
 212 else:
 213     # Find all dissector files from folder.
 214     files =  findDissectorFilesInFolder(os.path.join('epan', 'dissectors'))
 215     files += findDissectorFilesInFolder(os.path.join('plugins', 'epan'), recursive=True)
 216     files += findDissectorFilesInFolder(os.path.join('epan', 'dissectors', 'asn1'), recursive=True)
 217
 218
 219 # If scanning a subset of files, list them here.
 220 print('Examining:')
 221 if args.file or args.commits or args.open:
 222     if files:
 223         print(' '.join(files), '\n')
 224     else:
 225         print('No files to check.\n')
 226 else:
 227     print('All dissectors\n')
 228
 229
 230 # Now check the chosen files
 231 for f in files:
 232     if should_exit:
 233         exit(1)
 234     generated = isGeneratedFile(f)
 235     if args.generated or not generated:
 236         checkFile(f, generated)
 237
 238
 239 # Show summary.
 240 print(warnings_found, 'warnings found')
 241 if errors_found:
 242     print(errors_found, 'errors found')
 243     exit(1)