scripts/checkkconfigsymbols.py

   1 #!/usr/bin/env python2
   2
   3 """Find Kconfig symbols that are referenced but not defined."""
   4
   5 # (c) 2014-2015 Valentin Rothberg <Valentin.Rothberg@lip6.fr>
   6 # (c) 2014 Stefan Hengelein <stefan.hengelein@fau.de>
   7 #
   8 # Licensed under the terms of the GNU GPL License version 2
   9
  10
  11 import os
  12 import re
  13 import sys
  14 from subprocess import Popen, PIPE, STDOUT
  15 from optparse import OptionParser
  16
  17
  18 # regex expressions
  19 OPERATORS = r"&|\(|\)|\||\!"
  20 FEATURE = r"(?:\w*[A-Z0-9]\w*){2,}"
  21 DEF = r"^\s*(?:menu){,1}config\s+(" + FEATURE + r")\s*"
  22 EXPR = r"(?:" + OPERATORS + r"|\s|" + FEATURE + r")+"
  23 STMT = r"^\s*(?:if|select|depends\s+on)\s+" + EXPR
  24 SOURCE_FEATURE = r"(?:\W|\b)+[D]{,1}CONFIG_(" + FEATURE + r")"
  25
  26 # regex objects
  27 REGEX_FILE_KCONFIG = re.compile(r".*Kconfig[\.\w+\-]*$")
  28 REGEX_FEATURE = re.compile(r"(" + FEATURE + r")")
  29 REGEX_SOURCE_FEATURE = re.compile(SOURCE_FEATURE)
  30 REGEX_KCONFIG_DEF = re.compile(DEF)
  31 REGEX_KCONFIG_EXPR = re.compile(EXPR)
  32 REGEX_KCONFIG_STMT = re.compile(STMT)
  33 REGEX_KCONFIG_HELP = re.compile(r"^\s+(help|---help---)\s*$")
  34 REGEX_FILTER_FEATURES = re.compile(r"[A-Za-z0-9]$")
  35
  36
  37 def parse_options():
  38     """The user interface of this module."""
  39     usage = "%prog [options]\n\n"                                              \
  40             "Run this tool to detect Kconfig symbols that are referenced but " \
  41             "not defined in\nKconfig.  The output of this tool has the "       \
  42             "format \'Undefined symbol\\tFile list\'\n\n"                      \
  43             "If no option is specified, %prog will default to check your\n"    \
  44             "current tree.  Please note that specifying commits will "         \
  45             "\'git reset --hard\'\nyour current tree!  You may save "          \
  46             "uncommitted changes to avoid losing data."
  47
  48     parser = OptionParser(usage=usage)
  49
  50     parser.add_option('-c', '--commit', dest='commit', action='store',
  51                       default="",
  52                       help="Check if the specified commit (hash) introduces "
  53                            "undefined Kconfig symbols.")
  54
  55     parser.add_option('-d', '--diff', dest='diff', action='store',
  56                       default="",
  57                       help="Diff undefined symbols between two commits.  The "
  58                            "input format bases on Git log's "
  59                            "\'commmit1..commit2\'.")
  60
  61     parser.add_option('-i', '--ignore', dest='ignore', action='store',
  62                       default="",
  63                       help="Ignore files matching this pattern.  Note that "
  64                            "the pattern needs to be a Python regex.  To "
  65                            "ignore defconfigs, specify -i '.*defconfig'.")
  66
  67     parser.add_option('', '--force', dest='force', action='store_true',
  68                       default=False,
  69                       help="Reset current Git tree even when it's dirty.")
  70
  71     (opts, _) = parser.parse_args()
  72
  73     if opts.commit and opts.diff:
  74         sys.exit("Please specify only one option at once.")
  75
  76     if opts.diff and not re.match(r"^[\w\-\.]+\.\.[\w\-\.]+$", opts.diff):
  77         sys.exit("Please specify valid input in the following format: "
  78                  "\'commmit1..commit2\'")
  79
  80     if opts.commit or opts.diff:
  81         if not opts.force and tree_is_dirty():
  82             sys.exit("The current Git tree is dirty (see 'git status').  "
  83                      "Running this script may\ndelete important data since it "
  84                      "calls 'git reset --hard' for some performance\nreasons. "
  85                      " Please run this script in a clean Git tree or pass "
  86                      "'--force' if you\nwant to ignore this warning and "
  87                      "continue.")
  88
  89     if opts.ignore:
  90         try:
  91             re.match(opts.ignore, "this/is/just/a/test.c")
  92         except:
  93             sys.exit("Please specify a valid Python regex.")
  94
  95     return opts
  96
  97
  98 def main():
  99     """Main function of this module."""
 100     opts = parse_options()
 101
 102     if opts.commit or opts.diff:
 103         head = get_head()
 104
 105         # get commit range
 106         commit_a = None
 107         commit_b = None
 108         if opts.commit:
 109             commit_a = opts.commit + "~"
 110             commit_b = opts.commit
 111         elif opts.diff:
 112             split = opts.diff.split("..")
 113             commit_a = split[0]
 114             commit_b = split[1]
 115             undefined_a = {}
 116             undefined_b = {}
 117
 118         # get undefined items before the commit
 119         execute("git reset --hard %s" % commit_a)
 120         undefined_a = check_symbols(opts.ignore)
 121
 122         # get undefined items for the commit
 123         execute("git reset --hard %s" % commit_b)
 124         undefined_b = check_symbols(opts.ignore)
 125
 126         # report cases that are present for the commit but not before
 127         for feature in sorted(undefined_b):
 128             # feature has not been undefined before
 129             if not feature in undefined_a:
 130                 files = sorted(undefined_b.get(feature))
 131                 print "%s\t%s" % (feature, ", ".join(files))
 132             # check if there are new files that reference the undefined feature
 133             else:
 134                 files = sorted(undefined_b.get(feature) -
 135                                undefined_a.get(feature))
 136                 if files:
 137                     print "%s\t%s" % (feature, ", ".join(files))
 138
 139         # reset to head
 140         execute("git reset --hard %s" % head)
 141
 142     # default to check the entire tree
 143     else:
 144         undefined = check_symbols(opts.ignore)
 145         for feature in sorted(undefined):
 146             files = sorted(undefined.get(feature))
 147             print "%s\t%s" % (feature, ", ".join(files))
 148
 149
 150 def execute(cmd):
 151     """Execute %cmd and return stdout.  Exit in case of error."""
 152     pop = Popen(cmd, stdout=PIPE, stderr=STDOUT, shell=True)
 153     (stdout, _) = pop.communicate()  # wait until finished
 154     if pop.returncode != 0:
 155         sys.exit(stdout)
 156     return stdout
 157
 158
 159 def tree_is_dirty():
 160     """Return true if the current working tree is dirty (i.e., if any file has
 161     been added, deleted, modified, renamed or copied but not committed)."""
 162     stdout = execute("git status --porcelain")
 163     for line in stdout:
 164         if re.findall(r"[URMADC]{1}", line[:2]):
 165             return True
 166     return False
 167
 168
 169 def get_head():
 170     """Return commit hash of current HEAD."""
 171     stdout = execute("git rev-parse HEAD")
 172     return stdout.strip('\n')
 173
 174
 175 def check_symbols(ignore):
 176     """Find undefined Kconfig symbols and return a dict with the symbol as key
 177     and a list of referencing files as value.  Files matching %ignore are not
 178     checked for undefined symbols."""
 179     source_files = []
 180     kconfig_files = []
 181     defined_features = set()
 182     referenced_features = dict()  # {feature: [files]}
 183
 184     # use 'git ls-files' to get the worklist
 185     stdout = execute("git ls-files")
 186     if len(stdout) > 0 and stdout[-1] == "\n":
 187         stdout = stdout[:-1]
 188
 189     for gitfile in stdout.rsplit("\n"):
 190         if ".git" in gitfile or "ChangeLog" in gitfile or      \
 191                 ".log" in gitfile or os.path.isdir(gitfile) or \
 192                 gitfile.startswith("tools/"):
 193             continue
 194         if REGEX_FILE_KCONFIG.match(gitfile):
 195             kconfig_files.append(gitfile)
 196         else:
 197             # all non-Kconfig files are checked for consistency
 198             source_files.append(gitfile)
 199
 200     for sfile in source_files:
 201         if ignore and re.match(ignore, sfile):
 202             # do not check files matching %ignore
 203             continue
 204         parse_source_file(sfile, referenced_features)
 205
 206     for kfile in kconfig_files:
 207         if ignore and re.match(ignore, kfile):
 208             # do not collect references for files matching %ignore
 209             parse_kconfig_file(kfile, defined_features, dict())
 210         else:
 211             parse_kconfig_file(kfile, defined_features, referenced_features)
 212
 213     undefined = {}  # {feature: [files]}
 214     for feature in sorted(referenced_features):
 215         # filter some false positives
 216         if feature == "FOO" or feature == "BAR" or \
 217                 feature == "FOO_BAR" or feature == "XXX":
 218             continue
 219         if feature not in defined_features:
 220             if feature.endswith("_MODULE"):
 221                 # avoid false positives for kernel modules
 222                 if feature[:-len("_MODULE")] in defined_features:
 223                     continue
 224             undefined[feature] = referenced_features.get(feature)
 225     return undefined
 226
 227
 228 def parse_source_file(sfile, referenced_features):
 229     """Parse @sfile for referenced Kconfig features."""
 230     lines = []
 231     with open(sfile, "r") as stream:
 232         lines = stream.readlines()
 233
 234     for line in lines:
 235         if not "CONFIG_" in line:
 236             continue
 237         features = REGEX_SOURCE_FEATURE.findall(line)
 238         for feature in features:
 239             if not REGEX_FILTER_FEATURES.search(feature):
 240                 continue
 241             sfiles = referenced_features.get(feature, set())
 242             sfiles.add(sfile)
 243             referenced_features[feature] = sfiles
 244
 245
 246 def get_features_in_line(line):
 247     """Return mentioned Kconfig features in @line."""
 248     return REGEX_FEATURE.findall(line)
 249
 250
 251 def parse_kconfig_file(kfile, defined_features, referenced_features):
 252     """Parse @kfile and update feature definitions and references."""
 253     lines = []
 254     skip = False
 255
 256     with open(kfile, "r") as stream:
 257         lines = stream.readlines()
 258
 259     for i in range(len(lines)):
 260         line = lines[i]
 261         line = line.strip('\n')
 262         line = line.split("#")[0]  # ignore comments
 263
 264         if REGEX_KCONFIG_DEF.match(line):
 265             feature_def = REGEX_KCONFIG_DEF.findall(line)
 266             defined_features.add(feature_def[0])
 267             skip = False
 268         elif REGEX_KCONFIG_HELP.match(line):
 269             skip = True
 270         elif skip:
 271             # ignore content of help messages
 272             pass
 273         elif REGEX_KCONFIG_STMT.match(line):
 274             features = get_features_in_line(line)
 275             # multi-line statements
 276             while line.endswith("\\"):
 277                 i += 1
 278                 line = lines[i]
 279                 line = line.strip('\n')
 280                 features.extend(get_features_in_line(line))
 281             for feature in set(features):
 282                 paths = referenced_features.get(feature, set())
 283                 paths.add(kfile)
 284                 referenced_features[feature] = paths
 285
 286
 287 if __name__ == "__main__":
 288     main()