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 <valentinrothberg@gmail.com>
   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 difflib
  12 import os
  13 import re
  14 import signal
  15 import sys
  16 from multiprocessing import Pool, cpu_count
  17 from optparse import OptionParser
  18 from subprocess import Popen, PIPE, STDOUT
  19
  20
  21 # regex expressions
  22 OPERATORS = r"&|\(|\)|\||\!"
  23 FEATURE = r"(?:\w*[A-Z0-9]\w*){2,}"
  24 DEF = r"^\s*(?:menu){,1}config\s+(" + FEATURE + r")\s*"
  25 EXPR = r"(?:" + OPERATORS + r"|\s|" + FEATURE + r")+"
  26 DEFAULT = r"default\s+.*?(?:if\s.+){,1}"
  27 STMT = r"^\s*(?:if|select|depends\s+on|(?:" + DEFAULT + r"))\s+" + EXPR
  28 SOURCE_FEATURE = r"(?:\W|\b)+[D]{,1}CONFIG_(" + FEATURE + r")"
  29
  30 # regex objects
  31 REGEX_FILE_KCONFIG = re.compile(r".*Kconfig[\.\w+\-]*$")
  32 REGEX_FEATURE = re.compile(r'(?!\B)' + FEATURE + r'(?!\B)')
  33 REGEX_SOURCE_FEATURE = re.compile(SOURCE_FEATURE)
  34 REGEX_KCONFIG_DEF = re.compile(DEF)
  35 REGEX_KCONFIG_EXPR = re.compile(EXPR)
  36 REGEX_KCONFIG_STMT = re.compile(STMT)
  37 REGEX_KCONFIG_HELP = re.compile(r"^\s+(help|---help---)\s*$")
  38 REGEX_FILTER_FEATURES = re.compile(r"[A-Za-z0-9]$")
  39 REGEX_NUMERIC = re.compile(r"0[xX][0-9a-fA-F]+|[0-9]+")
  40 REGEX_QUOTES = re.compile("(\"(.*?)\")")
  41
  42
  43 def parse_options():
  44     """The user interface of this module."""
  45     usage = "%prog [options]\n\n"                                              \
  46             "Run this tool to detect Kconfig symbols that are referenced but " \
  47             "not defined in\nKconfig.  The output of this tool has the "       \
  48             "format \'Undefined symbol\\tFile list\'\n\n"                      \
  49             "If no option is specified, %prog will default to check your\n"    \
  50             "current tree.  Please note that specifying commits will "         \
  51             "\'git reset --hard\'\nyour current tree!  You may save "          \
  52             "uncommitted changes to avoid losing data."
  53
  54     parser = OptionParser(usage=usage)
  55
  56     parser.add_option('-c', '--commit', dest='commit', action='store',
  57                       default="",
  58                       help="Check if the specified commit (hash) introduces "
  59                            "undefined Kconfig symbols.")
  60
  61     parser.add_option('-d', '--diff', dest='diff', action='store',
  62                       default="",
  63                       help="Diff undefined symbols between two commits.  The "
  64                            "input format bases on Git log's "
  65                            "\'commmit1..commit2\'.")
  66
  67     parser.add_option('-f', '--find', dest='find', action='store_true',
  68                       default=False,
  69                       help="Find and show commits that may cause symbols to be "
  70                            "missing.  Required to run with --diff.")
  71
  72     parser.add_option('-i', '--ignore', dest='ignore', action='store',
  73                       default="",
  74                       help="Ignore files matching this pattern.  Note that "
  75                            "the pattern needs to be a Python regex.  To "
  76                            "ignore defconfigs, specify -i '.*defconfig'.")
  77
  78     parser.add_option('-s', '--sim', dest='sim', action='store', default="",
  79                       help="Print a list of maximum 10 string-similar symbols.")
  80
  81     parser.add_option('', '--force', dest='force', action='store_true',
  82                       default=False,
  83                       help="Reset current Git tree even when it's dirty.")
  84
  85     (opts, _) = parser.parse_args()
  86
  87     if opts.commit and opts.diff:
  88         sys.exit("Please specify only one option at once.")
  89
  90     if opts.diff and not re.match(r"^[\w\-\.]+\.\.[\w\-\.]+$", opts.diff):
  91         sys.exit("Please specify valid input in the following format: "
  92                  "\'commmit1..commit2\'")
  93
  94     if opts.commit or opts.diff:
  95         if not opts.force and tree_is_dirty():
  96             sys.exit("The current Git tree is dirty (see 'git status').  "
  97                      "Running this script may\ndelete important data since it "
  98                      "calls 'git reset --hard' for some performance\nreasons. "
  99                      " Please run this script in a clean Git tree or pass "
 100                      "'--force' if you\nwant to ignore this warning and "
 101                      "continue.")
 102
 103     if opts.commit:
 104         opts.find = False
 105
 106     if opts.ignore:
 107         try:
 108             re.match(opts.ignore, "this/is/just/a/test.c")
 109         except:
 110             sys.exit("Please specify a valid Python regex.")
 111
 112     return opts
 113
 114
 115 def main():
 116     """Main function of this module."""
 117     opts = parse_options()
 118
 119     if opts.sim and not opts.commit and not opts.diff:
 120         sims = find_sims(opts.sim, opts.ignore)
 121         if sims:
 122             print "%s: %s" % (yel("Similar symbols"), ', '.join(sims))
 123         else:
 124             print "%s: no similar symbols found" % yel("Similar symbols")
 125         sys.exit(0)
 126
 127     # dictionary of (un)defined symbols
 128     defined = {}
 129     undefined = {}
 130
 131     if opts.commit or opts.diff:
 132         head = get_head()
 133
 134         # get commit range
 135         commit_a = None
 136         commit_b = None
 137         if opts.commit:
 138             commit_a = opts.commit + "~"
 139             commit_b = opts.commit
 140         elif opts.diff:
 141             split = opts.diff.split("..")
 142             commit_a = split[0]
 143             commit_b = split[1]
 144             undefined_a = {}
 145             undefined_b = {}
 146
 147         # get undefined items before the commit
 148         execute("git reset --hard %s" % commit_a)
 149         undefined_a, _ = check_symbols(opts.ignore)
 150
 151         # get undefined items for the commit
 152         execute("git reset --hard %s" % commit_b)
 153         undefined_b, defined = check_symbols(opts.ignore)
 154
 155         # report cases that are present for the commit but not before
 156         for feature in sorted(undefined_b):
 157             # feature has not been undefined before
 158             if not feature in undefined_a:
 159                 files = sorted(undefined_b.get(feature))
 160                 undefined[feature] = files
 161             # check if there are new files that reference the undefined feature
 162             else:
 163                 files = sorted(undefined_b.get(feature) -
 164                                undefined_a.get(feature))
 165                 if files:
 166                     undefined[feature] = files
 167
 168         # reset to head
 169         execute("git reset --hard %s" % head)
 170
 171     # default to check the entire tree
 172     else:
 173         undefined, defined = check_symbols(opts.ignore)
 174
 175     # now print the output
 176     for feature in sorted(undefined):
 177         print red(feature)
 178
 179         files = sorted(undefined.get(feature))
 180         print "%s: %s" % (yel("Referencing files"), ", ".join(files))
 181
 182         sims = find_sims(feature, opts.ignore, defined)
 183         sims_out = yel("Similar symbols")
 184         if sims:
 185             print "%s: %s" % (sims_out, ', '.join(sims))
 186         else:
 187             print "%s: %s" % (sims_out, "no similar symbols found")
 188
 189         if opts.find:
 190             print "%s:" % yel("Commits changing symbol")
 191             commits = find_commits(feature, opts.diff)
 192             if commits:
 193                 for commit in commits:
 194                     commit = commit.split(" ", 1)
 195                     print "\t- %s (\"%s\")" % (yel(commit[0]), commit[1])
 196             else:
 197                 print "\t- no commit found"
 198         print  #  new line
 199
 200
 201 def yel(string):
 202     """
 203     Color %string yellow.
 204     """
 205     return "\033[33m%s\033[0m" % string
 206
 207
 208 def red(string):
 209     """
 210     Color %string red.
 211     """
 212     return "\033[31m%s\033[0m" % string
 213
 214
 215 def execute(cmd):
 216     """Execute %cmd and return stdout.  Exit in case of error."""
 217     pop = Popen(cmd, stdout=PIPE, stderr=STDOUT, shell=True)
 218     (stdout, _) = pop.communicate()  # wait until finished
 219     if pop.returncode != 0:
 220         sys.exit(stdout)
 221     return stdout
 222
 223
 224 def find_commits(symbol, diff):
 225     """Find commits changing %symbol in the given range of %diff."""
 226     commits = execute("git log --pretty=oneline --abbrev-commit -G %s %s"
 227                       % (symbol, diff))
 228     return [x for x in commits.split("\n") if x]
 229
 230
 231 def tree_is_dirty():
 232     """Return true if the current working tree is dirty (i.e., if any file has
 233     been added, deleted, modified, renamed or copied but not committed)."""
 234     stdout = execute("git status --porcelain")
 235     for line in stdout:
 236         if re.findall(r"[URMADC]{1}", line[:2]):
 237             return True
 238     return False
 239
 240
 241 def get_head():
 242     """Return commit hash of current HEAD."""
 243     stdout = execute("git rev-parse HEAD")
 244     return stdout.strip('\n')
 245
 246
 247 def partition(lst, size):
 248     """Partition list @lst into eveni-sized lists of size @size."""
 249     return [lst[i::size] for i in xrange(size)]
 250
 251
 252 def init_worker():
 253     """Set signal handler to ignore SIGINT."""
 254     signal.signal(signal.SIGINT, signal.SIG_IGN)
 255
 256
 257 def find_sims(symbol, ignore, defined = []):
 258     """Return a list of max. ten Kconfig symbols that are string-similar to
 259     @symbol."""
 260     if defined:
 261         return sorted(difflib.get_close_matches(symbol, set(defined), 10))
 262
 263     pool = Pool(cpu_count(), init_worker)
 264     kfiles = []
 265     for gitfile in get_files():
 266         if REGEX_FILE_KCONFIG.match(gitfile):
 267             kfiles.append(gitfile)
 268
 269     arglist = []
 270     for part in partition(kfiles, cpu_count()):
 271         arglist.append((part, ignore))
 272
 273     for res in pool.map(parse_kconfig_files, arglist):
 274         defined.extend(res[0])
 275
 276     return sorted(difflib.get_close_matches(symbol, set(defined), 10))
 277
 278
 279 def get_files():
 280     """Return a list of all files in the current git directory."""
 281     # use 'git ls-files' to get the worklist
 282     stdout = execute("git ls-files")
 283     if len(stdout) > 0 and stdout[-1] == "\n":
 284         stdout = stdout[:-1]
 285
 286     files = []
 287     for gitfile in stdout.rsplit("\n"):
 288         if ".git" in gitfile or "ChangeLog" in gitfile or      \
 289                 ".log" in gitfile or os.path.isdir(gitfile) or \
 290                 gitfile.startswith("tools/"):
 291             continue
 292         files.append(gitfile)
 293     return files
 294
 295
 296 def check_symbols(ignore):
 297     """Find undefined Kconfig symbols and return a dict with the symbol as key
 298     and a list of referencing files as value.  Files matching %ignore are not
 299     checked for undefined symbols."""
 300     pool = Pool(cpu_count(), init_worker)
 301     try:
 302         return check_symbols_helper(pool, ignore)
 303     except KeyboardInterrupt:
 304         pool.terminate()
 305         pool.join()
 306         sys.exit(1)
 307
 308
 309 def check_symbols_helper(pool, ignore):
 310     """Helper method for check_symbols().  Used to catch keyboard interrupts in
 311     check_symbols() in order to properly terminate running worker processes."""
 312     source_files = []
 313     kconfig_files = []
 314     defined_features = []
 315     referenced_features = dict()  # {file: [features]}
 316
 317     for gitfile in get_files():
 318         if REGEX_FILE_KCONFIG.match(gitfile):
 319             kconfig_files.append(gitfile)
 320         else:
 321             if ignore and not re.match(ignore, gitfile):
 322                 continue
 323             # add source files that do not match the ignore pattern
 324             source_files.append(gitfile)
 325
 326     # parse source files
 327     arglist = partition(source_files, cpu_count())
 328     for res in pool.map(parse_source_files, arglist):
 329         referenced_features.update(res)
 330
 331
 332     # parse kconfig files
 333     arglist = []
 334     for part in partition(kconfig_files, cpu_count()):
 335         arglist.append((part, ignore))
 336     for res in pool.map(parse_kconfig_files, arglist):
 337         defined_features.extend(res[0])
 338         referenced_features.update(res[1])
 339     defined_features = set(defined_features)
 340
 341     # inverse mapping of referenced_features to dict(feature: [files])
 342     inv_map = dict()
 343     for _file, features in referenced_features.iteritems():
 344         for feature in features:
 345             inv_map[feature] = inv_map.get(feature, set())
 346             inv_map[feature].add(_file)
 347     referenced_features = inv_map
 348
 349     undefined = {}  # {feature: [files]}
 350     for feature in sorted(referenced_features):
 351         # filter some false positives
 352         if feature == "FOO" or feature == "BAR" or \
 353                 feature == "FOO_BAR" or feature == "XXX":
 354             continue
 355         if feature not in defined_features:
 356             if feature.endswith("_MODULE"):
 357                 # avoid false positives for kernel modules
 358                 if feature[:-len("_MODULE")] in defined_features:
 359                     continue
 360             undefined[feature] = referenced_features.get(feature)
 361     return undefined, defined_features
 362
 363
 364 def parse_source_files(source_files):
 365     """Parse each source file in @source_files and return dictionary with source
 366     files as keys and lists of references Kconfig symbols as values."""
 367     referenced_features = dict()
 368     for sfile in source_files:
 369         referenced_features[sfile] = parse_source_file(sfile)
 370     return referenced_features
 371
 372
 373 def parse_source_file(sfile):
 374     """Parse @sfile and return a list of referenced Kconfig features."""
 375     lines = []
 376     references = []
 377
 378     if not os.path.exists(sfile):
 379         return references
 380
 381     with open(sfile, "r") as stream:
 382         lines = stream.readlines()
 383
 384     for line in lines:
 385         if not "CONFIG_" in line:
 386             continue
 387         features = REGEX_SOURCE_FEATURE.findall(line)
 388         for feature in features:
 389             if not REGEX_FILTER_FEATURES.search(feature):
 390                 continue
 391             references.append(feature)
 392
 393     return references
 394
 395
 396 def get_features_in_line(line):
 397     """Return mentioned Kconfig features in @line."""
 398     return REGEX_FEATURE.findall(line)
 399
 400
 401 def parse_kconfig_files(args):
 402     """Parse kconfig files and return tuple of defined and references Kconfig
 403     symbols.  Note, @args is a tuple of a list of files and the @ignore
 404     pattern."""
 405     kconfig_files = args[0]
 406     ignore = args[1]
 407     defined_features = []
 408     referenced_features = dict()
 409
 410     for kfile in kconfig_files:
 411         defined, references = parse_kconfig_file(kfile)
 412         defined_features.extend(defined)
 413         if ignore and re.match(ignore, kfile):
 414             # do not collect references for files that match the ignore pattern
 415             continue
 416         referenced_features[kfile] = references
 417     return (defined_features, referenced_features)
 418
 419
 420 def parse_kconfig_file(kfile):
 421     """Parse @kfile and update feature definitions and references."""
 422     lines = []
 423     defined = []
 424     references = []
 425     skip = False
 426
 427     if not os.path.exists(kfile):
 428         return defined, references
 429
 430     with open(kfile, "r") as stream:
 431         lines = stream.readlines()
 432
 433     for i in range(len(lines)):
 434         line = lines[i]
 435         line = line.strip('\n')
 436         line = line.split("#")[0]  # ignore comments
 437
 438         if REGEX_KCONFIG_DEF.match(line):
 439             feature_def = REGEX_KCONFIG_DEF.findall(line)
 440             defined.append(feature_def[0])
 441             skip = False
 442         elif REGEX_KCONFIG_HELP.match(line):
 443             skip = True
 444         elif skip:
 445             # ignore content of help messages
 446             pass
 447         elif REGEX_KCONFIG_STMT.match(line):
 448             line = REGEX_QUOTES.sub("", line)
 449             features = get_features_in_line(line)
 450             # multi-line statements
 451             while line.endswith("\\"):
 452                 i += 1
 453                 line = lines[i]
 454                 line = line.strip('\n')
 455                 features.extend(get_features_in_line(line))
 456             for feature in set(features):
 457                 if REGEX_NUMERIC.match(feature):
 458                     # ignore numeric values
 459                     continue
 460                 references.append(feature)
 461
 462     return defined, references
 463
 464
 465 if __name__ == "__main__":
 466     main()