]>
Commit | Line | Data |
---|---|---|
4b6fda0b | 1 | #!/usr/bin/env python2 |
24fe1f03 | 2 | |
b1a3f243 | 3 | """Find Kconfig symbols that are referenced but not defined.""" |
24fe1f03 | 4 | |
f175ba17 | 5 | # (c) 2014-2016 Valentin Rothberg <valentinrothberg@gmail.com> |
cc641d55 | 6 | # (c) 2014 Stefan Hengelein <stefan.hengelein@fau.de> |
24fe1f03 | 7 | # |
cc641d55 | 8 | # Licensed under the terms of the GNU GPL License version 2 |
24fe1f03 VR |
9 | |
10 | ||
1b2c8414 | 11 | import difflib |
24fe1f03 VR |
12 | import os |
13 | import re | |
e2042a8a | 14 | import signal |
f175ba17 | 15 | import subprocess |
b1a3f243 | 16 | import sys |
e2042a8a | 17 | from multiprocessing import Pool, cpu_count |
b1a3f243 | 18 | from optparse import OptionParser |
e2042a8a | 19 | from subprocess import Popen, PIPE, STDOUT |
24fe1f03 | 20 | |
cc641d55 VR |
21 | |
22 | # regex expressions | |
24fe1f03 | 23 | OPERATORS = r"&|\(|\)|\||\!" |
cc641d55 VR |
24 | FEATURE = r"(?:\w*[A-Z0-9]\w*){2,}" |
25 | DEF = r"^\s*(?:menu){,1}config\s+(" + FEATURE + r")\s*" | |
24fe1f03 | 26 | EXPR = r"(?:" + OPERATORS + r"|\s|" + FEATURE + r")+" |
0bd38ae3 VR |
27 | DEFAULT = r"default\s+.*?(?:if\s.+){,1}" |
28 | STMT = r"^\s*(?:if|select|depends\s+on|(?:" + DEFAULT + r"))\s+" + EXPR | |
cc641d55 | 29 | SOURCE_FEATURE = r"(?:\W|\b)+[D]{,1}CONFIG_(" + FEATURE + r")" |
24fe1f03 | 30 | |
cc641d55 | 31 | # regex objects |
24fe1f03 | 32 | REGEX_FILE_KCONFIG = re.compile(r".*Kconfig[\.\w+\-]*$") |
e2042a8a | 33 | REGEX_FEATURE = re.compile(r'(?!\B)' + FEATURE + r'(?!\B)') |
cc641d55 VR |
34 | REGEX_SOURCE_FEATURE = re.compile(SOURCE_FEATURE) |
35 | REGEX_KCONFIG_DEF = re.compile(DEF) | |
24fe1f03 VR |
36 | REGEX_KCONFIG_EXPR = re.compile(EXPR) |
37 | REGEX_KCONFIG_STMT = re.compile(STMT) | |
38 | REGEX_KCONFIG_HELP = re.compile(r"^\s+(help|---help---)\s*$") | |
39 | REGEX_FILTER_FEATURES = re.compile(r"[A-Za-z0-9]$") | |
0bd38ae3 | 40 | REGEX_NUMERIC = re.compile(r"0[xX][0-9a-fA-F]+|[0-9]+") |
e2042a8a | 41 | REGEX_QUOTES = re.compile("(\"(.*?)\")") |
24fe1f03 VR |
42 | |
43 | ||
b1a3f243 VR |
44 | def parse_options(): |
45 | """The user interface of this module.""" | |
46 | usage = "%prog [options]\n\n" \ | |
47 | "Run this tool to detect Kconfig symbols that are referenced but " \ | |
48 | "not defined in\nKconfig. The output of this tool has the " \ | |
49 | "format \'Undefined symbol\\tFile list\'\n\n" \ | |
50 | "If no option is specified, %prog will default to check your\n" \ | |
51 | "current tree. Please note that specifying commits will " \ | |
52 | "\'git reset --hard\'\nyour current tree! You may save " \ | |
53 | "uncommitted changes to avoid losing data." | |
54 | ||
55 | parser = OptionParser(usage=usage) | |
56 | ||
57 | parser.add_option('-c', '--commit', dest='commit', action='store', | |
58 | default="", | |
59 | help="Check if the specified commit (hash) introduces " | |
60 | "undefined Kconfig symbols.") | |
61 | ||
62 | parser.add_option('-d', '--diff', dest='diff', action='store', | |
63 | default="", | |
64 | help="Diff undefined symbols between two commits. The " | |
65 | "input format bases on Git log's " | |
66 | "\'commmit1..commit2\'.") | |
67 | ||
a42fa92c VR |
68 | parser.add_option('-f', '--find', dest='find', action='store_true', |
69 | default=False, | |
70 | help="Find and show commits that may cause symbols to be " | |
71 | "missing. Required to run with --diff.") | |
72 | ||
cf132e4a VR |
73 | parser.add_option('-i', '--ignore', dest='ignore', action='store', |
74 | default="", | |
75 | help="Ignore files matching this pattern. Note that " | |
76 | "the pattern needs to be a Python regex. To " | |
77 | "ignore defconfigs, specify -i '.*defconfig'.") | |
78 | ||
1b2c8414 VR |
79 | parser.add_option('-s', '--sim', dest='sim', action='store', default="", |
80 | help="Print a list of maximum 10 string-similar symbols.") | |
81 | ||
b1a3f243 VR |
82 | parser.add_option('', '--force', dest='force', action='store_true', |
83 | default=False, | |
84 | help="Reset current Git tree even when it's dirty.") | |
85 | ||
4c73c088 AD |
86 | parser.add_option('', '--no-color', dest='color', action='store_false', |
87 | default=True, | |
88 | help="Don't print colored output. Default when not " | |
89 | "outputting to a terminal.") | |
90 | ||
b1a3f243 VR |
91 | (opts, _) = parser.parse_args() |
92 | ||
93 | if opts.commit and opts.diff: | |
94 | sys.exit("Please specify only one option at once.") | |
95 | ||
96 | if opts.diff and not re.match(r"^[\w\-\.]+\.\.[\w\-\.]+$", opts.diff): | |
97 | sys.exit("Please specify valid input in the following format: " | |
38cbfe4f | 98 | "\'commit1..commit2\'") |
b1a3f243 VR |
99 | |
100 | if opts.commit or opts.diff: | |
101 | if not opts.force and tree_is_dirty(): | |
102 | sys.exit("The current Git tree is dirty (see 'git status'). " | |
103 | "Running this script may\ndelete important data since it " | |
104 | "calls 'git reset --hard' for some performance\nreasons. " | |
105 | " Please run this script in a clean Git tree or pass " | |
106 | "'--force' if you\nwant to ignore this warning and " | |
107 | "continue.") | |
108 | ||
a42fa92c VR |
109 | if opts.commit: |
110 | opts.find = False | |
111 | ||
cf132e4a VR |
112 | if opts.ignore: |
113 | try: | |
114 | re.match(opts.ignore, "this/is/just/a/test.c") | |
115 | except: | |
116 | sys.exit("Please specify a valid Python regex.") | |
117 | ||
b1a3f243 VR |
118 | return opts |
119 | ||
120 | ||
24fe1f03 VR |
121 | def main(): |
122 | """Main function of this module.""" | |
b1a3f243 VR |
123 | opts = parse_options() |
124 | ||
4c73c088 AD |
125 | global color |
126 | color = opts.color and sys.stdout.isatty() | |
127 | ||
1b2c8414 VR |
128 | if opts.sim and not opts.commit and not opts.diff: |
129 | sims = find_sims(opts.sim, opts.ignore) | |
130 | if sims: | |
131 | print "%s: %s" % (yel("Similar symbols"), ', '.join(sims)) | |
132 | else: | |
133 | print "%s: no similar symbols found" % yel("Similar symbols") | |
134 | sys.exit(0) | |
135 | ||
136 | # dictionary of (un)defined symbols | |
137 | defined = {} | |
138 | undefined = {} | |
139 | ||
b1a3f243 VR |
140 | if opts.commit or opts.diff: |
141 | head = get_head() | |
142 | ||
143 | # get commit range | |
144 | commit_a = None | |
145 | commit_b = None | |
146 | if opts.commit: | |
147 | commit_a = opts.commit + "~" | |
148 | commit_b = opts.commit | |
149 | elif opts.diff: | |
150 | split = opts.diff.split("..") | |
151 | commit_a = split[0] | |
152 | commit_b = split[1] | |
153 | undefined_a = {} | |
154 | undefined_b = {} | |
155 | ||
156 | # get undefined items before the commit | |
157 | execute("git reset --hard %s" % commit_a) | |
1b2c8414 | 158 | undefined_a, _ = check_symbols(opts.ignore) |
b1a3f243 VR |
159 | |
160 | # get undefined items for the commit | |
161 | execute("git reset --hard %s" % commit_b) | |
1b2c8414 | 162 | undefined_b, defined = check_symbols(opts.ignore) |
b1a3f243 VR |
163 | |
164 | # report cases that are present for the commit but not before | |
e9533ae5 | 165 | for feature in sorted(undefined_b): |
b1a3f243 VR |
166 | # feature has not been undefined before |
167 | if not feature in undefined_a: | |
e9533ae5 | 168 | files = sorted(undefined_b.get(feature)) |
1b2c8414 | 169 | undefined[feature] = files |
b1a3f243 VR |
170 | # check if there are new files that reference the undefined feature |
171 | else: | |
e9533ae5 VR |
172 | files = sorted(undefined_b.get(feature) - |
173 | undefined_a.get(feature)) | |
b1a3f243 | 174 | if files: |
1b2c8414 | 175 | undefined[feature] = files |
b1a3f243 VR |
176 | |
177 | # reset to head | |
178 | execute("git reset --hard %s" % head) | |
179 | ||
180 | # default to check the entire tree | |
181 | else: | |
1b2c8414 VR |
182 | undefined, defined = check_symbols(opts.ignore) |
183 | ||
184 | # now print the output | |
185 | for feature in sorted(undefined): | |
186 | print red(feature) | |
187 | ||
188 | files = sorted(undefined.get(feature)) | |
189 | print "%s: %s" % (yel("Referencing files"), ", ".join(files)) | |
190 | ||
191 | sims = find_sims(feature, opts.ignore, defined) | |
192 | sims_out = yel("Similar symbols") | |
193 | if sims: | |
194 | print "%s: %s" % (sims_out, ', '.join(sims)) | |
195 | else: | |
196 | print "%s: %s" % (sims_out, "no similar symbols found") | |
197 | ||
198 | if opts.find: | |
199 | print "%s:" % yel("Commits changing symbol") | |
200 | commits = find_commits(feature, opts.diff) | |
201 | if commits: | |
202 | for commit in commits: | |
203 | commit = commit.split(" ", 1) | |
204 | print "\t- %s (\"%s\")" % (yel(commit[0]), commit[1]) | |
205 | else: | |
206 | print "\t- no commit found" | |
207 | print # new line | |
c7455663 VR |
208 | |
209 | ||
210 | def yel(string): | |
211 | """ | |
212 | Color %string yellow. | |
213 | """ | |
4c73c088 | 214 | return "\033[33m%s\033[0m" % string if color else string |
c7455663 VR |
215 | |
216 | ||
217 | def red(string): | |
218 | """ | |
219 | Color %string red. | |
220 | """ | |
4c73c088 | 221 | return "\033[31m%s\033[0m" % string if color else string |
b1a3f243 VR |
222 | |
223 | ||
224 | def execute(cmd): | |
225 | """Execute %cmd and return stdout. Exit in case of error.""" | |
f175ba17 VR |
226 | try: |
227 | cmdlist = cmd.split(" ") | |
228 | stdout = subprocess.check_output(cmdlist, stderr=STDOUT, shell=False) | |
229 | except subprocess.CalledProcessError as fail: | |
230 | exit("Failed to execute %s\n%s" % (cmd, fail)) | |
b1a3f243 VR |
231 | return stdout |
232 | ||
233 | ||
a42fa92c VR |
234 | def find_commits(symbol, diff): |
235 | """Find commits changing %symbol in the given range of %diff.""" | |
236 | commits = execute("git log --pretty=oneline --abbrev-commit -G %s %s" | |
237 | % (symbol, diff)) | |
1b2c8414 | 238 | return [x for x in commits.split("\n") if x] |
a42fa92c VR |
239 | |
240 | ||
b1a3f243 VR |
241 | def tree_is_dirty(): |
242 | """Return true if the current working tree is dirty (i.e., if any file has | |
243 | been added, deleted, modified, renamed or copied but not committed).""" | |
244 | stdout = execute("git status --porcelain") | |
245 | for line in stdout: | |
246 | if re.findall(r"[URMADC]{1}", line[:2]): | |
247 | return True | |
248 | return False | |
249 | ||
250 | ||
251 | def get_head(): | |
252 | """Return commit hash of current HEAD.""" | |
253 | stdout = execute("git rev-parse HEAD") | |
254 | return stdout.strip('\n') | |
255 | ||
256 | ||
e2042a8a VR |
257 | def partition(lst, size): |
258 | """Partition list @lst into eveni-sized lists of size @size.""" | |
259 | return [lst[i::size] for i in xrange(size)] | |
260 | ||
261 | ||
262 | def init_worker(): | |
263 | """Set signal handler to ignore SIGINT.""" | |
264 | signal.signal(signal.SIGINT, signal.SIG_IGN) | |
265 | ||
266 | ||
1b2c8414 VR |
267 | def find_sims(symbol, ignore, defined = []): |
268 | """Return a list of max. ten Kconfig symbols that are string-similar to | |
269 | @symbol.""" | |
270 | if defined: | |
271 | return sorted(difflib.get_close_matches(symbol, set(defined), 10)) | |
272 | ||
273 | pool = Pool(cpu_count(), init_worker) | |
274 | kfiles = [] | |
275 | for gitfile in get_files(): | |
276 | if REGEX_FILE_KCONFIG.match(gitfile): | |
277 | kfiles.append(gitfile) | |
278 | ||
279 | arglist = [] | |
280 | for part in partition(kfiles, cpu_count()): | |
281 | arglist.append((part, ignore)) | |
282 | ||
283 | for res in pool.map(parse_kconfig_files, arglist): | |
284 | defined.extend(res[0]) | |
285 | ||
286 | return sorted(difflib.get_close_matches(symbol, set(defined), 10)) | |
287 | ||
288 | ||
289 | def get_files(): | |
290 | """Return a list of all files in the current git directory.""" | |
291 | # use 'git ls-files' to get the worklist | |
292 | stdout = execute("git ls-files") | |
293 | if len(stdout) > 0 and stdout[-1] == "\n": | |
294 | stdout = stdout[:-1] | |
295 | ||
296 | files = [] | |
297 | for gitfile in stdout.rsplit("\n"): | |
298 | if ".git" in gitfile or "ChangeLog" in gitfile or \ | |
299 | ".log" in gitfile or os.path.isdir(gitfile) or \ | |
300 | gitfile.startswith("tools/"): | |
301 | continue | |
302 | files.append(gitfile) | |
303 | return files | |
304 | ||
305 | ||
cf132e4a | 306 | def check_symbols(ignore): |
b1a3f243 | 307 | """Find undefined Kconfig symbols and return a dict with the symbol as key |
cf132e4a VR |
308 | and a list of referencing files as value. Files matching %ignore are not |
309 | checked for undefined symbols.""" | |
e2042a8a VR |
310 | pool = Pool(cpu_count(), init_worker) |
311 | try: | |
312 | return check_symbols_helper(pool, ignore) | |
313 | except KeyboardInterrupt: | |
314 | pool.terminate() | |
315 | pool.join() | |
316 | sys.exit(1) | |
317 | ||
318 | ||
319 | def check_symbols_helper(pool, ignore): | |
320 | """Helper method for check_symbols(). Used to catch keyboard interrupts in | |
321 | check_symbols() in order to properly terminate running worker processes.""" | |
24fe1f03 VR |
322 | source_files = [] |
323 | kconfig_files = [] | |
e2042a8a VR |
324 | defined_features = [] |
325 | referenced_features = dict() # {file: [features]} | |
24fe1f03 | 326 | |
1b2c8414 | 327 | for gitfile in get_files(): |
24fe1f03 VR |
328 | if REGEX_FILE_KCONFIG.match(gitfile): |
329 | kconfig_files.append(gitfile) | |
330 | else: | |
e2042a8a VR |
331 | if ignore and not re.match(ignore, gitfile): |
332 | continue | |
333 | # add source files that do not match the ignore pattern | |
24fe1f03 VR |
334 | source_files.append(gitfile) |
335 | ||
e2042a8a VR |
336 | # parse source files |
337 | arglist = partition(source_files, cpu_count()) | |
338 | for res in pool.map(parse_source_files, arglist): | |
339 | referenced_features.update(res) | |
24fe1f03 | 340 | |
e2042a8a VR |
341 | |
342 | # parse kconfig files | |
343 | arglist = [] | |
344 | for part in partition(kconfig_files, cpu_count()): | |
345 | arglist.append((part, ignore)) | |
346 | for res in pool.map(parse_kconfig_files, arglist): | |
347 | defined_features.extend(res[0]) | |
348 | referenced_features.update(res[1]) | |
349 | defined_features = set(defined_features) | |
350 | ||
351 | # inverse mapping of referenced_features to dict(feature: [files]) | |
352 | inv_map = dict() | |
353 | for _file, features in referenced_features.iteritems(): | |
354 | for feature in features: | |
355 | inv_map[feature] = inv_map.get(feature, set()) | |
356 | inv_map[feature].add(_file) | |
357 | referenced_features = inv_map | |
24fe1f03 | 358 | |
b1a3f243 | 359 | undefined = {} # {feature: [files]} |
24fe1f03 | 360 | for feature in sorted(referenced_features): |
cc641d55 VR |
361 | # filter some false positives |
362 | if feature == "FOO" or feature == "BAR" or \ | |
363 | feature == "FOO_BAR" or feature == "XXX": | |
364 | continue | |
24fe1f03 VR |
365 | if feature not in defined_features: |
366 | if feature.endswith("_MODULE"): | |
cc641d55 | 367 | # avoid false positives for kernel modules |
24fe1f03 VR |
368 | if feature[:-len("_MODULE")] in defined_features: |
369 | continue | |
b1a3f243 | 370 | undefined[feature] = referenced_features.get(feature) |
1b2c8414 | 371 | return undefined, defined_features |
24fe1f03 VR |
372 | |
373 | ||
e2042a8a VR |
374 | def parse_source_files(source_files): |
375 | """Parse each source file in @source_files and return dictionary with source | |
376 | files as keys and lists of references Kconfig symbols as values.""" | |
377 | referenced_features = dict() | |
378 | for sfile in source_files: | |
379 | referenced_features[sfile] = parse_source_file(sfile) | |
380 | return referenced_features | |
381 | ||
382 | ||
383 | def parse_source_file(sfile): | |
384 | """Parse @sfile and return a list of referenced Kconfig features.""" | |
24fe1f03 | 385 | lines = [] |
e2042a8a VR |
386 | references = [] |
387 | ||
388 | if not os.path.exists(sfile): | |
389 | return references | |
390 | ||
24fe1f03 VR |
391 | with open(sfile, "r") as stream: |
392 | lines = stream.readlines() | |
393 | ||
394 | for line in lines: | |
395 | if not "CONFIG_" in line: | |
396 | continue | |
397 | features = REGEX_SOURCE_FEATURE.findall(line) | |
398 | for feature in features: | |
399 | if not REGEX_FILTER_FEATURES.search(feature): | |
400 | continue | |
e2042a8a VR |
401 | references.append(feature) |
402 | ||
403 | return references | |
24fe1f03 VR |
404 | |
405 | ||
406 | def get_features_in_line(line): | |
407 | """Return mentioned Kconfig features in @line.""" | |
408 | return REGEX_FEATURE.findall(line) | |
409 | ||
410 | ||
e2042a8a VR |
411 | def parse_kconfig_files(args): |
412 | """Parse kconfig files and return tuple of defined and references Kconfig | |
413 | symbols. Note, @args is a tuple of a list of files and the @ignore | |
414 | pattern.""" | |
415 | kconfig_files = args[0] | |
416 | ignore = args[1] | |
417 | defined_features = [] | |
418 | referenced_features = dict() | |
419 | ||
420 | for kfile in kconfig_files: | |
421 | defined, references = parse_kconfig_file(kfile) | |
422 | defined_features.extend(defined) | |
423 | if ignore and re.match(ignore, kfile): | |
424 | # do not collect references for files that match the ignore pattern | |
425 | continue | |
426 | referenced_features[kfile] = references | |
427 | return (defined_features, referenced_features) | |
428 | ||
429 | ||
430 | def parse_kconfig_file(kfile): | |
24fe1f03 VR |
431 | """Parse @kfile and update feature definitions and references.""" |
432 | lines = [] | |
e2042a8a VR |
433 | defined = [] |
434 | references = [] | |
24fe1f03 VR |
435 | skip = False |
436 | ||
e2042a8a VR |
437 | if not os.path.exists(kfile): |
438 | return defined, references | |
439 | ||
24fe1f03 VR |
440 | with open(kfile, "r") as stream: |
441 | lines = stream.readlines() | |
442 | ||
443 | for i in range(len(lines)): | |
444 | line = lines[i] | |
445 | line = line.strip('\n') | |
cc641d55 | 446 | line = line.split("#")[0] # ignore comments |
24fe1f03 VR |
447 | |
448 | if REGEX_KCONFIG_DEF.match(line): | |
449 | feature_def = REGEX_KCONFIG_DEF.findall(line) | |
e2042a8a | 450 | defined.append(feature_def[0]) |
24fe1f03 VR |
451 | skip = False |
452 | elif REGEX_KCONFIG_HELP.match(line): | |
453 | skip = True | |
454 | elif skip: | |
cc641d55 | 455 | # ignore content of help messages |
24fe1f03 VR |
456 | pass |
457 | elif REGEX_KCONFIG_STMT.match(line): | |
e2042a8a | 458 | line = REGEX_QUOTES.sub("", line) |
24fe1f03 | 459 | features = get_features_in_line(line) |
cc641d55 | 460 | # multi-line statements |
24fe1f03 VR |
461 | while line.endswith("\\"): |
462 | i += 1 | |
463 | line = lines[i] | |
464 | line = line.strip('\n') | |
465 | features.extend(get_features_in_line(line)) | |
466 | for feature in set(features): | |
0bd38ae3 VR |
467 | if REGEX_NUMERIC.match(feature): |
468 | # ignore numeric values | |
469 | continue | |
e2042a8a VR |
470 | references.append(feature) |
471 | ||
472 | return defined, references | |
24fe1f03 VR |
473 | |
474 | ||
475 | if __name__ == "__main__": | |
476 | main() |