1#!/usr/bin/env python3
2
3"""Find Kconfig symbols that are referenced but not defined."""
4
5# (c) 2014-2016 Valentin Rothberg <[email protected]>
6# (c) 2014 Stefan Hengelein <[email protected]>
7#
8# Licensed under the terms of the GNU GPL License version 2
9
10
11import difflib
12import os
13import re
14import signal
15import subprocess
16import sys
17from multiprocessing import Pool, cpu_count
18from optparse import OptionParser
19from subprocess import Popen, PIPE, STDOUT
20
21
22# regex expressions
23OPERATORS = r"&|\(|\)|\||\!"
24FEATURE = r"(?:\w*[A-Z0-9]\w*){2,}"
25DEF = r"^\s*(?:menu){,1}config\s+(" + FEATURE + r")\s*"
26EXPR = r"(?:" + OPERATORS + r"|\s|" + FEATURE + r")+"
27DEFAULT = r"default\s+.*?(?:if\s.+){,1}"
28STMT = r"^\s*(?:if|select|depends\s+on|(?:" + DEFAULT + r"))\s+" + EXPR
29SOURCE_FEATURE = r"(?:\W|\b)+[D]{,1}CONFIG_(" + FEATURE + r")"
30
31# regex objects
32REGEX_FILE_KCONFIG = re.compile(r".*Kconfig[\.\w+\-]*$")
33REGEX_FEATURE = re.compile(r'(?!\B)' + FEATURE + r'(?!\B)')
34REGEX_SOURCE_FEATURE = re.compile(SOURCE_FEATURE)
35REGEX_KCONFIG_DEF = re.compile(DEF)
36REGEX_KCONFIG_EXPR = re.compile(EXPR)
37REGEX_KCONFIG_STMT = re.compile(STMT)
38REGEX_KCONFIG_HELP = re.compile(r"^\s+(help|---help---)\s*$")
39REGEX_FILTER_FEATURES = re.compile(r"[A-Za-z0-9]$")
40REGEX_NUMERIC = re.compile(r"0[xX][0-9a-fA-F]+|[0-9]+")
41REGEX_QUOTES = re.compile("(\"(.*?)\")")
42
43
44def 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
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
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
79    parser.add_option('-s', '--sim', dest='sim', action='store', default="",
80                      help="Print a list of maximum 10 string-similar symbols.")
81
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
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
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: "
98                 "\'commit1..commit2\'")
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
109    if opts.commit:
110        opts.find = False
111
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
118    return opts
119
120
121def main():
122    """Main function of this module."""
123    opts = parse_options()
124
125    global color
126    color = opts.color and sys.stdout.isatty()
127
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
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)
158        undefined_a, _ = check_symbols(opts.ignore)
159
160        # get undefined items for the commit
161        execute("git reset --hard %s" % commit_b)
162        undefined_b, defined = check_symbols(opts.ignore)
163
164        # report cases that are present for the commit but not before
165        for feature in sorted(undefined_b):
166            # feature has not been undefined before
167            if not feature in undefined_a:
168                files = sorted(undefined_b.get(feature))
169                undefined[feature] = files
170            # check if there are new files that reference the undefined feature
171            else:
172                files = sorted(undefined_b.get(feature) -
173                               undefined_a.get(feature))
174                if files:
175                    undefined[feature] = files
176
177        # reset to head
178        execute("git reset --hard %s" % head)
179
180    # default to check the entire tree
181    else:
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
208
209
210def yel(string):
211    """
212    Color %string yellow.
213    """
214    return "\033[33m%s\033[0m" % string if color else string
215
216
217def red(string):
218    """
219    Color %string red.
220    """
221    return "\033[31m%s\033[0m" % string if color else string
222
223
224def execute(cmd):
225    """Execute %cmd and return stdout.  Exit in case of error."""
226    try:
227        cmdlist = cmd.split(" ")
228        stdout = subprocess.check_output(cmdlist, stderr=subprocess.STDOUT, shell=False)
229        stdout = stdout.decode(errors='replace')
230    except subprocess.CalledProcessError as fail:
231        exit("Failed to execute %s\n%s" % (cmd, fail))
232    return stdout
233
234
235def find_commits(symbol, diff):
236    """Find commits changing %symbol in the given range of %diff."""
237    commits = execute("git log --pretty=oneline --abbrev-commit -G %s %s"
238                      % (symbol, diff))
239    return [x for x in commits.split("\n") if x]
240
241
242def tree_is_dirty():
243    """Return true if the current working tree is dirty (i.e., if any file has
244    been added, deleted, modified, renamed or copied but not committed)."""
245    stdout = execute("git status --porcelain")
246    for line in stdout:
247        if re.findall(r"[URMADC]{1}", line[:2]):
248            return True
249    return False
250
251
252def get_head():
253    """Return commit hash of current HEAD."""
254    stdout = execute("git rev-parse HEAD")
255    return stdout.strip('\n')
256
257
258def partition(lst, size):
259    """Partition list @lst into eveni-sized lists of size @size."""
260    return [lst[i::size] for i in range(size)]
261
262
263def init_worker():
264    """Set signal handler to ignore SIGINT."""
265    signal.signal(signal.SIGINT, signal.SIG_IGN)
266
267
268def find_sims(symbol, ignore, defined = []):
269    """Return a list of max. ten Kconfig symbols that are string-similar to
270    @symbol."""
271    if defined:
272        return sorted(difflib.get_close_matches(symbol, set(defined), 10))
273
274    pool = Pool(cpu_count(), init_worker)
275    kfiles = []
276    for gitfile in get_files():
277        if REGEX_FILE_KCONFIG.match(gitfile):
278            kfiles.append(gitfile)
279
280    arglist = []
281    for part in partition(kfiles, cpu_count()):
282        arglist.append((part, ignore))
283
284    for res in pool.map(parse_kconfig_files, arglist):
285        defined.extend(res[0])
286
287    return sorted(difflib.get_close_matches(symbol, set(defined), 10))
288
289
290def get_files():
291    """Return a list of all files in the current git directory."""
292    # use 'git ls-files' to get the worklist
293    stdout = execute("git ls-files")
294    if len(stdout) > 0 and stdout[-1] == "\n":
295        stdout = stdout[:-1]
296
297    files = []
298    for gitfile in stdout.rsplit("\n"):
299        if ".git" in gitfile or "ChangeLog" in gitfile or      \
300                ".log" in gitfile or os.path.isdir(gitfile) or \
301                gitfile.startswith("tools/"):
302            continue
303        files.append(gitfile)
304    return files
305
306
307def check_symbols(ignore):
308    """Find undefined Kconfig symbols and return a dict with the symbol as key
309    and a list of referencing files as value.  Files matching %ignore are not
310    checked for undefined symbols."""
311    pool = Pool(cpu_count(), init_worker)
312    try:
313        return check_symbols_helper(pool, ignore)
314    except KeyboardInterrupt:
315        pool.terminate()
316        pool.join()
317        sys.exit(1)
318
319
320def check_symbols_helper(pool, ignore):
321    """Helper method for check_symbols().  Used to catch keyboard interrupts in
322    check_symbols() in order to properly terminate running worker processes."""
323    source_files = []
324    kconfig_files = []
325    defined_features = []
326    referenced_features = dict()  # {file: [features]}
327
328    for gitfile in get_files():
329        if REGEX_FILE_KCONFIG.match(gitfile):
330            kconfig_files.append(gitfile)
331        else:
332            if ignore and not re.match(ignore, gitfile):
333                continue
334            # add source files that do not match the ignore pattern
335            source_files.append(gitfile)
336
337    # parse source files
338    arglist = partition(source_files, cpu_count())
339    for res in pool.map(parse_source_files, arglist):
340        referenced_features.update(res)
341
342
343    # parse kconfig files
344    arglist = []
345    for part in partition(kconfig_files, cpu_count()):
346        arglist.append((part, ignore))
347    for res in pool.map(parse_kconfig_files, arglist):
348        defined_features.extend(res[0])
349        referenced_features.update(res[1])
350    defined_features = set(defined_features)
351
352    # inverse mapping of referenced_features to dict(feature: [files])
353    inv_map = dict()
354    for _file, features in referenced_features.items():
355        for feature in features:
356            inv_map[feature] = inv_map.get(feature, set())
357            inv_map[feature].add(_file)
358    referenced_features = inv_map
359
360    undefined = {}  # {feature: [files]}
361    for feature in sorted(referenced_features):
362        # filter some false positives
363        if feature == "FOO" or feature == "BAR" or \
364                feature == "FOO_BAR" or feature == "XXX":
365            continue
366        if feature not in defined_features:
367            if feature.endswith("_MODULE"):
368                # avoid false positives for kernel modules
369                if feature[:-len("_MODULE")] in defined_features:
370                    continue
371            undefined[feature] = referenced_features.get(feature)
372    return undefined, defined_features
373
374
375def parse_source_files(source_files):
376    """Parse each source file in @source_files and return dictionary with source
377    files as keys and lists of references Kconfig symbols as values."""
378    referenced_features = dict()
379    for sfile in source_files:
380        referenced_features[sfile] = parse_source_file(sfile)
381    return referenced_features
382
383
384def parse_source_file(sfile):
385    """Parse @sfile and return a list of referenced Kconfig features."""
386    lines = []
387    references = []
388
389    if not os.path.exists(sfile):
390        return references
391
392    with open(sfile, "r", encoding='utf-8', errors='replace') as stream:
393        lines = stream.readlines()
394
395    for line in lines:
396        if not "CONFIG_" in line:
397            continue
398        features = REGEX_SOURCE_FEATURE.findall(line)
399        for feature in features:
400            if not REGEX_FILTER_FEATURES.search(feature):
401                continue
402            references.append(feature)
403
404    return references
405
406
407def get_features_in_line(line):
408    """Return mentioned Kconfig features in @line."""
409    return REGEX_FEATURE.findall(line)
410
411
412def parse_kconfig_files(args):
413    """Parse kconfig files and return tuple of defined and references Kconfig
414    symbols.  Note, @args is a tuple of a list of files and the @ignore
415    pattern."""
416    kconfig_files = args[0]
417    ignore = args[1]
418    defined_features = []
419    referenced_features = dict()
420
421    for kfile in kconfig_files:
422        defined, references = parse_kconfig_file(kfile)
423        defined_features.extend(defined)
424        if ignore and re.match(ignore, kfile):
425            # do not collect references for files that match the ignore pattern
426            continue
427        referenced_features[kfile] = references
428    return (defined_features, referenced_features)
429
430
431def parse_kconfig_file(kfile):
432    """Parse @kfile and update feature definitions and references."""
433    lines = []
434    defined = []
435    references = []
436    skip = False
437
438    if not os.path.exists(kfile):
439        return defined, references
440
441    with open(kfile, "r", encoding='utf-8', errors='replace') as stream:
442        lines = stream.readlines()
443
444    for i in range(len(lines)):
445        line = lines[i]
446        line = line.strip('\n')
447        line = line.split("#")[0]  # ignore comments
448
449        if REGEX_KCONFIG_DEF.match(line):
450            feature_def = REGEX_KCONFIG_DEF.findall(line)
451            defined.append(feature_def[0])
452            skip = False
453        elif REGEX_KCONFIG_HELP.match(line):
454            skip = True
455        elif skip:
456            # ignore content of help messages
457            pass
458        elif REGEX_KCONFIG_STMT.match(line):
459            line = REGEX_QUOTES.sub("", line)
460            features = get_features_in_line(line)
461            # multi-line statements
462            while line.endswith("\\"):
463                i += 1
464                line = lines[i]
465                line = line.strip('\n')
466                features.extend(get_features_in_line(line))
467            for feature in set(features):
468                if REGEX_NUMERIC.match(feature):
469                    # ignore numeric values
470                    continue
471                references.append(feature)
472
473    return defined, references
474
475
476if __name__ == "__main__":
477    main()
478