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