]>
Commit | Line | Data |
---|---|---|
2dbf1bf8 | 1 | #!/usr/bin/env perl |
a35beedf BB |
2 | # |
3 | # CDDL HEADER START | |
4 | # | |
5 | # The contents of this file are subject to the terms of the | |
6 | # Common Development and Distribution License (the "License"). | |
7 | # You may not use this file except in compliance with the License. | |
8 | # | |
9 | # You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE | |
1d3ba0bf | 10 | # or https://opensource.org/licenses/CDDL-1.0. |
a35beedf BB |
11 | # See the License for the specific language governing permissions |
12 | # and limitations under the License. | |
13 | # | |
14 | # When distributing Covered Code, include this CDDL HEADER in each | |
15 | # file and include the License file at usr/src/OPENSOLARIS.LICENSE. | |
16 | # If applicable, add the following below this CDDL HEADER, with the | |
17 | # fields enclosed by brackets "[]" replaced with your own identifying | |
18 | # information: Portions Copyright [yyyy] [name of copyright owner] | |
19 | # | |
20 | # CDDL HEADER END | |
21 | # | |
d96e5439 | 22 | # Copyright 2016 Nexenta Systems, Inc. |
a35beedf BB |
23 | # |
24 | # Copyright 2008 Sun Microsystems, Inc. All rights reserved. | |
25 | # Use is subject to license terms. | |
26 | # | |
27 | # @(#)cstyle 1.58 98/09/09 (from shannon) | |
28 | #ident "%Z%%M% %I% %E% SMI" | |
29 | # | |
30 | # cstyle - check for some common stylistic errors. | |
31 | # | |
32 | # cstyle is a sort of "lint" for C coding style. | |
33 | # It attempts to check for the style used in the | |
34 | # kernel, sometimes known as "Bill Joy Normal Form". | |
35 | # | |
36 | # There's a lot this can't check for, like proper indentation | |
37 | # of code blocks. There's also a lot more this could check for. | |
38 | # | |
39 | # A note to the non perl literate: | |
40 | # | |
41 | # perl regular expressions are pretty much like egrep | |
42 | # regular expressions, with the following special symbols | |
43 | # | |
44 | # \s any space character | |
45 | # \S any non-space character | |
46 | # \w any "word" character [a-zA-Z0-9_] | |
47 | # \W any non-word character | |
48 | # \d a digit [0-9] | |
49 | # \D a non-digit | |
50 | # \b word boundary (between \w and \W) | |
51 | # \B non-word boundary | |
52 | # | |
53 | ||
54 | require 5.0; | |
2dbf1bf8 | 55 | use warnings; |
a35beedf BB |
56 | use IO::File; |
57 | use Getopt::Std; | |
58 | use strict; | |
59 | ||
60 | my $usage = | |
95babbe5 | 61 | "usage: cstyle [-cgpvP] file... |
a35beedf | 62 | -c check continuation indentation inside functions |
42d4a8e5 | 63 | -g print github actions' workflow commands |
a35beedf BB |
64 | -p perform some of the more picky checks |
65 | -v verbose | |
a35beedf | 66 | -P check for use of non-POSIX types |
a35beedf BB |
67 | "; |
68 | ||
69 | my %opts; | |
70 | ||
7dc782e5 | 71 | if (!getopts("cghpvCP", \%opts)) { |
a35beedf BB |
72 | print $usage; |
73 | exit 2; | |
74 | } | |
75 | ||
76 | my $check_continuation = $opts{'c'}; | |
42d4a8e5 | 77 | my $github_workflow = $opts{'g'} || $ENV{'CI'}; |
a35beedf BB |
78 | my $picky = $opts{'p'}; |
79 | my $verbose = $opts{'v'}; | |
a35beedf BB |
80 | my $check_posix_types = $opts{'P'}; |
81 | ||
a35beedf BB |
82 | my ($filename, $line, $prev); # shared globals |
83 | ||
84 | my $fmt; | |
85 | my $hdr_comment_start; | |
86 | ||
87 | if ($verbose) { | |
88 | $fmt = "%s: %d: %s\n%s\n"; | |
89 | } else { | |
90 | $fmt = "%s: %d: %s\n"; | |
91 | } | |
92 | ||
7dc782e5 | 93 | $hdr_comment_start = qr/^\s*\/\*$/; |
a35beedf BB |
94 | |
95 | # Note, following must be in single quotes so that \s and \w work right. | |
96 | my $typename = '(int|char|short|long|unsigned|float|double' . | |
97 | '|\w+_t|struct\s+\w+|union\s+\w+|FILE)'; | |
98 | ||
99 | # mapping of old types to POSIX compatible types | |
100 | my %old2posix = ( | |
101 | 'unchar' => 'uchar_t', | |
102 | 'ushort' => 'ushort_t', | |
103 | 'uint' => 'uint_t', | |
104 | 'ulong' => 'ulong_t', | |
105 | 'u_int' => 'uint_t', | |
106 | 'u_short' => 'ushort_t', | |
107 | 'u_long' => 'ulong_t', | |
108 | 'u_char' => 'uchar_t', | |
109 | 'quad' => 'quad_t' | |
110 | ); | |
111 | ||
112 | my $lint_re = qr/\/\*(?: | |
15b98249 | 113 | NOTREACHED|LINTLIBRARY|VARARGS[0-9]*| |
a35beedf BB |
114 | CONSTCOND|CONSTANTCOND|CONSTANTCONDITION|EMPTY| |
115 | FALLTHRU|FALLTHROUGH|LINTED.*?|PRINTFLIKE[0-9]*| | |
116 | PROTOLIB[0-9]*|SCANFLIKE[0-9]*|CSTYLED.*? | |
117 | )\*\//x; | |
118 | ||
a35beedf BB |
119 | my $warlock_re = qr/\/\*\s*(?: |
120 | VARIABLES\ PROTECTED\ BY| | |
121 | MEMBERS\ PROTECTED\ BY| | |
122 | ALL\ MEMBERS\ PROTECTED\ BY| | |
123 | READ-ONLY\ VARIABLES:| | |
124 | READ-ONLY\ MEMBERS:| | |
125 | VARIABLES\ READABLE\ WITHOUT\ LOCK:| | |
126 | MEMBERS\ READABLE\ WITHOUT\ LOCK:| | |
127 | LOCKS\ COVERED\ BY| | |
128 | LOCK\ UNNEEDED\ BECAUSE| | |
129 | LOCK\ NEEDED:| | |
130 | LOCK\ HELD\ ON\ ENTRY:| | |
131 | READ\ LOCK\ HELD\ ON\ ENTRY:| | |
132 | WRITE\ LOCK\ HELD\ ON\ ENTRY:| | |
133 | LOCK\ ACQUIRED\ AS\ SIDE\ EFFECT:| | |
134 | READ\ LOCK\ ACQUIRED\ AS\ SIDE\ EFFECT:| | |
135 | WRITE\ LOCK\ ACQUIRED\ AS\ SIDE\ EFFECT:| | |
136 | LOCK\ RELEASED\ AS\ SIDE\ EFFECT:| | |
137 | LOCK\ UPGRADED\ AS\ SIDE\ EFFECT:| | |
138 | LOCK\ DOWNGRADED\ AS\ SIDE\ EFFECT:| | |
139 | FUNCTIONS\ CALLED\ THROUGH\ POINTER| | |
140 | FUNCTIONS\ CALLED\ THROUGH\ MEMBER| | |
141 | LOCK\ ORDER: | |
142 | )/x; | |
143 | ||
144 | my $err_stat = 0; # exit status | |
145 | ||
146 | if ($#ARGV >= 0) { | |
147 | foreach my $arg (@ARGV) { | |
148 | my $fh = new IO::File $arg, "r"; | |
149 | if (!defined($fh)) { | |
150 | printf "%s: can not open\n", $arg; | |
151 | } else { | |
152 | &cstyle($arg, $fh); | |
153 | close $fh; | |
154 | } | |
155 | } | |
156 | } else { | |
157 | &cstyle("<stdin>", *STDIN); | |
158 | } | |
159 | exit $err_stat; | |
160 | ||
161 | my $no_errs = 0; # set for CSTYLED-protected lines | |
162 | ||
163 | sub err($) { | |
164 | my ($error) = @_; | |
165 | unless ($no_errs) { | |
f866a4ea BB |
166 | if ($verbose) { |
167 | printf $fmt, $filename, $., $error, $line; | |
168 | } else { | |
169 | printf $fmt, $filename, $., $error; | |
42d4a8e5 GM |
170 | } |
171 | if ($github_workflow) { | |
172 | printf "::error file=%s,line=%s::%s\n", $filename, $., $error; | |
173 | } | |
a35beedf BB |
174 | $err_stat = 1; |
175 | } | |
176 | } | |
177 | ||
178 | sub err_prefix($$) { | |
179 | my ($prevline, $error) = @_; | |
180 | my $out = $prevline."\n".$line; | |
181 | unless ($no_errs) { | |
f866a4ea BB |
182 | if ($verbose) { |
183 | printf $fmt, $filename, $., $error, $out; | |
184 | } else { | |
185 | printf $fmt, $filename, $., $error; | |
186 | } | |
a35beedf BB |
187 | $err_stat = 1; |
188 | } | |
189 | } | |
190 | ||
191 | sub err_prev($) { | |
192 | my ($error) = @_; | |
193 | unless ($no_errs) { | |
f866a4ea BB |
194 | if ($verbose) { |
195 | printf $fmt, $filename, $. - 1, $error, $prev; | |
196 | } else { | |
197 | printf $fmt, $filename, $. - 1, $error; | |
198 | } | |
a35beedf BB |
199 | $err_stat = 1; |
200 | } | |
201 | } | |
202 | ||
203 | sub cstyle($$) { | |
204 | ||
205 | my ($fn, $filehandle) = @_; | |
206 | $filename = $fn; # share it globally | |
207 | ||
208 | my $in_cpp = 0; | |
209 | my $next_in_cpp = 0; | |
210 | ||
211 | my $in_comment = 0; | |
a35beedf BB |
212 | my $comment_done = 0; |
213 | my $in_warlock_comment = 0; | |
214 | my $in_function = 0; | |
215 | my $in_function_header = 0; | |
ec441a9c | 216 | my $function_header_full_indent = 0; |
a35beedf BB |
217 | my $in_declaration = 0; |
218 | my $note_level = 0; | |
219 | my $nextok = 0; | |
220 | my $nocheck = 0; | |
221 | ||
222 | my $in_string = 0; | |
223 | ||
224 | my ($okmsg, $comment_prefix); | |
225 | ||
226 | $line = ''; | |
227 | $prev = ''; | |
228 | reset_indent(); | |
229 | ||
230 | line: while (<$filehandle>) { | |
231 | s/\r?\n$//; # strip return and newline | |
232 | ||
233 | # save the original line, then remove all text from within | |
234 | # double or single quotes, we do not want to check such text. | |
235 | ||
236 | $line = $_; | |
237 | ||
238 | # | |
239 | # C allows strings to be continued with a backslash at the end of | |
240 | # the line. We translate that into a quoted string on the previous | |
241 | # line followed by an initial quote on the next line. | |
242 | # | |
243 | # (we assume that no-one will use backslash-continuation with character | |
244 | # constants) | |
245 | # | |
246 | $_ = '"' . $_ if ($in_string && !$nocheck && !$in_comment); | |
247 | ||
248 | # | |
249 | # normal strings and characters | |
250 | # | |
251 | s/'([^\\']|\\[^xX0]|\\0[0-9]*|\\[xX][0-9a-fA-F]*)'/''/g; | |
252 | s/"([^\\"]|\\.)*"/\"\"/g; | |
253 | ||
254 | # | |
255 | # detect string continuation | |
256 | # | |
257 | if ($nocheck || $in_comment) { | |
258 | $in_string = 0; | |
259 | } else { | |
260 | # | |
261 | # Now that all full strings are replaced with "", we check | |
262 | # for unfinished strings continuing onto the next line. | |
263 | # | |
264 | $in_string = | |
265 | (s/([^"](?:"")*)"([^\\"]|\\.)*\\$/$1""/ || | |
266 | s/^("")*"([^\\"]|\\.)*\\$/""/); | |
267 | } | |
268 | ||
269 | # | |
270 | # figure out if we are in a cpp directive | |
271 | # | |
272 | $in_cpp = $next_in_cpp || /^\s*#/; # continued or started | |
273 | $next_in_cpp = $in_cpp && /\\$/; # only if continued | |
274 | ||
275 | # strip off trailing backslashes, which appear in long macros | |
276 | s/\s*\\$//; | |
277 | ||
278 | # an /* END CSTYLED */ comment ends a no-check block. | |
279 | if ($nocheck) { | |
280 | if (/\/\* *END *CSTYLED *\*\//) { | |
281 | $nocheck = 0; | |
282 | } else { | |
283 | reset_indent(); | |
284 | next line; | |
285 | } | |
286 | } | |
287 | ||
288 | # a /*CSTYLED*/ comment indicates that the next line is ok. | |
289 | if ($nextok) { | |
290 | if ($okmsg) { | |
291 | err($okmsg); | |
292 | } | |
293 | $nextok = 0; | |
294 | $okmsg = 0; | |
295 | if (/\/\* *CSTYLED.*\*\//) { | |
296 | /^.*\/\* *CSTYLED *(.*) *\*\/.*$/; | |
297 | $okmsg = $1; | |
298 | $nextok = 1; | |
299 | } | |
300 | $no_errs = 1; | |
301 | } elsif ($no_errs) { | |
302 | $no_errs = 0; | |
303 | } | |
304 | ||
305 | # check length of line. | |
306 | # first, a quick check to see if there is any chance of being too long. | |
307 | if (($line =~ tr/\t/\t/) * 7 + length($line) > 80) { | |
308 | # yes, there is a chance. | |
309 | # replace tabs with spaces and check again. | |
310 | my $eline = $line; | |
311 | 1 while $eline =~ | |
312 | s/\t+/' ' x (length($&) * 8 - length($`) % 8)/e; | |
313 | if (length($eline) > 80) { | |
314 | err("line > 80 characters"); | |
315 | } | |
316 | } | |
317 | ||
318 | # ignore NOTE(...) annotations (assumes NOTE is on lines by itself). | |
319 | if ($note_level || /\b_?NOTE\s*\(/) { # if in NOTE or this is NOTE | |
320 | s/[^()]//g; # eliminate all non-parens | |
321 | $note_level += s/\(//g - length; # update paren nest level | |
322 | next; | |
323 | } | |
324 | ||
325 | # a /* BEGIN CSTYLED */ comment starts a no-check block. | |
326 | if (/\/\* *BEGIN *CSTYLED *\*\//) { | |
327 | $nocheck = 1; | |
328 | } | |
329 | ||
330 | # a /*CSTYLED*/ comment indicates that the next line is ok. | |
331 | if (/\/\* *CSTYLED.*\*\//) { | |
332 | /^.*\/\* *CSTYLED *(.*) *\*\/.*$/; | |
333 | $okmsg = $1; | |
334 | $nextok = 1; | |
335 | } | |
336 | if (/\/\/ *CSTYLED/) { | |
337 | /^.*\/\/ *CSTYLED *(.*)$/; | |
338 | $okmsg = $1; | |
339 | $nextok = 1; | |
340 | } | |
341 | ||
342 | # universal checks; apply to everything | |
343 | if (/\t +\t/) { | |
344 | err("spaces between tabs"); | |
345 | } | |
346 | if (/ \t+ /) { | |
347 | err("tabs between spaces"); | |
348 | } | |
349 | if (/\s$/) { | |
350 | err("space or tab at end of line"); | |
351 | } | |
352 | if (/[^ \t(]\/\*/ && !/\w\(\/\*.*\*\/\);/) { | |
353 | err("comment preceded by non-blank"); | |
354 | } | |
15b98249 AZ |
355 | if (/ARGSUSED/) { |
356 | err("ARGSUSED directive"); | |
357 | } | |
a35beedf BB |
358 | |
359 | # is this the beginning or ending of a function? | |
360 | # (not if "struct foo\n{\n") | |
2e7c1bb3 | 361 | if (/^\{$/ && $prev =~ /\)\s*(const\s*)?(\/\*.*\*\/\s*)?\\?$/) { |
a35beedf BB |
362 | $in_function = 1; |
363 | $in_declaration = 1; | |
364 | $in_function_header = 0; | |
ec441a9c | 365 | $function_header_full_indent = 0; |
a35beedf BB |
366 | $prev = $line; |
367 | next line; | |
368 | } | |
2e7c1bb3 | 369 | if (/^\}\s*(\/\*.*\*\/\s*)*$/) { |
a35beedf BB |
370 | if ($prev =~ /^\s*return\s*;/) { |
371 | err_prev("unneeded return at end of function"); | |
372 | } | |
373 | $in_function = 0; | |
374 | reset_indent(); # we don't check between functions | |
375 | $prev = $line; | |
376 | next line; | |
377 | } | |
d96e5439 | 378 | if ($in_function_header && ! /^ (\w|\.)/ ) { |
2e7c1bb3 | 379 | if (/^\{\}$/ # empty functions |
ec441a9c PD |
380 | || /;/ #run function with multiline arguments |
381 | || /#/ #preprocessor commands | |
382 | || /^[^\s\\]*\(.*\)$/ #functions without ; at the end | |
383 | || /^$/ #function declaration can't have empty line | |
384 | ) { | |
385 | $in_function_header = 0; | |
386 | $function_header_full_indent = 0; | |
387 | } elsif ($prev =~ /^__attribute__/) { #__attribute__((*)) | |
388 | $in_function_header = 0; | |
389 | $function_header_full_indent = 0; | |
390 | $prev = $line; | |
391 | next line; | |
392 | } elsif ($picky && ! (/^\t/ && $function_header_full_indent != 0)) { | |
42d4a8e5 | 393 | |
ec441a9c PD |
394 | err("continuation line should be indented by 4 spaces"); |
395 | } | |
396 | } | |
397 | ||
398 | # | |
399 | # If this matches something of form "foo(", it's probably a function | |
400 | # definition, unless it ends with ") bar;", in which case it's a declaration | |
401 | # that uses a macro to generate the type. | |
402 | # | |
403 | if (/^\w+\(/ && !/\) \w+;/) { | |
a35beedf | 404 | $in_function_header = 1; |
ec441a9c PD |
405 | if (/\($/) { |
406 | $function_header_full_indent = 1; | |
407 | } | |
408 | } | |
2e7c1bb3 | 409 | if ($in_function_header && /^\{$/) { |
ec441a9c PD |
410 | $in_function_header = 0; |
411 | $function_header_full_indent = 0; | |
412 | $in_function = 1; | |
413 | } | |
414 | if ($in_function_header && /\);$/) { | |
415 | $in_function_header = 0; | |
416 | $function_header_full_indent = 0; | |
417 | } | |
2e7c1bb3 | 418 | if ($in_function_header && /\{$/ ) { |
d4e00407 | 419 | if ($picky) { |
ec441a9c PD |
420 | err("opening brace on same line as function header"); |
421 | } | |
422 | $in_function_header = 0; | |
423 | $function_header_full_indent = 0; | |
424 | $in_function = 1; | |
425 | next line; | |
a35beedf BB |
426 | } |
427 | ||
428 | if ($in_warlock_comment && /\*\//) { | |
429 | $in_warlock_comment = 0; | |
430 | $prev = $line; | |
431 | next line; | |
432 | } | |
433 | ||
434 | # a blank line terminates the declarations within a function. | |
435 | # XXX - but still a problem in sub-blocks. | |
436 | if ($in_declaration && /^$/) { | |
437 | $in_declaration = 0; | |
438 | } | |
439 | ||
440 | if ($comment_done) { | |
441 | $in_comment = 0; | |
a35beedf BB |
442 | $comment_done = 0; |
443 | } | |
444 | # does this looks like the start of a block comment? | |
445 | if (/$hdr_comment_start/) { | |
446 | if (!/^\t*\/\*/) { | |
447 | err("block comment not indented by tabs"); | |
448 | } | |
449 | $in_comment = 1; | |
450 | /^(\s*)\//; | |
451 | $comment_prefix = $1; | |
a35beedf BB |
452 | $prev = $line; |
453 | next line; | |
454 | } | |
455 | # are we still in the block comment? | |
456 | if ($in_comment) { | |
457 | if (/^$comment_prefix \*\/$/) { | |
458 | $comment_done = 1; | |
459 | } elsif (/\*\//) { | |
460 | $comment_done = 1; | |
93f3a351 | 461 | err("improper block comment close"); |
a35beedf BB |
462 | } elsif (!/^$comment_prefix \*[ \t]/ && |
463 | !/^$comment_prefix \*$/) { | |
93f3a351 | 464 | err("improper block comment"); |
a35beedf BB |
465 | } |
466 | } | |
467 | ||
a35beedf BB |
468 | # check for errors that might occur in comments and in code. |
469 | ||
8ffef572 BB |
470 | # allow spaces to be used to draw pictures in all comments. |
471 | if (/[^ ] / && !/".* .*"/ && !$in_comment) { | |
a35beedf BB |
472 | err("spaces instead of tabs"); |
473 | } | |
474 | if (/^ / && !/^ \*[ \t\/]/ && !/^ \*$/ && | |
d96e5439 | 475 | (!/^ (\w|\.)/ || $in_function != 0)) { |
a35beedf BB |
476 | err("indent by spaces instead of tabs"); |
477 | } | |
478 | if (/^\t+ [^ \t\*]/ || /^\t+ \S/ || /^\t+ \S/) { | |
479 | err("continuation line not indented by 4 spaces"); | |
480 | } | |
481 | if (/$warlock_re/ && !/\*\//) { | |
482 | $in_warlock_comment = 1; | |
483 | $prev = $line; | |
484 | next line; | |
485 | } | |
486 | if (/^\s*\/\*./ && !/^\s*\/\*.*\*\// && !/$hdr_comment_start/) { | |
487 | err("improper first line of block comment"); | |
488 | } | |
489 | ||
490 | if ($in_comment) { # still in comment, don't do further checks | |
491 | $prev = $line; | |
492 | next line; | |
493 | } | |
494 | ||
7dc782e5 | 495 | if ((/[^(]\/\*\S/ || /^\/\*\S/) && !/$lint_re/) { |
a35beedf BB |
496 | err("missing blank after open comment"); |
497 | } | |
7dc782e5 | 498 | if (/\S\*\/[^)]|\S\*\/$/ && !/$lint_re/) { |
a35beedf BB |
499 | err("missing blank before close comment"); |
500 | } | |
501 | if (/\/\/\S/) { # C++ comments | |
502 | err("missing blank after start comment"); | |
503 | } | |
504 | # check for unterminated single line comments, but allow them when | |
505 | # they are used to comment out the argument list of a function | |
506 | # declaration. | |
507 | if (/\S.*\/\*/ && !/\S.*\/\*.*\*\// && !/\(\/\*/) { | |
508 | err("unterminated single line comment"); | |
509 | } | |
510 | ||
511 | if (/^(#else|#endif|#include)(.*)$/) { | |
512 | $prev = $line; | |
513 | if ($picky) { | |
514 | my $directive = $1; | |
515 | my $clause = $2; | |
516 | # Enforce ANSI rules for #else and #endif: no noncomment | |
517 | # identifiers are allowed after #endif or #else. Allow | |
518 | # C++ comments since they seem to be a fact of life. | |
519 | if ((($1 eq "#endif") || ($1 eq "#else")) && | |
520 | ($clause ne "") && | |
521 | (!($clause =~ /^\s+\/\*.*\*\/$/)) && | |
522 | (!($clause =~ /^\s+\/\/.*$/))) { | |
523 | err("non-comment text following " . | |
524 | "$directive (or malformed $directive " . | |
525 | "directive)"); | |
526 | } | |
527 | } | |
528 | next line; | |
529 | } | |
530 | ||
531 | # | |
532 | # delete any comments and check everything else. Note that | |
533 | # ".*?" is a non-greedy match, so that we don't get confused by | |
534 | # multiple comments on the same line. | |
535 | # | |
536 | s/\/\*.*?\*\//\ 1/g; | |
537 | s/\/\/.*$/\ 1/; # C++ comments | |
538 | ||
539 | # delete any trailing whitespace; we have already checked for that. | |
540 | s/\s*$//; | |
541 | ||
542 | # following checks do not apply to text in comments. | |
543 | ||
544 | if (/[^<>\s][!<>=]=/ || /[^<>][!<>=]=[^\s,]/ || | |
545 | (/[^->]>[^,=>\s]/ && !/[^->]>$/) || | |
546 | (/[^<]<[^,=<\s]/ && !/[^<]<$/) || | |
547 | /[^<\s]<[^<]/ || /[^->\s]>[^>]/) { | |
548 | err("missing space around relational operator"); | |
549 | } | |
550 | if (/\S>>=/ || /\S<<=/ || />>=\S/ || /<<=\S/ || /\S[-+*\/&|^%]=/ || | |
551 | (/[^-+*\/&|^%!<>=\s]=[^=]/ && !/[^-+*\/&|^%!<>=\s]=$/) || | |
552 | (/[^!<>=]=[^=\s]/ && !/[^!<>=]=$/)) { | |
553 | # XXX - should only check this for C++ code | |
554 | # XXX - there are probably other forms that should be allowed | |
555 | if (!/\soperator=/) { | |
556 | err("missing space around assignment operator"); | |
557 | } | |
558 | } | |
559 | if (/[,;]\S/ && !/\bfor \(;;\)/) { | |
560 | err("comma or semicolon followed by non-blank"); | |
561 | } | |
562 | # allow "for" statements to have empty "while" clauses | |
563 | if (/\s[,;]/ && !/^[\t]+;$/ && !/^\s*for \([^;]*; ;[^;]*\)/) { | |
564 | err("comma or semicolon preceded by blank"); | |
565 | } | |
566 | if (/^\s*(&&|\|\|)/) { | |
567 | err("improper boolean continuation"); | |
568 | } | |
569 | if (/\S *(&&|\|\|)/ || /(&&|\|\|) *\S/) { | |
570 | err("more than one space around boolean operator"); | |
571 | } | |
572 | if (/\b(for|if|while|switch|sizeof|return|case)\(/) { | |
573 | err("missing space between keyword and paren"); | |
574 | } | |
575 | if (/(\b(for|if|while|switch|return)\b.*){2,}/ && !/^#define/) { | |
576 | # multiple "case" and "sizeof" allowed | |
577 | err("more than one keyword on line"); | |
578 | } | |
579 | if (/\b(for|if|while|switch|sizeof|return|case)\s\s+\(/ && | |
580 | !/^#if\s+\(/) { | |
581 | err("extra space between keyword and paren"); | |
582 | } | |
583 | # try to detect "func (x)" but not "if (x)" or | |
584 | # "#define foo (x)" or "int (*func)();" | |
585 | if (/\w\s\(/) { | |
586 | my $s = $_; | |
587 | # strip off all keywords on the line | |
588 | s/\b(for|if|while|switch|return|case|sizeof)\s\(/XXX(/g; | |
589 | s/#elif\s\(/XXX(/g; | |
590 | s/^#define\s+\w+\s+\(/XXX(/; | |
591 | # do not match things like "void (*f)();" | |
592 | # or "typedef void (func_t)();" | |
593 | s/\w\s\(+\*/XXX(*/g; | |
594 | s/\b($typename|void)\s+\(+/XXX(/og; | |
595 | if (/\w\s\(/) { | |
596 | err("extra space between function name and left paren"); | |
597 | } | |
598 | $_ = $s; | |
599 | } | |
600 | # try to detect "int foo(x)", but not "extern int foo(x);" | |
601 | # XXX - this still trips over too many legitimate things, | |
602 | # like "int foo(x,\n\ty);" | |
603 | # if (/^(\w+(\s|\*)+)+\w+\(/ && !/\)[;,](\s|\ 1)*$/ && | |
604 | # !/^(extern|static)\b/) { | |
605 | # err("return type of function not on separate line"); | |
606 | # } | |
607 | # this is a close approximation | |
608 | if (/^(\w+(\s|\*)+)+\w+\(.*\)(\s|\ 1)*$/ && | |
609 | !/^(extern|static)\b/) { | |
610 | err("return type of function not on separate line"); | |
611 | } | |
612 | if (/^#define /) { | |
613 | err("#define followed by space instead of tab"); | |
614 | } | |
615 | if (/^\s*return\W[^;]*;/ && !/^\s*return\s*\(.*\);/) { | |
616 | err("unparenthesized return expression"); | |
617 | } | |
618 | if (/\bsizeof\b/ && !/\bsizeof\s*\(.*\)/) { | |
619 | err("unparenthesized sizeof expression"); | |
620 | } | |
621 | if (/\(\s/) { | |
622 | err("whitespace after left paren"); | |
623 | } | |
50240467 NB |
624 | # Allow "for" statements to have empty "continue" clauses. |
625 | # Allow right paren on its own line unless we're being picky (-p). | |
626 | if (/\s\)/ && !/^\s*for \([^;]*;[^;]*; \)/ && ($picky || !/^\s*\)/)) { | |
a35beedf BB |
627 | err("whitespace before right paren"); |
628 | } | |
629 | if (/^\s*\(void\)[^ ]/) { | |
630 | err("missing space after (void) cast"); | |
631 | } | |
f866a4ea | 632 | if (/\S\{/ && !/\{\{/) { |
a35beedf BB |
633 | err("missing space before left brace"); |
634 | } | |
2e7c1bb3 | 635 | if ($in_function && /^\s+\{/ && |
a35beedf BB |
636 | ($prev =~ /\)\s*$/ || $prev =~ /\bstruct\s+\w+$/)) { |
637 | err("left brace starting a line"); | |
638 | } | |
2e7c1bb3 | 639 | if (/\}(else|while)/) { |
a35beedf BB |
640 | err("missing space after right brace"); |
641 | } | |
2e7c1bb3 | 642 | if (/\}\s\s+(else|while)/) { |
a35beedf BB |
643 | err("extra space after right brace"); |
644 | } | |
645 | if (/\b_VOID\b|\bVOID\b|\bSTATIC\b/) { | |
646 | err("obsolete use of VOID or STATIC"); | |
647 | } | |
648 | if (/\b$typename\*/o) { | |
649 | err("missing space between type name and *"); | |
650 | } | |
651 | if (/^\s+#/) { | |
652 | err("preprocessor statement not in column 1"); | |
653 | } | |
654 | if (/^#\s/) { | |
655 | err("blank after preprocessor #"); | |
656 | } | |
657 | if (/!\s*(strcmp|strncmp|bcmp)\s*\(/) { | |
658 | err("don't use boolean ! with comparison functions"); | |
659 | } | |
660 | ||
661 | # | |
662 | # We completely ignore, for purposes of indentation: | |
663 | # * lines outside of functions | |
664 | # * preprocessor lines | |
665 | # | |
666 | if ($check_continuation && $in_function && !$in_cpp) { | |
667 | process_indent($_); | |
668 | } | |
669 | if ($picky) { | |
670 | # try to detect spaces after casts, but allow (e.g.) | |
671 | # "sizeof (int) + 1", "void (*funcptr)(int) = foo;", and | |
672 | # "int foo(int) __NORETURN;" | |
673 | if ((/^\($typename( \*+)?\)\s/o || | |
674 | /\W\($typename( \*+)?\)\s/o) && | |
675 | !/sizeof\s*\($typename( \*)?\)\s/o && | |
676 | !/\($typename( \*+)?\)\s+=[^=]/o) { | |
677 | err("space after cast"); | |
678 | } | |
679 | if (/\b$typename\s*\*\s/o && | |
680 | !/\b$typename\s*\*\s+const\b/o) { | |
681 | err("unary * followed by space"); | |
682 | } | |
683 | } | |
684 | if ($check_posix_types) { | |
685 | # try to detect old non-POSIX types. | |
686 | # POSIX requires all non-standard typedefs to end in _t, | |
687 | # but historically these have been used. | |
688 | if (/\b(unchar|ushort|uint|ulong|u_int|u_short|u_long|u_char|quad)\b/) { | |
689 | err("non-POSIX typedef $1 used: use $old2posix{$1} instead"); | |
690 | } | |
691 | } | |
a35beedf | 692 | if (/^\s*else\W/) { |
2e7c1bb3 | 693 | if ($prev =~ /^\s*\}$/) { |
a35beedf BB |
694 | err_prefix($prev, |
695 | "else and right brace should be on same line"); | |
696 | } | |
697 | } | |
698 | $prev = $line; | |
699 | } | |
700 | ||
701 | if ($prev eq "") { | |
702 | err("last line in file is blank"); | |
703 | } | |
704 | ||
705 | } | |
706 | ||
707 | # | |
708 | # Continuation-line checking | |
709 | # | |
710 | # The rest of this file contains the code for the continuation checking | |
711 | # engine. It's a pretty simple state machine which tracks the expression | |
712 | # depth (unmatched '('s and '['s). | |
713 | # | |
714 | # Keep in mind that the argument to process_indent() has already been heavily | |
715 | # processed; all comments have been replaced by control-A, and the contents of | |
716 | # strings and character constants have been elided. | |
717 | # | |
718 | ||
719 | my $cont_in; # currently inside of a continuation | |
720 | my $cont_off; # skipping an initializer or definition | |
721 | my $cont_noerr; # suppress cascading errors | |
722 | my $cont_start; # the line being continued | |
723 | my $cont_base; # the base indentation | |
724 | my $cont_first; # this is the first line of a statement | |
725 | my $cont_multiseg; # this continuation has multiple segments | |
726 | ||
727 | my $cont_special; # this is a C statement (if, for, etc.) | |
728 | my $cont_macro; # this is a macro | |
729 | my $cont_case; # this is a multi-line case | |
730 | ||
731 | my @cont_paren; # the stack of unmatched ( and [s we've seen | |
732 | ||
733 | sub | |
734 | reset_indent() | |
735 | { | |
736 | $cont_in = 0; | |
737 | $cont_off = 0; | |
738 | } | |
739 | ||
740 | sub | |
741 | delabel($) | |
742 | { | |
743 | # | |
744 | # replace labels with tabs. Note that there may be multiple | |
745 | # labels on a line. | |
746 | # | |
747 | local $_ = $_[0]; | |
748 | ||
749 | while (/^(\t*)( *(?:(?:\w+\s*)|(?:case\b[^:]*)): *)(.*)$/) { | |
750 | my ($pre_tabs, $label, $rest) = ($1, $2, $3); | |
751 | $_ = $pre_tabs; | |
752 | while ($label =~ s/^([^\t]*)(\t+)//) { | |
753 | $_ .= "\t" x (length($2) + length($1) / 8); | |
754 | } | |
755 | $_ .= ("\t" x (length($label) / 8)).$rest; | |
756 | } | |
757 | ||
758 | return ($_); | |
759 | } | |
760 | ||
761 | sub | |
762 | process_indent($) | |
763 | { | |
764 | require strict; | |
765 | local $_ = $_[0]; # preserve the global $_ | |
766 | ||
767 | s/\ 1//g; # No comments | |
768 | s/\s+$//; # Strip trailing whitespace | |
769 | ||
770 | return if (/^$/); # skip empty lines | |
771 | ||
772 | # regexps used below; keywords taking (), macros, and continued cases | |
773 | my $special = '(?:(?:\}\s*)?else\s+)?(?:if|for|while|switch)\b'; | |
774 | my $macro = '[A-Z_][A-Z_0-9]*\('; | |
775 | my $case = 'case\b[^:]*$'; | |
776 | ||
777 | # skip over enumerations, array definitions, initializers, etc. | |
778 | if ($cont_off <= 0 && !/^\s*$special/ && | |
2e7c1bb3 DH |
779 | (/(?:(?:\b(?:enum|struct|union)\s*[^\{]*)|(?:\s+=\s*))\{/ || |
780 | (/^\s*\{/ && $prev =~ /=\s*(?:\/\*.*\*\/\s*)*$/))) { | |
a35beedf BB |
781 | $cont_in = 0; |
782 | $cont_off = tr/{/{/ - tr/}/}/; | |
783 | return; | |
784 | } | |
785 | if ($cont_off) { | |
786 | $cont_off += tr/{/{/ - tr/}/}/; | |
787 | return; | |
788 | } | |
789 | ||
790 | if (!$cont_in) { | |
791 | $cont_start = $line; | |
792 | ||
793 | if (/^\t* /) { | |
794 | err("non-continuation indented 4 spaces"); | |
795 | $cont_noerr = 1; # stop reporting | |
796 | } | |
797 | $_ = delabel($_); # replace labels with tabs | |
798 | ||
799 | # check if the statement is complete | |
800 | return if (/^\s*\}?$/); | |
801 | return if (/^\s*\}?\s*else\s*\{?$/); | |
802 | return if (/^\s*do\s*\{?$/); | |
2e7c1bb3 DH |
803 | return if (/\{$/); |
804 | return if (/\}[,;]?$/); | |
a35beedf BB |
805 | |
806 | # Allow macros on their own lines | |
807 | return if (/^\s*[A-Z_][A-Z_0-9]*$/); | |
808 | ||
809 | # cases we don't deal with, generally non-kosher | |
2e7c1bb3 | 810 | if (/\{/) { |
a35beedf BB |
811 | err("stuff after {"); |
812 | return; | |
813 | } | |
814 | ||
815 | # Get the base line, and set up the state machine | |
816 | /^(\t*)/; | |
817 | $cont_base = $1; | |
818 | $cont_in = 1; | |
819 | @cont_paren = (); | |
820 | $cont_first = 1; | |
821 | $cont_multiseg = 0; | |
822 | ||
823 | # certain things need special processing | |
824 | $cont_special = /^\s*$special/? 1 : 0; | |
825 | $cont_macro = /^\s*$macro/? 1 : 0; | |
826 | $cont_case = /^\s*$case/? 1 : 0; | |
827 | } else { | |
828 | $cont_first = 0; | |
829 | ||
830 | # Strings may be pulled back to an earlier (half-)tabstop | |
831 | unless ($cont_noerr || /^$cont_base / || | |
832 | (/^\t*(?: )?(?:gettext\()?\"/ && !/^$cont_base\t/)) { | |
833 | err_prefix($cont_start, | |
834 | "continuation should be indented 4 spaces"); | |
835 | } | |
836 | } | |
837 | ||
838 | my $rest = $_; # keeps the remainder of the line | |
839 | ||
840 | # | |
841 | # The split matches 0 characters, so that each 'special' character | |
842 | # is processed separately. Parens and brackets are pushed and | |
843 | # popped off the @cont_paren stack. For normal processing, we wait | |
844 | # until a ; or { terminates the statement. "special" processing | |
845 | # (if/for/while/switch) is allowed to stop when the stack empties, | |
846 | # as is macro processing. Case statements are terminated with a : | |
847 | # and an empty paren stack. | |
848 | # | |
849 | foreach $_ (split /[^\(\)\[\]\{\}\;\:]*/) { | |
850 | next if (length($_) == 0); | |
851 | ||
852 | # rest contains the remainder of the line | |
853 | my $rxp = "[^\Q$_\E]*\Q$_\E"; | |
854 | $rest =~ s/^$rxp//; | |
855 | ||
856 | if (/\(/ || /\[/) { | |
857 | push @cont_paren, $_; | |
858 | } elsif (/\)/ || /\]/) { | |
859 | my $cur = $_; | |
860 | tr/\)\]/\(\[/; | |
861 | ||
862 | my $old = (pop @cont_paren); | |
863 | if (!defined($old)) { | |
864 | err("unexpected '$cur'"); | |
865 | $cont_in = 0; | |
866 | last; | |
867 | } elsif ($old ne $_) { | |
868 | err("'$cur' mismatched with '$old'"); | |
869 | $cont_in = 0; | |
870 | last; | |
871 | } | |
872 | ||
873 | # | |
874 | # If the stack is now empty, do special processing | |
875 | # for if/for/while/switch and macro statements. | |
876 | # | |
877 | next if (@cont_paren != 0); | |
878 | if ($cont_special) { | |
2e7c1bb3 | 879 | if ($rest =~ /^\s*\{?$/) { |
a35beedf BB |
880 | $cont_in = 0; |
881 | last; | |
882 | } | |
883 | if ($rest =~ /^\s*;$/) { | |
884 | err("empty if/for/while body ". | |
885 | "not on its own line"); | |
886 | $cont_in = 0; | |
887 | last; | |
888 | } | |
889 | if (!$cont_first && $cont_multiseg == 1) { | |
890 | err_prefix($cont_start, | |
891 | "multiple statements continued ". | |
892 | "over multiple lines"); | |
893 | $cont_multiseg = 2; | |
894 | } elsif ($cont_multiseg == 0) { | |
895 | $cont_multiseg = 1; | |
896 | } | |
897 | # We've finished this section, start | |
898 | # processing the next. | |
899 | goto section_ended; | |
900 | } | |
901 | if ($cont_macro) { | |
902 | if ($rest =~ /^$/) { | |
903 | $cont_in = 0; | |
904 | last; | |
905 | } | |
906 | } | |
907 | } elsif (/\;/) { | |
908 | if ($cont_case) { | |
909 | err("unexpected ;"); | |
910 | } elsif (!$cont_special) { | |
911 | err("unexpected ;") if (@cont_paren != 0); | |
912 | if (!$cont_first && $cont_multiseg == 1) { | |
913 | err_prefix($cont_start, | |
914 | "multiple statements continued ". | |
915 | "over multiple lines"); | |
916 | $cont_multiseg = 2; | |
917 | } elsif ($cont_multiseg == 0) { | |
918 | $cont_multiseg = 1; | |
919 | } | |
920 | if ($rest =~ /^$/) { | |
921 | $cont_in = 0; | |
922 | last; | |
923 | } | |
924 | if ($rest =~ /^\s*special/) { | |
925 | err("if/for/while/switch not started ". | |
926 | "on its own line"); | |
927 | } | |
928 | goto section_ended; | |
929 | } | |
930 | } elsif (/\{/) { | |
931 | err("{ while in parens/brackets") if (@cont_paren != 0); | |
932 | err("stuff after {") if ($rest =~ /[^\s}]/); | |
933 | $cont_in = 0; | |
934 | last; | |
935 | } elsif (/\}/) { | |
936 | err("} while in parens/brackets") if (@cont_paren != 0); | |
937 | if (!$cont_special && $rest !~ /^\s*(while|else)\b/) { | |
938 | if ($rest =~ /^$/) { | |
939 | err("unexpected }"); | |
940 | } else { | |
941 | err("stuff after }"); | |
942 | } | |
943 | $cont_in = 0; | |
944 | last; | |
945 | } | |
946 | } elsif (/\:/ && $cont_case && @cont_paren == 0) { | |
947 | err("stuff after multi-line case") if ($rest !~ /$^/); | |
948 | $cont_in = 0; | |
949 | last; | |
950 | } | |
951 | next; | |
952 | section_ended: | |
953 | # End of a statement or if/while/for loop. Reset | |
954 | # cont_special and cont_macro based on the rest of the | |
955 | # line. | |
956 | $cont_special = ($rest =~ /^\s*$special/)? 1 : 0; | |
957 | $cont_macro = ($rest =~ /^\s*$macro/)? 1 : 0; | |
958 | $cont_case = 0; | |
959 | next; | |
960 | } | |
961 | $cont_noerr = 0 if (!$cont_in); | |
962 | } |