scripts/cstyle.pl

   1 #!/usr/bin/env perl
   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
  10 # or http://www.opensolaris.org/os/licensing.
  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 #
  22 #
  23 # Copyright 2008 Sun Microsystems, Inc.  All rights reserved.
  24 # Use is subject to license terms.
  25 #
  26 # @(#)cstyle 1.58 98/09/09 (from shannon)
  27 #ident  "%Z%%M% %I%     %E% SMI"
  28 #
  29 # cstyle - check for some common stylistic errors.
  30 #
  31 #       cstyle is a sort of "lint" for C coding style.
  32 #       It attempts to check for the style used in the
  33 #       kernel, sometimes known as "Bill Joy Normal Form".
  34 #
  35 #       There's a lot this can't check for, like proper indentation
  36 #       of code blocks.  There's also a lot more this could check for.
  37 #
  38 #       A note to the non perl literate:
  39 #
  40 #               perl regular expressions are pretty much like egrep
  41 #               regular expressions, with the following special symbols
  42 #
  43 #               \s      any space character
  44 #               \S      any non-space character
  45 #               \w      any "word" character [a-zA-Z0-9_]
  46 #               \W      any non-word character
  47 #               \d      a digit [0-9]
  48 #               \D      a non-digit
  49 #               \b      word boundary (between \w and \W)
  50 #               \B      non-word boundary
  51 #
  52
  53 require 5.0;
  54 use warnings;
  55 use IO::File;
  56 use Getopt::Std;
  57 use strict;
  58
  59 my $usage =
  60 "usage: cstyle [-chpvCP] [-o constructs] file ...
  61         -c      check continuation indentation inside functions
  62         -h      perform heuristic checks that are sometimes wrong
  63         -p      perform some of the more picky checks
  64         -v      verbose
  65         -C      don't check anything in header block comments
  66         -P      check for use of non-POSIX types
  67         -o constructs
  68                 allow a comma-separated list of optional constructs:
  69                     doxygen     allow doxygen-style block comments (/** /*!)
  70                     splint      allow splint-style lint comments (/*@ ... @*/)
  71 ";
  72
  73 my %opts;
  74
  75 if (!getopts("cho:pvCP", \%opts)) {
  76         print $usage;
  77         exit 2;
  78 }
  79
  80 my $check_continuation = $opts{'c'};
  81 my $heuristic = $opts{'h'};
  82 my $picky = $opts{'p'};
  83 my $verbose = $opts{'v'};
  84 my $ignore_hdr_comment = $opts{'C'};
  85 my $check_posix_types = $opts{'P'};
  86
  87 my $doxygen_comments = 0;
  88 my $splint_comments = 0;
  89
  90 if (defined($opts{'o'})) {
  91         for my $x (split /,/, $opts{'o'}) {
  92                 if ($x eq "doxygen") {
  93                         $doxygen_comments = 1;
  94                 } elsif ($x eq "splint") {
  95                         $splint_comments = 1;
  96                 } else {
  97                         print "cstyle: unrecognized construct \"$x\"\n";
  98                         print $usage;
  99                         exit 2;
 100                 }
 101         }
 102 }
 103
 104 my ($filename, $line, $prev);           # shared globals
 105
 106 my $fmt;
 107 my $hdr_comment_start;
 108
 109 if ($verbose) {
 110         $fmt = "%s: %d: %s\n%s\n";
 111 } else {
 112         $fmt = "%s: %d: %s\n";
 113 }
 114
 115 if ($doxygen_comments) {
 116         # doxygen comments look like "/*!" or "/**"; allow them.
 117         $hdr_comment_start = qr/^\s*\/\*[\!\*]?$/;
 118 } else {
 119         $hdr_comment_start = qr/^\s*\/\*$/;
 120 }
 121
 122 # Note, following must be in single quotes so that \s and \w work right.
 123 my $typename = '(int|char|short|long|unsigned|float|double' .
 124     '|\w+_t|struct\s+\w+|union\s+\w+|FILE)';
 125
 126 # mapping of old types to POSIX compatible types
 127 my %old2posix = (
 128         'unchar' => 'uchar_t',
 129         'ushort' => 'ushort_t',
 130         'uint' => 'uint_t',
 131         'ulong' => 'ulong_t',
 132         'u_int' => 'uint_t',
 133         'u_short' => 'ushort_t',
 134         'u_long' => 'ulong_t',
 135         'u_char' => 'uchar_t',
 136         'quad' => 'quad_t'
 137 );
 138
 139 my $lint_re = qr/\/\*(?:
 140         ARGSUSED[0-9]*|NOTREACHED|LINTLIBRARY|VARARGS[0-9]*|
 141         CONSTCOND|CONSTANTCOND|CONSTANTCONDITION|EMPTY|
 142         FALLTHRU|FALLTHROUGH|LINTED.*?|PRINTFLIKE[0-9]*|
 143         PROTOLIB[0-9]*|SCANFLIKE[0-9]*|CSTYLED.*?
 144     )\*\//x;
 145
 146 my $splint_re = qr/\/\*@.*?@\*\//x;
 147
 148 my $warlock_re = qr/\/\*\s*(?:
 149         VARIABLES\ PROTECTED\ BY|
 150         MEMBERS\ PROTECTED\ BY|
 151         ALL\ MEMBERS\ PROTECTED\ BY|
 152         READ-ONLY\ VARIABLES:|
 153         READ-ONLY\ MEMBERS:|
 154         VARIABLES\ READABLE\ WITHOUT\ LOCK:|
 155         MEMBERS\ READABLE\ WITHOUT\ LOCK:|
 156         LOCKS\ COVERED\ BY|
 157         LOCK\ UNNEEDED\ BECAUSE|
 158         LOCK\ NEEDED:|
 159         LOCK\ HELD\ ON\ ENTRY:|
 160         READ\ LOCK\ HELD\ ON\ ENTRY:|
 161         WRITE\ LOCK\ HELD\ ON\ ENTRY:|
 162         LOCK\ ACQUIRED\ AS\ SIDE\ EFFECT:|
 163         READ\ LOCK\ ACQUIRED\ AS\ SIDE\ EFFECT:|
 164         WRITE\ LOCK\ ACQUIRED\ AS\ SIDE\ EFFECT:|
 165         LOCK\ RELEASED\ AS\ SIDE\ EFFECT:|
 166         LOCK\ UPGRADED\ AS\ SIDE\ EFFECT:|
 167         LOCK\ DOWNGRADED\ AS\ SIDE\ EFFECT:|
 168         FUNCTIONS\ CALLED\ THROUGH\ POINTER|
 169         FUNCTIONS\ CALLED\ THROUGH\ MEMBER|
 170         LOCK\ ORDER:
 171     )/x;
 172
 173 my $err_stat = 0;               # exit status
 174
 175 if ($#ARGV >= 0) {
 176         foreach my $arg (@ARGV) {
 177                 my $fh = new IO::File $arg, "r";
 178                 if (!defined($fh)) {
 179                         printf "%s: can not open\n", $arg;
 180                 } else {
 181                         &cstyle($arg, $fh);
 182                         close $fh;
 183                 }
 184         }
 185 } else {
 186         &cstyle("<stdin>", *STDIN);
 187 }
 188 exit $err_stat;
 189
 190 my $no_errs = 0;                # set for CSTYLED-protected lines
 191
 192 sub err($) {
 193         my ($error) = @_;
 194         unless ($no_errs) {
 195                 if ($verbose) {
 196                         printf $fmt, $filename, $., $error, $line;
 197                 } else {
 198                         printf $fmt, $filename, $., $error;
 199                 }
 200                 $err_stat = 1;
 201         }
 202 }
 203
 204 sub err_prefix($$) {
 205         my ($prevline, $error) = @_;
 206         my $out = $prevline."\n".$line;
 207         unless ($no_errs) {
 208                 if ($verbose) {
 209                         printf $fmt, $filename, $., $error, $out;
 210                 } else {
 211                         printf $fmt, $filename, $., $error;
 212                 }
 213                 $err_stat = 1;
 214         }
 215 }
 216
 217 sub err_prev($) {
 218         my ($error) = @_;
 219         unless ($no_errs) {
 220                 if ($verbose) {
 221                         printf $fmt, $filename, $. - 1, $error, $prev;
 222                 } else {
 223                         printf $fmt, $filename, $. - 1, $error;
 224                 }
 225                 $err_stat = 1;
 226         }
 227 }
 228
 229 sub cstyle($$) {
 230
 231 my ($fn, $filehandle) = @_;
 232 $filename = $fn;                        # share it globally
 233
 234 my $in_cpp = 0;
 235 my $next_in_cpp = 0;
 236
 237 my $in_comment = 0;
 238 my $in_header_comment = 0;
 239 my $comment_done = 0;
 240 my $in_warlock_comment = 0;
 241 my $in_function = 0;
 242 my $in_function_header = 0;
 243 my $function_header_full_indent = 0;
 244 my $in_declaration = 0;
 245 my $note_level = 0;
 246 my $nextok = 0;
 247 my $nocheck = 0;
 248
 249 my $in_string = 0;
 250
 251 my ($okmsg, $comment_prefix);
 252
 253 $line = '';
 254 $prev = '';
 255 reset_indent();
 256
 257 line: while (<$filehandle>) {
 258         s/\r?\n$//;     # strip return and newline
 259
 260         # save the original line, then remove all text from within
 261         # double or single quotes, we do not want to check such text.
 262
 263         $line = $_;
 264
 265         #
 266         # C allows strings to be continued with a backslash at the end of
 267         # the line.  We translate that into a quoted string on the previous
 268         # line followed by an initial quote on the next line.
 269         #
 270         # (we assume that no-one will use backslash-continuation with character
 271         # constants)
 272         #
 273         $_ = '"' . $_           if ($in_string && !$nocheck && !$in_comment);
 274
 275         #
 276         # normal strings and characters
 277         #
 278         s/'([^\\']|\\[^xX0]|\\0[0-9]*|\\[xX][0-9a-fA-F]*)'/''/g;
 279         s/"([^\\"]|\\.)*"/\"\"/g;
 280
 281         #
 282         # detect string continuation
 283         #
 284         if ($nocheck || $in_comment) {
 285                 $in_string = 0;
 286         } else {
 287                 #
 288                 # Now that all full strings are replaced with "", we check
 289                 # for unfinished strings continuing onto the next line.
 290                 #
 291                 $in_string =
 292                     (s/([^"](?:"")*)"([^\\"]|\\.)*\\$/$1""/ ||
 293                     s/^("")*"([^\\"]|\\.)*\\$/""/);
 294         }
 295
 296         #
 297         # figure out if we are in a cpp directive
 298         #
 299         $in_cpp = $next_in_cpp || /^\s*#/;      # continued or started
 300         $next_in_cpp = $in_cpp && /\\$/;        # only if continued
 301
 302         # strip off trailing backslashes, which appear in long macros
 303         s/\s*\\$//;
 304
 305         # an /* END CSTYLED */ comment ends a no-check block.
 306         if ($nocheck) {
 307                 if (/\/\* *END *CSTYLED *\*\//) {
 308                         $nocheck = 0;
 309                 } else {
 310                         reset_indent();
 311                         next line;
 312                 }
 313         }
 314
 315         # a /*CSTYLED*/ comment indicates that the next line is ok.
 316         if ($nextok) {
 317                 if ($okmsg) {
 318                         err($okmsg);
 319                 }
 320                 $nextok = 0;
 321                 $okmsg = 0;
 322                 if (/\/\* *CSTYLED.*\*\//) {
 323                         /^.*\/\* *CSTYLED *(.*) *\*\/.*$/;
 324                         $okmsg = $1;
 325                         $nextok = 1;
 326                 }
 327                 $no_errs = 1;
 328         } elsif ($no_errs) {
 329                 $no_errs = 0;
 330         }
 331
 332         # check length of line.
 333         # first, a quick check to see if there is any chance of being too long.
 334         if (($line =~ tr/\t/\t/) * 7 + length($line) > 80) {
 335                 # yes, there is a chance.
 336                 # replace tabs with spaces and check again.
 337                 my $eline = $line;
 338                 1 while $eline =~
 339                     s/\t+/' ' x (length($&) * 8 - length($`) % 8)/e;
 340                 if (length($eline) > 80) {
 341                         err("line > 80 characters");
 342                 }
 343         }
 344
 345         # ignore NOTE(...) annotations (assumes NOTE is on lines by itself).
 346         if ($note_level || /\b_?NOTE\s*\(/) { # if in NOTE or this is NOTE
 347                 s/[^()]//g;                       # eliminate all non-parens
 348                 $note_level += s/\(//g - length;  # update paren nest level
 349                 next;
 350         }
 351
 352         # a /* BEGIN CSTYLED */ comment starts a no-check block.
 353         if (/\/\* *BEGIN *CSTYLED *\*\//) {
 354                 $nocheck = 1;
 355         }
 356
 357         # a /*CSTYLED*/ comment indicates that the next line is ok.
 358         if (/\/\* *CSTYLED.*\*\//) {
 359                 /^.*\/\* *CSTYLED *(.*) *\*\/.*$/;
 360                 $okmsg = $1;
 361                 $nextok = 1;
 362         }
 363         if (/\/\/ *CSTYLED/) {
 364                 /^.*\/\/ *CSTYLED *(.*)$/;
 365                 $okmsg = $1;
 366                 $nextok = 1;
 367         }
 368
 369         # universal checks; apply to everything
 370         if (/\t +\t/) {
 371                 err("spaces between tabs");
 372         }
 373         if (/ \t+ /) {
 374                 err("tabs between spaces");
 375         }
 376         if (/\s$/) {
 377                 err("space or tab at end of line");
 378         }
 379         if (/[^ \t(]\/\*/ && !/\w\(\/\*.*\*\/\);/) {
 380                 err("comment preceded by non-blank");
 381         }
 382
 383         # is this the beginning or ending of a function?
 384         # (not if "struct foo\n{\n")
 385         if (/^{$/ && $prev =~ /\)\s*(const\s*)?(\/\*.*\*\/\s*)?\\?$/) {
 386                 $in_function = 1;
 387                 $in_declaration = 1;
 388                 $in_function_header = 0;
 389                 $function_header_full_indent = 0;
 390                 $prev = $line;
 391                 next line;
 392         }
 393         if (/^}\s*(\/\*.*\*\/\s*)*$/) {
 394                 if ($prev =~ /^\s*return\s*;/) {
 395                         err_prev("unneeded return at end of function");
 396                 }
 397                 $in_function = 0;
 398                 reset_indent();         # we don't check between functions
 399                 $prev = $line;
 400                 next line;
 401         }
 402         if ($in_function_header && ! /^    ./ ) {
 403                 if (/^{}$/ # empty functions
 404                 || /;/ #run function with multiline arguments
 405                 || /#/ #preprocessor commands
 406                 || /^[^\s\\]*\(.*\)$/ #functions without ; at the end
 407                 || /^$/ #function declaration can't have empty line
 408                 ) {
 409                         $in_function_header = 0;
 410                         $function_header_full_indent = 0;
 411                 } elsif ($prev =~ /^__attribute__/) { #__attribute__((*))
 412                         $in_function_header = 0;
 413                         $function_header_full_indent = 0;
 414                         $prev = $line;
 415                         next line;
 416                 } elsif ($picky && ! (/^\t/ && $function_header_full_indent != 0)) {
 417
 418                         err("continuation line should be indented by 4 spaces");
 419                 }
 420         }
 421
 422         #
 423         # If this matches something of form "foo(", it's probably a function
 424         # definition, unless it ends with ") bar;", in which case it's a declaration
 425         # that uses a macro to generate the type.
 426         #
 427         if (/^\w+\(/ && !/\) \w+;/) {
 428                 $in_function_header = 1;
 429                 if (/\($/) {
 430                         $function_header_full_indent = 1;
 431                 }
 432         }
 433         if ($in_function_header && /^{$/) {
 434                 $in_function_header = 0;
 435                 $function_header_full_indent = 0;
 436                 $in_function = 1;
 437         }
 438         if ($in_function_header && /\);$/) {
 439                 $in_function_header = 0;
 440                 $function_header_full_indent = 0;
 441         }
 442         if ($in_function_header && /{$/ ) {
 443                 if ($picky == 1) {
 444                         err("opening brace on same line as function header");
 445                 }
 446                 $in_function_header = 0;
 447                 $function_header_full_indent = 0;
 448                 $in_function = 1;
 449                 next line;
 450         }
 451
 452         if ($in_warlock_comment && /\*\//) {
 453                 $in_warlock_comment = 0;
 454                 $prev = $line;
 455                 next line;
 456         }
 457
 458         # a blank line terminates the declarations within a function.
 459         # XXX - but still a problem in sub-blocks.
 460         if ($in_declaration && /^$/) {
 461                 $in_declaration = 0;
 462         }
 463
 464         if ($comment_done) {
 465                 $in_comment = 0;
 466                 $in_header_comment = 0;
 467                 $comment_done = 0;
 468         }
 469         # does this looks like the start of a block comment?
 470         if (/$hdr_comment_start/) {
 471                 if (!/^\t*\/\*/) {
 472                         err("block comment not indented by tabs");
 473                 }
 474                 $in_comment = 1;
 475                 /^(\s*)\//;
 476                 $comment_prefix = $1;
 477                 if ($comment_prefix eq "") {
 478                         $in_header_comment = 1;
 479                 }
 480                 $prev = $line;
 481                 next line;
 482         }
 483         # are we still in the block comment?
 484         if ($in_comment) {
 485                 if (/^$comment_prefix \*\/$/) {
 486                         $comment_done = 1;
 487                 } elsif (/\*\//) {
 488                         $comment_done = 1;
 489                         err("improper block comment close")
 490                             unless ($ignore_hdr_comment && $in_header_comment);
 491                 } elsif (!/^$comment_prefix \*[ \t]/ &&
 492                     !/^$comment_prefix \*$/) {
 493                         err("improper block comment")
 494                             unless ($ignore_hdr_comment && $in_header_comment);
 495                 }
 496         }
 497
 498         if ($in_header_comment && $ignore_hdr_comment) {
 499                 $prev = $line;
 500                 next line;
 501         }
 502
 503         # check for errors that might occur in comments and in code.
 504
 505         # allow spaces to be used to draw pictures in all comments.
 506         if (/[^ ]     / && !/".*     .*"/ && !$in_comment) {
 507                 err("spaces instead of tabs");
 508         }
 509         if (/^ / && !/^ \*[ \t\/]/ && !/^ \*$/ &&
 510             (!/^    \w/ || $in_function != 0)) {
 511                 err("indent by spaces instead of tabs");
 512         }
 513         if (/^\t+ [^ \t\*]/ || /^\t+  \S/ || /^\t+   \S/) {
 514                 err("continuation line not indented by 4 spaces");
 515         }
 516         if (/$warlock_re/ && !/\*\//) {
 517                 $in_warlock_comment = 1;
 518                 $prev = $line;
 519                 next line;
 520         }
 521         if (/^\s*\/\*./ && !/^\s*\/\*.*\*\// && !/$hdr_comment_start/) {
 522                 err("improper first line of block comment");
 523         }
 524
 525         if ($in_comment) {      # still in comment, don't do further checks
 526                 $prev = $line;
 527                 next line;
 528         }
 529
 530         if ((/[^(]\/\*\S/ || /^\/\*\S/) &&
 531             !(/$lint_re/ || ($splint_comments && /$splint_re/))) {
 532                 err("missing blank after open comment");
 533         }
 534         if (/\S\*\/[^)]|\S\*\/$/ &&
 535             !(/$lint_re/ || ($splint_comments && /$splint_re/))) {
 536                 err("missing blank before close comment");
 537         }
 538         if (/\/\/\S/) {         # C++ comments
 539                 err("missing blank after start comment");
 540         }
 541         # check for unterminated single line comments, but allow them when
 542         # they are used to comment out the argument list of a function
 543         # declaration.
 544         if (/\S.*\/\*/ && !/\S.*\/\*.*\*\// && !/\(\/\*/) {
 545                 err("unterminated single line comment");
 546         }
 547
 548         if (/^(#else|#endif|#include)(.*)$/) {
 549                 $prev = $line;
 550                 if ($picky) {
 551                         my $directive = $1;
 552                         my $clause = $2;
 553                         # Enforce ANSI rules for #else and #endif: no noncomment
 554                         # identifiers are allowed after #endif or #else.  Allow
 555                         # C++ comments since they seem to be a fact of life.
 556                         if ((($1 eq "#endif") || ($1 eq "#else")) &&
 557                             ($clause ne "") &&
 558                             (!($clause =~ /^\s+\/\*.*\*\/$/)) &&
 559                             (!($clause =~ /^\s+\/\/.*$/))) {
 560                                 err("non-comment text following " .
 561                                     "$directive (or malformed $directive " .
 562                                     "directive)");
 563                         }
 564                 }
 565                 next line;
 566         }
 567
 568         #
 569         # delete any comments and check everything else.  Note that
 570         # ".*?" is a non-greedy match, so that we don't get confused by
 571         # multiple comments on the same line.
 572         #
 573         s/\/\*.*?\*\//\ 1/g;
 574         s/\/\/.*$/\ 1/;           # C++ comments
 575
 576         # delete any trailing whitespace; we have already checked for that.
 577         s/\s*$//;
 578
 579         # following checks do not apply to text in comments.
 580
 581         if (/[^<>\s][!<>=]=/ || /[^<>][!<>=]=[^\s,]/ ||
 582             (/[^->]>[^,=>\s]/ && !/[^->]>$/) ||
 583             (/[^<]<[^,=<\s]/ && !/[^<]<$/) ||
 584             /[^<\s]<[^<]/ || /[^->\s]>[^>]/) {
 585                 err("missing space around relational operator");
 586         }
 587         if (/\S>>=/ || /\S<<=/ || />>=\S/ || /<<=\S/ || /\S[-+*\/&|^%]=/ ||
 588             (/[^-+*\/&|^%!<>=\s]=[^=]/ && !/[^-+*\/&|^%!<>=\s]=$/) ||
 589             (/[^!<>=]=[^=\s]/ && !/[^!<>=]=$/)) {
 590                 # XXX - should only check this for C++ code
 591                 # XXX - there are probably other forms that should be allowed
 592                 if (!/\soperator=/) {
 593                         err("missing space around assignment operator");
 594                 }
 595         }
 596         if (/[,;]\S/ && !/\bfor \(;;\)/) {
 597                 err("comma or semicolon followed by non-blank");
 598         }
 599         # allow "for" statements to have empty "while" clauses
 600         if (/\s[,;]/ && !/^[\t]+;$/ && !/^\s*for \([^;]*; ;[^;]*\)/) {
 601                 err("comma or semicolon preceded by blank");
 602         }
 603         if (/^\s*(&&|\|\|)/) {
 604                 err("improper boolean continuation");
 605         }
 606         if (/\S   *(&&|\|\|)/ || /(&&|\|\|)   *\S/) {
 607                 err("more than one space around boolean operator");
 608         }
 609         if (/\b(for|if|while|switch|sizeof|return|case)\(/) {
 610                 err("missing space between keyword and paren");
 611         }
 612         if (/(\b(for|if|while|switch|return)\b.*){2,}/ && !/^#define/) {
 613                 # multiple "case" and "sizeof" allowed
 614                 err("more than one keyword on line");
 615         }
 616         if (/\b(for|if|while|switch|sizeof|return|case)\s\s+\(/ &&
 617             !/^#if\s+\(/) {
 618                 err("extra space between keyword and paren");
 619         }
 620         # try to detect "func (x)" but not "if (x)" or
 621         # "#define foo (x)" or "int (*func)();"
 622         if (/\w\s\(/) {
 623                 my $s = $_;
 624                 # strip off all keywords on the line
 625                 s/\b(for|if|while|switch|return|case|sizeof)\s\(/XXX(/g;
 626                 s/#elif\s\(/XXX(/g;
 627                 s/^#define\s+\w+\s+\(/XXX(/;
 628                 # do not match things like "void (*f)();"
 629                 # or "typedef void (func_t)();"
 630                 s/\w\s\(+\*/XXX(*/g;
 631                 s/\b($typename|void)\s+\(+/XXX(/og;
 632                 if (/\w\s\(/) {
 633                         err("extra space between function name and left paren");
 634                 }
 635                 $_ = $s;
 636         }
 637         # try to detect "int foo(x)", but not "extern int foo(x);"
 638         # XXX - this still trips over too many legitimate things,
 639         # like "int foo(x,\n\ty);"
 640 #               if (/^(\w+(\s|\*)+)+\w+\(/ && !/\)[;,](\s|\ 1)*$/ &&
 641 #                   !/^(extern|static)\b/) {
 642 #                       err("return type of function not on separate line");
 643 #               }
 644         # this is a close approximation
 645         if (/^(\w+(\s|\*)+)+\w+\(.*\)(\s|\ 1)*$/ &&
 646             !/^(extern|static)\b/) {
 647                 err("return type of function not on separate line");
 648         }
 649         if (/^#define /) {
 650                 err("#define followed by space instead of tab");
 651         }
 652         if (/^\s*return\W[^;]*;/ && !/^\s*return\s*\(.*\);/) {
 653                 err("unparenthesized return expression");
 654         }
 655         if (/\bsizeof\b/ && !/\bsizeof\s*\(.*\)/) {
 656                 err("unparenthesized sizeof expression");
 657         }
 658         if (/\(\s/) {
 659                 err("whitespace after left paren");
 660         }
 661         # Allow "for" statements to have empty "continue" clauses.
 662         # Allow right paren on its own line unless we're being picky (-p).
 663         if (/\s\)/ && !/^\s*for \([^;]*;[^;]*; \)/ && ($picky || !/^\s*\)/)) {
 664                 err("whitespace before right paren");
 665         }
 666         if (/^\s*\(void\)[^ ]/) {
 667                 err("missing space after (void) cast");
 668         }
 669         if (/\S\{/ && !/\{\{/) {
 670                 err("missing space before left brace");
 671         }
 672         if ($in_function && /^\s+{/ &&
 673             ($prev =~ /\)\s*$/ || $prev =~ /\bstruct\s+\w+$/)) {
 674                 err("left brace starting a line");
 675         }
 676         if (/}(else|while)/) {
 677                 err("missing space after right brace");
 678         }
 679         if (/}\s\s+(else|while)/) {
 680                 err("extra space after right brace");
 681         }
 682         if (/\b_VOID\b|\bVOID\b|\bSTATIC\b/) {
 683                 err("obsolete use of VOID or STATIC");
 684         }
 685         if (/\b$typename\*/o) {
 686                 err("missing space between type name and *");
 687         }
 688         if (/^\s+#/) {
 689                 err("preprocessor statement not in column 1");
 690         }
 691         if (/^#\s/) {
 692                 err("blank after preprocessor #");
 693         }
 694         if (/!\s*(strcmp|strncmp|bcmp)\s*\(/) {
 695                 err("don't use boolean ! with comparison functions");
 696         }
 697
 698         #
 699         # We completely ignore, for purposes of indentation:
 700         #  * lines outside of functions
 701         #  * preprocessor lines
 702         #
 703         if ($check_continuation && $in_function && !$in_cpp) {
 704                 process_indent($_);
 705         }
 706         if ($picky) {
 707                 # try to detect spaces after casts, but allow (e.g.)
 708                 # "sizeof (int) + 1", "void (*funcptr)(int) = foo;", and
 709                 # "int foo(int) __NORETURN;"
 710                 if ((/^\($typename( \*+)?\)\s/o ||
 711                     /\W\($typename( \*+)?\)\s/o) &&
 712                     !/sizeof\s*\($typename( \*)?\)\s/o &&
 713                     !/\($typename( \*+)?\)\s+=[^=]/o) {
 714                         err("space after cast");
 715                 }
 716                 if (/\b$typename\s*\*\s/o &&
 717                     !/\b$typename\s*\*\s+const\b/o) {
 718                         err("unary * followed by space");
 719                 }
 720         }
 721         if ($check_posix_types) {
 722                 # try to detect old non-POSIX types.
 723                 # POSIX requires all non-standard typedefs to end in _t,
 724                 # but historically these have been used.
 725                 if (/\b(unchar|ushort|uint|ulong|u_int|u_short|u_long|u_char|quad)\b/) {
 726                         err("non-POSIX typedef $1 used: use $old2posix{$1} instead");
 727                 }
 728         }
 729         if ($heuristic) {
 730                 # cannot check this everywhere due to "struct {\n...\n} foo;"
 731                 if ($in_function && !$in_declaration &&
 732                     /}./ && !/}\s+=/ && !/{.*}[;,]$/ && !/}(\s|\ 1)*$/ &&
 733                     !/} (else|while)/ && !/}}/) {
 734                         err("possible bad text following right brace");
 735                 }
 736                 # cannot check this because sub-blocks in
 737                 # the middle of code are ok
 738                 if ($in_function && /^\s+{/) {
 739                         err("possible left brace starting a line");
 740                 }
 741         }
 742         if (/^\s*else\W/) {
 743                 if ($prev =~ /^\s*}$/) {
 744                         err_prefix($prev,
 745                             "else and right brace should be on same line");
 746                 }
 747         }
 748         $prev = $line;
 749 }
 750
 751 if ($prev eq "") {
 752         err("last line in file is blank");
 753 }
 754
 755 }
 756
 757 #
 758 # Continuation-line checking
 759 #
 760 # The rest of this file contains the code for the continuation checking
 761 # engine.  It's a pretty simple state machine which tracks the expression
 762 # depth (unmatched '('s and '['s).
 763 #
 764 # Keep in mind that the argument to process_indent() has already been heavily
 765 # processed; all comments have been replaced by control-A, and the contents of
 766 # strings and character constants have been elided.
 767 #
 768
 769 my $cont_in;            # currently inside of a continuation
 770 my $cont_off;           # skipping an initializer or definition
 771 my $cont_noerr;         # suppress cascading errors
 772 my $cont_start;         # the line being continued
 773 my $cont_base;          # the base indentation
 774 my $cont_first;         # this is the first line of a statement
 775 my $cont_multiseg;      # this continuation has multiple segments
 776
 777 my $cont_special;       # this is a C statement (if, for, etc.)
 778 my $cont_macro;         # this is a macro
 779 my $cont_case;          # this is a multi-line case
 780
 781 my @cont_paren;         # the stack of unmatched ( and [s we've seen
 782
 783 sub
 784 reset_indent()
 785 {
 786         $cont_in = 0;
 787         $cont_off = 0;
 788 }
 789
 790 sub
 791 delabel($)
 792 {
 793         #
 794         # replace labels with tabs.  Note that there may be multiple
 795         # labels on a line.
 796         #
 797         local $_ = $_[0];
 798
 799         while (/^(\t*)( *(?:(?:\w+\s*)|(?:case\b[^:]*)): *)(.*)$/) {
 800                 my ($pre_tabs, $label, $rest) = ($1, $2, $3);
 801                 $_ = $pre_tabs;
 802                 while ($label =~ s/^([^\t]*)(\t+)//) {
 803                         $_ .= "\t" x (length($2) + length($1) / 8);
 804                 }
 805                 $_ .= ("\t" x (length($label) / 8)).$rest;
 806         }
 807
 808         return ($_);
 809 }
 810
 811 sub
 812 process_indent($)
 813 {
 814         require strict;
 815         local $_ = $_[0];                       # preserve the global $_
 816
 817         s/\ 1//g; # No comments
 818         s/\s+$//;       # Strip trailing whitespace
 819
 820         return                  if (/^$/);      # skip empty lines
 821
 822         # regexps used below; keywords taking (), macros, and continued cases
 823         my $special = '(?:(?:\}\s*)?else\s+)?(?:if|for|while|switch)\b';
 824         my $macro = '[A-Z_][A-Z_0-9]*\(';
 825         my $case = 'case\b[^:]*$';
 826
 827         # skip over enumerations, array definitions, initializers, etc.
 828         if ($cont_off <= 0 && !/^\s*$special/ &&
 829             (/(?:(?:\b(?:enum|struct|union)\s*[^\{]*)|(?:\s+=\s*)){/ ||
 830             (/^\s*{/ && $prev =~ /=\s*(?:\/\*.*\*\/\s*)*$/))) {
 831                 $cont_in = 0;
 832                 $cont_off = tr/{/{/ - tr/}/}/;
 833                 return;
 834         }
 835         if ($cont_off) {
 836                 $cont_off += tr/{/{/ - tr/}/}/;
 837                 return;
 838         }
 839
 840         if (!$cont_in) {
 841                 $cont_start = $line;
 842
 843                 if (/^\t* /) {
 844                         err("non-continuation indented 4 spaces");
 845                         $cont_noerr = 1;                # stop reporting
 846                 }
 847                 $_ = delabel($_);       # replace labels with tabs
 848
 849                 # check if the statement is complete
 850                 return          if (/^\s*\}?$/);
 851                 return          if (/^\s*\}?\s*else\s*\{?$/);
 852                 return          if (/^\s*do\s*\{?$/);
 853                 return          if (/{$/);
 854                 return          if (/}[,;]?$/);
 855
 856                 # Allow macros on their own lines
 857                 return          if (/^\s*[A-Z_][A-Z_0-9]*$/);
 858
 859                 # cases we don't deal with, generally non-kosher
 860                 if (/{/) {
 861                         err("stuff after {");
 862                         return;
 863                 }
 864
 865                 # Get the base line, and set up the state machine
 866                 /^(\t*)/;
 867                 $cont_base = $1;
 868                 $cont_in = 1;
 869                 @cont_paren = ();
 870                 $cont_first = 1;
 871                 $cont_multiseg = 0;
 872
 873                 # certain things need special processing
 874                 $cont_special = /^\s*$special/? 1 : 0;
 875                 $cont_macro = /^\s*$macro/? 1 : 0;
 876                 $cont_case = /^\s*$case/? 1 : 0;
 877         } else {
 878                 $cont_first = 0;
 879
 880                 # Strings may be pulled back to an earlier (half-)tabstop
 881                 unless ($cont_noerr || /^$cont_base    / ||
 882                     (/^\t*(?:    )?(?:gettext\()?\"/ && !/^$cont_base\t/)) {
 883                         err_prefix($cont_start,
 884                             "continuation should be indented 4 spaces");
 885                 }
 886         }
 887
 888         my $rest = $_;                  # keeps the remainder of the line
 889
 890         #
 891         # The split matches 0 characters, so that each 'special' character
 892         # is processed separately.  Parens and brackets are pushed and
 893         # popped off the @cont_paren stack.  For normal processing, we wait
 894         # until a ; or { terminates the statement.  "special" processing
 895         # (if/for/while/switch) is allowed to stop when the stack empties,
 896         # as is macro processing.  Case statements are terminated with a :
 897         # and an empty paren stack.
 898         #
 899         foreach $_ (split /[^\(\)\[\]\{\}\;\:]*/) {
 900                 next            if (length($_) == 0);
 901
 902                 # rest contains the remainder of the line
 903                 my $rxp = "[^\Q$_\E]*\Q$_\E";
 904                 $rest =~ s/^$rxp//;
 905
 906                 if (/\(/ || /\[/) {
 907                         push @cont_paren, $_;
 908                 } elsif (/\)/ || /\]/) {
 909                         my $cur = $_;
 910                         tr/\)\]/\(\[/;
 911
 912                         my $old = (pop @cont_paren);
 913                         if (!defined($old)) {
 914                                 err("unexpected '$cur'");
 915                                 $cont_in = 0;
 916                                 last;
 917                         } elsif ($old ne $_) {
 918                                 err("'$cur' mismatched with '$old'");
 919                                 $cont_in = 0;
 920                                 last;
 921                         }
 922
 923                         #
 924                         # If the stack is now empty, do special processing
 925                         # for if/for/while/switch and macro statements.
 926                         #
 927                         next            if (@cont_paren != 0);
 928                         if ($cont_special) {
 929                                 if ($rest =~ /^\s*{?$/) {
 930                                         $cont_in = 0;
 931                                         last;
 932                                 }
 933                                 if ($rest =~ /^\s*;$/) {
 934                                         err("empty if/for/while body ".
 935                                             "not on its own line");
 936                                         $cont_in = 0;
 937                                         last;
 938                                 }
 939                                 if (!$cont_first && $cont_multiseg == 1) {
 940                                         err_prefix($cont_start,
 941                                             "multiple statements continued ".
 942                                             "over multiple lines");
 943                                         $cont_multiseg = 2;
 944                                 } elsif ($cont_multiseg == 0) {
 945                                         $cont_multiseg = 1;
 946                                 }
 947                                 # We've finished this section, start
 948                                 # processing the next.
 949                                 goto section_ended;
 950                         }
 951                         if ($cont_macro) {
 952                                 if ($rest =~ /^$/) {
 953                                         $cont_in = 0;
 954                                         last;
 955                                 }
 956                         }
 957                 } elsif (/\;/) {
 958                         if ($cont_case) {
 959                                 err("unexpected ;");
 960                         } elsif (!$cont_special) {
 961                                 err("unexpected ;")     if (@cont_paren != 0);
 962                                 if (!$cont_first && $cont_multiseg == 1) {
 963                                         err_prefix($cont_start,
 964                                             "multiple statements continued ".
 965                                             "over multiple lines");
 966                                         $cont_multiseg = 2;
 967                                 } elsif ($cont_multiseg == 0) {
 968                                         $cont_multiseg = 1;
 969                                 }
 970                                 if ($rest =~ /^$/) {
 971                                         $cont_in = 0;
 972                                         last;
 973                                 }
 974                                 if ($rest =~ /^\s*special/) {
 975                                         err("if/for/while/switch not started ".
 976                                             "on its own line");
 977                                 }
 978                                 goto section_ended;
 979                         }
 980                 } elsif (/\{/) {
 981                         err("{ while in parens/brackets") if (@cont_paren != 0);
 982                         err("stuff after {")            if ($rest =~ /[^\s}]/);
 983                         $cont_in = 0;
 984                         last;
 985                 } elsif (/\}/) {
 986                         err("} while in parens/brackets") if (@cont_paren != 0);
 987                         if (!$cont_special && $rest !~ /^\s*(while|else)\b/) {
 988                                 if ($rest =~ /^$/) {
 989                                         err("unexpected }");
 990                                 } else {
 991                                         err("stuff after }");
 992                                 }
 993                                 $cont_in = 0;
 994                                 last;
 995                         }
 996                 } elsif (/\:/ && $cont_case && @cont_paren == 0) {
 997                         err("stuff after multi-line case") if ($rest !~ /$^/);
 998                         $cont_in = 0;
 999                         last;
1000                 }
1001                 next;
1002 section_ended:
1003                 # End of a statement or if/while/for loop.  Reset
1004                 # cont_special and cont_macro based on the rest of the
1005                 # line.
1006                 $cont_special = ($rest =~ /^\s*$special/)? 1 : 0;
1007                 $cont_macro = ($rest =~ /^\s*$macro/)? 1 : 0;
1008                 $cont_case = 0;
1009                 next;
1010         }
1011         $cont_noerr = 0                 if (!$cont_in);
1012 }