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