1#! /usr/bin/env perl 2# 3# Copyright 2020-2024 The OpenSSL Project Authors. All Rights Reserved. 4# Copyright Siemens AG 2019-2022 5# 6# Licensed under the Apache License 2.0 (the "License"). 7# You may not use this file except in compliance with the License. 8# You can obtain a copy in the file LICENSE in the source distribution 9# or at https://www.openssl.org/source/license.html 10# 11# check-format.pl 12# - check formatting of C source according to OpenSSL coding style 13# 14# usage: 15# check-format.pl [-l|--strict-len] [-b|--sloppy-bodylen] 16# [-s|--sloppy-space] [-c|--sloppy-comment] 17# [-m|--sloppy-macro] [-h|--sloppy-hang] 18# [-e|--eol-comment] [-1|--1-stmt] 19# <files> 20# 21# run self-tests: 22# util/check-format.pl util/check-format-test-positives.c 23# util/check-format.pl util/check-format-test-negatives.c 24# 25# checks adherence to the formatting rules of the OpenSSL coding guidelines 26# assuming that the input files contain syntactically correct C code. 27# This pragmatic tool is incomplete and yields some false positives. 28# Still it should be useful for detecting most typical glitches. 29# 30# options: 31# -l | --strict-len decrease accepted max line length from 100 to 80 32# -b | --sloppy-bodylen do not report function body length > 200 33# -s | --sloppy-space do not report whitespace nits 34# -c | --sloppy-comment do not report indentation of comments 35# Otherwise for each multi-line comment the indentation of 36# its lines is checked for consistency. For each comment 37# that does not begin to the right of normal code its 38# indentation must be as for normal code, while in case it 39# also has no normal code to its right it is considered to 40# refer to the following line and may be indented equally. 41# -m | --sloppy-macro allow missing extra indentation of macro bodies 42# -h | --sloppy-hang when checking hanging indentation, do not report 43# * same indentation as on line before 44# * same indentation as non-hanging indent level 45# * indentation moved left (not beyond non-hanging indent) 46# just to fit contents within the line length limit 47# -e | --eol-comment report needless intermediate multiple consecutive spaces also before end-of-line comments 48# -1 | --1-stmt do more aggressive checks for { 1 stmt } - see below 49# 50# There are non-trivial false positives and negatives such as the following. 51# 52# * When a line contains several issues of the same kind only one is reported. 53# 54# * When a line contains more than one statement this is (correctly) reported 55# but in some situations the indentation checks for subsequent lines go wrong. 56# 57# * There is the special OpenSSL rule not to unnecessarily use braces around 58# single statements: 59# { 60# stmt; 61# } 62# except within if ... else constructs where some branch contains more than one 63# statement. Since the exception is hard to recognize when such branches occur 64# after the current position (such that false positives would be reported) 65# the tool by checks for this rule by default only for do/while/for bodies. 66# Yet with the --1-stmt option false positives are preferred over negatives. 67# False negatives occur if the braces are more than two non-blank lines apart. 68# 69# * The presence of multiple consecutive spaces is regarded a coding style nit 70# except when this is before end-of-line comments (unless the --eol-comment is given) and 71# except when done in order to align certain columns over multiple lines, e.g.: 72# # define AB 1 73# # define CDE 22 74# # define F 3333 75# This pattern is recognized - and consequently extra space not reported - 76# for a given line if in the non-blank line before or after (if existing) 77# for each occurrence of " \S" (where \S means non-space) in the given line 78# there is " \S" in the other line in the respective column position. 79# This may lead to both false negatives (in case of coincidental " \S") 80# and false positives (in case of more complex multi-column alignment). 81# 82# * When just part of control structures depend on #if(n)(def), which can be 83# considered bad programming style, indentation false positives occur, e.g.: 84# #if X 85# if (1) /* bad style */ 86# #else 87# if (2) /* bad style resulting in false positive */ 88# #endif 89# c; /* resulting further false positive */ 90 91use strict; 92# use List::Util qw[min max]; 93use POSIX; 94 95use constant INDENT_LEVEL => 4; 96use constant MAX_LINE_LENGTH => 100; 97use constant STRICT_LINE_LENGTH => 80; 98use constant MAX_BODY_LENGTH => 200; 99 100# global variables @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ 101 102# command-line options 103my $max_length = MAX_LINE_LENGTH; 104my $sloppy_bodylen = 0; 105my $sloppy_SPC = 0; 106my $sloppy_hang = 0; 107my $sloppy_cmt = 0; 108my $sloppy_macro = 0; 109my $eol_cmt = 0; 110my $extended_1_stmt = 0; 111 112while ($ARGV[0] =~ m/^-(\w|-[\w\-]+)$/) { 113 my $arg = $1; shift; 114 if ($arg =~ m/^(l|-strict-len)$/) { 115 $max_length = STRICT_LINE_LENGTH; 116 } elsif ($arg =~ m/^(b|-sloppy-bodylen)$/) { 117 $sloppy_bodylen = 1; 118 } elsif ($arg =~ m/^(s|-sloppy-space)$/) { 119 $sloppy_SPC= 1; 120 } elsif ($arg =~ m/^(c|-sloppy-comment)$/) { 121 $sloppy_cmt = 1; 122 } elsif ($arg =~ m/^(m|-sloppy-macro)$/) { 123 $sloppy_macro = 1; 124 } elsif ($arg =~ m/^(h|-sloppy-hang)$/) { 125 $sloppy_hang = 1; 126 } elsif ($arg =~ m/^(e|-eol-comment)$/) { 127 $eol_cmt = 1; 128 } elsif ($arg =~ m/^(1|-1-stmt)$/) { 129 $extended_1_stmt = 1; 130 } else { 131 die("unknown option: -$arg"); 132 } 133} 134 135# state variables 136my $self_test; # whether the current input file is regarded to contain (positive/negative) self-tests 137 138my $in_comment; # number of lines so far within multi-line comment, 0 if no comment, < 0 when end is on current line 139my $leading_comment; # multi-line comment has no code before its beginning delimiter, if $in_comment != 0 140my $formatted_comment; # multi-line comment beginning with "/*-", which indicates/allows special formatting, if $in_comment != 0 141my $comment_indent; # comment indent, if $in_comment != 0 142 143my $ifdef__cplusplus; # line before contained '#ifdef __cplusplus' (used in header files) 144my $preproc_if_nesting; # currently required indentation of preprocessor directive according to #if(n)(def) 145my $in_preproc; # 0 or number of lines so far within preprocessor directive, e.g., macro definition 146my $preproc_directive; # name of current preprocessor directive, if $in_preproc != 0 147my $preproc_offset; # offset to $block_indent within multi-line preprocessor directive, else 0 148my $in_macro_header; # number of open parentheses + 1 in (multi-line) header of #define, if $in_preproc != 0 149 150my $line; # current line number 151my $line_before; # number of previous not essentially blank line (containing at most whitespace and '\') 152my $line_before2; # number of not essentially blank line before previous not essentially blank line 153 154# indentation state 155my $contents; # contents of current line (without blinding) 156# $_ # current line, where comments etc. get blinded 157my $code_contents_before; # contents of previous non-comment non-preprocessor-directive line (without blinding), initially "" 158my $contents_before; # contents of $line_before (without blinding), if $line_before > 0 159my $contents_before_; # contents of $line_before after blinding comments etc., if $line_before > 0 160my $contents_before2; # contents of $line_before2 (without blinding), if $line_before2 > 0 161my $contents_before_2; # contents of $line_before2 after blinding comments etc., if $line_before2 > 0 162my $in_multiline_string; # line starts within multi-line string literal 163my $count; # -1 or number of leading whitespace characters (except newline) in current line, 164 # which should be $block_indent + $hanging_offset + $local_offset or $expr_indent 165my $count_before; # number of leading whitespace characters (except line ending chars) in $contents_before 166my $has_label; # current line contains label 167my $local_offset; # current extra indent due to label, switch case/default, or leading closing brace(s) 168my $line_body_start; # number of line where last function body started, or 0 169my $line_function_start; # number of line where last function definition started, used for $line_body_start 170my $last_function_header; # header containing name of last function defined, used if $line_body_start != 0 171my $line_opening_brace; # number of previous line with opening brace after if/do/while/for, optionally for 'else' 172 173my $keyword_opening_brace; # name of previous keyword, used if $line_opening_brace != 0 174my $block_indent; # currently required normal indentation at block/statement level 175my $hanging_offset; # extra indent, which may be nested, for just one hanging statement or expr or typedef 176my @in_do_hanging_offsets; # stack of hanging offsets for nested 'do' ... 'while' 177my @in_if_hanging_offsets; # stack of hanging offsets for nested 'if' (but not its potential 'else' branch) 178my $if_maybe_terminated; # 'if' ends and $hanging_offset should be reset unless the next line starts with 'else' 179my @nested_block_indents; # stack of indentations at block/statement level, needed due to hanging statements 180my @nested_hanging_offsets;# stack of nested $hanging_offset values, in parallel to @nested_block_indents 181my @nested_in_typedecl; # stack of nested $in_typedecl values, partly in parallel to @nested_block_indents 182my @nested_indents; # stack of hanging indents due to parentheses, braces, brackets, or conditionals 183my @nested_symbols; # stack of hanging symbols '(', '{', '[', or '?', in parallel to @nested_indents 184my @nested_conds_indents; # stack of hanging indents due to conditionals ('?' ... ':') 185my $expr_indent; # resulting hanging indent within (multi-line) expressions including type exprs, else 0 186my $hanging_symbol; # character ('(', '{', '[', not: '?') responsible for $expr_indent, if $expr_indent != 0 187my $in_block_decls; # number of local declaration lines after block opening before normal statements, or -1 if no block opening 188my $in_expr; # in expression after if/while/for/switch/return/enum/LHS of assignment 189my $in_paren_expr; # in parenthesized if/while/for condition and switch expression, if $expr_indent != 0 190my $in_typedecl; # nesting level of typedef/struct/union/enum 191 192my $num_reports_line = 0; # number of issues found on current line 193my $num_reports = 0; # total number of issues found 194my $num_indent_reports = 0;# total number of indentation issues found 195my $num_nesting_issues = 0;# total number of preprocessor #if nesting issues found 196my $num_syntax_issues = 0; # total number of syntax issues found during sanity checks 197my $num_SPC_reports = 0; # total number of whitespace issues found 198my $num_length_reports = 0;# total number of line length issues found 199 200sub reset_file_state { 201 $in_comment = 0; 202 $ifdef__cplusplus = 0; 203 $preproc_if_nesting = 0; 204 $in_preproc = 0; 205 $line = 0; 206 $line_before = 0; 207 $line_before2 = 0; 208 reset_indentation_state(); 209} 210sub reset_indentation_state { 211 $code_contents_before = ""; 212 @nested_block_indents = (); 213 @nested_hanging_offsets = (); 214 @nested_in_typedecl = (); 215 @nested_symbols = (); 216 @nested_indents = (); 217 @nested_conds_indents = (); 218 $expr_indent = 0; 219 $in_block_decls = -1; 220 $in_expr = 0; 221 $in_paren_expr = 0; 222 $hanging_offset = 0; 223 @in_do_hanging_offsets = (); 224 @in_if_hanging_offsets = (); 225 $if_maybe_terminated = 0; 226 $block_indent = 0; 227 $in_multiline_string = 0; 228 $line_body_start = 0; 229 $line_opening_brace = 0; 230 $in_typedecl = 0; 231} 232my $bak_line_before; 233my $bak_line_before2; 234my $bak_code_contents_before; 235my @bak_nested_block_indents; 236my @bak_nested_hanging_offsets; 237my @bak_nested_in_typedecl; 238my @bak_nested_symbols; 239my @bak_nested_indents; 240my @bak_nested_conds_indents; 241my $bak_expr_indent; 242my $bak_in_block_decls; 243my $bak_in_expr; 244my $bak_in_paren_expr; 245my $bak_hanging_offset; 246my @bak_in_do_hanging_offsets; 247my @bak_in_if_hanging_offsets; 248my $bak_if_maybe_terminated; 249my $bak_block_indent; 250my $bak_in_multiline_string; 251my $bak_line_body_start; 252my $bak_line_opening_brace; 253my $bak_in_typedecl; 254sub backup_indentation_state { 255 $bak_code_contents_before = $code_contents_before; 256 @bak_nested_block_indents = @nested_block_indents; 257 @bak_nested_hanging_offsets = @nested_hanging_offsets; 258 @bak_nested_in_typedecl = @nested_in_typedecl; 259 @bak_nested_symbols = @nested_symbols; 260 @bak_nested_indents = @nested_indents; 261 @bak_nested_conds_indents = @nested_conds_indents; 262 $bak_expr_indent = $expr_indent; 263 $bak_in_block_decls = $in_block_decls; 264 $bak_in_expr = $in_expr; 265 $bak_in_paren_expr = $in_paren_expr; 266 $bak_hanging_offset = $hanging_offset; 267 @bak_in_do_hanging_offsets = @in_do_hanging_offsets; 268 @bak_in_if_hanging_offsets = @in_if_hanging_offsets; 269 $bak_if_maybe_terminated = $if_maybe_terminated; 270 $bak_block_indent = $block_indent; 271 $bak_in_multiline_string = $in_multiline_string; 272 $bak_line_body_start = $line_body_start; 273 $bak_line_opening_brace = $line_opening_brace; 274 $bak_in_typedecl = $in_typedecl; 275} 276sub restore_indentation_state { 277 $code_contents_before = $bak_code_contents_before; 278 @nested_block_indents = @bak_nested_block_indents; 279 @nested_hanging_offsets = @bak_nested_hanging_offsets; 280 @nested_in_typedecl = @bak_nested_in_typedecl; 281 @nested_symbols = @bak_nested_symbols; 282 @nested_indents = @bak_nested_indents; 283 @nested_conds_indents = @bak_nested_conds_indents; 284 $expr_indent = $bak_expr_indent; 285 $in_block_decls = $bak_in_block_decls; 286 $in_expr = $bak_in_expr; 287 $in_paren_expr = $bak_in_paren_expr; 288 $hanging_offset = $bak_hanging_offset; 289 @in_do_hanging_offsets = @bak_in_do_hanging_offsets; 290 @in_if_hanging_offsets = @bak_in_if_hanging_offsets; 291 $if_maybe_terminated = $bak_if_maybe_terminated; 292 $block_indent = $bak_block_indent; 293 $in_multiline_string = $bak_in_multiline_string; 294 $line_body_start = $bak_line_body_start; 295 $line_opening_brace = $bak_line_opening_brace; 296 $in_typedecl = $bak_in_typedecl; 297} 298 299# auxiliary submodules @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ 300 301sub report_flexibly { 302 my $line = shift; 303 my $msg = shift; 304 my $contents = shift; 305 my $report_SPC = $msg =~ /space|blank/; 306 return if $report_SPC && $sloppy_SPC; 307 308 print "$ARGV:$line:$msg:$contents" unless $self_test; 309 $num_reports_line++; 310 $num_reports++; 311 $num_indent_reports++ if $msg =~ m/:indent /; 312 $num_nesting_issues++ if $msg =~ m/ nesting indent /; 313 $num_syntax_issues++ if $msg =~ m/unclosed|unexpected/; 314 $num_SPC_reports++ if $report_SPC; 315 $num_length_reports++ if $msg =~ m/length/; 316} 317 318sub report { 319 my $msg = shift; 320 report_flexibly($line, $msg, $contents); 321} 322 323sub parens_balance { # count balance of opening parentheses - closing parentheses 324 my $str = shift; 325 return $str =~ tr/\(// - $str =~ tr/\)//; 326} 327 328sub blind_nonspace { # blind non-space text of comment as @, preserving length and spaces 329 # the @ character is used because it cannot occur in normal program code so there is no confusion 330 # comment text is not blinded to whitespace in order to be able to check extra SPC also in comments 331 my $comment_text = shift; 332 $comment_text =~ s/([\.\?\!])\s\s/$1. /g; # in extra SPC checks allow one extra SPC after period '.', '?', or '!' in comments 333 return $comment_text =~ tr/ /@/cr; 334} 335 336# submodule for indentation checking/reporting @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ 337 338sub check_indent { # used for lines outside multi-line string literals 339 my $stmt_indent = $block_indent + $hanging_offset + $local_offset; 340 # print "DEBUG: expr_indent $expr_indent; stmt_indent $stmt_indent = block_indent $block_indent + hanging_offset $hanging_offset + local_offset $local_offset\n"; 341 $stmt_indent = 0 if $stmt_indent < 0; # TODO maybe give warning/error 342 my $stmt_desc = $contents =~ 343 m/^\s*\/\*/ ? "intra-line comment" : 344 $has_label ? "label" : 345 ($hanging_offset != 0 ? "hanging " : ""). 346 ($hanging_offset != 0 ? "stmt/expr" : "stmt/decl"); # $in_typedecl is not fully to the point here 347 my ($ref_desc, $ref_indent) = $expr_indent == 0 ? ($stmt_desc, $stmt_indent) 348 : ("hanging '$hanging_symbol'", $expr_indent); 349 my ($alt_desc, $alt_indent) = ("", $ref_indent); 350 351 # allow indent 1 for labels - this cannot happen for leading ':' 352 ($alt_desc, $alt_indent) = ("outermost position", 1) if $expr_indent == 0 && $has_label; 353 354 if (@nested_conds_indents != 0 && substr($_, $count, 1) eq ":") { 355 # leading ':' within stmt/expr/decl - this cannot happen for labels, leading '&&', or leading '||' 356 # allow special indent at level of corresponding "?" 357 ($alt_desc, $alt_indent) = ("leading ':'", @nested_conds_indents[-1]); 358 } 359 # allow extra indent offset leading '&&' or '||' - this cannot happen for leading ":" 360 ($alt_desc, $alt_indent) = ("leading '$1'", $ref_indent + INDENT_LEVEL) if $contents =~ m/^[\s@]*(\&\&|\|\|)/; 361 362 if ($expr_indent < 0) { # implies @nested_symbols != 0 && @nested_symbols[0] eq "{" && @nested_indents[-1] < 0 363 # allow normal stmt indentation level for hanging initializer/enum expressions after trailing '{' 364 # this cannot happen for labels and overrides special treatment of ':', '&&' and '||' for this line 365 ($alt_desc, $alt_indent) = ("lines after '{'", $stmt_indent); 366 # decide depending on current actual indentation, preventing forth and back 367 @nested_indents[-1] = $count == $stmt_indent ? $stmt_indent : -@nested_indents[-1]; # allow $stmt_indent 368 $ref_indent = $expr_indent = @nested_indents[-1]; 369 } 370 371 # check consistency of indentation within multi-line comment (i.e., between its first, inner, and last lines) 372 if ($in_comment != 0 && $in_comment != 1) { # in multi-line comment but not on its first line 373 if (!$sloppy_cmt) { 374 if ($in_comment > 0) { # not at its end 375 report("indent = $count != $comment_indent within multi-line comment") 376 if $count != $comment_indent; 377 } else { 378 my $tweak = $in_comment == -2 ? 1 : 0; 379 report("indent = ".($count + $tweak)." != $comment_indent at end of multi-line comment") 380 if $count + $tweak != $comment_indent; 381 } 382 } 383 # do not check indentation of last line of non-leading multi-line comment 384 if ($in_comment < 0 && !$leading_comment) { 385 s/^(\s*)@/$1*/; # blind first '@' as '*' to prevent below delayed check for the line before 386 return; 387 } 388 return if $in_comment > 0; # not on its last line 389 # $comment_indent will be checked by the below checks for end of multi-line comment 390 } 391 392 # else check indentation of entire-line comment or entire-line end of multi-line comment 393 # ... w.r.t. indent of the following line by delayed check for the line before 394 if (($in_comment == 0 || $in_comment == 1) # no comment, intra-line comment, or begin of multi-line comment 395 && $line_before > 0 # there is a line before 396 && $contents_before_ =~ m/^(\s*)@[\s@]*$/) { # line before begins with '@', no code follows (except '\') 397 report_flexibly($line_before, "entire-line comment indent = $count_before != $count (of following line)", 398 $contents_before) if !$sloppy_cmt && $count_before != -1 && $count_before != $count; 399 } 400 # ... but allow normal indentation for the current line, else above check will be done for the line before 401 if (($in_comment == 0 || $in_comment < 0) # (no comment,) intra-line comment or end of multi-line comment 402 && m/^(\s*)@[\s@]*$/) { # line begins with '@', no code follows (except '\') 403 if ($count == $ref_indent) { # indentation is like for (normal) code in this line 404 s/^(\s*)@/$1*/; # blind first '@' as '*' to prevent above delayed check for the line before 405 return; 406 } 407 return if !eof; # defer check of entire-line comment to next line 408 } 409 410 # else check indentation of leading intra-line comment or end of multi-line comment 411 if (m/^(\s*)@/) { # line begins with '@', i.e., any (remaining type of) comment 412 if (!$sloppy_cmt && $count != $ref_indent) { 413 report("intra-line comment indent = $count != $ref_indent") if $in_comment == 0; 414 report("multi-line comment indent = $count != $ref_indent") if $in_comment < 0; 415 } 416 return; 417 } 418 419 if ($sloppy_hang && ($hanging_offset != 0 || $expr_indent != 0)) { 420 # do not report same indentation as on the line before (potentially due to same violations) 421 return if $line_before > 0 && $count == $count_before; 422 423 # do not report indentation at normal indentation level while hanging expression indent would be required 424 return if $expr_indent != 0 && $count == $stmt_indent; 425 426 # do not report if contents have been shifted left of nested expr indent (but not as far as stmt indent) 427 # apparently aligned to the right in order to fit within line length limit 428 return if $stmt_indent < $count && $count < $expr_indent && 429 length($contents) == $max_length + length("\n"); 430 } 431 432 report("indent = $count != $ref_indent for $ref_desc". 433 ($alt_desc eq "" 434 || $alt_indent == $ref_indent # prevent showing alternative that happens to have equal value 435 ? "" : " or $alt_indent for $alt_desc")) 436 if $count != $ref_indent && $count != $alt_indent; 437} 438 439# submodules handling indentation within expressions @@@@@@@@@@@@@@@@@@@@@@@@@@@ 440 441sub update_nested_indents { # may reset $in_paren_expr and in this case also resets $in_expr 442 my $str = shift; 443 my $start = shift; # defaults to 0 444 my $terminator_position = -1; 445 for (my $i = $start; $i < length($str); $i++) { 446 my $c; 447 my $curr = substr($str, $i); 448 if ($curr =~ m/^(.*?)([{}()?:;\[\]])(.*)$/) { # match from position $i the first {}()?:;[] 449 $c = $2; 450 } else { 451 last; 452 } 453 my ($head, $tail) = (substr($str, 0, $i).$1, $3); 454 $i += length($1) + length($2) - 1; 455 456 # stop at terminator outside 'for (..;..;..)', assuming that 'for' is followed by '(' 457 return $i if $c eq ";" && (!$in_paren_expr || @nested_indents == 0); 458 459 my $in_stmt = $in_expr || @nested_symbols != 0; # not: || $in_typedecl != 0 460 if ($c =~ m/[{([?]/) { # $c is '{', '(', '[', or '?' 461 if ($c eq "{") { # '{' in any context 462 $in_block_decls = 0 if !$in_expr && $in_typedecl == 0; 463 # cancel newly hanging_offset if opening brace '{' is after non-whitespace non-comment: 464 $hanging_offset -= INDENT_LEVEL if $hanging_offset > 0 && $head =~ m/[^\s\@]/; 465 push @nested_block_indents, $block_indent; 466 push @nested_hanging_offsets, $in_expr ? $hanging_offset : 0; 467 push @nested_in_typedecl, $in_typedecl if $in_typedecl != 0; 468 $block_indent += INDENT_LEVEL + $hanging_offset; 469 $hanging_offset = 0; 470 } 471 if ($c ne "{" || $in_stmt) { # for '{' inside stmt/expr (not: decl), for '(', '[', or '?' anywhere 472 $tail =~ m/^([\s@]*)([^\s\@])/; 473 push @nested_indents, defined $2 474 ? $i + 1 + length($1) # actual indentation of following non-space non-comment 475 : $c ne "{" ? +($i + 1) # just after '(' or '[' if only whitespace thereafter 476 : -($i + 1); # allow also $stmt_indent if '{' with only whitespace thereafter 477 push @nested_symbols, $c; # done also for '?' to be able to check correct nesting 478 push @nested_conds_indents, $i if $c eq "?"; # remember special alternative indent for ':' 479 } 480 } elsif ($c =~ m/[})\]:]/) { # $c is '}', ')', ']', or ':' 481 my $opening_c = ($c =~ tr/})]:/{([/r); 482 if (($c ne ":" || $in_stmt # ignore ':' outside stmt/expr/decl 483 # in the presence of ':', one could add this sanity check: 484 # && !(# ':' after initial label/case/default 485 # $head =~ m/^([\s@]*)(case\W.*$|\w+$)/ || # this matching would not work for 486 # # multi-line expr after 'case' 487 # # bitfield length within unsigned type decl 488 # $tail =~ m/^[\s@]*\d+/ # this matching would need improvement 489 # ) 490 )) { 491 if ($c ne "}" || $in_stmt) { # for '}' inside stmt/expr/decl, ')', ']', or ':' 492 if (@nested_symbols != 0 && 493 @nested_symbols[-1] == $opening_c) { # for $c there was a corresponding $opening_c 494 pop @nested_indents; 495 pop @nested_symbols; 496 pop @nested_conds_indents if $opening_c eq "?"; 497 } else { 498 report("unexpected '$c' @ ".($in_paren_expr ? "(expr)" : "expr")); 499 next; 500 } 501 } 502 if ($c eq "}") { # '}' at block level but also inside stmt/expr/decl 503 if (@nested_block_indents == 0) { 504 report("unexpected '}'"); 505 } else { 506 $block_indent = pop @nested_block_indents; 507 $hanging_offset = pop @nested_hanging_offsets; 508 $in_typedecl = pop @nested_in_typedecl if @nested_in_typedecl != 0; 509 } 510 } 511 if ($in_paren_expr && !grep(/\(/, @nested_symbols)) { # end of (expr) 512 check_nested_nonblock_indents("(expr)"); 513 $in_paren_expr = $in_expr = 0; 514 report("code after (expr)") 515 if $tail =~ m/^([^{]*)/ && $1 =~ m/[^\s\@;]/; # non-space non-';' before any '{' 516 } 517 } 518 } 519 } 520 return -1; 521} 522 523sub check_nested_nonblock_indents { 524 my $position = shift; 525 while (@nested_symbols != 0) { 526 my $symbol = pop @nested_symbols; 527 report("unclosed '$symbol' in $position"); 528 if ($symbol eq "{") { # repair stack of blocks 529 $block_indent = pop @nested_block_indents; 530 $hanging_offset = pop @nested_hanging_offsets; 531 $in_typedecl = pop @nested_in_typedecl if @nested_in_typedecl != 0; 532 } 533 } 534 @nested_indents = (); 535 @nested_conds_indents = (); 536} 537 538# start of main program @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ 539 540reset_file_state(); 541 542while (<>) { # loop over all lines of all input files 543 $self_test = $ARGV =~ m/check-format-test/; 544 $_ = "" if $self_test && m/ blank line within local decls /; 545 $line++; 546 s/\r$//; # strip any trailing CR '\r' (which are typical on Windows systems) 547 $contents = $_; 548 549 # check for illegal characters 550 if (m/(.*?)([\x00-\x09\x0B-\x1F\x7F-\xFF])/) { 551 my $col = length($1); 552 report(($2 eq "\x09" ? "TAB" : $2 eq "\x0D" ? "CR " : $2 =~ m/[\x00-\x1F]/ ? "non-printable" 553 : "non-7bit char") . " at column $col") ; 554 } 555 556 # check for whitespace at EOL 557 report("trailing whitespace at EOL") if m/\s\n$/; 558 559 # assign to $count the actual indentation level of the current line 560 chomp; # remove trailing NL '\n' 561 m/^(\s*)/; 562 $count = length($1); # actual indentation 563 $has_label = 0; 564 $local_offset = 0; 565 566 # character/string literals @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ 567 568 s/\\["']/@@/g; # blind all '"' and "'" escaped by '\' (typically within character literals or string literals) 569 570 # handle multi-line string literals to avoid confusion on starting/ending '"' and trailing '\' 571 if ($in_multiline_string) { 572 if (s#^([^"]*)"#($1 =~ tr/"/@/cr).'@'#e) { # string literal terminated by '"' 573 # string contents and its terminating '"' have been blinded as '@' 574 $count = -1; # do not check indentation 575 } else { 576 report("multi-line string literal not terminated by '\"' and trailing '\' is missing") 577 unless s#^([^\\]*)\s*\\\s*$#$1#; # strip trailing '\' plus any whitespace around 578 goto LINE_FINISHED; 579 } 580 } 581 582 # blind contents of character and string literals as @, preserving length (but not spaces) 583 # this prevents confusing any of the matching below, e.g., of whitespace and comment delimiters 584 s#('[^']*')#$1 =~ tr/'/@/cr#eg; # handle all intra-line character literals 585 s#("[^"]*")#$1 =~ tr/"/@/cr#eg; # handle all intra-line string literals 586 $in_multiline_string = # handle trailing string literal terminated by '\' 587 s#^(([^"]*"[^"]*")*[^"]*)("[^"]*)\\(\s*)$#$1.($3 =~ tr/"/@/cr).'"'.$4#e; 588 # its contents have been blinded and the trailing '\' replaced by '"' 589 590 # strip any other trailing '\' along with any whitespace around it such that it does not interfere with various matching below 591 my $trailing_backslash = s#^(.*?)\s*\\\s*$#$1#; # trailing '\' possibly preceded or followed by whitespace 592 my $essentially_blank_line = m/^\s*$/; # just whitespace and maybe a '\' 593 594 # comments @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ 595 596 # do/prepare checks within multi-line comments 597 my $self_test_exception = $self_test ? "@" : ""; 598 if ($in_comment > 0) { # this still includes the last line of multi-line comment 599 my ($head, $any_symbol, $cmt_text) = m/^(\s*)(.?)(.*)$/; 600 if ($any_symbol eq "*") { 601 report("missing space or '*' after leading '*' in multi-line comment") if $cmt_text =~ m|^[^*\s/$self_test_exception]|; 602 } else { 603 report("missing leading '*' in multi-line comment"); 604 } 605 $in_comment++; 606 } 607 608 # detect end of comment, must be within multi-line comment, check if it is preceded by non-whitespace text 609 if ((my ($head, $tail) = m|^(.*?)\*/(.*)$|) && $1 ne '/') { # ending comment: '*/' 610 report("missing space or '*' before '*/'") if $head =~ m/[^*\s]$/; 611 report("missing space (or ',', ';', ')', '}', ']') after '*/'") if $tail =~ m/^[^\s,;)}\]]/; # no space or ,;)}] after '*/' 612 if (!($head =~ m|/\*|)) { # not begin of comment '/*', which is is handled below 613 if ($in_comment == 0) { 614 report("unexpected '*/' outside comment"); 615 $_ = "$head@@".$tail; # blind the "*/" 616 } else { 617 report("text before '*/' in multi-line comment") if ($head =~ m/[^*\s]/); # non-SPC before '*/' 618 $in_comment = -1; # indicate that multi-line comment ends on current line 619 if ($count > 0) { 620 # make indentation of end of multi-line comment appear like of leading intra-line comment 621 $head =~ s/^(\s*)\s/$1@/; # replace the last leading space by '@' 622 $count--; 623 $in_comment = -2; # indicate that multi-line comment ends on current line, with tweak 624 } 625 my $cmt_text = $head; 626 $_ = blind_nonspace($cmt_text)."@@".$tail; 627 } 628 } 629 } 630 631 # detect begin of comment, check if it is followed by non-space text 632 MATCH_COMMENT: 633 if (my ($head, $opt_minus, $tail) = m|^(.*?)/\*(-?)(.*)$|) { # begin of comment: '/*' 634 report("missing space before '/*'") 635 if $head =~ m/[^\s(\*]$/; # not space, '(', or or '*' (needed to allow '*/') before comment delimiter 636 report("missing space, '*', or '!' after '/*$opt_minus'") if $tail =~ m/^[^\s*!$self_test_exception]/; 637 my $cmt_text = $opt_minus.$tail; # preliminary 638 if ($in_comment > 0) { 639 report("unexpected '/*' inside multi-line comment"); 640 } elsif ($tail =~ m|^(.*?)\*/(.*)$|) { # comment end: */ on same line 641 report("unexpected '/*' inside intra-line comment") if $1 =~ /\/\*/; 642 # blind comment text, preserving length and spaces 643 ($cmt_text, my $rest) = ($opt_minus.$1, $2); 644 $_ = "$head@@".blind_nonspace($cmt_text)."@@".$rest; 645 goto MATCH_COMMENT; 646 } else { # begin of multi-line comment 647 my $self_test_exception = $self_test ? "(@\d?)?" : ""; 648 report("text after '/*' in multi-line comment") 649 unless $tail =~ m/^$self_test_exception.?[*\s]*$/; 650 # tail not essentially blank, first char already checked 651 # adapt to actual indentation of first line 652 $comment_indent = length($head) + 1; 653 $_ = "$head@@".blind_nonspace($cmt_text); 654 $in_comment = 1; 655 $leading_comment = $head =~ m/^\s*$/; # there is code before beginning delimiter 656 $formatted_comment = $opt_minus eq "-"; 657 } 658 } elsif (($head, $tail) = m|^\{-(.*)$|) { # begin of Perl pragma: '{-' 659 } 660 661 if ($in_comment > 1) { # still inside multi-line comment (not at its begin or end) 662 m/^(\s*)\*?(\s*)(.*)$/; 663 $_ = $1."@".$2.blind_nonspace($3); 664 } 665 666 # handle special case of line after '#ifdef __cplusplus' (which typically appears in header files) 667 if ($ifdef__cplusplus) { 668 $ifdef__cplusplus = 0; 669 $_ = "$1 $2" if $contents =~ m/^(\s*extern\s*"C"\s*)\{(\s*)$/; # ignore opening brace in 'extern "C" {' 670 goto LINE_FINISHED if m/^\s*\}\s*$/; # ignore closing brace '}' 671 } 672 673 # check for over-long lines, 674 # while allowing trailing (also multi-line) string literals to go past $max_length 675 my $len = length; # total line length (without trailing '\n') 676 if ($len > $max_length && 677 !(m/^(.*)"[^"]*"\s*[\)\}\]]*[,;]?\s*$/ # string literal terminated by '"' (or '\'), then maybe )}],; 678 && length($1) < $max_length) 679 # this allows over-long trailing string literals with beginning col before $max_length 680 ) { 681 report("line length = $len > ".$max_length); 682 } 683 684 # handle C++ / C99 - style end-of-line comments 685 if (my ($head, $cmt_text) = m|^(.*?)//(.*$)|) { 686 report("'//' end-of-line comment"); # the '//' comment style is not allowed for C90 687 # blind comment text, preserving length and spaces 688 $_ = "$head@@".blind_nonspace($cmt_text); 689 } 690 691 # at this point all non-space portions of any types of comments have been blinded as @ 692 693 goto LINE_FINISHED if $essentially_blank_line; 694 695 # handle preprocessor directives @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ 696 697 if (s/^(\s*#)(\s*)(\w+)//) { # line beginning with '#' and directive name; 698 # blank these portions to prevent confusion with C-level 'if', 'else', etc. 699 my ($lead, $space) = ($1, $2); 700 $preproc_directive = $3; 701 $_ = "$lead$space$preproc_directive$_" if $preproc_directive =~ m/^(define|include)$/; # yet do not blank #define or #include to prevent confusing the indentation or whitespace checks, resp. 702 $_ = blind_nonspace($_) if $preproc_directive eq "error"; # blind error message 703 if ($in_preproc != 0) { 704 report("preprocessor directive within multi-line directive"); 705 reset_indentation_state(); 706 } 707 $in_preproc++; 708 report("indent = $count != 0 for '#'") if $count != 0; 709 report("'#$preproc_directive' with constant condition") 710 if $preproc_directive =~ m/^(if|elif)$/ && m/^[\W0-9]+$/ && !$trailing_backslash; 711 $preproc_if_nesting-- if $preproc_directive =~ m/^(else|elif|endif)$/; 712 if ($preproc_if_nesting < 0) { 713 $preproc_if_nesting = 0; 714 report("unexpected '#$preproc_directive' according to '#if' nesting"); 715 } 716 my $space_count = length($space); # maybe could also use indentation before '#' 717 report("'#if' nesting indent = $space_count != $preproc_if_nesting") if $space_count != $preproc_if_nesting; 718 $preproc_if_nesting++ if $preproc_directive =~ m/^(if|ifdef|ifndef|else|elif)$/; 719 $ifdef__cplusplus = $preproc_directive eq "ifdef" && m/\s+__cplusplus\s*$/; 720 721 # handle indentation of preprocessor directive independently of surrounding normal code 722 $count = -1; # do not check indentation of first line of preprocessor directive 723 backup_indentation_state(); 724 reset_indentation_state(); 725 } 726 727 # intra-line whitespace nits @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ 728 729 my $in_multiline_comment = ($in_comment > 1 || $in_comment < 0); # $in_multiline_comment refers to line before 730 if (!$sloppy_SPC && !($in_multiline_comment && $formatted_comment)) { 731 sub extra_SPC { 732 my $intra_line = shift; 733 return "extra space".($intra_line =~ m/@\s\s/ ? 734 $in_comment != 0 ? " in multi-line comment" 735 : " in intra-line comment" : ""); 736 } 737 sub split_line_head { # split line contents into header containing leading spaces and the first non-space char, and the rest of the line 738 my $comment_symbol = 739 $in_comment != 0 ? "@" : ""; # '@' will match the blinded leading '*' in multi-line comment 740 # $in_comment may pertain to the following line due to delayed check 741 # do not check for extra SPC in leading spaces including any '#' (or '*' within multi-line comment) 742 shift =~ m/^(\s*([#$comment_symbol]\s*)?)(.*?)\s*$/; 743 return ($1, $3); 744 } 745 my ($head , $intra_line ) = split_line_head($_); 746 my ($head1, $intra_line1) = split_line_head($contents_before_ ) if $line_before > 0; 747 my ($head2, $intra_line2) = split_line_head($contents_before_2) if $line_before2 > 0; 748 if ($line_before > 0) { # check with one line delay, such that at least $contents_before is available 749 sub column_alignments_only { # return 1 if the given line has multiple consecutive spaces only at columns that match the reference line 750 # all parameter strings are assumed to contain contents after blinding comments etc. 751 my $head = shift; # leading spaces and the first non-space char 752 my $intra = shift; # the rest of the line contents 753 my $contents = shift; # reference line 754 # check if all extra SPC in $intra is used only for multi-line column alignment with $contents 755 my $offset = length($head); 756 for (my $col = 0; $col < length($intra) - 2; $col++) { 757 my $substr = substr($intra, $col); 758 next unless $substr =~ m/^\s\s\S/; # extra SPC (but not in leading spaces of the line) 759 next if !$eol_cmt && $substr =~ m/^[@\s]+$/; # end-of-line comment 760 return 0 unless substr($contents, $col + $offset + 1, 2) =~ m/\s\S/; # reference line contents do not match 761 } 762 return 1; 763 } 764 report_flexibly($line_before, extra_SPC($intra_line1), $contents_before) if $intra_line1 =~ m/\s\s\S/ && 765 !( column_alignments_only($head1, $intra_line1, $_ ) # compare with $line 766 || ($line_before2 > 0 && 767 column_alignments_only($head1, $intra_line1, $contents_before_2))); # compare w/ $line_before2 768 report(extra_SPC($intra_line)) if $intra_line =~ m/\s\s\S/ && eof 769 && ! column_alignments_only($head , $intra_line , $contents_before_ ) ; # compare w/ $line_before 770 } elsif (eof) { # special case: just one line exists 771 report(extra_SPC($intra_line)) if $intra_line =~ m/\s\s\S/; 772 } 773 # ignore paths in #include 774 $intra_line =~ s/^(include\s*)(".*?"|<.*?>)/$1/e if $head =~ m/#/; 775 report("missing space before '$2'") 776 if $intra_line =~ m/(\S)((<<|>>)=)/ # '<<=' or >>=' without preceding space 777 || ($intra_line =~ m/(\S)([\+\-\*\/\/%\&\|\^\!<>=]=)/ 778 && "$1$2" ne "<<=" && "$1$2" ne ">>=") # other <op>= or (in)equality without preceding space 779 || ($intra_line =~ m/(\S)=/ 780 && !($1 =~ m/[\+\-\*\/\/%\&\|\^\!<>=]/) 781 && $intra_line =~ m/(\S)(=)/); # otherwise, '=' without preceding space 782 # treat op= and comparison operators as simple '=', simplifying matching below 783 $intra_line =~ s/(<<|>>|[\+\-\*\/\/%\&\|\^\!<>=])=/=/g; 784 # treat (type) variables within macro, indicated by trailing '\', as 'int' simplifying matching below 785 $intra_line =~ s/[A-Z_]+/int/g if $trailing_backslash; 786 # treat double &&, ||, <<, and >> as single ones, simplifying matching below 787 $intra_line =~ s/(&&|\|\||<<|>>)/substr($1, 0, 1)/eg; 788 # remove blinded comments etc. directly after [{( 789 while ($intra_line =~ s/([\[\{\(])@+\s?/$1/e) {} # /g does not work here 790 # remove blinded comments etc. directly before ,;)}] 791 while ($intra_line =~ s/\s?@+([,;\)\}\]])/$1/e) {} # /g does not work here 792 # treat remaining blinded comments and string literal contents as (single) space during matching below 793 $intra_line =~ s/@+/ /g; # note that extra SPC has already been handled above 794 $intra_line =~ s/\s+$//; # strip any (resulting) space at EOL 795 # replace ';;' or '; ;' by ';' in "for (;;)" and in "for (...)" unless "..." contains just SPC and ';' characters: 796 $intra_line =~ s/((^|\W)for\s*\()([^;]*?)(\s*)(;\s?);(\s*)([^;]*)(\))/ 797 "$1$3$4".("$3$4$5$6$7" eq ";" || $3 ne "" || $7 ne "" ? "" : $5).";$6$7$8"/eg; 798 # strip trailing ';' or '; ' in "for (...)" except in "for (;;)" or "for (;; )": 799 $intra_line =~ s/((^|\W)for\s*\()([^;]*(;[^;]*)?)(;\s?)(\))/ 800 "$1$3".($3 eq ";" ? $5 : "")."$6"/eg; 801 $intra_line =~ s/(=\s*)\{ /"$1@ "/eg; # do not report {SPC in initializers such as ' = { 0, };' 802 $intra_line =~ s/, \};/, @;/g; # do not report SPC} in initializers such as ' = { 0, };' 803 report("space before '$1'") if $intra_line =~ m/[\w)\]]\s+(\+\+|--)/; # postfix ++/-- with preceding space 804 report("space after '$1'") if $intra_line =~ m/(\+\+|--)\s+[a-zA-Z_(]/; # prefix ++/-- with following space 805 $intra_line =~ s/\.\.\./@/g; # blind '...' 806 report("space before '$1'") if $intra_line =~ m/\s(\.|->)/; # '.' or '->' with preceding space 807 report("space after '$1'") if $intra_line =~ m/(\.|->)\s/; # '.' or '->' with following space 808 $intra_line =~ s/\-\>|\+\+|\-\-/@/g; # blind '->,', '++', and '--' 809 report("space before '$1'") if $intra_line =~ m/[^:)]\s+(;)/; # space before ';' but not after ':' or ')' # note that 810 # exceptions for "for (;; )" are handled above 811 report("space before '$1'") if $intra_line =~ m/\s([,)\]])/; # space before ,)] 812 report("space after '$1'") if $intra_line =~ m/([(\[~!])\s/; # space after ([~! 813 report("space after '$1'") if $intra_line =~ m/(defined)\s/; # space after 'defined' 814 report("missing space before '$1'") if $intra_line =~ m/\S([|\/%<>^\?])/; # |/%<>^? without preceding space 815 # TODO ternary ':' without preceding SPC, while allowing no SPC before ':' after 'case' 816 report("missing space before binary '$2'") if $intra_line =~ m/([^\s{()\[e])([+\-])/; # '+'/'-' without preceding space or {()[e 817 # ')' may be used for type casts or before "->", 'e' may be used for numerical literals such as "1e-6" 818 report("missing space before binary '$1'") if $intra_line =~ m/[^\s{()\[*!]([*])/; # '*' without preceding space or {()[*! 819 report("missing space before binary '$1'") if $intra_line =~ m/[^\s{()\[]([&])/; # '&' without preceding space or {()[ 820 report("missing space after ternary '$1'") if $intra_line =~ m/(:)[^\s\d]/; # ':' without following space or digit 821 report("missing space after '$1'") if $intra_line =~ m/([,;=|\/%<>^\?])\S/; # ,;=|/%<>^? without following space 822 report("missing space after binary '$1'") if $intra_line=~m/[^{(\[]([*])[^\sa-zA-Z_(),*]/;# '*' w/o space or \w(),* after 823 # TODO unary '*' must not be followed by SPC 824 report("missing space after binary '$1'") if $intra_line=~m/([&])[^\sa-zA-Z_(]/; # '&' w/o following space or \w( 825 # TODO unary '&' must not be followed by SPC 826 report("missing space after binary '$1'") if $intra_line=~m/[^{(\[]([+\-])[^\s\d(]/; # +/- w/o following space or \d( 827 # TODO unary '+' and '-' must not be followed by SPC 828 report("missing space after '$2'") if $intra_line =~ m/(^|\W)(if|while|for|switch|case)[^\w\s]/; # kw w/o SPC 829 report("missing space after '$2'") if $intra_line =~ m/(^|\W)(return)[^\w\s;]/; # return w/o SPC or ';' 830 report("space after function/macro name") 831 if $intra_line =~ m/(\w+)\s+\(/ # fn/macro name with space before '(' 832 && !($1 =~ m/^(sizeof|if|else|while|do|for|switch|case|default|break|continue|goto|return|void|char|signed|unsigned|int|short|long|float|double|typedef|enum|struct|union|auto|extern|static|const|volatile|register)$/) # not keyword 833 && !(m/^\s*#\s*define\s+\w+\s+\(/); # not a macro without parameters having a body that starts with '(' 834 report("missing space before '{'") if $intra_line =~ m/[^\s{(\[]\{/; # '{' without preceding space or {([ 835 report("missing space after '}'") if $intra_line =~ m/\}[^\s,;\])}]/; # '}' without following space or ,;])} 836 } 837 838 # adapt required indentation @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ 839 840 s/(\w*ASN1_[A-Z_]+END\w*([^(]|\(.*?\)|$))/$1;/g; # treat *ASN1_*END*(..) macro calls as if followed by ';' 841 842 my $nested_indents_position = 0; 843 844 # update indents according to leading closing brace(s) '}' or label or switch case 845 my $in_stmt = $in_expr || @nested_symbols != 0 || $in_typedecl != 0; 846 if ($in_stmt) { # expr/stmt/type decl/var def/fn hdr, i.e., not at block level 847 if (m/^([\s@]*\})/) { # leading '}' within stmt, any preceding blinded comment must not be matched 848 $in_block_decls = -1; 849 my $head = $1; 850 update_nested_indents($head); 851 $nested_indents_position = length($head); 852 if (@nested_symbols >= 1) { 853 $hanging_symbol = @nested_symbols[-1]; 854 $expr_indent = @nested_indents[-1]; 855 } else { # typically end of initialiizer expr or enum 856 $expr_indent = 0; 857 } 858 } elsif (m/^([\s@]*)(static_)?ASN1_ITEM_TEMPLATE_END(\W|$)/) { # workaround for ASN1 macro indented as '}' 859 $local_offset = -INDENT_LEVEL; 860 $expr_indent = 0; 861 } elsif (m/;.*?\}/) { # expr ends with ';' before '}' 862 report("code before '}'"); 863 } 864 } 865 if (@in_do_hanging_offsets != 0 && # note there is nothing like "unexpected 'while'" 866 m/^[\s@]*while(\W|$)/) { # leading 'while' 867 $hanging_offset = pop @in_do_hanging_offsets; 868 } 869 if ($if_maybe_terminated) { 870 if (m/(^|\W)else(\W|$)/) { # (not necessarily leading) 'else' 871 if (@in_if_hanging_offsets == 0) { 872 report("unexpected 'else'"); 873 } else { 874 $hanging_offset = pop @in_if_hanging_offsets; 875 } 876 } else { 877 @in_if_hanging_offsets = (); # note there is nothing like "unclosed 'if'" 878 $hanging_offset = 0; 879 } 880 } 881 if (!$in_stmt) { # at block level, i.e., outside expr/stmt/type decl/var def/fn hdr 882 $if_maybe_terminated = 0; 883 if (my ($head, $before, $tail) = m/^([\s@]*([^{}]*)\})[\s@]*(.*)$/) { # leading closing '}', but possibly 884 # with non-whitespace non-'{' before 885 report("code after '}'") unless $tail eq "" || $tail =~ m/(else|while|OSSL_TRACE_END)(\W|$)/; 886 my $outermost_level = @nested_block_indents == 1 && @nested_block_indents[0] == 0; 887 if (!$sloppy_bodylen && $outermost_level && $line_body_start != 0) { 888 my $body_len = $line - $line_body_start - 1; 889 report_flexibly($line_function_start, "function body length = $body_len > ".MAX_BODY_LENGTH." lines", 890 $last_function_header) if $body_len > MAX_BODY_LENGTH; 891 $line_body_start = 0; 892 } 893 if ($before ne "") { # non-whitespace non-'{' before '}' 894 report("code before '}'"); 895 } else { # leading '}' outside stmt, any preceding blinded comment must not be matched 896 $in_block_decls = -1; 897 $local_offset = $block_indent + $hanging_offset - INDENT_LEVEL; 898 update_nested_indents($head); 899 $nested_indents_position = length($head); 900 $local_offset -= ($block_indent + $hanging_offset); 901 # in effect $local_offset = -INDENT_LEVEL relative to $block_indent + $hanging_offset values before 902 } 903 } 904 905 # handle opening brace '{' after if/else/while/for/switch/do on line before 906 if ($hanging_offset > 0 && m/^[\s@]*{/ && # leading opening '{' 907 $line_before > 0 && 908 $contents_before_ =~ m/(^|^.*\W)(if|else|while|for|(OSSL_)?LIST_FOREACH(_\w+)?|switch|do)(\W.*$|$)/) { 909 $keyword_opening_brace = $1; 910 $hanging_offset -= INDENT_LEVEL; # cancel newly hanging_offset 911 } 912 913 if (m/^[\s@]*(case|default)(\W.*$|$)/) { # leading 'case' or 'default' 914 my $keyword = $1; 915 report("code after $keyword: ") if $2 =~ /:.*[^\s@].*$/; 916 $local_offset = -INDENT_LEVEL; 917 } else { 918 if (m/^([\s@]*)(\w+):/) { # (leading) label, cannot be "default" 919 $local_offset = -INDENT_LEVEL; 920 $has_label = 1; 921 } 922 } 923 } 924 925 # potential adaptations of indent in first line of macro body in multi-line macro definition 926 if ($in_preproc != 0 && $in_macro_header > 0) { 927 if ($in_macro_header > 1) { # still in macro definition header 928 $in_macro_header += parens_balance($_); 929 } else { # begin of macro body 930 $in_macro_header = 0; 931 if ($count == $block_indent - $preproc_offset # body began with same indentation as preceding code 932 && $sloppy_macro) { # workaround for this situation is enabled 933 $block_indent -= $preproc_offset; 934 $preproc_offset = 0; 935 } 936 } 937 } 938 939 # check required indentation @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ 940 941 check_indent() if $count >= 0; # not for start of preprocessor directive and not if multi-line string literal is continued 942 943 # check for blank lines within/after local decls @@@@@@@@@@@@@@@@@@@@@@@@@@@ 944 945 if ($in_block_decls >= 0 && 946 $in_comment == 0 && !m/^\s*\*?@/ && # not in a multi-line or intra-line comment 947 !$in_expr && $expr_indent == 0 && $in_typedecl == 0) { 948 my $blank_line_before = $line > 1 && $code_contents_before =~ m/^\s*(\\\s*)?$/; 949 # essentially blank line before: just whitespace and maybe a '\' 950 if (m/^[\s(]*(char|signed|unsigned|int|short|long|float|double|enum|struct|union|auto|extern|static|const|volatile|register)(\W|$)/ # clear start of local decl 951 || (m/^(\s*(\w+|\[\]|[\*()]))+?\s+[\*\(]*\w+(\s*(\)|\[[^\]]*\]))*\s*[;,=]/ # weak check for decl involving user-defined type 952 && !m/^\s*(\}|sizeof|if|else|while|do|for|switch|case|default|break|continue|goto|return)(\W|$)/)) { 953 $in_block_decls++; 954 report_flexibly($line - 1, "blank line within local decls, before", $contents) if $blank_line_before; 955 } else { 956 report_flexibly($line, "missing blank line after local decls", "\n$contents_before$contents") 957 if $in_block_decls > 0 && !$blank_line_before; 958 $in_block_decls = -1 unless 959 m/^\s*(\\\s*)?$/ # essentially blank line: just whitespace (and maybe a trailing '\') 960 || $in_comment != 0 || m/^\s*\*?@/; # in multi-line comment or an intra-line comment 961 } 962 } 963 964 $in_comment = 0 if $in_comment < 0; # multi-line comment has ended 965 966 # do some further checks @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ 967 968 my $outermost_level = $block_indent - $preproc_offset == 0; 969 970 report("more than one stmt") if !m/(^|\W)(for|(OSSL_)?LIST_FOREACH(_\w+)?)(\W.*|$)/ && # no 'for' - TODO improve matching 971 m/;.*;/; # two or more terminators ';', so more than one statement 972 973 # check for code block containing a single line/statement 974 if ($line_before2 > 0 && !$outermost_level && # within function body 975 $in_typedecl == 0 && @nested_indents == 0 && # neither within type declaration nor inside stmt/expr 976 m/^[\s@]*\}\s*(\w*)/) { # leading closing brace '}', any preceding blinded comment must not be matched 977 # TODO extend detection from single-line to potentially multi-line statement 978 my $next_word = $1; 979 if ($line_opening_brace > 0 && 980 ($keyword_opening_brace ne "if" || 981 $extended_1_stmt || $next_word ne "else") && 982 ($line_opening_brace == $line_before2 || 983 $line_opening_brace == $line_before) 984 && $contents_before =~ m/;/) { # there is at least one terminator ';', so there is some stmt 985 # TODO do not report cases where a further else branch 986 # follows with a block containing more than one line/statement 987 report_flexibly($line_before, "'$keyword_opening_brace' { 1 stmt }", $contents_before); 988 } 989 } 990 991 report("single-letter name '$2'") if (m/(^|.*\W)([IO])(\W.*|$)/); # single-letter name 'I' or 'O' # maybe re-add 'l'? 992 # constant on LHS of comparison or assignment, e.g., NULL != x or 'a' < c, but not a + 1 == b 993 report("constant on LHS of '$3'") 994 if (m/(['"]|([\+\-\*\/\/%\&\|\^<>]\s*)?\W[0-9]+L?|\WNULL)\s*([\!<>=]=|[<=>])([<>]?)/ && 995 $2 eq "" && (($3 ne "<" && $3 ne "='" && $3 ne ">") || $4 eq "")); 996 997 # TODO report needless use of parentheses, while 998 # macro parameters should always be in parens (except when passed on), e.g., '#define ID(x) (x)' 999 1000 # adapt required indentation for following lines @@@@@@@@@@@@@@@@@@@@@@@@@@@ 1001 1002 # set $in_expr, $in_paren_expr, and $hanging_offset for if/while/for/switch, return/enum, and assignment RHS 1003 my $paren_expr_start = 0; 1004 my $return_enum_start = 0; 1005 my $assignment_start = 0; 1006 my $tmp = $_; 1007 $tmp =~ s/[\!<>=]=/@@/g; # blind (in-)equality symbols like '<=' as '@@' to prevent matching them as '=' below 1008 if (m/^((^|.*\W)(if|while|for|(OSSL_)?LIST_FOREACH(_\w+)?|switch))(\W.*|$)$/) { # (last) if/for/while/switch 1009 $paren_expr_start = 1; 1010 } elsif (m/^((^|.*\W)(return|enum))(\W.*|$)/ # (last) return/enum 1011 && !$in_expr && @nested_indents == 0 && parens_balance($1) == 0) { # not nested enum 1012 $return_enum_start = 1; 1013 } elsif ($tmp =~ m/^(([^=]*)(=))(.*)$/ # (last) '=', i.e., assignment 1014 && !$in_expr && @nested_indents == 0 && parens_balance($1) == 0) { # not nested assignment 1015 $assignment_start = 1; 1016 } 1017 if ($paren_expr_start || $return_enum_start || $assignment_start) 1018 { 1019 my ($head, $mid, $tail) = ($1, $3, $4); 1020 $keyword_opening_brace = $mid if $mid ne "="; 1021 # to cope with multi-line expressions, do this also if !($tail =~ m/\{/) 1022 push @in_if_hanging_offsets, $hanging_offset if $mid eq "if"; 1023 1024 # already handle $head, i.e., anything before expression 1025 update_nested_indents($head, $nested_indents_position); 1026 $nested_indents_position = length($head); 1027 # now can set $in_expr and $in_paren_expr 1028 $in_expr = 1; 1029 $in_paren_expr = 1 if $paren_expr_start; 1030 if ($mid eq "while" && @in_do_hanging_offsets != 0) { 1031 $hanging_offset = pop @in_do_hanging_offsets; 1032 } else { 1033 $hanging_offset += INDENT_LEVEL; # tentatively set hanging_offset, may be canceled by following '{' 1034 } 1035 } 1036 1037 # set $hanging_offset and $keyword_opening_brace for do/else 1038 if (my ($head, $mid, $tail) = m/(^|^.*\W)(else|do)(\W.*|$)$/) { # last else/do, where 'do' is preferred, but not #else 1039 my $code_before = $head =~ m/[^\s\@}]/; # leading non-whitespace non-comment non-'}' 1040 report("code before '$mid'") if $code_before; 1041 report("code after '$mid'" ) if $tail =~ m/[^\s\@{]/# trailing non-whitespace non-comment non-'{' (non-'\') 1042 && !($mid eq "else" && $tail =~ m/[\s@]*if(\W|$)/); 1043 if ($mid eq "do") { # workarounds for code before 'do' 1044 if ($head =~ m/(^|^.*\W)(else)(\W.*$|$)/) { # 'else' ... 'do' 1045 $hanging_offset += INDENT_LEVEL; # tentatively set hanging_offset, may be canceled by following '{' 1046 } 1047 if ($head =~ m/;/) { # terminator ';' ... 'do' 1048 @in_if_hanging_offsets = (); # note there is nothing like "unclosed 'if'" 1049 $hanging_offset = 0; 1050 } 1051 } 1052 push @in_do_hanging_offsets, $hanging_offset if $mid eq "do"; 1053 if ($code_before && $mid eq "do") { 1054 $hanging_offset = length($head) - $block_indent; 1055 } 1056 if (!$in_paren_expr) { 1057 $keyword_opening_brace = $mid if $tail =~ m/\{/; 1058 $hanging_offset += INDENT_LEVEL; 1059 } 1060 } 1061 1062 # set $in_typedecl and potentially $hanging_offset for type declaration 1063 if (!$in_expr && @nested_indents == 0 # not in expression 1064 && m/(^|^.*\W)(typedef|enum|struct|union)(\W.*|$)$/ 1065 && parens_balance($1) == 0 # not in newly started expression or function arg list 1066 && ($2 eq "typedef" || !($3 =~ m/\s*\w++\s*(.)/ && $1 ne "{")) # 'struct'/'union'/'enum' <name> not followed by '{' 1067 # not needed: && $keyword_opening_brace = $2 if $3 =~ m/\{/; 1068 ) { 1069 $in_typedecl++; 1070 $hanging_offset += INDENT_LEVEL if m/\*.*\(/; # '*' followed by '(' - seems consistent with Emacs C mode 1071 } 1072 1073 my $local_in_expr = $in_expr; 1074 my $terminator_position = update_nested_indents($_, $nested_indents_position); 1075 1076 if ($local_in_expr) { 1077 # on end of non-if/while/for/switch (multi-line) expression (i.e., return/enum/assignment) and 1078 # on end of statement/type declaration/variable definition/function header 1079 if ($terminator_position >= 0 && ($in_typedecl == 0 || @nested_indents == 0)) { 1080 check_nested_nonblock_indents("expr"); 1081 $in_expr = 0; 1082 } 1083 } else { 1084 check_nested_nonblock_indents($in_typedecl == 0 ? "stmt" : "decl") if $terminator_position >= 0; 1085 } 1086 1087 # on ';', which terminates the current statement/type declaration/variable definition/function declaration 1088 if ($terminator_position >= 0) { 1089 my $tail = substr($_, $terminator_position + 1); 1090 if (@in_if_hanging_offsets != 0) { 1091 if ($tail =~ m/\s*else(\W|$)/) { 1092 pop @in_if_hanging_offsets; 1093 $hanging_offset -= INDENT_LEVEL; 1094 } elsif ($tail =~ m/[^\s@]/) { # code (not just comment) follows 1095 @in_if_hanging_offsets = (); # note there is nothing like "unclosed 'if'" 1096 $hanging_offset = 0; 1097 } else { 1098 $if_maybe_terminated = 1; 1099 } 1100 } elsif ($tail =~ m/^[\s@]*$/) { # ';' has been trailing, i.e. there is nothing but whitespace and comments 1101 $hanging_offset = 0; # reset in case of terminated assignment ('=') etc. 1102 } 1103 $in_typedecl-- if $in_typedecl != 0 && @nested_in_typedecl == 0; # TODO handle multiple type decls per line 1104 m/(;[^;]*)$/; # match last ';' 1105 $terminator_position = length($_) - length($1) if $1; 1106 # new $terminator_position value may be after the earlier one in case multiple terminators on current line 1107 # TODO check treatment in case of multiple terminators on current line 1108 update_nested_indents($_, $terminator_position + 1); 1109 } 1110 1111 # set hanging expression indent according to nested indents - TODO maybe do better in update_nested_indents() 1112 # also if $in_expr is 0: in statement/type declaration/variable definition/function header 1113 $expr_indent = 0; 1114 for (my $i = -1; $i >= -@nested_symbols; $i--) { 1115 if (@nested_symbols[$i] ne "?") { # conditionals '?' ... ':' are treated specially in check_indent() 1116 $hanging_symbol = @nested_symbols[$i]; 1117 $expr_indent = $nested_indents[$i]; 1118 # $expr_indent is guaranteed to be != 0 unless @nested_indents contains just outer conditionals 1119 last; 1120 } 1121 } 1122 1123 # remember line number and header containing name of last function defined for reports w.r.t. MAX_BODY_LENGTH 1124 if ($in_preproc == 0 && $outermost_level && m/(\w+)\s*\(/ && $1 ne "STACK_OF") { 1125 $line_function_start = $line; 1126 $last_function_header = $contents; 1127 } 1128 1129 # special checks for last, typically trailing opening brace '{' in line 1130 if (my ($head, $tail) = m/^(.*)\{(.*)$/) { # match last ... '{' 1131 if (!$in_expr && $in_typedecl == 0) { 1132 if ($outermost_level) { 1133 if (!$assignment_start && !$local_in_expr) { 1134 # at end of function definition header (or stmt or var definition) 1135 report("'{' not at line start") if length($head) != $preproc_offset && $head =~ m/\)\s*/; # at end of function definition header 1136 $line_body_start = $contents =~ m/LONG BODY/ ? 0 : $line if $line_function_start != 0; 1137 } 1138 } else { 1139 $line_opening_brace = $line if $keyword_opening_brace =~ m/if|do|while|for|(OSSL_)?LIST_FOREACH(_\w+)?/; 1140 # using, not assigning, $keyword_opening_brace here because it could be on an earlier line 1141 $line_opening_brace = $line if $keyword_opening_brace eq "else" && $extended_1_stmt && 1142 # TODO prevent false positives for if/else where braces around single-statement branches 1143 # should be avoided but only if all branches have just single statements 1144 # The following helps detecting the exception when handling multiple 'if ... else' branches: 1145 !($keyword_opening_brace eq "else" && $line_opening_brace < $line_before2); 1146 } 1147 report("code after '{'") if $tail=~ m/[^\s\@]/ && # trailing non-whitespace non-comment (non-'\') 1148 !($tail=~ m/\}/); # missing '}' after last '{' 1149 } 1150 } 1151 1152 # check for opening brace after if/while/for/switch/do missing on same line 1153 # note that "missing '{' on same line after '} else'" is handled further below 1154 if (/^[\s@]*{/ && # leading '{' 1155 $line_before > 0 && !($contents_before_ =~ m/^\s*#/) && # not preprocessor directive '#if 1156 (my ($head, $mid, $tail) = ($contents_before_ =~ m/(^|^.*\W)(if|while|for|(OSSL_)?LIST_FOREACH(_\w+)?|switch|do)(\W.*$|$)/))) { 1157 my $brace_after = $tail =~ /^[\s@]*{/; # any whitespace or comments then '{' 1158 report("'{' not on same line as preceding '$mid'") if !$brace_after; 1159 } 1160 # check for closing brace on line before 'else' not followed by leading '{' 1161 elsif (my ($head, $tail) = m/(^|^.*\W)else(\W.*$|$)/) { 1162 if (parens_balance($tail) == 0 && # avoid false positive due to unfinished expr on current line 1163 !($tail =~ m/{/) && # after 'else' missing '{' on same line 1164 !($head =~ m/}[\s@]*$/) && # not: '}' then any whitespace or comments before 'else' 1165 $line_before > 0 && $contents_before_ =~ /}[\s@]*$/) { # trailing '}' on line before 1166 report("missing '{' on same line after '} else'"); 1167 } 1168 } 1169 1170 # check for closing brace before 'while' not on same line 1171 if (my ($head, $tail) = m/(^|^.*\W)while(\W.*$|$)/) { 1172 my $brace_before = $head =~ m/}[\s@]*$/; # '}' then any whitespace or comments 1173 # possibly 'if (...)' (with potentially inner '(' and ')') then any whitespace or comments then '{' 1174 if (!$brace_before && 1175 # does not work here: @in_do_hanging_offsets != 0 && #'while' terminates loop 1176 parens_balance($tail) == 0 && # avoid false positive due to unfinished expr on current line 1177 $tail =~ /;/ && # 'while' terminates loop (by ';') 1178 $line_before > 0 && 1179 $contents_before_ =~ /}[\s@]*$/) { # on line before: '}' then any whitespace or comments 1180 report("'while' not on same line as preceding '}'"); 1181 } 1182 } 1183 1184 # check for missing brace on same line before or after 'else' 1185 if (my ($head, $tail) = m/(^|^.*\W)else(\W.*$|$)/) { 1186 my $brace_before = $head =~ /}[\s@]*$/; # '}' then any whitespace or comments 1187 my $brace_after = $tail =~ /^[\s@]*if[\s@]*\(.*\)[\s@]*{|[\s@]*{/; 1188 # possibly 'if (...)' (with potentially inner '(' and ')') then any whitespace or comments then '{' 1189 if (!$brace_before) { 1190 if ($line_before > 0 && $contents_before_ =~ /}[\s@]*$/) { 1191 report("'else' not on same line as preceding '}'"); 1192 } elsif (parens_balance($tail) == 0) { # avoid false positive due to unfinished expr on current line 1193 report("missing '}' on same line before 'else ... {'") if $brace_after; 1194 } 1195 } elsif (parens_balance($tail) == 0) { # avoid false positive due to unfinished expr on current line 1196 report("missing '{' on same line after '} else'") if $brace_before && !$brace_after; 1197 } 1198 } 1199 1200 # on begin of multi-line preprocessor directive, adapt indent 1201 if ($in_comment == 0 && $trailing_backslash) { 1202 # trailing '\'typically used in preprocessor directive like '#define' 1203 if ($in_preproc == 1) { # start of multi-line preprocessor directive 1204 # note that backup+reset_indentation_state() has already been called 1205 $in_macro_header = m/^\s*#\s*define(\W|$)?(.*)/ ? 1 + parens_balance($2) : 0; # '#define' is beginning 1206 $preproc_offset = INDENT_LEVEL; 1207 $block_indent = $preproc_offset; 1208 } 1209 $in_preproc += 1; 1210 } 1211 1212 # post-processing at end of line @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ 1213 1214 LINE_FINISHED: 1215 $code_contents_before = $contents if 1216 !m/^\s*#(\s*)(\w+)/ && # not single-line preprocessor directive 1217 $in_comment == 0 && !m/^\s*\*?@/; # not in a multi-line comment nor in an intra-line comment 1218 1219 # on end of (possibly multi-line) preprocessor directive, adapt indent 1220 if ($in_preproc != 0 && !$trailing_backslash) { # no trailing '\' 1221 $in_preproc = 0; 1222 $preproc_offset = 0; 1223 restore_indentation_state(); 1224 } 1225 1226 if ($essentially_blank_line) { 1227 report("leading ".($1 eq "" ? "blank" :"whitespace")." line") if $line == 1 && !$sloppy_SPC; 1228 } else { 1229 if ($line_before > 0) { 1230 my $linediff = $line - $line_before - 1; 1231 report("$linediff blank lines before") if $linediff > 1 && !$sloppy_SPC; 1232 } 1233 $line_before2 = $line_before; 1234 $contents_before2 = $contents_before; 1235 $contents_before_2 = $contents_before_; 1236 $line_before = $line; 1237 $contents_before = $contents; 1238 $contents_before_ = $_; 1239 $count_before = $count; 1240 } 1241 1242 if ($self_test) { # debugging 1243 my $should_report = $contents =~ m/\*@(\d)?/ ? 1 : 0; 1244 $should_report = +$1 if $should_report != 0 && defined $1; 1245 print("$ARGV:$line:$num_reports_line reports on:$contents") 1246 if $num_reports_line != $should_report; 1247 } 1248 $num_reports_line = 0; 1249 1250 # post-processing at end of file @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ 1251 1252 if (eof) { 1253 # check for essentially blank line (which may include a '\') just before EOF 1254 report(($1 eq "\n" ? "blank line" : $2 ne "" ? "'\\'" : "whitespace")." at EOF") 1255 if $contents =~ m/^(\s*(\\?)\s*)$/ && !$sloppy_SPC; 1256 1257 # report unclosed expression-level nesting 1258 check_nested_nonblock_indents("expr at EOF"); # also adapts @nested_block_indents 1259 1260 # sanity-check balance of block-level { ... } via final $block_indent at end of file 1261 report_flexibly($line, +@nested_block_indents." unclosed '{'", "(EOF)\n") if @nested_block_indents != 0; 1262 1263 # sanity-check balance of #if ... #endif via final preprocessor directive indent at end of file 1264 report_flexibly($line, "$preproc_if_nesting unclosed '#if'", "(EOF)\n") if $preproc_if_nesting != 0; 1265 1266 reset_file_state(); 1267 } 1268} 1269 1270# final summary report @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ 1271 1272my $num_other_reports = $num_reports - $num_indent_reports - $num_nesting_issues 1273 - $num_syntax_issues - $num_SPC_reports - $num_length_reports; 1274print "$num_reports ($num_indent_reports indentation, $num_nesting_issues '#if' nesting indent, ". 1275 "$num_syntax_issues syntax, $num_SPC_reports whitespace, $num_length_reports length, $num_other_reports other)". 1276 " issues have been found by $0\n" if $num_reports != 0 && !$self_test; 1277
