VirtualBox

source: vbox/trunk/src/libs/openssl-3.1.5/util/check-format.pl

Last change on this file was 104078, checked in by vboxsync, 2 months ago

openssl-3.1.5: Applied and adjusted our OpenSSL changes to 3.1.4. bugref:10638

File size: 68.4 KB
Line 
1#! /usr/bin/env perl
2#
3# Copyright 2020-2023 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|--sloppy-len] [-l|--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 | --sloppy-len increase accepted max line length from 80 to 84
32# -l | --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 => 80;
97use constant MAX_BODY_LENGTH => 200;
98
99# global variables @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
100
101# command-line options
102my $max_length = MAX_LINE_LENGTH;
103my $sloppy_bodylen = 0;
104my $sloppy_SPC = 0;
105my $sloppy_hang = 0;
106my $sloppy_cmt = 0;
107my $sloppy_macro = 0;
108my $eol_cmt = 0;
109my $extended_1_stmt = 0;
110
111while ($ARGV[0] =~ m/^-(\w|-[\w\-]+)$/) {
112 my $arg = $1; shift;
113 if ($arg =~ m/^(l|-sloppy-len)$/) {
114 $max_length += INDENT_LEVEL;
115 } elsif ($arg =~ m/^(b|-sloppy-bodylen)$/) {
116 $sloppy_bodylen = 1;
117 } elsif ($arg =~ m/^(s|-sloppy-space)$/) {
118 $sloppy_SPC= 1;
119 } elsif ($arg =~ m/^(c|-sloppy-comment)$/) {
120 $sloppy_cmt = 1;
121 } elsif ($arg =~ m/^(m|-sloppy-macro)$/) {
122 $sloppy_macro = 1;
123 } elsif ($arg =~ m/^(h|-sloppy-hang)$/) {
124 $sloppy_hang = 1;
125 } elsif ($arg =~ m/^(e|-eol-comment)$/) {
126 $eol_cmt = 1;
127 } elsif ($arg =~ m/^(1|-1-stmt)$/) {
128 $extended_1_stmt = 1;
129 } else {
130 die("unknown option: -$arg");
131 }
132}
133
134# state variables
135my $self_test; # whether the current input file is regarded to contain (positive/negative) self-tests
136
137my $in_comment; # number of lines so far within multi-line comment, 0 if no comment, < 0 when end is on current line
138my $leading_comment; # multi-line comment has no code before its beginning delimiter, if $in_comment != 0
139my $formatted_comment; # multi-line comment beginning with "/*-", which indicates/allows special formatting, if $in_comment != 0
140my $comment_indent; # comment indent, if $in_comment != 0
141
142my $ifdef__cplusplus; # line before contained '#ifdef __cplusplus' (used in header files)
143my $preproc_if_nesting; # currently required indentation of preprocessor directive according to #if(n)(def)
144my $in_preproc; # 0 or number of lines so far within preprocessor directive, e.g., macro definition
145my $preproc_directive; # name of current preprocessor directive, if $in_preproc != 0
146my $preproc_offset; # offset to $block_indent within multi-line preprocessor directive, else 0
147my $in_macro_header; # number of open parentheses + 1 in (multi-line) header of #define, if $in_preproc != 0
148
149my $line; # current line number
150my $line_before; # number of previous not essentially blank line (containing at most whitespace and '\')
151my $line_before2; # number of not essentially blank line before previous not essentially blank line
152
153# indentation state
154my $contents; # contents of current line (without blinding)
155# $_ # current line, where comments etc. get blinded
156my $code_contents_before; # contents of previous non-comment non-preprocessor-directive line (without blinding), initially ""
157my $contents_before; # contents of $line_before (without blinding), if $line_before > 0
158my $contents_before_; # contents of $line_before after blinding comments etc., if $line_before > 0
159my $contents_before2; # contents of $line_before2 (without blinding), if $line_before2 > 0
160my $contents_before_2; # contents of $line_before2 after blinding comments etc., if $line_before2 > 0
161my $in_multiline_string; # line starts within multi-line string literal
162my $count; # -1 or number of leading whitespace characters (except newline) in current line,
163 # which should be $block_indent + $hanging_offset + $local_offset or $expr_indent
164my $count_before; # number of leading whitespace characters (except line ending chars) in $contents_before
165my $has_label; # current line contains label
166my $local_offset; # current extra indent due to label, switch case/default, or leading closing brace(s)
167my $line_body_start; # number of line where last function body started, or 0
168my $line_function_start; # number of line where last function definition started, used for $line_body_start
169my $last_function_header; # header containing name of last function defined, used if $line_body_start != 0
170my $line_opening_brace; # number of previous line with opening brace after do/while/for, optionally for if/else
171
172my $keyword_opening_brace; # name of previous keyword, used if $line_opening_brace != 0
173my $block_indent; # currently required normal indentation at block/statement level
174my $hanging_offset; # extra indent, which may be nested, for just one hanging statement or expr or typedef
175my @in_do_hanging_offsets; # stack of hanging offsets for nested 'do' ... 'while'
176my @in_if_hanging_offsets; # stack of hanging offsets for nested 'if' (but not its potential 'else' branch)
177my $if_maybe_terminated; # 'if' ends and $hanging_offset should be reset unless the next line starts with 'else'
178my @nested_block_indents; # stack of indentations at block/statement level, needed due to hanging statements
179my @nested_hanging_offsets;# stack of nested $hanging_offset values, in parallel to @nested_block_indents
180my @nested_in_typedecl; # stack of nested $in_typedecl values, partly in parallel to @nested_block_indents
181my @nested_indents; # stack of hanging indents due to parentheses, braces, brackets, or conditionals
182my @nested_symbols; # stack of hanging symbols '(', '{', '[', or '?', in parallel to @nested_indents
183my @nested_conds_indents; # stack of hanging indents due to conditionals ('?' ... ':')
184my $expr_indent; # resulting hanging indent within (multi-line) expressions including type exprs, else 0
185my $hanging_symbol; # character ('(', '{', '[', not: '?') responsible for $expr_indent, if $expr_indent != 0
186my $in_block_decls; # number of local declaration lines after block opening before normal statements, or -1 if no block opening
187my $in_expr; # in expression after if/while/for/switch/return/enum/LHS of assignment
188my $in_paren_expr; # in parenthesized if/while/for condition and switch expression, if $expr_indent != 0
189my $in_typedecl; # nesting level of typedef/struct/union/enum
190
191my $num_reports_line = 0; # number of issues found on current line
192my $num_reports = 0; # total number of issues found
193my $num_indent_reports = 0;# total number of indentation issues found
194my $num_nesting_issues = 0;# total number of preprocessor #if nesting issues found
195my $num_syntax_issues = 0; # total number of syntax issues found during sanity checks
196my $num_SPC_reports = 0; # total number of whitespace issues found
197my $num_length_reports = 0;# total number of line length issues found
198
199sub reset_file_state {
200 $in_comment = 0;
201 $ifdef__cplusplus = 0;
202 $preproc_if_nesting = 0;
203 $in_preproc = 0;
204 $line = 0;
205 $line_before = 0;
206 $line_before2 = 0;
207 reset_indentation_state();
208}
209sub reset_indentation_state {
210 $code_contents_before = "";
211 @nested_block_indents = ();
212 @nested_hanging_offsets = ();
213 @nested_in_typedecl = ();
214 @nested_symbols = ();
215 @nested_indents = ();
216 @nested_conds_indents = ();
217 $expr_indent = 0;
218 $in_block_decls = -1;
219 $in_expr = 0;
220 $in_paren_expr = 0;
221 $hanging_offset = 0;
222 @in_do_hanging_offsets = ();
223 @in_if_hanging_offsets = ();
224 $if_maybe_terminated = 0;
225 $block_indent = 0;
226 $in_multiline_string = 0;
227 $line_body_start = 0;
228 $line_opening_brace = 0;
229 $in_typedecl = 0;
230}
231my $bak_line_before;
232my $bak_line_before2;
233my $bak_code_contents_before;
234my @bak_nested_block_indents;
235my @bak_nested_hanging_offsets;
236my @bak_nested_in_typedecl;
237my @bak_nested_symbols;
238my @bak_nested_indents;
239my @bak_nested_conds_indents;
240my $bak_expr_indent;
241my $bak_in_block_decls;
242my $bak_in_expr;
243my $bak_in_paren_expr;
244my $bak_hanging_offset;
245my @bak_in_do_hanging_offsets;
246my @bak_in_if_hanging_offsets;
247my $bak_if_maybe_terminated;
248my $bak_block_indent;
249my $bak_in_multiline_string;
250my $bak_line_body_start;
251my $bak_line_opening_brace;
252my $bak_in_typedecl;
253sub backup_indentation_state {
254 $bak_code_contents_before = $code_contents_before;
255 @bak_nested_block_indents = @nested_block_indents;
256 @bak_nested_hanging_offsets = @nested_hanging_offsets;
257 @bak_nested_in_typedecl = @nested_in_typedecl;
258 @bak_nested_symbols = @nested_symbols;
259 @bak_nested_indents = @nested_indents;
260 @bak_nested_conds_indents = @nested_conds_indents;
261 $bak_expr_indent = $expr_indent;
262 $bak_in_block_decls = $in_block_decls;
263 $bak_in_expr = $in_expr;
264 $bak_in_paren_expr = $in_paren_expr;
265 $bak_hanging_offset = $hanging_offset;
266 @bak_in_do_hanging_offsets = @in_do_hanging_offsets;
267 @bak_in_if_hanging_offsets = @in_if_hanging_offsets;
268 $bak_if_maybe_terminated = $if_maybe_terminated;
269 $bak_block_indent = $block_indent;
270 $bak_in_multiline_string = $in_multiline_string;
271 $bak_line_body_start = $line_body_start;
272 $bak_line_opening_brace = $line_opening_brace;
273 $bak_in_typedecl = $in_typedecl;
274}
275sub restore_indentation_state {
276 $code_contents_before = $bak_code_contents_before;
277 @nested_block_indents = @bak_nested_block_indents;
278 @nested_hanging_offsets = @bak_nested_hanging_offsets;
279 @nested_in_typedecl = @bak_nested_in_typedecl;
280 @nested_symbols = @bak_nested_symbols;
281 @nested_indents = @bak_nested_indents;
282 @nested_conds_indents = @bak_nested_conds_indents;
283 $expr_indent = $bak_expr_indent;
284 $in_block_decls = $bak_in_block_decls;
285 $in_expr = $bak_in_expr;
286 $in_paren_expr = $bak_in_paren_expr;
287 $hanging_offset = $bak_hanging_offset;
288 @in_do_hanging_offsets = @bak_in_do_hanging_offsets;
289 @in_if_hanging_offsets = @bak_in_if_hanging_offsets;
290 $if_maybe_terminated = $bak_if_maybe_terminated;
291 $block_indent = $bak_block_indent;
292 $in_multiline_string = $bak_in_multiline_string;
293 $line_body_start = $bak_line_body_start;
294 $line_opening_brace = $bak_line_opening_brace;
295 $in_typedecl = $bak_in_typedecl;
296}
297
298# auxiliary submodules @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
299
300sub report_flexibly {
301 my $line = shift;
302 my $msg = shift;
303 my $contents = shift;
304 my $report_SPC = $msg =~ /space|blank/;
305 return if $report_SPC && $sloppy_SPC;
306
307 print "$ARGV:$line:$msg:$contents" unless $self_test;
308 $num_reports_line++;
309 $num_reports++;
310 $num_indent_reports++ if $msg =~ m/:indent /;
311 $num_nesting_issues++ if $msg =~ m/ nesting indent /;
312 $num_syntax_issues++ if $msg =~ m/unclosed|unexpected/;
313 $num_SPC_reports++ if $report_SPC;
314 $num_length_reports++ if $msg =~ m/length/;
315}
316
317sub report {
318 my $msg = shift;
319 report_flexibly($line, $msg, $contents);
320}
321
322sub parens_balance { # count balance of opening parentheses - closing parentheses
323 my $str = shift;
324 return $str =~ tr/\(// - $str =~ tr/\)//;
325}
326
327sub blind_nonspace { # blind non-space text of comment as @, preserving length and spaces
328 # the @ character is used because it cannot occur in normal program code so there is no confusion
329 # comment text is not blinded to whitespace in order to be able to check extra SPC also in comments
330 my $comment_text = shift;
331 $comment_text =~ s/([\.\?\!])\s\s/$1. /g; # in extra SPC checks allow one extra SPC after period '.', '?', or '!' in comments
332 return $comment_text =~ tr/ /@/cr;
333}
334
335# submodule for indentation checking/reporting @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
336
337sub check_indent { # used for lines outside multi-line string literals
338 my $stmt_indent = $block_indent + $hanging_offset + $local_offset;
339 # print "DEBUG: expr_indent $expr_indent; stmt_indent $stmt_indent = block_indent $block_indent + hanging_offset $hanging_offset + local_offset $local_offset\n";
340 $stmt_indent = 0 if $stmt_indent < 0; # TODO maybe give warning/error
341 my $stmt_desc = $contents =~
342 m/^\s*\/\*/ ? "intra-line comment" :
343 $has_label ? "label" :
344 ($hanging_offset != 0 ? "hanging " : "").
345 ($hanging_offset != 0 ? "stmt/expr" : "stmt/decl"); # $in_typedecl is not fully to the point here
346 my ($ref_desc, $ref_indent) = $expr_indent == 0 ? ($stmt_desc, $stmt_indent)
347 : ("hanging '$hanging_symbol'", $expr_indent);
348 my ($alt_desc, $alt_indent) = ("", $ref_indent);
349
350 # allow indent 1 for labels - this cannot happen for leading ':'
351 ($alt_desc, $alt_indent) = ("outermost position", 1) if $expr_indent == 0 && $has_label;
352
353 if (@nested_conds_indents != 0 && substr($_, $count, 1) eq ":") {
354 # leading ':' within stmt/expr/decl - this cannot happen for labels, leading '&&', or leading '||'
355 # allow special indent at level of corresponding "?"
356 ($alt_desc, $alt_indent) = ("leading ':'", @nested_conds_indents[-1]);
357 }
358 # allow extra indent offset leading '&&' or '||' - this cannot happen for leading ":"
359 ($alt_desc, $alt_indent) = ("leading '$1'", $ref_indent + INDENT_LEVEL) if $contents =~ m/^[\s@]*(\&\&|\|\|)/;
360
361 if ($expr_indent < 0) { # implies @nested_symbols != 0 && @nested_symbols[0] eq "{" && @nested_indents[-1] < 0
362 # allow normal stmt indentation level for hanging initializer/enum expressions after trailing '{'
363 # this cannot happen for labels and overrides special treatment of ':', '&&' and '||' for this line
364 ($alt_desc, $alt_indent) = ("lines after '{'", $stmt_indent);
365 # decide depending on current actual indentation, preventing forth and back
366 @nested_indents[-1] = $count == $stmt_indent ? $stmt_indent : -@nested_indents[-1]; # allow $stmt_indent
367 $ref_indent = $expr_indent = @nested_indents[-1];
368 }
369
370 # check consistency of indentation within multi-line comment (i.e., between its first, inner, and last lines)
371 if ($in_comment != 0 && $in_comment != 1) { # in multi-line comment but not on its first line
372 if (!$sloppy_cmt) {
373 if ($in_comment > 0) { # not at its end
374 report("indent = $count != $comment_indent within multi-line comment")
375 if $count != $comment_indent;
376 } else {
377 my $tweak = $in_comment == -2 ? 1 : 0;
378 report("indent = ".($count + $tweak)." != $comment_indent at end of multi-line comment")
379 if $count + $tweak != $comment_indent;
380 }
381 }
382 # do not check indentation of last line of non-leading multi-line comment
383 if ($in_comment < 0 && !$leading_comment) {
384 s/^(\s*)@/$1*/; # blind first '@' as '*' to prevent below delayed check for the line before
385 return;
386 }
387 return if $in_comment > 0; # not on its last line
388 # $comment_indent will be checked by the below checks for end of multi-line comment
389 }
390
391 # else check indentation of entire-line comment or entire-line end of multi-line comment
392 # ... w.r.t. indent of the following line by delayed check for the line before
393 if (($in_comment == 0 || $in_comment == 1) # no comment, intra-line comment, or begin of multi-line comment
394 && $line_before > 0 # there is a line before
395 && $contents_before_ =~ m/^(\s*)@[\s@]*$/) { # line before begins with '@', no code follows (except '\')
396 report_flexibly($line_before, "entire-line comment indent = $count_before != $count (of following line)",
397 $contents_before) if !$sloppy_cmt && $count_before != -1 && $count_before != $count;
398 }
399 # ... but allow normal indentation for the current line, else above check will be done for the line before
400 if (($in_comment == 0 || $in_comment < 0) # (no comment,) intra-line comment or end of multi-line comment
401 && m/^(\s*)@[\s@]*$/) { # line begins with '@', no code follows (except '\')
402 if ($count == $ref_indent) { # indentation is like for (normal) code in this line
403 s/^(\s*)@/$1*/; # blind first '@' as '*' to prevent above delayed check for the line before
404 return;
405 }
406 return if !eof; # defer check of entire-line comment to next line
407 }
408
409 # else check indentation of leading intra-line comment or end of multi-line comment
410 if (m/^(\s*)@/) { # line begins with '@', i.e., any (remaining type of) comment
411 if (!$sloppy_cmt && $count != $ref_indent) {
412 report("intra-line comment indent = $count != $ref_indent") if $in_comment == 0;
413 report("multi-line comment indent = $count != $ref_indent") if $in_comment < 0;
414 }
415 return;
416 }
417
418 if ($sloppy_hang && ($hanging_offset != 0 || $expr_indent != 0)) {
419 # do not report same indentation as on the line before (potentially due to same violations)
420 return if $line_before > 0 && $count == $count_before;
421
422 # do not report indentation at normal indentation level while hanging expression indent would be required
423 return if $expr_indent != 0 && $count == $stmt_indent;
424
425 # do not report if contents have been shifted left of nested expr indent (but not as far as stmt indent)
426 # apparently aligned to the right in order to fit within line length limit
427 return if $stmt_indent < $count && $count < $expr_indent &&
428 length($contents) == MAX_LINE_LENGTH + length("\n");
429 }
430
431 report("indent = $count != $ref_indent for $ref_desc".
432 ($alt_desc eq ""
433 || $alt_indent == $ref_indent # prevent showing alternative that happens to have equal value
434 ? "" : " or $alt_indent for $alt_desc"))
435 if $count != $ref_indent && $count != $alt_indent;
436}
437
438# submodules handling indentation within expressions @@@@@@@@@@@@@@@@@@@@@@@@@@@
439
440sub update_nested_indents { # may reset $in_paren_expr and in this case also resets $in_expr
441 my $str = shift;
442 my $start = shift; # defaults to 0
443 my $terminator_position = -1;
444 for (my $i = $start; $i < length($str); $i++) {
445 my $c;
446 my $curr = substr($str, $i);
447 if ($curr =~ m/^(.*?)([{}()?:;\[\]])(.*)$/) { # match from position $i the first {}()?:;[]
448 $c = $2;
449 } else {
450 last;
451 }
452 my ($head, $tail) = (substr($str, 0, $i).$1, $3);
453 $i += length($1) + length($2) - 1;
454
455 # stop at terminator outside 'for (..;..;..)', assuming that 'for' is followed by '('
456 return $i if $c eq ";" && (!$in_paren_expr || @nested_indents == 0);
457
458 my $in_stmt = $in_expr || @nested_symbols != 0; # not: || $in_typedecl != 0
459 if ($c =~ m/[{([?]/) { # $c is '{', '(', '[', or '?'
460 if ($c eq "{") { # '{' in any context
461 $in_block_decls = 0 if !$in_expr && $in_typedecl == 0;
462 # cancel newly hanging_offset if opening brace '{' is after non-whitespace non-comment:
463 $hanging_offset -= INDENT_LEVEL if $hanging_offset > 0 && $head =~ m/[^\s\@]/;
464 push @nested_block_indents, $block_indent;
465 push @nested_hanging_offsets, $in_expr ? $hanging_offset : 0;
466 push @nested_in_typedecl, $in_typedecl if $in_typedecl != 0;
467 $block_indent += INDENT_LEVEL + $hanging_offset;
468 $hanging_offset = 0;
469 }
470 if ($c ne "{" || $in_stmt) { # for '{' inside stmt/expr (not: decl), for '(', '[', or '?' anywhere
471 $tail =~ m/^([\s@]*)([^\s\@])/;
472 push @nested_indents, defined $2
473 ? $i + 1 + length($1) # actual indentation of following non-space non-comment
474 : $c ne "{" ? +($i + 1) # just after '(' or '[' if only whitespace thereafter
475 : -($i + 1); # allow also $stmt_indent if '{' with only whitespace thereafter
476 push @nested_symbols, $c; # done also for '?' to be able to check correct nesting
477 push @nested_conds_indents, $i if $c eq "?"; # remember special alternative indent for ':'
478 }
479 } elsif ($c =~ m/[})\]:]/) { # $c is '}', ')', ']', or ':'
480 my $opening_c = ($c =~ tr/})]:/{([/r);
481 if (($c ne ":" || $in_stmt # ignore ':' outside stmt/expr/decl
482 # in the presence of ':', one could add this sanity check:
483 # && !(# ':' after initial label/case/default
484 # $head =~ m/^([\s@]*)(case\W.*$|\w+$)/ || # this matching would not work for
485 # # multi-line expr after 'case'
486 # # bitfield length within unsigned type decl
487 # $tail =~ m/^[\s@]*\d+/ # this matching would need improvement
488 # )
489 )) {
490 if ($c ne "}" || $in_stmt) { # for '}' inside stmt/expr/decl, ')', ']', or ':'
491 if (@nested_symbols != 0 &&
492 @nested_symbols[-1] == $opening_c) { # for $c there was a corresponding $opening_c
493 pop @nested_indents;
494 pop @nested_symbols;
495 pop @nested_conds_indents if $opening_c eq "?";
496 } else {
497 report("unexpected '$c' @ ".($in_paren_expr ? "(expr)" : "expr"));
498 next;
499 }
500 }
501 if ($c eq "}") { # '}' at block level but also inside stmt/expr/decl
502 if (@nested_block_indents == 0) {
503 report("unexpected '}'");
504 } else {
505 $block_indent = pop @nested_block_indents;
506 $hanging_offset = pop @nested_hanging_offsets;
507 $in_typedecl = pop @nested_in_typedecl if @nested_in_typedecl != 0;
508 }
509 }
510 if ($in_paren_expr && !grep(/\(/, @nested_symbols)) { # end of (expr)
511 check_nested_nonblock_indents("(expr)");
512 $in_paren_expr = $in_expr = 0;
513 report("code after (expr)")
514 if $tail =~ m/^([^{]*)/ && $1 =~ m/[^\s\@;]/; # non-space non-';' before any '{'
515 }
516 }
517 }
518 }
519 return -1;
520}
521
522sub check_nested_nonblock_indents {
523 my $position = shift;
524 while (@nested_symbols != 0) {
525 my $symbol = pop @nested_symbols;
526 report("unclosed '$symbol' in $position");
527 if ($symbol eq "{") { # repair stack of blocks
528 $block_indent = pop @nested_block_indents;
529 $hanging_offset = pop @nested_hanging_offsets;
530 $in_typedecl = pop @nested_in_typedecl if @nested_in_typedecl != 0;
531 }
532 }
533 @nested_indents = ();
534 @nested_conds_indents = ();
535}
536
537# start of main program @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
538
539reset_file_state();
540
541while (<>) { # loop over all lines of all input files
542 $self_test = $ARGV =~ m/check-format-test/;
543 $_ = "" if $self_test && m/ blank line within local decls /;
544 $line++;
545 s/\r$//; # strip any trailing CR '\r' (which are typical on Windows systems)
546 $contents = $_;
547
548 # check for illegal characters
549 if (m/(.*?)([\x00-\x09\x0B-\x1F\x7F-\xFF])/) {
550 my $col = length($1);
551 report(($2 eq "\x09" ? "TAB" : $2 eq "\x0D" ? "CR " : $2 =~ m/[\x00-\x1F]/ ? "non-printable"
552 : "non-7bit char") . " at column $col") ;
553 }
554
555 # check for whitespace at EOL
556 report("trailing whitespace at EOL") if m/\s\n$/;
557
558 # assign to $count the actual indentation level of the current line
559 chomp; # remove trailing NL '\n'
560 m/^(\s*)/;
561 $count = length($1); # actual indentation
562 $has_label = 0;
563 $local_offset = 0;
564
565 # character/string literals @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
566
567 s/\\["']/@@/g; # blind all '"' and "'" escaped by '\' (typically within character literals or string literals)
568
569 # handle multi-line string literals to avoid confusion on starting/ending '"' and trailing '\'
570 if ($in_multiline_string) {
571 if (s#^([^"]*)"#($1 =~ tr/"/@/cr).'@'#e) { # string literal terminated by '"'
572 # string contents and its terminating '"' have been blinded as '@'
573 $count = -1; # do not check indentation
574 } else {
575 report("multi-line string literal not terminated by '\"' and trailing '\' is missing")
576 unless s#^([^\\]*)\s*\\\s*$#$1#; # strip trailing '\' plus any whitespace around
577 goto LINE_FINISHED;
578 }
579 }
580
581 # blind contents of character and string literals as @, preserving length (but not spaces)
582 # this prevents confusing any of the matching below, e.g., of whitespace and comment delimiters
583 s#('[^']*')#$1 =~ tr/'/@/cr#eg; # handle all intra-line character literals
584 s#("[^"]*")#$1 =~ tr/"/@/cr#eg; # handle all intra-line string literals
585 $in_multiline_string = # handle trailing string literal terminated by '\'
586 s#^(([^"]*"[^"]*")*[^"]*)("[^"]*)\\(\s*)$#$1.($3 =~ tr/"/@/cr).'"'.$4#e;
587 # its contents have been blinded and the trailing '\' replaced by '"'
588
589 # strip any other trailing '\' along with any whitespace around it such that it does not interfere with various matching below
590 my $trailing_backslash = s#^(.*?)\s*\\\s*$#$1#; # trailing '\' possibly preceded or followed by whitespace
591 my $essentially_blank_line = m/^\s*$/; # just whitespace and maybe a '\'
592
593 # comments @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
594
595 # do/prepare checks within multi-line comments
596 my $self_test_exception = $self_test ? "@" : "";
597 if ($in_comment > 0) { # this still includes the last line of multi-line comment
598 my ($head, $any_symbol, $cmt_text) = m/^(\s*)(.?)(.*)$/;
599 if ($any_symbol eq "*") {
600 report("missing space or '*' after leading '*' in multi-line comment") if $cmt_text =~ m|^[^*\s/$self_test_exception]|;
601 } else {
602 report("missing leading '*' in multi-line comment");
603 }
604 $in_comment++;
605 }
606
607 # detect end of comment, must be within multi-line comment, check if it is preceded by non-whitespace text
608 if ((my ($head, $tail) = m|^(.*?)\*/(.*)$|) && $1 ne '/') { # ending comment: '*/'
609 report("missing space or '*' before '*/'") if $head =~ m/[^*\s]$/;
610 report("missing space (or ',', ';', ')', '}', ']') after '*/'") if $tail =~ m/^[^\s,;)}\]]/; # no space or ,;)}] after '*/'
611 if (!($head =~ m|/\*|)) { # not begin of comment '/*', which is is handled below
612 if ($in_comment == 0) {
613 report("unexpected '*/' outside comment");
614 $_ = "$head@@".$tail; # blind the "*/"
615 } else {
616 report("text before '*/' in multi-line comment") if ($head =~ m/[^*\s]/); # non-SPC before '*/'
617 $in_comment = -1; # indicate that multi-line comment ends on current line
618 if ($count > 0) {
619 # make indentation of end of multi-line comment appear like of leading intra-line comment
620 $head =~ s/^(\s*)\s/$1@/; # replace the last leading space by '@'
621 $count--;
622 $in_comment = -2; # indicate that multi-line comment ends on current line, with tweak
623 }
624 my $cmt_text = $head;
625 $_ = blind_nonspace($cmt_text)."@@".$tail;
626 }
627 }
628 }
629
630 # detect begin of comment, check if it is followed by non-space text
631 MATCH_COMMENT:
632 if (my ($head, $opt_minus, $tail) = m|^(.*?)/\*(-?)(.*)$|) { # begin of comment: '/*'
633 report("missing space before '/*'")
634 if $head =~ m/[^\s(\*]$/; # not space, '(', or or '*' (needed to allow '*/') before comment delimiter
635 report("missing space, '*', or '!' after '/*$opt_minus'") if $tail =~ m/^[^\s*!$self_test_exception]/;
636 my $cmt_text = $opt_minus.$tail; # preliminary
637 if ($in_comment > 0) {
638 report("unexpected '/*' inside multi-line comment");
639 } elsif ($tail =~ m|^(.*?)\*/(.*)$|) { # comment end: */ on same line
640 report("unexpected '/*' inside intra-line comment") if $1 =~ /\/\*/;
641 # blind comment text, preserving length and spaces
642 ($cmt_text, my $rest) = ($opt_minus.$1, $2);
643 $_ = "$head@@".blind_nonspace($cmt_text)."@@".$rest;
644 goto MATCH_COMMENT;
645 } else { # begin of multi-line comment
646 my $self_test_exception = $self_test ? "(@\d?)?" : "";
647 report("text after '/*' in multi-line comment")
648 unless $tail =~ m/^$self_test_exception.?[*\s]*$/;
649 # tail not essentially blank, first char already checked
650 # adapt to actual indentation of first line
651 $comment_indent = length($head) + 1;
652 $_ = "$head@@".blind_nonspace($cmt_text);
653 $in_comment = 1;
654 $leading_comment = $head =~ m/^\s*$/; # there is code before beginning delimiter
655 $formatted_comment = $opt_minus eq "-";
656 }
657 } elsif (($head, $tail) = m|^\{-(.*)$|) { # begin of Perl pragma: '{-'
658 }
659
660 if ($in_comment > 1) { # still inside multi-line comment (not at its begin or end)
661 m/^(\s*)\*?(\s*)(.*)$/;
662 $_ = $1."@".$2.blind_nonspace($3);
663 }
664
665 # handle special case of line after '#ifdef __cplusplus' (which typically appears in header files)
666 if ($ifdef__cplusplus) {
667 $ifdef__cplusplus = 0;
668 $_ = "$1 $2" if $contents =~ m/^(\s*extern\s*"C"\s*)\{(\s*)$/; # ignore opening brace in 'extern "C" {'
669 goto LINE_FINISHED if m/^\s*\}\s*$/; # ignore closing brace '}'
670 }
671
672 # check for over-long lines,
673 # while allowing trailing (also multi-line) string literals to go past $max_length
674 my $len = length; # total line length (without trailing '\n')
675 if ($len > $max_length &&
676 !(m/^(.*)"[^"]*"\s*[\)\}\]]*[,;]?\s*$/ # string literal terminated by '"' (or '\'), then maybe )}],;
677 && length($1) < $max_length)
678 # this allows over-long trailing string literals with beginning col before $max_length
679 ) {
680 report("line length = $len > ".MAX_LINE_LENGTH);
681 }
682
683 # handle C++ / C99 - style end-of-line comments
684 if (my ($head, $cmt_text) = m|^(.*?)//(.*$)|) {
685 report("'//' end-of-line comment"); # the '//' comment style is not allowed for C90
686 # blind comment text, preserving length and spaces
687 $_ = "$head@@".blind_nonspace($cmt_text);
688 }
689
690 # at this point all non-space portions of any types of comments have been blinded as @
691
692 goto LINE_FINISHED if $essentially_blank_line;
693
694 # handle preprocessor directives @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
695
696 if (s/^(\s*#)(\s*)(\w+)//) { # line beginning with '#' and directive name;
697 # blank these portions to prevent confusion with C-level 'if', 'else', etc.
698 my ($lead, $space) = ($1, $2);
699 $preproc_directive = $3;
700 $_ = "$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.
701 $_ = blind_nonspace($_) if $preproc_directive eq "error"; # blind error message
702 if ($in_preproc != 0) {
703 report("preprocessor directive within multi-line directive");
704 reset_indentation_state();
705 }
706 $in_preproc++;
707 report("indent = $count != 0 for '#'") if $count != 0;
708 report("'#$preproc_directive' with constant condition")
709 if $preproc_directive =~ m/^(if|elif)$/ && m/^[\W0-9]+$/ && !$trailing_backslash;
710 $preproc_if_nesting-- if $preproc_directive =~ m/^(else|elif|endif)$/;
711 if ($preproc_if_nesting < 0) {
712 $preproc_if_nesting = 0;
713 report("unexpected '#$preproc_directive' according to '#if' nesting");
714 }
715 my $space_count = length($space); # maybe could also use indentation before '#'
716 report("'#if' nesting indent = $space_count != $preproc_if_nesting") if $space_count != $preproc_if_nesting;
717 $preproc_if_nesting++ if $preproc_directive =~ m/^(if|ifdef|ifndef|else|elif)$/;
718 $ifdef__cplusplus = $preproc_directive eq "ifdef" && m/\s+__cplusplus\s*$/;
719
720 # handle indentation of preprocessor directive independently of surrounding normal code
721 $count = -1; # do not check indentation of first line of preprocessor directive
722 backup_indentation_state();
723 reset_indentation_state();
724 }
725
726 # intra-line whitespace nits @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
727
728 my $in_multiline_comment = ($in_comment > 1 || $in_comment < 0); # $in_multiline_comment refers to line before
729 if (!$sloppy_SPC && !($in_multiline_comment && $formatted_comment)) {
730 sub extra_SPC {
731 my $intra_line = shift;
732 return "extra space".($intra_line =~ m/@\s\s/ ?
733 $in_comment != 0 ? " in multi-line comment"
734 : " in intra-line comment" : "");
735 }
736 sub split_line_head { # split line contents into header containing leading spaces and the first non-space char, and the rest of the line
737 my $comment_symbol =
738 $in_comment != 0 ? "@" : ""; # '@' will match the blinded leading '*' in multi-line comment
739 # $in_comment may pertain to the following line due to delayed check
740 # do not check for extra SPC in leading spaces including any '#' (or '*' within multi-line comment)
741 shift =~ m/^(\s*([#$comment_symbol]\s*)?)(.*?)\s*$/;
742 return ($1, $3);
743 }
744 my ($head , $intra_line ) = split_line_head($_);
745 my ($head1, $intra_line1) = split_line_head($contents_before_ ) if $line_before > 0;
746 my ($head2, $intra_line2) = split_line_head($contents_before_2) if $line_before2 > 0;
747 if ($line_before > 0) { # check with one line delay, such that at least $contents_before is available
748 sub column_alignments_only { # return 1 if the given line has multiple consecutive spaces only at columns that match the reference line
749 # all parameter strings are assumed to contain contents after blinding comments etc.
750 my $head = shift; # leading spaces and the first non-space char
751 my $intra = shift; # the rest of the line contents
752 my $contents = shift; # reference line
753 # check if all extra SPC in $intra is used only for multi-line column alignment with $contents
754 my $offset = length($head);
755 for (my $col = 0; $col < length($intra) - 2; $col++) {
756 my $substr = substr($intra, $col);
757 next unless $substr =~ m/^\s\s\S/; # extra SPC (but not in leading spaces of the line)
758 next if !$eol_cmt && $substr =~ m/^[@\s]+$/; # end-of-line comment
759 return 0 unless substr($contents, $col + $offset + 1, 2) =~ m/\s\S/; # reference line contents do not match
760 }
761 return 1;
762 }
763 report_flexibly($line_before, extra_SPC($intra_line1), $contents_before) if $intra_line1 =~ m/\s\s\S/ &&
764 !( column_alignments_only($head1, $intra_line1, $_ ) # compare with $line
765 || ($line_before2 > 0 &&
766 column_alignments_only($head1, $intra_line1, $contents_before_2))); # compare w/ $line_before2
767 report(extra_SPC($intra_line)) if $intra_line =~ m/\s\s\S/ && eof
768 && ! column_alignments_only($head , $intra_line , $contents_before_ ) ; # compare w/ $line_before
769 } elsif (eof) { # special case: just one line exists
770 report(extra_SPC($intra_line)) if $intra_line =~ m/\s\s\S/;
771 }
772 # ignore paths in #include
773 $intra_line =~ s/^(include\s*)(".*?"|<.*?>)/$1/e if $head =~ m/#/;
774 report("missing space before '$2'")
775 if $intra_line =~ m/(\S)((<<|>>)=)/ # '<<=' or >>=' without preceding space
776 || ($intra_line =~ m/(\S)([\+\-\*\/\/%\&\|\^\!<>=]=)/
777 && "$1$2" ne "<<=" && "$1$2" ne ">>=") # other <op>= or (in)equality without preceding space
778 || ($intra_line =~ m/(\S)=/
779 && !($1 =~ m/[\+\-\*\/\/%\&\|\^\!<>=]/)
780 && $intra_line =~ m/(\S)(=)/); # otherwise, '=' without preceding space
781 # treat op= and comparison operators as simple '=', simplifying matching below
782 $intra_line =~ s/(<<|>>|[\+\-\*\/\/%\&\|\^\!<>=])=/=/g;
783 # treat (type) variables within macro, indicated by trailing '\', as 'int' simplifying matching below
784 $intra_line =~ s/[A-Z_]+/int/g if $trailing_backslash;
785 # treat double &&, ||, <<, and >> as single ones, simplifying matching below
786 $intra_line =~ s/(&&|\|\||<<|>>)/substr($1, 0, 1)/eg;
787 # remove blinded comments etc. directly after [{(
788 while ($intra_line =~ s/([\[\{\(])@+\s?/$1/e) {} # /g does not work here
789 # remove blinded comments etc. directly before ,;)}]
790 while ($intra_line =~ s/\s?@+([,;\)\}\]])/$1/e) {} # /g does not work here
791 # treat remaining blinded comments and string literal contents as (single) space during matching below
792 $intra_line =~ s/@+/ /g; # note that extra SPC has already been handled above
793 $intra_line =~ s/\s+$//; # strip any (resulting) space at EOL
794 # replace ';;' or '; ;' by ';' in "for(;;)" and in "for (...)" unless "..." contains just SPC and ';' characters:
795 $intra_line =~ s/((^|\W)for\s*\()([^;]*?)(\s*)(;\s?);(\s*)([^;]*)(\))/
796 "$1$3$4".("$3$4$5$6$7" eq ";" || $3 ne "" || $7 ne "" ? "" : $5).";$6$7$8"/eg;
797 # strip trailing ';' or '; ' in "for (...)" except in "for (;;)" or "for (;; )":
798 $intra_line =~ s/((^|\W)for\s*\()([^;]*(;[^;]*)?)(;\s?)(\))/
799 "$1$3".($3 eq ";" ? $5 : "")."$6"/eg;
800 $intra_line =~ s/(=\s*)\{ /"$1@ "/eg; # do not report {SPC in initializers such as ' = { 0, };'
801 $intra_line =~ s/, \};/, @;/g; # do not report SPC} in initializers such as ' = { 0, };'
802 report("space before '$1'") if $intra_line =~ m/[\w)\]]\s+(\+\+|--)/; # postfix ++/-- with preceding space
803 report("space after '$1'") if $intra_line =~ m/(\+\+|--)\s+[a-zA-Z_(]/; # prefix ++/-- with following space
804 $intra_line =~ s/\.\.\./@/g; # blind '...'
805 report("space before '$1'") if $intra_line =~ m/\s(\.|->)/; # '.' or '->' with preceding space
806 report("space after '$1'") if $intra_line =~ m/(\.|->)\s/; # '.' or '->' with following space
807 $intra_line =~ s/\-\>|\+\+|\-\-/@/g; # blind '->,', '++', and '--'
808 report("space before '$1'") if $intra_line =~ m/[^:)]\s+(;)/; # space before ';' but not after ':' or ')' # note that
809 # exceptions for "for (;; )" are handled above
810 report("space before '$1'") if $intra_line =~ m/\s([,)\]])/; # space before ,)]
811 report("space after '$1'") if $intra_line =~ m/([(\[~!])\s/; # space after ([~!
812 report("space after '$1'") if $intra_line =~ m/(defined)\s/; # space after 'defined'
813 report("missing space before '$1'") if $intra_line =~ m/\S([|\/%<>^\?])/; # |/%<>^? without preceding space
814 # TODO ternary ':' without preceding SPC, while allowing no SPC before ':' after 'case'
815 report("missing space before binary '$2'") if $intra_line =~ m/([^\s{()\[e])([+\-])/; # '+'/'-' without preceding space or {()[e
816 # ')' may be used for type casts or before "->", 'e' may be used for numerical literals such as "1e-6"
817 report("missing space before binary '$1'") if $intra_line =~ m/[^\s{()\[*!]([*])/; # '*' without preceding space or {()[*!
818 report("missing space before binary '$1'") if $intra_line =~ m/[^\s{()\[]([&])/; # '&' without preceding space or {()[
819 report("missing space after ternary '$1'") if $intra_line =~ m/(:)[^\s\d]/; # ':' without following space or digit
820 report("missing space after '$1'") if $intra_line =~ m/([,;=|\/%<>^\?])\S/; # ,;=|/%<>^? without following space
821 report("missing space after binary '$1'") if $intra_line=~m/[^{(\[]([*])[^\sa-zA-Z_(),*]/;# '*' w/o space or \w(),* after
822 # TODO unary '*' must not be followed by SPC
823 report("missing space after binary '$1'") if $intra_line=~m/([&])[^\sa-zA-Z_(]/; # '&' w/o following space or \w(
824 # TODO unary '&' must not be followed by SPC
825 report("missing space after binary '$1'") if $intra_line=~m/[^{(\[]([+\-])[^\s\d(]/; # +/- w/o following space or \d(
826 # TODO unary '+' and '-' must not be followed by SPC
827 report("missing space after '$2'") if $intra_line =~ m/(^|\W)(if|while|for|switch|case)[^\w\s]/; # kw w/o SPC
828 report("missing space after '$2'") if $intra_line =~ m/(^|\W)(return)[^\w\s;]/; # return w/o SPC or ';'
829 report("space after function/macro name")
830 if $intra_line =~ m/(\w+)\s+\(/ # fn/macro name with space before '('
831 && !($1 =~ m/^(sizeof|if|else|while|do|for|switch|case|default|break|continue|goto|return|void|char|signed|unsigned|int|short|long|float|double|typedef|enum|struct|union|auto|extern|static|const|volatile|register)$/) # not keyword
832 && !(m/^\s*#\s*define\s+\w+\s+\(/); # not a macro without parameters having a body that starts with '('
833 report("missing space before '{'") if $intra_line =~ m/[^\s{(\[]\{/; # '{' without preceding space or {([
834 report("missing space after '}'") if $intra_line =~ m/\}[^\s,;\])}]/; # '}' without following space or ,;])}
835 }
836
837 # adapt required indentation @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
838
839 s/(\w*ASN1_[A-Z_]+END\w*([^(]|\(.*?\)|$))/$1;/g; # treat *ASN1_*END*(..) macro calls as if followed by ';'
840
841 my $nested_indents_position = 0;
842
843 # update indents according to leading closing brace(s) '}' or label or switch case
844 my $in_stmt = $in_expr || @nested_symbols != 0 || $in_typedecl != 0;
845 if ($in_stmt) { # expr/stmt/type decl/var def/fn hdr, i.e., not at block level
846 if (m/^([\s@]*\})/) { # leading '}' within stmt, any preceding blinded comment must not be matched
847 $in_block_decls = -1;
848 my $head = $1;
849 update_nested_indents($head);
850 $nested_indents_position = length($head);
851 if (@nested_symbols >= 1) {
852 $hanging_symbol = @nested_symbols[-1];
853 $expr_indent = @nested_indents[-1];
854 } else { # typically end of initialiizer expr or enum
855 $expr_indent = 0;
856 }
857 } elsif (m/^([\s@]*)(static_)?ASN1_ITEM_TEMPLATE_END(\W|$)/) { # workaround for ASN1 macro indented as '}'
858 $local_offset = -INDENT_LEVEL;
859 $expr_indent = 0;
860 } elsif (m/;.*?\}/) { # expr ends with ';' before '}'
861 report("code before '}'");
862 }
863 }
864 if (@in_do_hanging_offsets != 0 && # note there is nothing like "unexpected 'while'"
865 m/^[\s@]*while(\W|$)/) { # leading 'while'
866 $hanging_offset = pop @in_do_hanging_offsets;
867 }
868 if ($if_maybe_terminated) {
869 if (m/(^|\W)else(\W|$)/) { # (not necessarily leading) 'else'
870 if (@in_if_hanging_offsets == 0) {
871 report("unexpected 'else'");
872 } else {
873 $hanging_offset = pop @in_if_hanging_offsets;
874 }
875 } else {
876 @in_if_hanging_offsets = (); # note there is nothing like "unclosed 'if'"
877 $hanging_offset = 0;
878 }
879 }
880 if (!$in_stmt) { # at block level, i.e., outside expr/stmt/type decl/var def/fn hdr
881 $if_maybe_terminated = 0;
882 if (my ($head, $before, $tail) = m/^([\s@]*([^{}]*)\})[\s@]*(.*)$/) { # leading closing '}', but possibly
883 # with non-whitespace non-'{' before
884 report("code after '}'") unless $tail eq "" || $tail =~ m/(else|while|OSSL_TRACE_END)(\W|$)/;
885 my $outermost_level = @nested_block_indents == 1 && @nested_block_indents[0] == 0;
886 if (!$sloppy_bodylen && $outermost_level && $line_body_start != 0) {
887 my $body_len = $line - $line_body_start - 1;
888 report_flexibly($line_function_start, "function body length = $body_len > ".MAX_BODY_LENGTH." lines",
889 $last_function_header) if $body_len > MAX_BODY_LENGTH;
890 $line_body_start = 0;
891 }
892 if ($before ne "") { # non-whitespace non-'{' before '}'
893 report("code before '}'");
894 } else { # leading '}' outside stmt, any preceding blinded comment must not be matched
895 $in_block_decls = -1;
896 $local_offset = $block_indent + $hanging_offset - INDENT_LEVEL;
897 update_nested_indents($head);
898 $nested_indents_position = length($head);
899 $local_offset -= ($block_indent + $hanging_offset);
900 # in effect $local_offset = -INDENT_LEVEL relative to $block_indent + $hanging_offset values before
901 }
902 }
903
904 # handle opening brace '{' after if/else/while/for/switch/do on line before
905 if ($hanging_offset > 0 && m/^[\s@]*{/ && # leading opening '{'
906 $line_before > 0 &&
907 $contents_before_ =~ m/(^|^.*\W)(if|else|while|for|switch|do)(\W.*$|$)/) {
908 $keyword_opening_brace = $1;
909 $hanging_offset -= INDENT_LEVEL; # cancel newly hanging_offset
910 }
911
912 if (m/^[\s@]*(case|default)(\W.*$|$)/) { # leading 'case' or 'default'
913 my $keyword = $1;
914 report("code after $keyword: ") if $2 =~ /:.*[^\s@].*$/;
915 $local_offset = -INDENT_LEVEL;
916 } else {
917 if (m/^([\s@]*)(\w+):/) { # (leading) label, cannot be "default"
918 $local_offset = -INDENT_LEVEL;
919 $has_label = 1;
920 }
921 }
922 }
923
924 # potential adaptations of indent in first line of macro body in multi-line macro definition
925 if ($in_preproc != 0 && $in_macro_header > 0) {
926 if ($in_macro_header > 1) { # still in macro definition header
927 $in_macro_header += parens_balance($_);
928 } else { # begin of macro body
929 $in_macro_header = 0;
930 if ($count == $block_indent - $preproc_offset # body began with same indentation as preceding code
931 && $sloppy_macro) { # workaround for this situation is enabled
932 $block_indent -= $preproc_offset;
933 $preproc_offset = 0;
934 }
935 }
936 }
937
938 # check required indentation @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
939
940 check_indent() if $count >= 0; # not for start of preprocessor directive and not if multi-line string literal is continued
941
942 # check for blank lines within/after local decls @@@@@@@@@@@@@@@@@@@@@@@@@@@
943
944 if ($in_block_decls >= 0 &&
945 $in_comment == 0 && !m/^\s*\*?@/ && # not in a multi-line or intra-line comment
946 !$in_expr && $expr_indent == 0 && $in_typedecl == 0) {
947 my $blank_line_before = $line > 1 && $code_contents_before =~ m/^\s*(\\\s*)?$/;
948 # essentially blank line before: just whitespace and maybe a '\'
949 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
950 || (m/^(\s*(\w+|\[\]|[\*()]))+?\s+[\*\(]*\w+(\s*(\)|\[[^\]]*\]))*\s*[;,=]/ # weak check for decl involving user-defined type
951 && !m/^\s*(\}|sizeof|if|else|while|do|for|switch|case|default|break|continue|goto|return)(\W|$)/)) {
952 $in_block_decls++;
953 report_flexibly($line - 1, "blank line within local decls, before", $contents) if $blank_line_before;
954 } else {
955 report_flexibly($line, "missing blank line after local decls", "\n$contents_before$contents")
956 if $in_block_decls > 0 && !$blank_line_before;
957 $in_block_decls = -1 unless
958 m/^\s*(\\\s*)?$/ # essentially blank line: just whitespace (and maybe a trailing '\')
959 || $in_comment != 0 || m/^\s*\*?@/; # in multi-line comment or an intra-line comment
960 }
961 }
962
963 $in_comment = 0 if $in_comment < 0; # multi-line comment has ended
964
965 # do some further checks @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
966
967 my $outermost_level = $block_indent - $preproc_offset == 0;
968
969 report("more than one stmt") if !m/(^|\W)for(\W.*|$)/ && # no 'for' - TODO improve matching
970 m/;.*;/; # two or more terminators ';', so more than one statement
971
972 # check for code block containing a single line/statement
973 if ($line_before2 > 0 && !$outermost_level && # within function body
974 $in_typedecl == 0 && @nested_indents == 0 && # neither within type declaration nor inside stmt/expr
975 m/^[\s@]*\}/) { # leading closing brace '}', any preceding blinded comment must not be matched
976 # TODO extend detection from single-line to potentially multi-line statement
977 if ($line_opening_brace > 0 &&
978 ($line_opening_brace == $line_before2 ||
979 $line_opening_brace == $line_before)
980 && $contents_before =~ m/;/) { # there is at least one terminator ';', so there is some stmt
981 # TODO do not report cases where a further else branch
982 # follows with a block containing more than one line/statement
983 report_flexibly($line_before, "'$keyword_opening_brace' { 1 stmt }", $contents_before);
984 }
985 }
986
987 report("single-letter name '$2'") if (m/(^|.*\W)([IO])(\W.*|$)/); # single-letter name 'I' or 'O' # maybe re-add 'l'?
988 # constant on LHS of comparison or assignment, e.g., NULL != x or 'a' < c, but not a + 1 == b
989 report("constant on LHS of '$3'")
990 if (m/(['"]|([\+\-\*\/\/%\&\|\^<>]\s*)?\W[0-9]+L?|\WNULL)\s*([\!<>=]=|[<=>])([<>]?)/ &&
991 $2 eq "" && (($3 ne "<" && $3 ne "='" && $3 ne ">") || $4 eq ""));
992
993 # TODO report needless use of parentheses, while
994 # macro parameters should always be in parens (except when passed on), e.g., '#define ID(x) (x)'
995
996 # adapt required indentation for following lines @@@@@@@@@@@@@@@@@@@@@@@@@@@
997
998 # set $in_expr, $in_paren_expr, and $hanging_offset for if/while/for/switch, return/enum, and assignment RHS
999 my $paren_expr_start = 0;
1000 my $return_enum_start = 0;
1001 my $assignment_start = 0;
1002 my $tmp = $_;
1003 $tmp =~ s/[\!<>=]=/@@/g; # blind (in-)equality symbols like '<=' as '@@' to prevent matching them as '=' below
1004 if (m/^((^|.*\W)(if|while|for|switch))(\W.*|$)$/) { # (last) if/for/while/switch
1005 $paren_expr_start = 1;
1006 } elsif (m/^((^|.*\W)(return|enum))(\W.*|$)/ # (last) return/enum
1007 && !$in_expr && @nested_indents == 0 && parens_balance($1) == 0) { # not nested enum
1008 $return_enum_start = 1;
1009 } elsif ($tmp =~ m/^(([^=]*)(=))(.*)$/ # (last) '=', i.e., assignment
1010 && !$in_expr && @nested_indents == 0 && parens_balance($1) == 0) { # not nested assignment
1011 $assignment_start = 1;
1012 }
1013 if ($paren_expr_start || $return_enum_start || $assignment_start)
1014 {
1015 my ($head, $mid, $tail) = ($1, $3, $4);
1016 $keyword_opening_brace = $mid if $mid ne "=";
1017 # to cope with multi-line expressions, do this also if !($tail =~ m/\{/)
1018 push @in_if_hanging_offsets, $hanging_offset if $mid eq "if";
1019
1020 # already handle $head, i.e., anything before expression
1021 update_nested_indents($head, $nested_indents_position);
1022 $nested_indents_position = length($head);
1023 # now can set $in_expr and $in_paren_expr
1024 $in_expr = 1;
1025 $in_paren_expr = 1 if $paren_expr_start;
1026 if ($mid eq "while" && @in_do_hanging_offsets != 0) {
1027 $hanging_offset = pop @in_do_hanging_offsets;
1028 } else {
1029 $hanging_offset += INDENT_LEVEL; # tentatively set hanging_offset, may be canceled by following '{'
1030 }
1031 }
1032
1033 # set $hanging_offset and $keyword_opening_brace for do/else
1034 if (my ($head, $mid, $tail) = m/(^|^.*\W)(else|do)(\W.*|$)$/) { # last else/do, where 'do' is preferred, but not #else
1035 my $code_before = $head =~ m/[^\s\@}]/; # leading non-whitespace non-comment non-'}'
1036 report("code before '$mid'") if $code_before;
1037 report("code after '$mid'" ) if $tail =~ m/[^\s\@{]/# trailing non-whitespace non-comment non-'{' (non-'\')
1038 && !($mid eq "else" && $tail =~ m/[\s@]*if(\W|$)/);
1039 if ($mid eq "do") { # workarounds for code before 'do'
1040 if ($head =~ m/(^|^.*\W)(else)(\W.*$|$)/) { # 'else' ... 'do'
1041 $hanging_offset += INDENT_LEVEL; # tentatively set hanging_offset, may be canceled by following '{'
1042 }
1043 if ($head =~ m/;/) { # terminator ';' ... 'do'
1044 @in_if_hanging_offsets = (); # note there is nothing like "unclosed 'if'"
1045 $hanging_offset = 0;
1046 }
1047 }
1048 push @in_do_hanging_offsets, $hanging_offset if $mid eq "do";
1049 if ($code_before && $mid eq "do") {
1050 $hanging_offset = length($head) - $block_indent;
1051 }
1052 if (!$in_paren_expr) {
1053 $keyword_opening_brace = $mid if $tail =~ m/\{/;
1054 $hanging_offset += INDENT_LEVEL;
1055 }
1056 }
1057
1058 # set $in_typedecl and potentially $hanging_offset for type declaration
1059 if (!$in_expr && @nested_indents == 0 # not in expression
1060 && m/(^|^.*\W)(typedef|enum|struct|union)(\W.*|$)$/
1061 && parens_balance($1) == 0 # not in newly started expression or function arg list
1062 && ($2 eq "typedef" || !($3 =~ m/\s*\w++\s*(.)/ && $1 ne "{")) # 'struct'/'union'/'enum' <name> not followed by '{'
1063 # not needed: && $keyword_opening_brace = $2 if $3 =~ m/\{/;
1064 ) {
1065 $in_typedecl++;
1066 $hanging_offset += INDENT_LEVEL if m/\*.*\(/; # '*' followed by '(' - seems consistent with Emacs C mode
1067 }
1068
1069 my $local_in_expr = $in_expr;
1070 my $terminator_position = update_nested_indents($_, $nested_indents_position);
1071
1072 if ($local_in_expr) {
1073 # on end of non-if/while/for/switch (multi-line) expression (i.e., return/enum/assignment) and
1074 # on end of statement/type declaration/variable definition/function header
1075 if ($terminator_position >= 0 && ($in_typedecl == 0 || @nested_indents == 0)) {
1076 check_nested_nonblock_indents("expr");
1077 $in_expr = 0;
1078 }
1079 } else {
1080 check_nested_nonblock_indents($in_typedecl == 0 ? "stmt" : "decl") if $terminator_position >= 0;
1081 }
1082
1083 # on ';', which terminates the current statement/type declaration/variable definition/function declaration
1084 if ($terminator_position >= 0) {
1085 my $tail = substr($_, $terminator_position + 1);
1086 if (@in_if_hanging_offsets != 0) {
1087 if ($tail =~ m/\s*else(\W|$)/) {
1088 pop @in_if_hanging_offsets;
1089 $hanging_offset -= INDENT_LEVEL;
1090 } elsif ($tail =~ m/[^\s@]/) { # code (not just comment) follows
1091 @in_if_hanging_offsets = (); # note there is nothing like "unclosed 'if'"
1092 $hanging_offset = 0;
1093 } else {
1094 $if_maybe_terminated = 1;
1095 }
1096 } elsif ($tail =~ m/^[\s@]*$/) { # ';' has been trailing, i.e. there is nothing but whitespace and comments
1097 $hanging_offset = 0; # reset in case of terminated assignment ('=') etc.
1098 }
1099 $in_typedecl-- if $in_typedecl != 0 && @nested_in_typedecl == 0; # TODO handle multiple type decls per line
1100 m/(;[^;]*)$/; # match last ';'
1101 $terminator_position = length($_) - length($1) if $1;
1102 # new $terminator_position value may be after the earlier one in case multiple terminators on current line
1103 # TODO check treatment in case of multiple terminators on current line
1104 update_nested_indents($_, $terminator_position + 1);
1105 }
1106
1107 # set hanging expression indent according to nested indents - TODO maybe do better in update_nested_indents()
1108 # also if $in_expr is 0: in statement/type declaration/variable definition/function header
1109 $expr_indent = 0;
1110 for (my $i = -1; $i >= -@nested_symbols; $i--) {
1111 if (@nested_symbols[$i] ne "?") { # conditionals '?' ... ':' are treated specially in check_indent()
1112 $hanging_symbol = @nested_symbols[$i];
1113 $expr_indent = $nested_indents[$i];
1114 # $expr_indent is guaranteed to be != 0 unless @nested_indents contains just outer conditionals
1115 last;
1116 }
1117 }
1118
1119 # remember line number and header containing name of last function defined for reports w.r.t. MAX_BODY_LENGTH
1120 if ($in_preproc == 0 && $outermost_level && m/(\w+)\s*\(/ && $1 ne "STACK_OF") {
1121 $line_function_start = $line;
1122 $last_function_header = $contents;
1123 }
1124
1125 # special checks for last, typically trailing opening brace '{' in line
1126 if (my ($head, $tail) = m/^(.*)\{(.*)$/) { # match last ... '{'
1127 if (!$in_expr && $in_typedecl == 0) {
1128 if ($outermost_level) {
1129 if (!$assignment_start && !$local_in_expr) {
1130 # at end of function definition header (or stmt or var definition)
1131 report("'{' not at line start") if length($head) != $preproc_offset && $head =~ m/\)\s*/; # at end of function definition header
1132 $line_body_start = $contents =~ m/LONG BODY/ ? 0 : $line if $line_function_start != 0;
1133 }
1134 } else {
1135 $line_opening_brace = $line if $keyword_opening_brace =~ m/do|while|for/;
1136 # using, not assigning, $keyword_opening_brace here because it could be on an earlier line
1137 $line_opening_brace = $line if $keyword_opening_brace =~ m/if|else/ && $extended_1_stmt &&
1138 # TODO prevent false positives for if/else where braces around single-statement branches
1139 # should be avoided but only if all branches have just single statements
1140 # The following helps detecting the exception when handling multiple 'if ... else' branches:
1141 !($keyword_opening_brace eq "else" && $line_opening_brace < $line_before2);
1142 }
1143 report("code after '{'") if $tail=~ m/[^\s\@]/ && # trailing non-whitespace non-comment (non-'\')
1144 !($tail=~ m/\}/); # missing '}' after last '{'
1145 }
1146 }
1147
1148 # check for opening brace after if/while/for/switch/do not on same line
1149 # note that "missing '{' on same line after '} else'" is handled further below
1150 if (/^[\s@]*{/ && # leading '{'
1151 $line_before > 0 && !($contents_before_ =~ m/^\s*#/) && # not preprocessor directive '#if
1152 (my ($head, $mid, $tail) = ($contents_before_ =~ m/(^|^.*\W)(if|while|for|switch|do)(\W.*$|$)/))) {
1153 my $brace_after = $tail =~ /^[\s@]*{/; # any whitespace or comments then '{'
1154 report("'{' not on same line as preceding '$mid'") if !$brace_after;
1155 }
1156 # check for closing brace on line before 'else' not followed by leading '{'
1157 elsif (my ($head, $tail) = m/(^|^.*\W)else(\W.*$|$)/) {
1158 if (parens_balance($tail) == 0 && # avoid false positive due to unfinished expr on current line
1159 !($tail =~ m/{/) && # after 'else' missing '{' on same line
1160 !($head =~ m/}[\s@]*$/) && # not: '}' then any whitespace or comments before 'else'
1161 $line_before > 0 && $contents_before_ =~ /}[\s@]*$/) { # trailing '}' on line before
1162 report("missing '{' on same line after '} else'");
1163 }
1164 }
1165
1166 # check for closing brace before 'while' not on same line
1167 if (my ($head, $tail) = m/(^|^.*\W)while(\W.*$|$)/) {
1168 my $brace_before = $head =~ m/}[\s@]*$/; # '}' then any whitespace or comments
1169 # possibly 'if (...)' (with potentially inner '(' and ')') then any whitespace or comments then '{'
1170 if (!$brace_before &&
1171 # does not work here: @in_do_hanging_offsets != 0 && #'while' terminates loop
1172 parens_balance($tail) == 0 && # avoid false positive due to unfinished expr on current line
1173 $tail =~ /;/ && # 'while' terminates loop (by ';')
1174 $line_before > 0 &&
1175 $contents_before_ =~ /}[\s@]*$/) { # on line before: '}' then any whitespace or comments
1176 report("'while' not on same line as preceding '}'");
1177 }
1178 }
1179
1180 # check for missing brace on same line before or after 'else'
1181 if (my ($head, $tail) = m/(^|^.*\W)else(\W.*$|$)/) {
1182 my $brace_before = $head =~ /}[\s@]*$/; # '}' then any whitespace or comments
1183 my $brace_after = $tail =~ /^[\s@]*if[\s@]*\(.*\)[\s@]*{|[\s@]*{/;
1184 # possibly 'if (...)' (with potentially inner '(' and ')') then any whitespace or comments then '{'
1185 if (!$brace_before) {
1186 if ($line_before > 0 && $contents_before_ =~ /}[\s@]*$/) {
1187 report("'else' not on same line as preceding '}'");
1188 } elsif (parens_balance($tail) == 0) { # avoid false positive due to unfinished expr on current line
1189 report("missing '}' on same line before 'else ... {'") if $brace_after;
1190 }
1191 } elsif (parens_balance($tail) == 0) { # avoid false positive due to unfinished expr on current line
1192 report("missing '{' on same line after '} else'") if $brace_before && !$brace_after;
1193 }
1194 }
1195
1196 # on begin of multi-line preprocessor directive, adapt indent
1197 if ($in_comment == 0 && $trailing_backslash) {
1198 # trailing '\'typically used in preprocessor directive like '#define'
1199 if ($in_preproc == 1) { # start of multi-line preprocessor directive
1200 # note that backup+reset_indentation_state() has already been called
1201 $in_macro_header = m/^\s*#\s*define(\W|$)?(.*)/ ? 1 + parens_balance($2) : 0; # '#define' is beginning
1202 $preproc_offset = INDENT_LEVEL;
1203 $block_indent = $preproc_offset;
1204 }
1205 $in_preproc += 1;
1206 }
1207
1208 # post-processing at end of line @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
1209
1210 LINE_FINISHED:
1211 $code_contents_before = $contents if
1212 !m/^\s*#(\s*)(\w+)/ && # not single-line preprocessor directive
1213 $in_comment == 0 && !m/^\s*\*?@/; # not in a multi-line comment nor in an intra-line comment
1214
1215 # on end of (possibly multi-line) preprocessor directive, adapt indent
1216 if ($in_preproc != 0 && !$trailing_backslash) { # no trailing '\'
1217 $in_preproc = 0;
1218 $preproc_offset = 0;
1219 restore_indentation_state();
1220 }
1221
1222 if ($essentially_blank_line) {
1223 report("leading ".($1 eq "" ? "blank" :"whitespace")." line") if $line == 1 && !$sloppy_SPC;
1224 } else {
1225 if ($line_before > 0) {
1226 my $linediff = $line - $line_before - 1;
1227 report("$linediff blank lines before") if $linediff > 1 && !$sloppy_SPC;
1228 }
1229 $line_before2 = $line_before;
1230 $contents_before2 = $contents_before;
1231 $contents_before_2 = $contents_before_;
1232 $line_before = $line;
1233 $contents_before = $contents;
1234 $contents_before_ = $_;
1235 $count_before = $count;
1236 }
1237
1238 if ($self_test) { # debugging
1239 my $should_report = $contents =~ m/\*@(\d)?/ ? 1 : 0;
1240 $should_report = +$1 if $should_report != 0 && defined $1;
1241 print("$ARGV:$line:$num_reports_line reports on:$contents")
1242 if $num_reports_line != $should_report;
1243 }
1244 $num_reports_line = 0;
1245
1246 # post-processing at end of file @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
1247
1248 if (eof) {
1249 # check for essentially blank line (which may include a '\') just before EOF
1250 report(($1 eq "\n" ? "blank line" : $2 ne "" ? "'\\'" : "whitespace")." at EOF")
1251 if $contents =~ m/^(\s*(\\?)\s*)$/ && !$sloppy_SPC;
1252
1253 # report unclosed expression-level nesting
1254 check_nested_nonblock_indents("expr at EOF"); # also adapts @nested_block_indents
1255
1256 # sanity-check balance of block-level { ... } via final $block_indent at end of file
1257 report_flexibly($line, +@nested_block_indents." unclosed '{'", "(EOF)\n") if @nested_block_indents != 0;
1258
1259 # sanity-check balance of #if ... #endif via final preprocessor directive indent at end of file
1260 report_flexibly($line, "$preproc_if_nesting unclosed '#if'", "(EOF)\n") if $preproc_if_nesting != 0;
1261
1262 reset_file_state();
1263 }
1264}
1265
1266# final summary report @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
1267
1268my $num_other_reports = $num_reports - $num_indent_reports - $num_nesting_issues
1269 - $num_syntax_issues - $num_SPC_reports - $num_length_reports;
1270print "$num_reports ($num_indent_reports indentation, $num_nesting_issues '#if' nesting indent, ".
1271 "$num_syntax_issues syntax, $num_SPC_reports whitespace, $num_length_reports length, $num_other_reports other)".
1272 " issues have been found by $0\n" if $num_reports != 0 && !$self_test;
Note: See TracBrowser for help on using the repository browser.

© 2023 Oracle
ContactPrivacy policyTerms of Use