#!/usr/bin/env perl #*************************************************************************** # _ _ ____ _ # Project ___| | | | _ \| | # / __| | | | |_) | | # | (__| |_| | _ <| |___ # \___|\___/|_| \_\_____| # # Copyright (C) 2011 - 2020, Daniel Stenberg, , et al. # # This software is licensed as described in the file COPYING, which # you should have received as part of this distribution. The terms # are also available at https://curl.haxx.se/docs/copyright.html. # # You may opt to use, copy, modify, merge, publish, distribute and/or sell # copies of the Software, and permit persons to whom the Software is # furnished to do so, under the terms of the COPYING file. # # This software is distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY # KIND, either express or implied. # ########################################################################### use strict; use warnings; my $max_column = 79; my $indent = 2; my $warnings = 0; my $swarnings = 0; my $errors = 0; my $serrors = 0; my $suppressed; # skipped problems my $file; my $dir="."; my $wlist=""; my @alist; my $windows_os = $^O eq 'MSWin32' || $^O eq 'cygwin' || $^O eq 'msys'; my $verbose; my %skiplist; my %ignore; my %ignore_set; my %ignore_used; my @ignore_line; my %warnings_extended = ( 'COPYRIGHTYEAR' => 'copyright year incorrect', ); my %warnings = ( 'LONGLINE' => "Line longer than $max_column", 'TABS' => 'TAB characters not allowed', 'TRAILINGSPACE' => 'Trailing white space on the line', 'CPPCOMMENTS' => '// comment detected', 'SPACEBEFOREPAREN' => 'space before an open parenthesis', 'SPACEAFTERPAREN' => 'space after open parenthesis', 'SPACEBEFORECLOSE' => 'space before a close parenthesis', 'SPACEBEFORECOMMA' => 'space before a comma', 'RETURNNOSPACE' => 'return without space', 'COMMANOSPACE' => 'comma without following space', 'BRACEELSE' => '} else on the same line', 'PARENBRACE' => '){ without sufficient space', 'SPACESEMICOLON' => 'space before semicolon', 'BANNEDFUNC' => 'a banned function was used', 'FOPENMODE' => 'fopen needs a macro for the mode string', 'BRACEPOS' => 'wrong position for an open brace', 'INDENTATION' => 'wrong start column for code', 'COPYRIGHT' => 'file missing a copyright statement', 'BADCOMMAND' => 'bad !checksrc! instruction', 'UNUSEDIGNORE' => 'a warning ignore was not used', 'OPENCOMMENT' => 'file ended with a /* comment still "open"', 'ASTERISKSPACE' => 'pointer declared with space after asterisk', 'ASTERISKNOSPACE' => 'pointer declared without space before asterisk', 'ASSIGNWITHINCONDITION' => 'assignment within conditional expression', 'EQUALSNOSPACE' => 'equals sign without following space', 'NOSPACEEQUALS' => 'equals sign without preceding space', 'SEMINOSPACE' => 'semicolon without following space', 'MULTISPACE' => 'multiple spaces used when not suitable', 'SIZEOFNOPAREN' => 'use of sizeof without parentheses', 'SNPRINTF' => 'use of snprintf', 'ONELINECONDITION' => 'conditional block on the same line as the if()', 'TYPEDEFSTRUCT' => 'typedefed struct', ); sub readskiplist { open(W, "<$dir/checksrc.skip") or return; my @all=; for(@all) { $windows_os ? $_ =~ s/\r?\n$// : chomp; $skiplist{$_}=1; } close(W); } # Reads the .checksrc in $dir for any extended warnings to enable locally. # Currently there is no support for disabling warnings from the standard set, # and since that's already handled via !checksrc! commands there is probably # little use to add it. sub readlocalfile { my $i = 0; open(my $rcfile, "<", "$dir/.checksrc") or return; while(<$rcfile>) { $i++; # Lines starting with '#' are considered comments if (/^\s*(#.*)/) { next; } elsif (/^\s*enable ([A-Z]+)$/) { if(!defined($warnings_extended{$1})) { print STDERR "invalid warning specified in .checksrc: \"$1\"\n"; next; } $warnings{$1} = $warnings_extended{$1}; } elsif (/^\s*disable ([A-Z]+)$/) { if(!defined($warnings{$1})) { print STDERR "invalid warning specified in .checksrc: \"$1\"\n"; next; } # Accept-list push @alist, $1; } else { die "Invalid format in $dir/.checksrc on line $i\n"; } } close($rcfile); } sub checkwarn { my ($name, $num, $col, $file, $line, $msg, $error) = @_; my $w=$error?"error":"warning"; my $nowarn=0; #if(!$warnings{$name}) { # print STDERR "Dev! there's no description for $name!\n"; #} # checksrc.skip if($skiplist{$line}) { $nowarn = 1; } # !checksrc! controlled elsif($ignore{$name}) { $ignore{$name}--; $ignore_used{$name}++; $nowarn = 1; if(!$ignore{$name}) { # reached zero, enable again enable_warn($name, $num, $file, $line); } } if($nowarn) { $suppressed++; if($w) { $swarnings++; } else { $serrors++; } return; } if($w) { $warnings++; } else { $errors++; } $col++; print "$file:$num:$col: $w: $msg ($name)\n"; print " $line\n"; if($col < 80) { my $pref = (' ' x $col); print "${pref}^\n"; } } $file = shift @ARGV; while(defined $file) { if($file =~ /-D(.*)/) { $dir = $1; $file = shift @ARGV; next; } elsif($file =~ /-W(.*)/) { $wlist .= " $1 "; $file = shift @ARGV; next; } elsif($file =~ /-A(.+)/) { push @alist, $1; $file = shift @ARGV; next; } elsif($file =~ /-i([1-9])/) { $indent = $1 + 0; $file = shift @ARGV; next; } elsif($file =~ /-m([0-9]+)/) { $max_column = $1 + 0; $file = shift @ARGV; next; } elsif($file =~ /^(-h|--help)/) { undef $file; last; } last; } if(!$file) { print "checksrc.pl [option] [file2] ...\n"; print " Options:\n"; print " -A[rule] Accept this violation, can be used multiple times\n"; print " -D[DIR] Directory to prepend file names\n"; print " -h Show help output\n"; print " -W[file] Skip the given file - ignore all its flaws\n"; print " -i Indent spaces. Default: 2\n"; print " -m Maximum line length. Default: 79\n"; print "\nDetects and warns for these problems:\n"; for(sort keys %warnings) { printf (" %-18s: %s\n", $_, $warnings{$_}); } exit; } readskiplist(); readlocalfile(); do { if("$wlist" !~ / $file /) { my $fullname = $file; $fullname = "$dir/$file" if ($fullname !~ '^\.?\.?/'); scanfile($fullname); } $file = shift @ARGV; } while($file); sub accept_violations { for my $r (@alist) { if(!$warnings{$r}) { print "'$r' is not a warning to accept!\n"; exit; } $ignore{$r}=999999; $ignore_used{$r}=0; } } sub checksrc_clear { undef %ignore; undef %ignore_set; undef @ignore_line; } sub checksrc_endoffile { my ($file) = @_; for(keys %ignore_set) { if($ignore_set{$_} && !$ignore_used{$_}) { checkwarn("UNUSEDIGNORE", $ignore_set{$_}, length($_)+11, $file, $ignore_line[$ignore_set{$_}], "Unused ignore: $_"); } } } sub enable_warn { my ($what, $line, $file, $l) = @_; # switch it back on, but warn if not triggered! if(!$ignore_used{$what}) { checkwarn("UNUSEDIGNORE", $line, length($what) + 11, $file, $l, "No warning was inhibited!"); } $ignore_set{$what}=0; $ignore_used{$what}=0; $ignore{$what}=0; } sub checksrc { my ($cmd, $line, $file, $l) = @_; if($cmd =~ / *([^ ]*) *(.*)/) { my ($enable, $what) = ($1, $2); $what =~ s: *\*/$::; # cut off end of C comment # print "ENABLE $enable WHAT $what\n"; if($enable eq "disable") { my ($warn, $scope)=($1, $2); if($what =~ /([^ ]*) +(.*)/) { ($warn, $scope)=($1, $2); } else { $warn = $what; $scope = 1; } # print "IGNORE $warn for SCOPE $scope\n"; if($scope eq "all") { $scope=999999; } # Comparing for a literal zero rather than the scalar value zero # covers the case where $scope contains the ending '*' from the # comment. If we use a scalar comparison (==) we induce warnings # on non-scalar contents. if($scope eq "0") { checkwarn("BADCOMMAND", $line, 0, $file, $l, "Disable zero not supported, did you mean to enable?"); } elsif($ignore_set{$warn}) { checkwarn("BADCOMMAND", $line, 0, $file, $l, "$warn already disabled from line $ignore_set{$warn}"); } else { $ignore{$warn}=$scope; $ignore_set{$warn}=$line; $ignore_line[$line]=$l; } } elsif($enable eq "enable") { enable_warn($what, $line, $file, $l); } else { checkwarn("BADCOMMAND", $line, 0, $file, $l, "Illegal !checksrc! command"); } } } sub nostrings { my ($str) = @_; $str =~ s/\".*\"//g; return $str; } sub scanfile { my ($file) = @_; my $line = 1; my $prevl=""; my $l; open(R, "<$file") || die "failed to open $file"; my $incomment=0; my @copyright=(); checksrc_clear(); # for file based ignores accept_violations(); while() { $windows_os ? $_ =~ s/\r?\n$// : chomp; my $l = $_; my $ol = $l; # keep the unmodified line for error reporting my $column = 0; # check for !checksrc! commands if($l =~ /\!checksrc\! (.*)/) { my $cmd = $1; checksrc($cmd, $line, $file, $l) } # check for a copyright statement and save the years if($l =~ /\* +copyright .* \d\d\d\d/i) { while($l =~ /([\d]{4})/g) { push @copyright, { year => $1, line => $line, col => index($l, $1), code => $l }; } } # detect long lines if(length($l) > $max_column) { checkwarn("LONGLINE", $line, length($l), $file, $l, "Longer than $max_column columns"); } # detect TAB characters if($l =~ /^(.*)\t/) { checkwarn("TABS", $line, length($1), $file, $l, "Contains TAB character", 1); } # detect trailing white space if($l =~ /^(.*)[ \t]+\z/) { checkwarn("TRAILINGSPACE", $line, length($1), $file, $l, "Trailing whitespace"); } # ------------------------------------------------------------ # Above this marker, the checks were done on lines *including* # comments # ------------------------------------------------------------ # strip off C89 comments comment: if(!$incomment) { if($l =~ s/\/\*.*\*\// /g) { # full /* comments */ were removed! } if($l =~ s/\/\*.*//) { # start of /* comment was removed $incomment = 1; } } else { if($l =~ s/.*\*\///) { # end of comment */ was removed $incomment = 0; goto comment; } else { # still within a comment $l=""; } } # ------------------------------------------------------------ # Below this marker, the checks were done on lines *without* # comments # ------------------------------------------------------------ # crude attempt to detect // comments without too many false # positives if($l =~ /^([^"\*]*)[^:"]\/\//) { checkwarn("CPPCOMMENTS", $line, length($1), $file, $l, "\/\/ comment"); } my $nostr = nostrings($l); # check spaces after for/if/while/function call if($nostr =~ /^(.*)(for|if|while| ([a-zA-Z0-9_]+)) \((.)/) { if($1 =~ / *\#/) { # this is a #if, treat it differently } elsif(defined $3 && $3 eq "return") { # return must have a space } elsif(defined $3 && $3 eq "case") { # case must have a space } elsif($4 eq "*") { # (* beginning makes the space OK! } elsif($1 =~ / *typedef/) { # typedefs can use space-paren } else { checkwarn("SPACEBEFOREPAREN", $line, length($1)+length($2), $file, $l, "$2 with space"); } } if($nostr =~ /^((.*\s)(if) *\()(.*)\)(.*)/) { my $pos = length($1); my $postparen = $5; my $cond = $4; if($cond =~ / = /) { checkwarn("ASSIGNWITHINCONDITION", $line, $pos+1, $file, $l, "assignment within conditional expression"); } my $temp = $cond; $temp =~ s/\(//g; # remove open parens my $openc = length($cond) - length($temp); $temp = $cond; $temp =~ s/\)//g; # remove close parens my $closec = length($cond) - length($temp); my $even = $openc == $closec; if($l =~ / *\#/) { # this is a #if, treat it differently } elsif($even && $postparen && ($postparen !~ /^ *$/) && ($postparen !~ /^ *[,{&|\\]+/)) { print STDERR "5: '$postparen'\n"; checkwarn("ONELINECONDITION", $line, length($l)-length($postparen), $file, $l, "conditional block on the same line"); } } # check spaces after open parentheses if($l =~ /^(.*[a-z])\( /i) { checkwarn("SPACEAFTERPAREN", $line, length($1)+1, $file, $l, "space after open parenthesis"); } # check spaces before close parentheses, unless it was a space or a # close parenthesis! if($l =~ /(.*[^\) ]) \)/) { checkwarn("SPACEBEFORECLOSE", $line, length($1)+1, $file, $l, "space before close parenthesis"); } # check spaces before comma! if($l =~ /(.*[^ ]) ,/) { checkwarn("SPACEBEFORECOMMA", $line, length($1)+1, $file, $l, "space before comma"); } # check for "return(" without space if($l =~ /^(.*)return\(/) { if($1 =~ / *\#/) { # this is a #if, treat it differently } else { checkwarn("RETURNNOSPACE", $line, length($1)+6, $file, $l, "return without space before paren"); } } # check for "sizeof" without parenthesis if(($l =~ /^(.*)sizeof *([ (])/) && ($2 ne "(")) { if($1 =~ / *\#/) { # this is a #if, treat it differently } else { checkwarn("SIZEOFNOPAREN", $line, length($1)+6, $file, $l, "sizeof without parenthesis"); } } # check for comma without space if($l =~ /^(.*),[^ \n]/) { my $pref=$1; my $ign=0; if($pref =~ / *\#/) { # this is a #if, treat it differently $ign=1; } elsif($pref =~ /\/\*/) { # this is a comment $ign=1; } elsif($pref =~ /[\"\']/) { $ign = 1; # There is a quote here, figure out whether the comma is # within a string or '' or not. if($pref =~ /\"/) { # within a string } elsif($pref =~ /\'$/) { # a single letter } else { $ign = 0; } } if(!$ign) { checkwarn("COMMANOSPACE", $line, length($pref)+1, $file, $l, "comma without following space"); } } # check for "} else" if($l =~ /^(.*)\} *else/) { checkwarn("BRACEELSE", $line, length($1), $file, $l, "else after closing brace on same line"); } # check for "){" if($l =~ /^(.*)\)\{/) { checkwarn("PARENBRACE", $line, length($1)+1, $file, $l, "missing space after close paren"); } # check for space before the semicolon last in a line if($l =~ /^(.*[^ ].*) ;$/) { checkwarn("SPACESEMICOLON", $line, length($1), $file, $ol, "space before last semicolon"); } # scan for use of banned functions if($l =~ /^(.*\W) (gmtime|localtime| gets| strtok| v?sprintf| (str|_mbs|_tcs|_wcs)n?cat| LoadLibrary(Ex)?(A|W)?) \s*\( /x) { checkwarn("BANNEDFUNC", $line, length($1), $file, $ol, "use of $2 is banned"); } # scan for use of snprintf for curl-internals reasons if($l =~ /^(.*\W)(v?snprintf)\s*\(/x) { checkwarn("SNPRINTF", $line, length($1), $file, $ol, "use of $2 is banned"); } # scan for use of non-binary fopen without the macro if($l =~ /^(.*\W)fopen\s*\([^,]*, *\"([^"]*)/) { my $mode = $2; if($mode !~ /b/) { checkwarn("FOPENMODE", $line, length($1), $file, $ol, "use of non-binary fopen without FOPEN_* macro: $mode"); } } # check for open brace first on line but not first column # only alert if previous line ended with a close paren and wasn't a cpp # line if((($prevl =~ /\)\z/) && ($prevl !~ /^ *#/)) && ($l =~ /^( +)\{/)) { checkwarn("BRACEPOS", $line, length($1), $file, $ol, "badly placed open brace"); } # if the previous line starts with if/while/for AND ends with an open # brace, or an else statement, check that this line is indented $indent # more steps, if not a cpp line if($prevl =~ /^( *)((if|while|for)\(.*\{|else)\z/) { my $first = length($1); # this line has some character besides spaces if(($l !~ /^ *#/) && ($l =~ /^( *)[^ ]/)) { my $second = length($1); my $expect = $first+$indent; if($expect != $second) { my $diff = $second - $first; checkwarn("INDENTATION", $line, length($1), $file, $ol, "not indented $indent steps (uses $diff)"); } } } # check for 'char * name' if(($l =~ /(^.*(char|int|long|void|CURL|CURLM|CURLMsg|[cC]url_[A-Za-z_]+|struct [a-zA-Z_]+) *(\*+)) (\w+)/) && ($4 !~ /^(const|volatile)$/)) { checkwarn("ASTERISKSPACE", $line, length($1), $file, $ol, "space after declarative asterisk"); } # check for 'char*' if(($l =~ /(^.*(char|int|long|void|curl_slist|CURL|CURLM|CURLMsg|curl_httppost|sockaddr_in|FILE)\*)/)) { checkwarn("ASTERISKNOSPACE", $line, length($1)-1, $file, $ol, "no space before asterisk"); } # check for 'void func() {', but avoid false positives by requiring # both an open and closed parentheses before the open brace if($l =~ /^((\w).*)\{\z/) { my $k = $1; $k =~ s/const *//; $k =~ s/static *//; if($k =~ /\(.*\)/) { checkwarn("BRACEPOS", $line, length($l)-1, $file, $ol, "wrongly placed open brace"); } } # check for equals sign without spaces next to it if($nostr =~ /(.*)\=[a-z0-9]/i) { checkwarn("EQUALSNOSPACE", $line, length($1)+1, $file, $ol, "no space after equals sign"); } # check for equals sign without spaces before it elsif($nostr =~ /(.*)[a-z0-9]\=/i) { checkwarn("NOSPACEEQUALS", $line, length($1)+1, $file, $ol, "no space before equals sign"); } # check for plus signs without spaces next to it if($nostr =~ /(.*)[^+]\+[a-z0-9]/i) { checkwarn("PLUSNOSPACE", $line, length($1)+1, $file, $ol, "no space after plus sign"); } # check for plus sign without spaces before it elsif($nostr =~ /(.*)[a-z0-9]\+[^+]/i) { checkwarn("NOSPACEPLUS", $line, length($1)+1, $file, $ol, "no space before plus sign"); } # check for semicolons without space next to it if($nostr =~ /(.*)\;[a-z0-9]/i) { checkwarn("SEMINOSPACE", $line, length($1)+1, $file, $ol, "no space after semicolon"); } # typedef struct ... { if($nostr =~ /^(.*)typedef struct.*{/) { checkwarn("TYPEDEFSTRUCT", $line, length($1)+1, $file, $ol, "typedef'ed struct"); } # check for more than one consecutive space before open brace or # question mark. Skip lines containing strings since they make it hard # due to artificially getting multiple spaces if(($l eq $nostr) && $nostr =~ /^(.*(\S)) + [{?]/i) { checkwarn("MULTISPACE", $line, length($1)+1, $file, $ol, "multiple space"); print STDERR "L: $l\n"; print STDERR "nostr: $nostr\n"; } $line++; $prevl = $ol; } if(!scalar(@copyright)) { checkwarn("COPYRIGHT", 1, 0, $file, "", "Missing copyright statement", 1); } # COPYRIGHTYEAR is a extended warning so we must first see if it has been # enabled in .checksrc if(defined($warnings{"COPYRIGHTYEAR"})) { # The check for updated copyrightyear is overly complicated in order to # not punish current hacking for past sins. The copyright years are # right now a bit behind, so enforcing copyright year checking on all # files would cause hundreds of errors. Instead we only look at files # which are tracked in the Git repo and edited in the workdir, or # committed locally on the branch without being in upstream master. # # The simple and naive test is to simply check for the current year, # but updating the year even without an edit is against project policy # (and it would fail every file on January 1st). # # A rather more interesting, and correct, check would be to not test # only locally committed files but inspect all files wrt the year of # their last commit. Removing the `git rev-list origin/master..HEAD` # condition below will enfore copyright year checks against the year # the file was last committed (and thus edited to some degree). my $commityear = undef; @copyright = sort {$$b{year} cmp $$a{year}} @copyright; # if the file is modified, assume commit year this year if(`git status -s -- $file` =~ /^ [MARCU]/) { $commityear = (localtime(time))[5] + 1900; } else { # min-parents=1 to ignore wrong initial commit in truncated repos my $grl = `git rev-list --max-count=1 --min-parents=1 --timestamp HEAD -- $file`; if($grl) { chomp $grl; $commityear = (localtime((split(/ /, $grl))[0]))[5] + 1900; } } if(defined($commityear) && scalar(@copyright) && $copyright[0]{year} != $commityear) { checkwarn("COPYRIGHTYEAR", $copyright[0]{line}, $copyright[0]{col}, $file, $copyright[0]{code}, "Copyright year out of date, should be $commityear, " . "is $copyright[0]{year}", 1); } } if($incomment) { checkwarn("OPENCOMMENT", 1, 0, $file, "", "Missing closing comment", 1); } checksrc_endoffile($file); close(R); } if($errors || $warnings || $verbose) { printf "checksrc: %d errors and %d warnings\n", $errors, $warnings; if($suppressed) { printf "checksrc: %d errors and %d warnings suppressed\n", $serrors, $swarnings; } exit 5; # return failure }