#!/usr/bin/perl

$USAGE = "
  USAGE: $0 options dir
  SYNOPSIS: Find common problem / ambiguities in C programs:
            Recursively scans *.c *.h *.cpp files under dir.
  OPTIONS:
    -?               Help
    -v               Verbose.
    -W[0-9]          Warn level.
    -l               Log warns in c-check.log
    -lLOGFILE        Log warns in LOGFILE
    -f               Search all files (ie. =~ .*).
    -fPATTERN        Search only files matching PATTERN.
    -Ffilelist       Scan only files listed in filelist.
  NOTES:
  . Assignments inside: asserts(x=1), if, while.
  . Backslash followed by trailing spaces.
  . Wrong variables in: for(iap=1;iap<10;jap++){ ... }
  . Some Precedence Ambiguities. eg. (a<<b +c).
  . Semicolon at end of: if( .. );.
  . sizeof( LPDATA ) almost alway means sizeof( DATA ).
  . Possible to ignore c comments.
  . Need find or dir/s/b to get file list.
  EXAMPLE:
  . $0 -f
  AUTHOR: (c) GPL, Mohsin Ahmed, http://www.cs.albany.edu/~mosh
              MohsinA\@Microsoft.Com
";

# ====================================================================


unless( @ARGV ){
    print $USAGE;
    die "c-check: need arg\n";
}

$total_files = 0;       # Count of all files.
$scanned_files = 0;     # count of all files seen.
$warn_files   = 0;      # Count of files having with warnings.

$total_lines = 0;
$total_warns = 0;
$warn_level  = 4;
$filepat = '\.(c|h|cpp|cxx)$' ;

while( $_ = $ARGV[0], /^-/ ){
    shift; last if /^--$/;
    if(     /^-\?/     ){  print $USAGE; exit;
    }elsif( /^-W(\d*)/ ){  $warn_level = $1 || 4;
    }elsif( /^-v(\d*)/ ){  $verbose    = $1 || 1;
    }elsif( /^-l(.*)/  ){  $logfile    = $1 || 'c-check.log';
    }elsif( /^-f(.+)/  ){  $filepat    = $1;
    }elsif( /^-F(.+)/  ){  $filelist   = $1;
    }else{            die "Invalid option: '$_', see -? for help\n";
    }
}

# ====================================================================

$dir = shift || '.';


if( $ENV{ 'windir' } || $ENV{'WINDIR'} ){
    $findcmd = "dir/S/B $dir |";
    $skipfiles = '\b(prn|com|con|aux|lpt)\d?[:.]';
    $skipdirs  = '\b(obj|diff|etc)\b';
}else{
    $findcmd   = "find $dir -name '*' -print |";
}

if( $filelist ){
    $findcmd   = "< $filelist";
}


print STDERR "c-check start=\'$dir\' files=\'$filepat\' -W$warn_level ";
print STDERR "logfile=$logfile " if $logfile;
print STDERR "-v$verbose " if $verbose;
print STDERR "-skipfiles=$skipfiles " if $skipfiles;
print STDERR "\n";

open(FIND,  $findcmd ) || die $findcmd," failed.\n";

if( $logfile ){
    open( LOG, "> $logfile" ) || die "Cannot write log to $logfile\n";
}

while ( $file = <FIND> ){
    chop $file;
    print STDERR "FILE: $file.\n" if $verbose;
    $total_files++;
    next unless $filelist || $file =~ m/$filepat/i;
    next if( $skipfiles && ( $file =~ m/$skipfiles/i ));
    next if( $skipdirs  && ( $file =~ m/$skipdirs/i  ));
    next unless -T $file;

    checkfile( $file );
}

print STDERR "Scanned $scanned_files of $total_files files.\n";
print STDERR "Examined $total_lines lines with $total_warns warnings\n";
print STDERR "In $warn_files files.\n" if $warn_files;

if( $logfile ){
    print STDERR "Wrote $logfile.\n";
    close( LOG );
}

# ====================================================================
# Returns number of lines scanned.

sub checkfile {
    local ($file ) = @_ ;
    if( ! open( SRCFILE, $file ) ){
        print STDERR "Cannot open $file.\n";
        return 0;
    }
    $scanned_files++;
    print STDERR "[$scanned_files] Checking file $file.\n" if $verbose;
    while(<SRCFILE>){
        $warn = 0;
        $error ='';
        s,//.*,, ;
        if(  m/\bassert\b.*[^!<>=]=[^=]/ ){     # assert(i=1)
            $warn++;
            $error = 'assert=';
        }
        if(  m/\b(if|while)\s*
               \(.*[^!<>=]=[^=]\w+[^\w()]\)
              /x                               # if(i=1)
          && ! m/\+\+|--|==/                  # if( c=p++)   ok.
          && ! m/\(.*\w+\s*\(.*\)\)/          # if(c=get(x)) ok.
          && !m/\(.*\)\s*{?}?\n/              # if(c=2){  ok. }
          && !m/return/                       # if(1) return; ok.
           ){
            $warn++;
            $error = 'cond=';
        }

        # To handle do{ .. }\n while( cond );
        if(  m/^\s*(if)\s*\(.*\)\s*;\s*$/  # if( x==1 )
                                             # print x if $verbose; ok.
                                             # {...}while( Cond ); ok.
           && !m/return/                     # if(x) return;
           && ! m/--|\+\+/                   # while(*p++)  ok.
           && !m/\(.*\).*\w+\s*\(.*\)\s*;/   # if(x) abort(x); ok.
           && !m/[)]\s*[(]/                  # if(x) (f=y) ok.
           ){
            $warn++;
            $error = 'cond has no body';
            $print_following_lines = 1;
        }

        if(  m/\\s+$/ ){                       # Backslash space newline.
            $warn++;
            $error = 'backslash spaces';
        }

        # for loop bugs, test: for(iap=1;iap<10;jap++){... }

        if(  ($warn_level > 4) &&
           m/ for \s* \(
            \s* ([^;]*,)* ([\w_]+) \s* = \s* [^;\s]+\s*;
            \s* ([\w_]+) \s* [<>]=* \s* ([^;\s]+) \s*;
            \s* ([\w_]+) [+-]{2,2} \s* \)
          /x
           && ( ($2 ne $3 ) || ($2 ne $5)  ) ){
            $warn++;
            $error = "loop vars($2!=$3 or $2!=$5)";
        }

        # if( m/\bsizeof\(?\s*(lp\w*)/i  && ($badsize = $1) &&
        #     ($badsize !~ m/lpstr/i )   &&
        #     ($badsize !~ m/lpvoid/i ) &&
        #     ($badsize !~ m/lpbyte/i )
        # ){
        #     $warn++;
        #     $error = "sizeof pointer $badsize";
        # }

        # Trivial ambiguities.
        # if(  m/\*\w+\.\w+/ ){                       # *p.f is *(p.f)
        #     $warn++;
        #     $error = '?.';
        # }
        # if(  m/\*\w+\[.*\]/ ){                      # *a[] is *(a[])
        #     $warn++;
        #     $error = '?[]';
        # }
        # if(  m/\*\w+\(.*\)/ ){                      # *f() is *(f())
        #     $warn++;
        #     $error = '?*';
        # }

        if(  m/\w+\s*\&\s*\w+\s*[=!]=/ ){           # a&b ==c is a&(b==c)
            $warn++;
            $error = '?==';
        }
        if(  m/\w+\s*(<<|>>)\s*[+-]\s*\w/ ){     # a<<b +c is a<<(b+c)
            $warn++;
            $error = "?$1";
        }

        if( $warn ){
            $total_warns++;
            unless( $seen{ $file }++ ){
                $warn_files++;
                print     "[$warn_files] File: $file \n";
                print LOG "[$warn_files] === File: $file =====\n"
                    if $logfile;
            }
            print     "* $.:$error:\n", "\t $_" ;
            print LOG "* $.:$error:\n", "\t $_" if $logfile;

        }elsif( $print_following_lines > 0 ){
            if( m/\S/ ){
                print     "\t $_";
                print LOG "\t $_" if $logfile;
                --$print_following_lines;
            }
        }

        if( eof ){
            print STDERR "[$scanned_files] file: $file had $. lines.\n"
                if $verbose;
            $total_lines += $.;
            close(SRCFILE);
            return $.;
        }
    }
}

# ====================================================================