ACE/bin/split-cpp.pl

   1 #!/usr/bin/env perl
   2 eval '(exit $?0)' && eval 'exec perl -w -S $0 ${1+"$@"}'
   3     & eval 'exec perl -w -S $0 $argv:q'
   4     if 0;
   5
   6 #
   7 # Splits C++ source files into one file per function or data item.
   8 #
   9 # Author: David L. Levine, with much help and encouragment from
  10 #         Umar Syyid and Gonzalo A. Diethelm.
  11 #         Completed by Andrew Gilpin, July 2000
  12 # Date:   10 November 1998
  13 #
  14 # For each C++ source file:
  15 # 1) Extracts the "intro" code, i.e., #includes and declarations.
  16 # 2) Identifies function definitions, relying on {, and } at the
  17 #    beginning of a line, to delineate the function begin and
  18 #    end.
  19 #
  20 # Assumptions: (applies only to the files being split, i.e. .cpp files)
  21 # * Function definition bodies are terminated with } appearing at
  22 #   the beginning of a line.
  23 # * Free-standing (outside of functions) macro invocations must be
  24 #   followed by a blank line, or terminated with a semicolon.
  25 # * A function must not have a blank line between its header
  26 #   (signature) and its body.
  27 # * There aren't multiple C-style comments on one line, with code
  28 #   between them.
  29 # * typedefs are on a single line
  30 # * A #endif doesn't have a multi-line C comment starting on that line.
  31
  32 # The first three lines above let this script run without specifying the
  33 # full path to perl, as long as it is in the user's PATH.
  34 # Taken from perlrun man page.
  35
  36 # Changes made by Andrew Gilpin (June - July 2000)
  37 # * Added option -c to use .c extension instead of .cpp extension
  38 # * Prints message when no filenames are specified on the command line
  39 # * Changed -? option to -h so that it works properly in most shells
  40 # * Added option -s to skip certain files, but copy them to $split_dir,
  41 #   renaming them. (filename.cpp -> $split_dir/filename_S1.cpp). This is
  42 #   here so that ACE can selectively not split certain files (namely those
  43 #   that this script doesn't work with :)
  44 # * Added support for classes declared in the .cpp file.
  45
  46 $usage="usage: $0 [-h] [-d] [-v] [-c] [-s filename] filenames\n";
  47
  48 #### Configuration parameters.
  49 $verbose = 0;
  50 $debug = 0;
  51 $split_dir = 'SPLIT';
  52 $extension = 'cpp';
  53 @files_to_skip = ();
  54
  55 #### Constants.
  56 $DIR_SEPARATOR = $^O eq "MSWin32"  ?  '\\'  :  '/';
  57
  58
  59 ####
  60 #### Process command line args.
  61 ####
  62 while ( $#ARGV >= $[  &&  $ARGV[0] =~ /^-/ ) {
  63   if ( $ARGV[0] eq '-d' ) {
  64     $debug = 1;
  65   } elsif ( $ARGV[0] eq '-v' ) {
  66     $verbose = 1;
  67   } elsif ( $ARGV[0] eq '-c' ) {
  68     $extension = 'c';
  69   } elsif ( $ARGV[0] eq '-s' ) {
  70     push @files_to_skip, $ARGV[1];
  71     shift;
  72   } elsif ( $ARGV[0] eq '-h' ) {
  73     print "$usage";
  74     exit;
  75   } else {
  76     print STDERR "$0:  unknown option $ARGV[0]\n";
  77     die $usage;
  78   }
  79   shift;
  80 }
  81
  82
  83 &main ();
  84
  85
  86 ####
  87 #### Reset state, to process a new file starting with a clean slate.
  88 ####
  89 sub reset {
  90     #### Working data buffers.
  91   @intro = ();
  92   @current_comments = ();
  93   @current_code = ();
  94   @if = ();
  95   @save_if = ();
  96   @endif = ();
  97   @unknown = ();
  98   ####@unknown_s = ();
  99
 100   #### State variables.
 101   $current_file_number = 0;
 102   $top_of_file = 1;
 103   $in_braces = 0;
 104   $in_nonfunction_code = 0;
 105   $in_C_comment = 0;
 106   $intro_length = 0;
 107   $preprocessor_continuation = 0;
 108   $preserved_ifs = 0;
 109 }
 110
 111
 112 sub main {
 113   #### Print error message if no files are specified.
 114   #### We need to do this before we modify anything on disk.
 115   die "No files specified!\n$usage" if (@ARGV == 0);
 116
 117   #### Remove the destination subdirectory, if it exists.
 118   #### Attempts to clean it out using unlink may fail because
 119   #### it can have many files.
 120   if (-d "$split_dir") {
 121     system ("/bin/rm -r $split_dir") << 256  &&
 122       die "$0: unable to rm \"$split_dir\"\n";
 123   }
 124
 125   #### Create the destination subdirectory.
 126   mkdir "$split_dir", 0755  ||
 127     die "$0: unable to create $split_dir directory: $!\n";
 128
 129   MAIN_LOOP: foreach $file (@ARGV) {
 130     #### Strip off filename extension.
 131     ($basename = $file) =~ s/\.[^\.]+$//;
 132
 133     foreach $skip_file (@files_to_skip) {
 134       if ($skip_file eq $file) {
 135         system ("/bin/cp $file $split_dir/" . $basename. "_S1\.$extension");
 136         next MAIN_LOOP;
 137       }
 138     }
 139
 140     &reset ();
 141
 142     print "FILE: $file\n" if $verbose;
 143     open INPUT, "$file"  ||  die "$0: unable to open \"$file\"\n";
 144
 145     while (<INPUT>) {
 146       #### Strip comments from $line and use that for processing.
 147       #### But, use $_ for output, so that comments will be preserved.
 148       my $line = $_;
 149
 150       #### If we're in the midst of a multiline C comment, see
 151       #### if it's finished on this line.
 152       if ($in_C_comment) {
 153         if ($line =~ s%^.*\*/%%) {
 154           #### End C-style comment.
 155           $in_C_comment = 0;
 156
 157           if ($line =~ /^\s*$/  &&  ! $in_braces) {
 158             #### No code on the line.
 159             #&save_comment ($_);
 160             next;
 161           }
 162         } else {
 163           unless ($in_braces) {
 164             #&save_comment ($_);
 165             next;
 166           }
 167         }
 168       }
 169
 170       #### Strip C++-style comments.
 171       if ($line =~ s%\s*//.*$%%) {
 172         if ($line =~ /^\s*$/  &&  ! $in_braces) {
 173           #### C++-style comment, without any code on the line.
 174           #&save_comment ($_);
 175           next;
 176         }
 177       }
 178
 179       #### And C-style comments.
 180       if ($line =~ m%/\*%) {
 181         #### Begin C-style comment.  Strip any complete comment(s),
 182         #### then see what's left.
 183
 184         $line =~ s%\s*/\*.*\*/\s*%%g;
 185
 186         #### check to see if a preprocessor is on this line
 187         if (! $in_braces) {
 188           if ($line eq '') {
 189           #### The line just had comment(s).  Save it.
 190             #&save_comment ($_);
 191             next;
 192           } else {
 193             #### There's other text on the line.  See if it's just the
 194             #### start of a comment.
 195             if ($line =~ m%/\*%  &&  $line !~ m%\*/%) {
 196               #### The C-style comment isn't terminated on this line.
 197               $in_C_comment = 1;
 198               #&save_comment ($_);
 199               next;
 200             }
 201           }
 202         }
 203       }
 204
 205       #### For now, skip ACE_RCSID's.  Eventually, we might want to
 206       #### consider putting them in _every_ file, if they're enabled.
 207       next if $line =~ /^ACE_RCSID/;
 208
 209       if ($in_braces) {
 210         push @unknown, $_;
 211         if ($line =~ /{/) {
 212           ++$in_braces;
 213         } elsif ($line =~ /^};/) {
 214           #### }; at beginning of line could signify end of class
 215           --$in_braces;
 216           if ($in_braces == 0) {
 217             push @intro, @unknown;
 218             @unknown = ();
 219           }
 220         } elsif ($line =~ /^}/) {
 221           #### } at beginning of line signifies end of function.
 222           --$in_braces;
 223           push @current_code, @unknown;
 224           @unknown = ();
 225           &finish_current ($basename, ++$current_file_number);
 226         } elsif ($line =~ /};/) {
 227           #### end of multi-line data delcaration
 228           --$in_braces;
 229           if ($in_braces == 0) {
 230             push @current_code, @unknown;
 231             @unknown = ();
 232             &finish_current ($basename, ++$current_file_number);
 233           }
 234         }
 235       } else {
 236         #### Not in braces.
 237         if (($line =~ m%[^/]*{%) && (! $preprocessor_continuation)) {
 238           #### { signifies beginning of braces (obviously :).
 239           if ($line =~ /};/) {
 240             #### braces end on this line
 241             push @unknown, $_;
 242             push @current_code, @unknown;
 243             @unknown = ();
 244             &finish_current ($basename, ++$current_file_number);
 245           } else {
 246             push @unknown, $_;
 247             $in_braces = 1;
 248             $in_nonfunction_code = $top_of_file = 0;
 249           }
 250         } elsif ($line =~ /^}/) {
 251           warn "$0: skipping unexpected } on line $. of \"$file\"\n";
 252           next;
 253         } elsif ($line =~ /^typedef/) {
 254           push @intro, $_;
 255         } elsif ($line =~ /^\s*#/  ||  $preprocessor_continuation) {
 256           #### Preprocessor directive.
 257           if ($in_nonfunction_code) {
 258             push @unknown, $_;
 259           } else {
 260             push @intro, $_;
 261           }
 262           $top_of_file = 0;
 263           $preprocessor_continuation = /\\$/  ?  1  :  0;
 264
 265           if ($line =~ m%^\s*#\s*if\s*(.*)(/.*)*$%) {
 266             push @save_if, $_;
 267             unshift @endif, "#endif /* $1 [Added by split-cpp.] */\n";
 268
 269           } elsif ($line =~ /^\s*#\s*endif/) {
 270             #### End an #if/#else block.
 271             unless (defined pop @save_if) {
 272               pop @if;
 273               if ($preserved_ifs > 0) {
 274                 --$preserved_ifs;
 275               }
 276             }
 277             shift @endif;
 278
 279           #### } elsif ($line =~ /^\s*#/) {
 280             #### Any other preprocessor directive.
 281           }
 282
 283         } elsif ($line =~ /^\s*$/) {
 284           #### Whitespace only, or empty line..
 285           push @current_code, "\n";
 286           if ($in_nonfunction_code) {
 287             #### In the midst of non-function code, we reached a
 288             #### blank line.  Assume that we're done with it.
 289             &finish_current ($basename, ++$current_file_number);
 290           } else {
 291             #### Not in a function, so add to intro.  Just in case data or
 292             #### a function follow it, flush now.
 293             $preserved_ifs += $#save_if + 1;
 294             &flush_current (\@intro);
 295           }
 296
 297         } elsif ($line =~ /;/) {
 298           #### Data definition or semicolon-terminated macro invocation.
 299           push @unknown, $_;
 300           $top_of_file = 0;
 301
 302           #### Is it file-static?  Squash newlines out of @current_code.
 303           my $statement = join (' ', @current_code);
 304           if ($statement =~ /([^=[(]+)[=[(](.*)/) {
 305             if ($1 =~ /static/) {
 306               #### Move code to the intro.
 307               push @intro, @current_comments;
 308               @current_comments = ();
 309               &flush_current (\@intro);
 310
 311               #### Not separate code.
 312               $in_nonfunction_code = 0;
 313
 314               #### ???? Extract name from the left side and save for
 315               #### later matching.
 316             } else {
 317               if ($statement =~ /^USEUNIT\s*\(/) {
 318                 #### Special-case those Borland USEUNIT things.
 319                 &flush_current (\@intro);
 320               } else {
 321                 #### Non-static entity, with semicolon.  Wrap it up.
 322                 push @current_code, @unknown;
 323                 @unknown = ();
 324                 &finish_current ($basename, ++$current_file_number);
 325               }
 326             }
 327           } else {
 328             #### Dunno.  Wrap it up, anyways.
 329             push @current_code, @unknown;
 330             @unknown = ();
 331             &finish_current ($basename, ++$current_file_number);
 332           }
 333         } else {
 334           #### Beginning of data definition or function or class.
 335           push @unknown, $_;
 336           $in_nonfunction_code = 1;
 337           $top_of_file = 0;
 338         }
 339       }
 340
 341       if (eof) {
 342         close (ARGV);  #### To reset line number counter.
 343         if ($#intro > $intro_length) {
 344           #### Leftover prepreprocessor statement(s), such as #pragma
 345           #### instantiate.
 346           &finish_current ($basename, ++$current_file_number);
 347         }
 348       }
 349     }
 350
 351     close INPUT;
 352   }
 353 };
 354
 355
 356 ####
 357 #### Save a comment in the appropriate array.
 358 ####
 359 #sub save_comment {
 360 #  my ($comment) = @_;
 361 #
 362 #  if ($top_of_file) {
 363 #    push @intro, $comment;
 364 #  } else {
 365 #    push @current_comments, $comment;
 366 #  }
 367 #}
 368
 369
 370 ####
 371 #### Flush the contents of the @current_code array to the destination
 372 #### argument array.  It is passed by reference.
 373 ####
 374 sub flush_current {
 375   my ($destination) = @_;
 376
 377   push @$destination, @current_code;
 378   @current_code = ();
 379 }
 380
 381
 382 ####
 383 #### Flush what we've got now to an output (split) file.
 384 ####
 385 sub finish_current {
 386   my ($basename, $current_file_number) = @_;
 387
 388   my $current_file_name =
 389     sprintf "$split_dir$DIR_SEPARATOR${basename}_S%d.$extension",
 390       $current_file_number++;
 391
 392   if ($verbose) {
 393     print "CURRENT OUTPUT FILE: $current_file_name\n";
 394     print "INTRO:\n";
 395     print @intro;
 396     print @if;
 397     print @current_comments;
 398     print "CURRENT CODE:\n";
 399     print @current_code;
 400     print @endif;
 401   }
 402
 403   open OUTPUT, "> $current_file_name"  ||
 404     die "unable to open $current_file_name\n";
 405
 406   print OUTPUT "// Automatically generated by ACE's split-cpp.\n" .
 407                "// DO NOT EDIT!\n\n";
 408   if ($debug) {
 409       print OUTPUT "INTRO:\n", @intro, "IF:\n", @if,
 410                    "COMMENTS:\n", @current_comments,
 411                    "CURRENT:\n", @current_code, "ENDIF:\n", @endif;
 412   } else {
 413       print OUTPUT @intro, @if, @current_comments, @current_code, @endif;
 414   }
 415
 416   close OUTPUT;
 417
 418   #### For detection of leftover preprocessor statements and
 419   #### comments at end of file.
 420   $intro_length = $#intro;
 421
 422   @current_comments = @current_code = @save_if = ();
 423   $in_braces = $in_nonfunction_code = 0;
 424 }