Home | History | Annotate | Line # | Download | only in util 
      1 #! /usr/bin/env perl
      2 #
      3 # C source compressor.  This:
      4 #
      5 # - merges continuation lines
      6 # - removes comments (not in strings)
      7 # - removes empty lines (not in strings)
      8 
      9 use strict;
     10 use warnings;
     11 
     12 my $debug = defined $ENV{DEBUG};
     13 my $lang = shift @ARGV;
     14 
     15 # Slurp the file
     16 $/ = undef;
     17 $_ = <>;
     18 
     19 if ($lang eq 'C') {
     20     # Merge continuation lines
     21     s{\\\n}{}g;
     22 
     23     # Regexp for things that should be preserved
     24     my $preserved =
     25         qr{
     26               (?:
     27                   "                 # String start
     28                   (?: \\. | [^\"])* # Any character, including escaped ones
     29                   "                 # String end
     30               )
     31 
     32           |                         # OR
     33 
     34               (?:
     35                   '                 # Character start (multi-chars supported)
     36                   (?: \\. | [^\'])+ # Any character, including escaped ones
     37                   '                 # String end
     38               )
     39         }x;
     40 
     41     # Remove comments while preserving strings
     42     s{
     43          (?|                        # All things preserved end up in $1
     44 
     45              /\*                    # C comment start
     46              .*?                    # Contents up until
     47              \*/                    # C comment end
     48 
     49          |                          # OR
     50 
     51              (                      # Grouping for the replacement
     52                  $preserved
     53              )
     54 
     55          )
     56     }{
     57         if ($debug) {
     58             print STDERR "DEBUG: '$&' => '$1'\n" if defined $1;
     59             print STDERR "DEBUG: '$&' removed\n" unless defined $1;
     60         }
     61         defined $1 ? $1 : ""
     62     }gsxe;
     63 
     64     # Remove empty lines
     65     s{
     66          (?|                        # All things preserved end up in $1
     67 
     68              (^|\n)(?:\s*(?:\n|$))+ # Empty lines, preserve one newline
     69 
     70          |                          # OR
     71 
     72              (                      # Grouping for the replacement
     73                  $preserved
     74              )
     75 
     76          )
     77     }{$1}gsx;
     78 
     79     # Remove extra spaces
     80     s{
     81          (?|                        # All things preserved end up in $1
     82 
     83              \h+                    # Horizontal spaces replaced with one
     84 
     85          |                          # OR
     86 
     87              (                      # Grouping for the replacement
     88                  $preserved
     89              )
     90 
     91          )
     92     }{
     93         if ($debug) {
     94             print STDERR "DEBUG: '$&' => '$1'\n" if defined $1;
     95             print STDERR "DEBUG: '$&' => ' '\n" unless defined $1;
     96         }
     97         defined $1 ? $1 : " "
     98     }gsxe;
     99 
    100     # Clean up spaces at start and end of lines
    101     s/^ //mg;
    102     s/ $//mg;
    103 } elsif ($lang eq 'S') {
    104     # Because we use C++ style comments in our .S files, all we can do
    105     # is to drop them
    106     s{
    107          ^([^\n]*?)//[^\n]*?$   # Any line with a // comment
    108     }{
    109         if ($debug) {
    110             print STDERR "DEBUG: '$&' => '$1'\n" if defined $1;
    111             print STDERR "DEBUG: '$&' removed\n" unless defined $1;
    112         }
    113         defined $1 ? $1 : ""
    114     }mgsxe;
    115 
    116     # Drop all empty lines
    117     s{
    118          (^|\n)(?:\s*(?:\n|$))+ # Empty lines, preserve one newline
    119     }{$1}gsx;
    120 } elsif ($lang eq 'perl') {
    121     # Merge continuation lines
    122     s{\\\n}{}g;
    123 
    124     # Regexp for things that should be preserved
    125     my $preserved =
    126         qr{
    127               (?:
    128                   <<["']?(\w+)["']? # HERE document start
    129                   .*?               # Its contents
    130                   ^\g{-1}$
    131               )
    132           |
    133               (?:
    134                   "                 # Double quoted string start
    135                   (?: \\. | [^\"])* # Any character, including escaped ones
    136                   "                 # Double quoted string end
    137               )
    138 
    139           |                         # OR
    140 
    141               (?:
    142                   '                 # Single quoted string start
    143                   [^\']*            # Any character
    144                   '                 # Single quoted string end
    145               )
    146         }msx;
    147 
    148     # Remove comments while preserving strings
    149     s{
    150          (?|                        # All things preserved end up in $1
    151 
    152              \#.*?(\n|$)            # Perl comments
    153 
    154          |                          # OR
    155 
    156              (                      # Grouping for the replacement
    157                  $preserved
    158              )
    159 
    160          )
    161     }{
    162         if ($debug) {
    163             print STDERR "DEBUG: '$&' => '$1'\n" if defined $1;
    164             print STDERR "DEBUG: '$&' removed\n" unless defined $1;
    165         }
    166         defined $1 ? $1 : ""
    167     }gsxe;
    168 
    169     # Remove empty lines
    170     s{
    171          (?|                        # All things preserved end up in $1
    172 
    173              (^|\n)(?:\s*(?:\n|$))+ # Empty lines, preserve one newline
    174 
    175          |                          # OR
    176 
    177              (                      # Grouping for the replacement
    178                  $preserved
    179              )
    180 
    181          )
    182     }{$1}gsx;
    183 }
    184 
    185 print;
    186