sums

   1 #!/usr/bin/env perl
   2
   3 #=======================================================================
   4 # sums
   5 # File ID: 53bdb770-b77a-11de-90f3-00248cd5cf1e
   6 # Generates smsum hashes
   7 #
   8 # Character set: UTF-8
   9 # ©opyleft 2008– Øyvind A. Holm <sunny@sunbase.org>
  10 # License: GNU General Public License version 2 or later, see end of
  11 # file for legal stuff.
  12 #=======================================================================
  13
  14 use strict;
  15 use warnings;
  16 use Getopt::Long;
  17 use Digest::MD5 qw{ md5_hex };
  18 use Digest::SHA qw{ sha1_hex };
  19
  20 $| = 1;
  21
  22 our $Debug = 0;
  23
  24 our %Opt = (
  25
  26     'debug' => 0,
  27     'help' => 0,
  28     'verbose' => 0,
  29     'version' => 0,
  30     'with-mtime' => 0,
  31
  32 );
  33
  34 our $progname = $0;
  35 $progname =~ s/^.*\/(.*?)$/$1/;
  36 our $VERSION = "0.00";
  37
  38 Getopt::Long::Configure("bundling");
  39 GetOptions(
  40
  41     "debug" => \$Opt{'debug'},
  42     "help|h" => \$Opt{'help'},
  43     "verbose|v+" => \$Opt{'verbose'},
  44     "version" => \$Opt{'version'},
  45     "with-mtime|m" => \$Opt{'with-mtime'},
  46
  47 ) || die("$progname: Option error. Use -h for help.\n");
  48
  49 $Opt{'debug'} && ($Debug = 1);
  50 $Opt{'help'} && usage(0);
  51 if ($Opt{'version'}) {
  52     print_version();
  53     exit(0);
  54 }
  55
  56 while (<>) {
  57     chomp(my $Line = $_);
  58     printf("INSERT INTO sums (string, md5, sha1) VALUES (E'%s', '%s', '%s');\n",
  59     safe_sql($Line), md5_hex($Line), sha1_hex($Line));
  60 }
  61
  62 sub safe_sql {
  63     # {{{
  64     my $Text = shift;
  65     $Text =~ s/\\/\\\\\\\\/g; # I’m not kidding
  66     $Text =~ s/'/''/g;
  67     $Text =~ s/\n/\\n/g;
  68     $Text =~ s/\r/\\r/g;
  69     $Text =~ s/\t/\\t/g;
  70     $Text =~ s/([\x00-\x1f\x7e-\xff])/escape_char($1)/ge;
  71     return($Text);
  72     # }}}
  73 } # safe_sql()
  74
  75 sub escape_char {
  76     my $Char = shift;
  77     return(sprintf("\\\\%03o", ord($Char)));
  78 } # escape_char()
  79
  80 sub process_file {
  81     # {{{
  82     my $Filename = shift;
  83     my $Retval = "";
  84     my %Sum = ();
  85     D("process_file('$Filename')");
  86     my $use_stdin = ($Filename eq "-") ? 1 : 0;
  87     my @stat_array = ();
  88     if ($use_stdin || (@stat_array = stat($Filename))) {
  89         my ($Dev, $Inode, $Mode, $Nlinks, $Uid, $Gid, $Rdev, $Size,
  90             $Atime, $Mtime, $Ctime, $Blksize, $Blocks) = @stat_array;
  91         if ($use_stdin || -f $Filename) {
  92             local *FP;
  93             if ($use_stdin || open(FP, "<", $Filename)) {
  94                 my $sha1 = Digest::SHA->new;
  95                 my $md5 = Digest::MD5->new;
  96                 $use_stdin && (*FP = *STDIN, $Size = 0, $Mtime = time);
  97                 msg(2, sprintf("Reading %s...", safe_tab($Filename)));
  98                 while (my $Curr = <FP>) {
  99                     $sha1->add($Curr);
 100                     $md5->add($Curr);
 101                     $use_stdin && ($Size += length($Curr));
 102                 }
 103                 $Sum{'sha1'} = $sha1->hexdigest;
 104                 $Sum{'md5'} = $md5->hexdigest;
 105                 $Retval =
 106                     $Sum{'sha1'}        . "-" .
 107                     $Sum{'md5'}         . "-" .
 108                     $Size               . "\t" .
 109                     safe_tab($Filename) . (
 110                         $Opt{'with-mtime'}
 111                             ? "\t" . sec_to_string($Mtime)
 112                             : ""
 113                     ) . "\n";
 114             } else {
 115                 warn("$progname: $Filename: Cannot read file\n");
 116             }
 117         } else {
 118             msg(1, "$Filename: Ignoring non-file");
 119         }
 120     } else {
 121         warn("$progname: $Filename: Cannot read file status\n");
 122     }
 123     return($Retval);
 124     # }}}
 125 } # process_file()
 126
 127 sub safe_tab {
 128     # {{{
 129     my $Str = shift;
 130     $Str =~ s/\\/\\\\/gs;
 131     $Str =~ s/\n/\\n/gs;
 132     $Str =~ s/\r/\\r/gs;
 133     $Str =~ s/\t/\\t/gs;
 134     return($Str);
 135     # }}}
 136 } # safe_tab()
 137
 138 sub sec_to_string {
 139     # Convert seconds since 1970 to "yyyy-mm-ddThh:mm:ssZ" {{{
 140     my ($Seconds) = shift;
 141
 142     my @TA = gmtime($Seconds);
 143     my($DateString) = sprintf("%04u-%02u-%02uT%02u:%02u:%02uZ",
 144                               $TA[5]+1900, $TA[4]+1, $TA[3],
 145                               $TA[2], $TA[1], $TA[0]);
 146     return($DateString);
 147     # }}}
 148 } # sec_to_string()
 149
 150 sub print_version {
 151     # Print program version {{{
 152     print("$progname v$VERSION\n");
 153     # }}}
 154 } # print_version()
 155
 156 sub usage {
 157     # Send the help message to stdout {{{
 158     my $Retval = shift;
 159
 160     if ($Opt{'verbose'}) {
 161         print("\n");
 162         print_version();
 163     }
 164     print(<<END);
 165
 166 Usage: $progname [options] [file [files [...]]]
 167
 168 The program is based on the same principle as md5sum(1) and sha1sum(1),
 169 but combines the two hashes and also includes the file size:
 170
 171   [SHA1][-][MD5][-][SIZE][\\t][FILENAME][\\n]
 172
 173 or if the --with-mtime option is used:
 174
 175   [SHA1][-][MD5][-][SIZE][\\t][FILENAME][\\t][MTIME][\\n]
 176
 177 The reason for this approach, is that both hashing algoritms are well
 178 known and widely used. Both algorithms are good enogh for everyday
 179 content verification, but at least the MD5 algorithm is vulnerable to
 180 intentional collisions. Instead of inventing new algorithms which has to
 181 earn trust over the years, combining the two well examined algorithms
 182 and adding the size of the file will make a smsum hash collision much
 183 harder.
 184
 185 If no filenames are specified on the command line, stdin is used.
 186
 187 Special characters in filenames are escaped this way:
 188
 189 Horizontal Tab  (0x09): \\t
 190 Line feed       (0x0a): \\n
 191 Carriage return (0x0d): \\r
 192 Backslash  ('\\', 0x5c): \\\\
 193
 194 Options:
 195
 196   -h, --help
 197     Show this help.
 198   -m, --with-mtime
 199     Also include file modification time at the end of every line. The
 200     date uses the UTC timezone and has the format
 201     "yyyy-mm-ddThh:mm:ssZ". If stdin is read, the current time is used.
 202   -v, --verbose
 203     Increase level of verbosity. Can be repeated.
 204   --version
 205     Print version information.
 206   --debug
 207     Print debugging messages.
 208
 209 END
 210     exit($Retval);
 211     # }}}
 212 } # usage()
 213
 214 sub msg {
 215     # Print a status message to stderr based on verbosity level {{{
 216     my ($verbose_level, $Txt) = @_;
 217
 218     if ($Opt{'verbose'} >= $verbose_level) {
 219         print(STDERR "$progname: $Txt\n");
 220     }
 221     # }}}
 222 } # msg()
 223
 224 sub D {
 225     # Print a debugging message {{{
 226     $Debug || return;
 227     my @call_info = caller;
 228     chomp(my $Txt = shift);
 229     my $File = $call_info[1];
 230     $File =~ s#\\#/#g;
 231     $File =~ s#^.*/(.*?)$#$1#;
 232     print(STDERR "$File:$call_info[2] $$ $Txt\n");
 233     return("");
 234     # }}}
 235 } # D()
 236
 237 __END__
 238
 239 # Plain Old Documentation (POD) {{{
 240
 241 =pod
 242
 243 =head1 NAME
 244
 245
 246
 247 =head1 SYNOPSIS
 248
 249  [options] [file [files [...]]]
 250
 251 =head1 DESCRIPTION
 252
 253
 254
 255 =head1 OPTIONS
 256
 257 =over 4
 258
 259 =item B<-h>, B<--help>
 260
 261 Print a brief help summary.
 262
 263 =item B<-v>, B<--verbose>
 264
 265 Increase level of verbosity. Can be repeated.
 266
 267 =item B<--version>
 268
 269 Print version information.
 270
 271 =item B<--debug>
 272
 273 Print debugging messages.
 274
 275 =back
 276
 277 =head1 BUGS
 278
 279
 280
 281 =head1 AUTHOR
 282
 283 Made by Øyvind A. Holm S<E<lt>sunny@sunbase.orgE<gt>>.
 284
 285 =head1 COPYRIGHT
 286
 287 Copyleft © Øyvind A. Holm E<lt>sunny@sunbase.orgE<gt>
 288 This is free software; see the file F<COPYING> for legalese stuff.
 289
 290 =head1 LICENCE
 291
 292 This program is free software: you can redistribute it and/or modify it
 293 under the terms of the GNU General Public License as published by the
 294 Free Software Foundation, either version 2 of the License, or (at your
 295 option) any later version.
 296
 297 This program is distributed in the hope that it will be useful, but
 298 WITHOUT ANY WARRANTY; without even the implied warranty of
 299 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
 300 See the GNU General Public License for more details.
 301
 302 You should have received a copy of the GNU General Public License along
 303 with this program.
 304 If not, see L<http://www.gnu.org/licenses/>.
 305
 306 =head1 SEE ALSO
 307
 308 =cut
 309
 310 # }}}
 311
 312 # vim: set fenc=UTF-8 ft=perl fdm=marker ts=4 sw=4 sts=4 et fo+=w :