src/install/acquire/captivemodid-print.pl

   1 #! /usr/bin/perl
   2 #
   3 # $Id$
   4 # Scan downloaded Microsoft files to build the .captivemodid.xml file core.
   5 # Copyright (C) 2005 Jan Kratochvil <project-captive@jankratochvil.net>
   6 #
   7 # This program is free software; you can redistribute it and/or modify
   8 # it under the terms of the GNU General Public License as published by
   9 # the Free Software Foundation; exactly version 2 of June 1991 is required
  10 #
  11 # This program is distributed in the hope that it will be useful,
  12 # but WITHOUT ANY WARRANTY; without even the implied warranty of
  13 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  14 # GNU General Public License for more details.
  15 #
  16 # You should have received a copy of the GNU General Public License
  17 # along with this program; if not, write to the Free Software
  18 # Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
  19
  20
  21 use strict;
  22 use warnings;
  23
  24 use Cwd;
  25 use Carp qw(confess cluck);
  26 use Digest::MD5 qw(md5_hex);
  27 use File::Remove qw(remove);
  28 use Getopt::Long;
  29 use Data::Dumper;
  30
  31
  32 my $tmp="/tmp/captivemodid";
  33 my $out;
  34
  35
  36 my $opt_quiet;
  37 die if !GetOptions(
  38                 "q|quiet+",\$opt_quiet,
  39                 );
  40
  41
  42 our $dirnew=$ARGV[0];
  43 do { $dirnew="."; warn "Defaulting ARGV[0] to: $dirnew"; } if !$dirnew;
  44 our $out=$ARGV[1];
  45 warn "No output md5 directory, not copying." if !$out;
  46
  47
  48 my %type=(
  49         "ntoskrnl.exe"=>"Kernel",
  50         "ntkrnlpa.exe"=>"Kernel PA",
  51         "ntkrnlmp.exe"=>"Kernel SMP",
  52         "ntkrpamp.exe"=>"Kernel SMP PA",
  53         "cdfs.sys"    =>"CD-ROM/iso-9660",
  54         "fastfat.sys" =>"FastFAT/vfat",
  55         "ntfs.sys"    =>"NTFS",
  56         );
  57
  58 my %lang=(
  59         "ar"=>"Arabic",
  60         "hk"=>"Chinese (Hong Kong SAR)",
  61         "cn"=>"Chinese (Simplified)",
  62         "tw"=>"Chinese (Traditional)",
  63         "cs"=>"Czech",
  64         "da"=>"Danish",
  65         "nl"=>"Dutch",
  66         "en"=>"English",
  67         "fi"=>"Finnish",
  68         "fr"=>"French",
  69         "de"=>"German",
  70         "el"=>"Greek",
  71         "he"=>"Hebrew",
  72         "hu"=>"Hungarian",
  73         "it"=>"Italian",
  74         "ja"=>"Japanese",
  75         "ko"=>"Korean",
  76         "no"=>"Norwegian",
  77         "pl"=>"Polish",
  78         "br"=>"Portugese (Brazil)",
  79         "pt"=>"Portugese (Portugal)",
  80         "ru"=>"Russian",
  81         "es"=>"Spanish",
  82         "sv"=>"Swedish",
  83         "tr"=>"Turkish",
  84         "ara"=>"Arabic",
  85         "chh"=>"Chinese (Hong Kong SAR)",
  86         "chs"=>"Chinese (Simplified)",
  87         "cht"=>"Chinese (Traditional)",
  88         "csy"=>"Czech",
  89         "dan"=>"Danish",
  90         "nld"=>"Dutch",
  91         "enu"=>"English",
  92         "fin"=>"Finnish",
  93         "fra"=>"French",
  94         "deu"=>"German",
  95         "ell"=>"Greek",
  96         "heb"=>"Hebrew",
  97         "hun"=>"Hungarian",
  98         "ita"=>"Italian",
  99         "jpn"=>"Japanese",
 100         "kor"=>"Korean",
 101         "nor"=>"Norwegian",
 102         "plk"=>"Polish",
 103         "ptb"=>"Portugese (Brazil)",
 104         "ptg"=>"Portugese (Portugal)",
 105         "rus"=>"Russian",
 106         "esn"=>"Spanish",
 107         "sve"=>"Swedish",
 108         "trk"=>"Turkish",
 109         );
 110
 111
 112 sub name_to_lang($)
 113 {
 114 my($name)=@_;
 115
 116         $name=~tr/-/_/;
 117         $name=lc $name;
 118         my $r;
 119         while (my($subst,$long)=each(%lang)) {
 120                 next if $name!~/[^a-z]$subst[^a-z]/;
 121 # FIXME
 122                 cluck "$name: $subst" if $r;
 123                 $r=$long;
 124                 }
 125 # FIXME
 126         cluck $name if !$r;
 127         return $r;
 128 }
 129
 130 my $ver_prefix="5.1.2600.";
 131 my @stack_out;
 132 my $STDERR_needs_eol;
 133 my %md5sum_printed_global;
 134
 135 sub stack0_flush()
 136 {
 137         return if $opt_quiet;
 138         # FIXME: Proper nesting, not just these 2 levels:
 139         my %md5sum_printed_local;
 140         while (my $stack=shift @stack_out) {
 141                 for my $stacki (0..$#$stack) {
 142                         my $this=$stack->[$stacki];
 143                         next if $md5sum_printed_local{$this->{"md5"}}++;
 144                         my $dupe_global=$md5sum_printed_global{$this->{"md5"}}++;
 145                         my $print="";
 146                         $print.=<<"HERE";
 147 <!-- @{[ $this->{"relname"} ]} -->
 148 HERE
 149                         $print.=<<"HERE" if !$dupe_global;
 150 <module type="@{[ $this->{"type"} ]}" length="@{[ $this->{"length"} ]}" priority="@{[ $this->{"priority"} ]}" md5="@{[ $this->{"md5"} ]}"
 151                 id="MS-Windows XP @{[ $this->{"dist"} ]} @{[ $this->{"lang"} ]} $ver_prefix@{[ $this->{"ver"} ]} @{[ $this->{"this_name"} ]}" />
 152 HERE
 153                         $print=~s/^/"\t" x (1+$stacki)/egm;
 154                         print STDERR "\n" if $STDERR_needs_eol;
 155                         $STDERR_needs_eol=0;
 156                         print $print;
 157                         }
 158                 }
 159         print "\n";
 160 }
 161
 162 our @stack;
 163 my %md5sum_stored;
 164 my $last_stack0_filename;
 165
 166 sub check($$)
 167 {
 168 my($filename_unused,$basename_orig)=@_;
 169
 170         return if !$type{lc $basename_orig};
 171         my $final_type=$basename_orig;
 172         my $final_name=$type{$final_type} or confess;
 173         $final_type="ntoskrnl.exe" if $final_type=~/^nt.*[.]exe$/;
 174         (my $basename_orig_=$basename_orig)=~s/.$/_/;
 175         (my $basename0=$stack[0]->{"filename"})=~s{^.*/}{} or confess;
 176         stack0_flush() if $last_stack0_filename && $last_stack0_filename ne $stack[0]->{"filename"};
 177         $last_stack0_filename=$stack[0]->{"filename"};
 178         for my $stacki (0..$#stack) {
 179                 my $this=$stack[$stacki];
 180                 my $filename=$this->{"filename"} or confess;
 181                 my $relname=$this->{"relname"} or confess;
 182                 (my $basename=$filename)=~s{^.*/}{} or confess;
 183                 my $expanded=$stack[$#stack]->{"filename"};
 184                 my $this_name;
 185                 my $type;
 186                 if ($stacki==$#stack-1 && $basename eq $basename_orig_) {
 187                         $this->{"type"} eq "cabinet" or confess;
 188                         $type=$this->{"type"} or confess;
 189                         $this_name="$final_name Cabinet";
 190                         }
 191                 elsif ($stacki<$#stack) {
 192                         $type=$this->{"type"} or confess;
 193                         $this_name="Cabinet";
 194                         }
 195                 else {
 196                         $type=$final_type or confess;
 197                         $this_name=$final_name;
 198                         }
 199                 my $length=(stat $filename)[7] or confess;
 200                 my $dist=$stack[0]->{"filename"};
 201                 $dist=~s{^.*/}{} or confess;
 202                 my $lang=name_to_lang $basename0;
 203                 local *F;
 204                 open F,$expanded or confess;
 205                 my $buf="";
 206                 my $buf2;
 207                 my $ver;
 208                 while (my $got=read F,$buf2,0x1000) {
 209                         confess $expanded if !defined $got;
 210                         last if !$got;
 211                         confess $expanded if $got<0;
 212                         confess $expanded if length($buf2)!=$got;
 213                         $buf.=$buf2;
 214                         $ver=$buf;
 215                         $ver=~s/\x00\x00+/!/g;
 216                         $ver=~tr/\x00//d;
 217                         $ver=~tr/!/\x00/;
 218                         last if $ver=~s/^.*\x00\Q$ver_prefix\E(\d+)[ \x00].*$/$1/s;
 219                         $ver=undef();
 220                         $buf=substr($buf,-0x100);
 221                         }
 222                 confess $expanded if !$ver;
 223                 close F or confess;
 224                 $ver=~/^\d+$/ or confess "$expanded: $ver";
 225                 my $pri=sprintf "510%04u00",$ver;
 226                 $pri+=90;
 227                 $pri-=10 if $lang ne "English";
 228                 $pri-=20 if $basename_orig=~/^ntkrnlpa/;
 229                 $pri-=40 if $basename_orig=~/^ntkrnlmp/;
 230                 $pri-=60 if $basename_orig=~/^ntkrpamp/;
 231                 local *F;
 232                 open F,$filename or confess;
 233                 my $md5obj=Digest::MD5->new();
 234                 $md5obj->addfile(*F);
 235                 close F or confess;
 236                 my $md5sum=lc $md5obj->hexdigest();
 237                 spawn("cp -p '$filename' '$out/$md5sum'") if !$md5sum_stored{$md5sum}++ && $out && $stacki==$#stack;
 238
 239                 my %new=(
 240                         %$this,
 241                         "type"=>$type,
 242                         "length"=>$length,
 243                         "priority"=>$pri,
 244                         "md5"=>$md5sum,
 245                         "dist"=>$dist,
 246                         "lang"=>$lang,
 247                         "ver"=>$ver,
 248                         "this_name"=>$this_name,
 249                         );
 250                 # Highest priority/version of all the files in the cabinet:
 251                 for (qw(ver priority)) {
 252                         next if !$this->{$_};
 253                         $new{$_}=$this->{$_} if $new{$_}<$this->{$_};
 254                         delete $this->{$_} if $this->{$_}<$new{$_};
 255                         }
 256                 do { cluck Dumper($this,\%new) if $new{$_} ne $this->{$_}; } for keys(%$this);
 257                 %$this=%new;
 258                 }
 259         push @stack_out,[@stack];
 260 }
 261
 262 sub spawn($)
 263 {
 264 my($command)=@_;
 265
 266         my $code=system "($command)".' >&2';
 267         confess "$code: $command" if $code;
 268 }
 269
 270 sub rm_rf($)
 271 {
 272 my($dir)=@_;
 273
 274         return if !-e $dir;
 275         # Do not: Can't remove directory xyzzy: Directory not empty
 276         #         -r--r--r-- xyzzy
 277         #         remove \1,$dir or confess "$dir: $!";
 278         spawn "rm -rf '$dir'";
 279 }
 280
 281 sub mkdir_checked
 282 {
 283 my(@dirs)=@_;
 284
 285         for (@dirs) {
 286                 mkdir $_ or confess "$_: $!";
 287                 }
 288 }
 289
 290
 291 our $depth=-1;
 292
 293 sub prep()
 294 {
 295         confess if $dirnew;
 296         $dirnew="$tmp/$depth";
 297         chdir "/" or confess;
 298         rm_rf $dirnew;
 299         mkdir_checked $dirnew;
 300         chdir $dirnew or confess "chdir: $dirnew";
 301         return $dirnew;
 302 }
 303
 304 sub relname($)
 305 {
 306 my($filename)=@_;
 307
 308         return $filename if $filename=~s{^\Q$tmp\E/\d+/}{};
 309         (my $basename=$filename)=~s{^.*/}{} or confess;
 310         return $basename;
 311 }
 312
 313 my %unknown;
 314 sub process();
 315 sub process()
 316 {
 317         my $dirname=$dirnew or confess;
 318         $dirnew=undef();
 319         local $depth=$depth+1;
 320         my @stack_orig=@stack;
 321
 322         local *FIND;
 323         open FIND,"find '$dirname' -type f|sort|" or confess $dirname;
 324         local $_;
 325         while (<FIND>) {
 326                 next if $_ eq "";
 327                 if (!$depth) {
 328                         print STDERR ".";
 329                         $STDERR_needs_eol=1;
 330                         }
 331                 chomp;
 332                 my $filename=$_;
 333                 $filename=~m{^/} or confess $filename;
 334                 (my $basename=$filename)=~s{^.*/}{} or confess;
 335                 -r $filename or do { cluck $filename; system("ls -l '$filename'"); };
 336                 my $this={
 337                         "filename"=>$filename,
 338                         "relname"=>relname($filename),
 339                         };
 340                 @stack=(@stack_orig,$this);
 341
 342                 check $filename,$basename;
 343
 344                 local *MIME;
 345                 open MIME,"file -b -i '$filename'|" or confess;
 346                 my $mime=do { local $/; <MIME> or confess; };
 347                 close MIME or confess;
 348                 chomp $mime;
 349                 next if $mime=~m{^text/};
 350                 next if $mime=~m{^image/};
 351                 next if $mime=~m{^audio/};
 352                 next if $mime eq 'application/x-empty';
 353                 next if $mime eq 'application/msaccess';
 354
 355                 local *ID;
 356                 open ID,"file -b '$filename'|" or confess;
 357                 my $id=do { local $/; <ID> or confess; };
 358                 close ID or confess;
 359                 chomp $id;
 360                 next if $id eq 'Microsoft Office Document';
 361                 next if $id=~/^PDF document, version 1[.]\d+$/;
 362                 next if $id eq 'Rich Text Format data, version 1,';
 363                 next if $id=~/^x86 boot sector, /;
 364                 next if $id eq 'data';
 365                 next if $id eq 'MZ executable for MS-DOS';
 366                 next if $id eq 'XML document text';
 367                 next if $id eq 'MS Windows HtmlHelp Data';
 368                 next if $id eq 'MPEG sequence';
 369                 next if $id=~/^MSVC program database ver /;
 370                 next if $id eq 'Lotus 1-2-3';
 371                 next if $id eq 'MS Windows Help Data';
 372                 next if $id=~/^Macromedia Flash data, /;
 373                 next if $id eq 'Microsoft ASF'; # .wmv
 374                 next if $id eq 'GLF_BINARY_LSB_FIRST';  # .hlp
 375                 next if $id=~/^Sendmail frozen configuration /;
 376                 next if $id eq 'Assembler source';
 377                 next if $id=~/^DBase 3 data file/;
 378                 next if $id=~/^Macintosh HFS Extended version /;
 379                 next if $id=~/^MPEG ADTS, /;    # .txt?,.hlp?
 380                 next if $id eq 'MS-DOS executable (COM)';
 381                 next if $id eq 'NE executable for MS Windows 3.x (driver)';
 382                 next if $id eq 'TrueType font data';
 383                 next if $id eq 'Windows NT registry file';
 384                 next if $id eq 'lif file';      # .sig?
 385                 next if $id=~/^Infocom game data /;     # ._p?
 386                 next if $id=~/^Sun disk label /;        # .dl_?
 387                 next if $id=~/^compiled Java class data, /;
 388                 next if $id=~/^Minix filesystem/;       # .dl_?,_p?
 389                 next if $id eq 'DBase 3 index file';    # .nls?
 390                 next if $id=~/^Macintosh MFS data /;    # .cp_?
 391                 next if $id eq 'ACB archive data';      # .cat?
 392                 next if $id eq 'COM executable for MS-DOS';     # .edb?
 393                 next if $id=~/^Bio-Rad .PIC Image File/;        # ._p?
 394                 next if $id eq 'SysEx File - Gulbransen';       # .edb?
 395
 396                 # Do not:
 397                 #next if $id=~/^PE executable for MS Windows .* 32-bit$/;       # TODO
 398
 399                 if ($id eq 'PE executable for MS Windows (GUI) Intel 80386 32-bit, InnoSetup self-extracting archive') {
 400                         # FIXME: http://innounp.sourceforge.net/
 401                         next;
 402                         }
 403                 if (0
 404                                 || $id=~/^Microsoft Cabinet archive data, \d+ byte(?:s)?, \d+ file(?:s)?$/
 405                                 || $id eq 'PE executable for MS Windows (GUI) Intel 80386 32-bit, MS CAB-Installer self-extracting archive'
 406                                 # Catch all unidentified flying cabinets:
 407                                 || $id=~/^PE executable for MS Windows .* 32-bit$/
 408                                 ) {
 409                         $this->{"type"}="cabinet";
 410                         prep();
 411                         spawn q{cabextract -q '}.$filename.q{' 2>&1|grep -v ': \(}.join(q{\|},
 412                                         q{WARNING; possible [0-9]* extra bytes at end of file.},
 413                                         q{no valid cabinets found},
 414                                         q{checksum error},
 415                                         q{error in CAB data format},
 416                                         ).q{\)$';true};
 417                         process();
 418                         next;
 419                         }
 420                 # non-"32-bit" (amd64 in fact)
 421                 next if $id=~/^PE executable for MS Windows/;
 422                 if (0
 423                                 || $id eq 'PE executable for MS Windows (GUI) Intel 80386 32-bit, ZIP self-extracting archive (WinZip)'
 424                                 || $id=~/^Zip archive data, /
 425                                 ) {
 426                         prep();
 427                         spawn "unzip -q '$filename'";
 428                         process();
 429                         next;
 430                         }
 431                 if ($id=~m{^gzip compressed data, was "\S+", from Win/32}) {
 432                         prep();
 433                         spawn "gzip -d <'$filename' >'$basename'";
 434                         process();
 435                         next;
 436                         }
 437                 if ($id eq 'tar archive') {
 438                         prep();
 439                         spawn "tar xf '$filename'";
 440                         process();
 441                         next;
 442                         }
 443                 warn "$filename: $mime - $id" if !$unknown{$mime,$id}++;
 444                 }
 445         close FIND or confess;
 446         $dirnew=undef();
 447 }
 448
 449
 450 $dirnew=getcwd()."/".$dirnew if $dirnew!~m{^/};
 451 $out   =getcwd()."/".$out    if $out && $out!~m{^/};
 452 rm_rf $tmp;
 453 mkdir_checked $tmp;
 454 if ($out) {
 455         rm_rf $out;
 456         mkdir_checked $out;
 457         }
 458 process;
 459 stack0_flush() if $last_stack0_filename;
 460 print STDERR "\n" if $STDERR_needs_eol;
 461 $STDERR_needs_eol=0;