src/libcaptive/captivemodid/captivemodid-list.pl

   1 #! /usr/bin/perl
   2 #
   3 # $Id$
   4 # Extract the list of files to download from Microsoft.
   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 require LWP::Simple;
  25 use URI::Escape;
  26
  27
  28 sub unamp($)
  29 {
  30 my($ref)=@_;
  31
  32         die "non-amp amp: $$ref" if $$ref=~/&(?!amp;)/;
  33         $$ref=~s/&amp;/&/go;
  34 }
  35
  36 print STDERR "Downloading index: ";
  37 my $found_total;
  38 my @found;
  39 my $found_processed=0;
  40 my $next='results.aspx?freetext=&productID=4C937A02-BAE0-4317-A1A9-0C56CD979D05&categoryId=7&period=&sortCriteria=date&nr=50&DisplayLang=en&type=a';
  41 my $BASE='http://www.microsoft.com/downloads/';
  42 my $VALIDATION_CODE=$ARGV[0] or die "ARGV[0]=VALIDATION_CODE";
  43
  44 while ($next) {
  45         my $page=LWP::Simple::get($BASE.$next) or die;
  46         my($page_total,$page_first,$page_last)=($page=~m{<b>(\d+)</b> results found; results <b>(\d+)-(\d+)</b> shown.}) or die;
  47         die if !$page_total;
  48         $found_total=$page_total if !$found_total;
  49         $found_total==$page_total or die;
  50         $found_processed+1==$page_first or die "(found_processed+1)(".($found_processed+1)."!=page_first($page_first)";
  51         $page_last>=$page_first or die;
  52         if ($page_last<$found_total) {
  53                 $page_last==$page_first+50-1 or die;
  54                 ($next)=($page=~m{<a href="([^"]*)">Next\s+&gt;</a>}) or die;
  55                 unamp \$next;
  56                 $next=~/^info.aspx[?]/ or die;
  57                 }
  58         else {
  59                 $page_last==$found_total or die;
  60                 $next=undef();
  61                 }
  62
  63         while ($page=~m{<p><a href="([^"]*)">[^<]*</a>(?:&nbsp;<a href=[^>]*><img[^>]*\balt="Genuine Windows download"[^>]*></a>)?</p>}g) {
  64                 my($url)=($1);
  65                 $found_processed++;
  66                 unamp \$url;
  67                 push @found,$url;
  68                 }
  69         print STDERR ".";
  70         $found_processed==$page_last or die "found_processed(".($found_processed).")!=page_last($page_last)";
  71         }
  72 print STDERR " found: ".(0+@found)."\n";
  73
  74 my @download;
  75 print STDERR "Downloading product pages: ";
  76 for my $found (@found) {
  77         my $page=LWP::Simple::get($BASE.$found) or die;
  78         my($lang)=($page=~m{<select name="displaylang" [^>]*>((?:<option [^>]*>[^<>]*</option>)+)</select>});
  79         # <option value="en" selected>English</option><option value="fr">French</option></select>
  80         print STDERR "+";
  81         my @stage2;
  82         if (!$lang) {
  83                 @stage2=$found;
  84                 }
  85         else {
  86                 while ($lang=~s{<option value="([^"]*)"(?: selected)?>([^<>]*)</option>}{}) {
  87                         my($short,$long)=($1,$2);
  88                         (my $found_lang=$found)=~s/((?:\b|%26)DisplayLang(?:=|%3d))en\b/$1$short/ or die $found;
  89                         push @stage2,$found_lang;
  90                         }
  91                 !$lang or die;
  92                 die if !@stage2;
  93                 }
  94         for my $stage2 (@stage2) {
  95                 my $delim='&';
  96                 $delim='%26' if $stage2=~/%26DisplayLang/i;
  97                 $stage2.="${delim}Hash=$VALIDATION_CODE";
  98                 $stage2=$BASE.$stage2;
  99                 my $page=LWP::Simple::get($stage2) or die $stage2;
 100                 while ($page=~m{<h2>Object moved to <a href="([^"]*)">}) {
 101                         $stage2=$1;
 102                         unamp \$stage2;
 103                         $page=LWP::Simple::get($stage2) or die $stage2;
 104                         }
 105                 my $download;
 106                 while ($page=~m{window.open(?:.)'(http://download.microsoft.com/download/[^']*)',null,}g) {
 107                         die if $download;
 108                         $download=$1;
 109                         }
 110                 if ($download) {
 111                         push @download,$download;
 112                         print STDERR ".";
 113                         }
 114                 elsif ($page=~m{<table id="multiFileList"(.*?)</table>}) {
 115                         my $files=$1;
 116                         my $download_orig=@download;
 117                         while ($files=~m{<a href="([^"]*)">[^<>]*</a>}g) {
 118                                 my $url=$1;
 119                                 unamp \$url;
 120                                 push @download,$url;
 121                                 }
 122                         die if $download_orig==@download;
 123                         print STDERR "_";
 124                         }
 125                 elsif ($page=~m{<div id="regsysNotRegistered">}) {
 126                         print STDERR "!";
 127                         }
 128                 else {
 129                         die $stage2;
 130                         }
 131                 }
 132         }
 133 for (@download) {
 134         next if !/^info.aspx[?].*&u=(http%3a[^&]*)$/;
 135         $_=uri_unescape $1;
 136         }
 137 @download=sort keys(%{{ map(($_=>1),@download) }});
 138 print STDERR " found downloads: ".(0+@download)."\n";
 139 print "$_\n" for @download;