book_download

   1 #!/usr/bin/perl
   2 ###APPNAME:     book_download.pl
   3 ###APPAUTHOR:   xiaoranzzz
   4 ###APPDATE:     Tue Mar 11 04:58:02 2008
   5 ###APPVER:      0.1
   6 ###APPDESC:     download html using book_parser
   7 ###APPUSAGE:    indexURL [directory]
   8 ###APPEXAMPLE:  book_download http://book.sina.com.cn/nzt/lit/wodemingzijiaohong/index.shtml src
   9 use strict;
  10 use XML::Simple;
  11 use Data::Dumper;
  12 $ARGV[0]="-h" unless(@ARGV);
  13 foreach(@ARGV){
  14     exit(system("plhelp",$0,@ARGV)) if($_ eq "-h" || $_ eq "--help");
  15 }
  16
  17 sub buildURL($$) {
  18     my $base=shift;
  19     my $href=shift;
  20     return "$base\/$href" if($href =~ /^[^\/]/);
  21     $base =~ s/(http:\/\/[^\/]*)\/.*$/\1/;
  22     return "$base$href";
  23 }
  24
  25 my $url=shift;
  26 my $saveto=shift;
  27 open FI,"-|","book_parser '$url'";
  28 my $data = join("",<FI>);
  29 my $book = XMLin($data,SuppressEmpty=>1,NoAttr=>1,KeyAttr=>[]);
  30 close FI;
  31
  32 my $base = $url;
  33 $base =~ s/\/[^\/]*$//g;
  34 my $name = $url;
  35 $name =~ s/\/$//g;
  36 $name =~ s/^.*\///g;
  37
  38
  39 #print Dumper($book);
  40 $book->{title}=$name unless($book->{title});
  41
  42 if(!$saveto) {
  43     if($book->{author}) {
  44         $saveto=$book->{author} . "/" . $book->{title};
  45         if(! -d $book->{author}) {mkdir $book->{author} or die("$!\n");}
  46     }
  47     else {
  48         $saveto=$book->{title};
  49     }
  50     if(! -d $saveto) {mkdir $saveto or die("$!\n");}
  51     $saveto .= "/src";
  52 }
  53
  54 if(! -d $saveto) {
  55     mkdir $saveto or die("$!\n");
  56 }
  57 chdir $saveto;
  58
  59 open fO,">","bookinfo.xml";
  60 print fO $data;
  61 close fO;
  62
  63 open fO,"|-","batchget -n '$book->{title}' -r '$url'";
  64 print fO $url,"\n";
  65 if($book->{pages}) {
  66     for(my $i=0;$i<@{$book->{pages}};$i++) {
  67         print fO buildURL($base,$book->{pages}->[$i]->{url}),"\n";
  68     }
  69 }
  70 close fO;
  71 print STDERR "Task done! Saved in $saveto\n";
  72
  73