xrbook_writeinfo

   1 #!/usr/bin/perl
   2 ###APPNAME:     xrbook_writeinfo.pl
   3 ###APPAUTHOR:   xiaoranzzz
   4 ###APPDATE:     Tue Mar 11 04:58:02 2008
   5 ###APPVER:      0.1
   6 ###APPDESC:     create bookinfo.xml
   7 ###APPUSAGE:    indexURL [directory]
   8 ###APPEXAMPLE:  xrbook_writeinfo http://book.sina.com.cn/nzt/lit/wodemingzijiaohong/index.shtml src
   9 use strict;
  10 use XML::Simple;
  11 use Data::Dumper;
  12
  13
  14 $ARGV[0]="-h" unless(@ARGV);
  15 foreach(@ARGV){
  16     exit(system("plhelp",$0,@ARGV)) if($_ eq "-h" || $_ eq "--help");
  17 }
  18
  19 die("Usage: $0 indexURL [directory]\n") unless(@ARGV);
  20
  21 sub getTextBetween {
  22     my $text=shift;
  23     my $begin=shift;
  24     my $end=shift;
  25     $end='[\<\n]' unless($end);
  26     my @match = $text =~ m{$begin\s*([^\<\>]*)\s*$end}i;
  27     return $match[0] if(@match);
  28     return "";
  29 }
  30
  31 my %book;
  32 $book{URL}=$ARGV[0];
  33 $book{saveto}= $ARGV[1];
  34
  35 my @match = ($book{URL} =~ /^(http:\/\/|)(.*\/)catalog\.php\?book=([0-9]+)/);
  36 if(@match) {
  37         $book{base}=$match[1];
  38         $book{dirname}=$match[2];
  39 }
  40 else {
  41     if($book{URL} =~ /\/$/) {
  42         $book{URL} .= "index.shtml";
  43     }
  44     @match=($book{URL} =~ /^(http:\/\/|)([^\/]+)\/.*\/([^\/]+)\/[^\/]*$/);
  45     if(@match) {
  46     $book{base}=$match[1];
  47     $book{dirname}=$match[2];
  48     }
  49 }
  50
  51 die("Invalid index URL:\n$book{URL}\n") unless(@match);
  52
  53 open(SOURCE,"-|", "netcat '$book{URL}' | iconv -f gb2312 -t utf8 -c")
  54     or die("Cant't fork netcat | iconv\n");
  55 my $TEXT="[^\<\>]*";
  56 my @pages=(());
  57 while(<SOURCE>) {
  58     unless($book{title}) {
  59         $book{title} = getTextBetween($_,'\<\s*title\s*\>');
  60         $book{title} =~ s/_.*$//g;
  61         $book{title} =~ s/.*(代表作|作品)：//g;
  62     }
  63     unless($book{author}) {
  64         $book{author} = getTextBetween($_,'\<span class=td_m\>作者：\<\/span\>\<a[^\>\<]*\>\<span class=td_m\>');
  65         $book{author} = getTextBetween($_,"\<span\>作者：") unless($book{author});
  66     }
  67     unless($book{description}) {
  68         $book{description} = getTextBetween($_,
  69                 '\<meta\s*name=\"*description\"*\s*content=\"',
  70                 '\"\s*\>');
  71     }
  72     unless($book{about}) {
  73         $book{about} = getTextBetween($_,'\<span class=td_m\>');
  74     }
  75     unless($book{cover}) {
  76         my @match = $_ =~ /src\s*=\s*(.*$book{dirname}.*\.jpg)/;
  77         if(@match) {
  78             $book{cover}=$match[0];
  79         }
  80     }
  81     my @match = $_ =~ /\<a[^\<\>]*href\s*=\s*($TEXT$book{dirname}${TEXT}html)$TEXT\>($TEXT)\<\/a\>/i;
  82     @match = $_ =~ /\<a[^\<\>]*href\s*=\s*(\/longbook\/$TEXT\.shtml)$TEXT\>($TEXT)\<\/a\>/i unless(@match);
  83     if(@match) {
  84         next if($match[0] =~ /index/);
  85         $match[1] =~ s/^\s*[0-9]*//g;
  86         $match[1] =~ s/^．\s*//;
  87         $match[0] =~ s/\"//g;
  88         push(@pages,{"NAME"=>$match[1],"URL"=>$match[0]});
  89     }
  90 }
  91
  92 $book{pages}=\@pages;
  93 $book{title}=$book{dirname} unless($book{title});
  94 if(! $book{saveto}) {
  95     if($book{author}) {
  96         $book{saveto}=$book{author} . "/" . $book{title};
  97         if(! -d $book{author}) {mkdir $book{author} or die("$!\n");}
  98     }
  99     else {
 100         $book{saveto}=$book{title};
 101     }
 102     if(! -d $book{saveto}) {mkdir $book{saveto} or die("$!\n");}
 103     $book{saveto} .= "/src";
 104 }
 105
 106 if(! -d $book{saveto}) {
 107     mkdir $book{saveto} or die("$!\n");
 108 }
 109 chdir $book{saveto};
 110
 111 open fO,"|-","batchget -m 1 -n '$book{title}' -r '$book{URL}'";
 112 for(my $i=0;$i<@pages;$i++) {
 113     print fO "http://$book{base}$pages[$i]{URL}\n";
 114 }
 115 print fO $book{cover},"\n" if($book{cover});
 116 print fO $book{URL},"\n";
 117 close fO;
 118
 119 my $xml = XMLout(\%book,SuppressEmpty=>1,KeyAttr=>[],NoAttr=>1,NumericEscape=>0) or die("$!\n");
 120 open fO,">","bookinfo.xml";
 121 print fO $xml;
 122 close fO;
 123 print "Done saved in \"$book{saveto}\"\n";
 124
 125