sinabook_parsesearch

   1 #!/usr/bin/perl -w
   2 ###APPNAME:     sinabook_parsesearch
   3 ###APPAUTHOR:   xiaoranzzz
   4 ###APPDATE:     Tue Mar 11 16:56:49 2008
   5 ###APPVER:      0.1
   6 ###APPDESC:     [cmd] default as sinabook_justtext
   7 ###APPUSAGE:    URL
   8 ###APPOPTION:
   9 use strict;
  10 use URI::Escape;
  11 use HTML::TreeBuilder;
  12 use Data::Dumper;
  13
  14 $ARGV[0]="-h" unless(@ARGV);
  15 foreach(@ARGV){
  16     exit(system("plhelp",$0,@ARGV)) if($_ eq "-h" || $_ eq "--help");
  17 }
  18
  19 my $URL=shift;
  20
  21 open FI,"-|","netcat '$URL' | iconv -f gb2312 -t utf8" or die("Unable to fork netcat | iconv\n");
  22 my @tree;
  23 $tree[0] = HTML::TreeBuilder->new();
  24 my $pages=0;
  25 while(<FI>) {
  26     $tree[0]->parse($_);
  27     my @match = $_ =~ /(\/booksearch\/booksearch\.php\?page=)([0-9]+)(\&[^\"\' ]+)/;
  28     if(@match) {
  29         $pages = $match[1];
  30     }
  31 }
  32 close FI;
  33 $tree[0]->eof;
  34
  35 for(my $i=2;$i<=$pages;$i++) {
  36     $tree[$i-1] = HTML::TreeBuilder->new();
  37     open FI,"-|","netcat '$URL&page=$i' | iconv -f gb2312 -t utf8" or die("Unable to fork netcat | iconv\n");
  38     $tree[$i-1]->parse($_) while(<FI>);
  39     close FI;
  40     $tree[$i-1]->eof;
  41 }
  42
  43 open FO,"|-","sort -u";
  44 foreach my $tree(@tree) {
  45 foreach my $div($tree->look_down("_tag","div","class","des")) {
  46     foreach my $node($div->look_down("_tag","a")) {
  47         print FO $node->attr("href"),"\n";
  48     }
  49 }
  50 $tree->delete();
  51 }
  52 close FO
  53
  54