sinabook_search

   1 #!/usr/bin/perl -w
   2 ###APPNAME:     sinabook_search
   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:    <keyword> [cmd]
   8 ###APPEXAMPLE:  sinabook_search 我的名字
   9 ###APPOPTION:
  10 use strict;
  11 use URI::Escape;
  12 use HTML::TreeBuilder;
  13 use Data::Dumper;
  14
  15 $ARGV[0]="-h" unless(@ARGV);
  16 foreach(@ARGV){
  17     exit(system("plhelp",$0,@ARGV)) if($_ eq "-h" || $_ eq "--help");
  18 }
  19
  20 my $keyword=shift;
  21 print "searching club.book.sina.com.cn for \"$keyword\" ...\n";
  22
  23 my $cmd=shift;
  24 $cmd = "sinabook_justtext" unless($cmd);
  25
  26 $keyword=`echo $keyword | iconv -f utf8 -t gb2312 -c`;
  27 chomp($keyword);
  28 $keyword = uri_escape($keyword);
  29 my $URL="http://club.book.sina.com.cn/booksearch/booksearch.php?key1=&k=&textfield=&bookuser=&c=112&col=%B6%C1%CA%E9&item=writingname&area=hotkey&kw=$keyword";
  30
  31 open FI,"-|","netcat '$URL' | iconv -f gb2312 -t utf8" or die("Unable to fork netcat | iconv\n");
  32 my @tree;
  33 $tree[0] = HTML::TreeBuilder->new();
  34 my $pages=1;
  35 while(<FI>) {
  36     $tree[0]->parse($_);
  37     my @match = $_ =~ /(\/booksearch\/booksearch\.php\?page=)([0-9]+)(\&[^\"\' ]+)/;
  38     if(@match) {
  39         $pages = $match[1];
  40     }
  41 }
  42 close FI;
  43 $tree[0]->eof;
  44
  45 for(my $i=2;$i<=$pages;$i++) {
  46     $tree[$i-1] = HTML::TreeBuilder->new();
  47     open FI,"-|","netcat '$URL&page=$i' | iconv -f gb2312 -t utf8" or die("Unable to fork netcat | iconv\n");
  48     $tree[$i-1]->parse($_) while(<FI>);
  49     close FI;
  50     $tree[$i-1]->eof;
  51 }
  52
  53 sub gettext(@) {
  54     my @text;
  55     foreach my $child(@_) {
  56         next if(ref $child);
  57         push(@text,$child) if($child);
  58     }
  59     return @text;
  60 }
  61
  62 sub selectBook(@) {
  63     my @books = @_;
  64     print "=" x 40,"\n\n";
  65     for(my $i=0;$i<@books;$i++) {
  66         printf "%-2d. %s (%s)\n",$i + 1,$books[$i]{title},$books[$i]{author};
  67     }
  68     print "=" x 40,"\n";
  69     print "Your choice (0 to quit): ";
  70     my $result;
  71     until($result) {
  72         $result = <STDIN>;
  73         chomp($result);
  74         if($result =~ /^\d$/) {
  75            return $result-1 if($result >=0 && $result <=@books);
  76            $result=0;
  77         }
  78         print "\rInvalid input,Select again: ";
  79     }
  80 }
  81
  82 my @books=();
  83 foreach my $tree(@tree) {
  84 foreach my $div($tree->look_down("_tag","div","class","des")) {
  85     my %book;
  86     foreach my $node($div->look_down("_tag",qr/^(a|p)$/i)) {
  87         $node->normalize_content;
  88         if((!$book{author}) and $node->tag() eq "p") {
  89             my @text=gettext($node->content_list);
  90             $book{author} = $text[0] if(@text);
  91             $book{about} = $text[1];
  92         };
  93         unless($book{href}) {
  94             next unless($node->tag() eq "a");
  95             $book{href}=$node->attr("href");
  96             my @text=gettext($node->content_list);
  97             $book{title} = $text[0] if(@text);
  98         };
  99     }
 100     push(@books,\%book);
 101 }
 102 $tree->delete();
 103 }
 104
 105 if(!@books) {
 106     print "Empty result!\n";
 107     exit 0;
 108 }
 109 else {
 110     my $idx = selectBook(@books);
 111     exit 0 if($idx == -1);
 112     print("Selected ",$books[$idx]{href},"\n");
 113     system($cmd,$books[$idx]{href});
 114 }
 115
 116