tidywenku2

   1 #!/usr/bin/perl -w
   2 ###APPNAME:     tidywenku2
   3 ###APPAUTHOR:   xiaoranzzz
   4 ###APPDATE:     Fri Oct  5 12:37:53 2007
   5 ###APPVER:      0.2
   6 ###APPDESC:     将文心阁小说转为UTF8,并整理页面
   7 ###APPUSAGE:    [目录名]
   8 ###APPEXAMPLE:  tidywenk2 "在人间"
   9 ###APPOPTION:
  10 use strict;
  11 $ARGV[0]="-h" unless(@ARGV);
  12 foreach(@ARGV){
  13     exit(system("plhelp",$0,@ARGV)) if($_ eq "-h" || $_ eq "--help");
  14 }
  15
  16 my @PAT;
  17 my @REP;
  18
  19 sub buildRule($$) {
  20     push @PAT,shift;
  21     push @REP,shift;
  22 }
  23
  24 #my $SIGN="<div align=right id=sign>Reproduced by <b>Xiaoranzzz<\/b><br>" . `date` . "<\/div>";
  25 my $SIGN='<div align="right" id="sign">WenKuV2.0 Geek@MYPLACE</b><br>' . `date` . '</div>';
  26
  27 buildRule qr//,'';
  28 buildRule qr/--.*<\/title>/,'</title>';
  29 buildRule qr/<script src=.*<\/script>/,"";
  30 buildRule qr/gb2312/,"utf-8";
  31 buildRule qr/<body .*  marginwidth="20">/,'<body class="mainBody">' . "\n";
  32 buildRule qr/class='middle'/,'class="mtext"';
  33 buildRule qr/.*<a href='(http\:\/\/www\.wenku\.com|\/)'>.*$/i,"$SIGN";
  34 buildRule qr/<script language='JavaScript'>/,"\n" . '<!--/';
  35 buildRule qr/<\/script>/,"-->\n" .
  36                 '<script language="JavaScript" src="wenku.js"></script>' . "\n" .
  37                 '<script language="JavaScript" src="file:///c:/windows/wenku.js"></script>' . "\n" .
  38                 '<script language="JavaScript" src="file:///etc/wenku.js"></script>' . "\n";
  39 buildRule qr/<link[^<>]*new.css[^<>]*>/,
  40                 '<link rel="stylesheet" href="wenku.css" type="text/css">' . "\n" .
  41                 '<link rel="stylesheet" href="file:///c:/windows/wenku.css" type="text/css">' . "\n" .
  42                 '<link rel="stylesheet" href="file:///etc/wenku.css" type="text/css">' . "\n";
  43
  44 buildRule qr/<table[^<>]*width=[\'\"]*100%[^><]*>/,'<table id="headTable">';
  45 buildRule qr/<table[^<>]*width=[\'\"]*93%[^<>]*>/,'<table id="contentTable">';
  46 buildRule qr/<td[^<>]*width=[\"\']*80%[^<>]*>/,'<td id="contentText">';
  47 buildRule qr/class=\"title\"/,'id="bookTitle"';
  48 buildRule qr/<center><\/center>/,'';
  49 buildRule qr/<hr[^<>]*>/,'';
  50
  51 my $TEMPFILE=`mktemp`;
  52
  53 sub GB2UTF($) {
  54     my $filename=shift;
  55     return 0 unless(-f "$filename");
  56     if (system("iconv","-c","-f","gb18030","-t","utf8","-o","$TEMPFILE","--","$filename")==0){
  57         return 1;
  58     }
  59     else {
  60         print STDERR "Error while convert $filename from gb2312 to utf8\n";
  61         return 0;
  62     }
  63 }
  64 sub TidyFile($) {
  65     my $filename=shift;
  66     if(GB2UTF($filename)) {
  67         open(TEMP,"<",$TEMPFILE) or die("$!\n");
  68         open(FH,">",$filename) or die("$!\n");
  69         while(<TEMP>) {
  70             my $LINE=$_;
  71             $LINE =~ s#<a.*bookroom.php.*>(.*)<\/a>》#<b id="title">$1</b>#;
  72             for(my $i=0;$i<@PAT;$i++) {
  73                 $LINE =~ s/$PAT[$i]/$REP[$i]/g;
  74             }
  75             print FH ($LINE,"\n");
  76         }
  77        close(TEMP);
  78        close(FH);
  79     }
  80 }
  81
  82
  83 foreach my $cur(@ARGV) {
  84     if(-f $cur) {
  85         print STDERR ("$cur ...                                \r");
  86         TidyFile $cur;
  87         print STDERR ("$cur [Done]                                 \n");
  88     }
  89     elsif(-d $cur) {
  90         foreach my $sub(glob("$cur/*")) {
  91             if( -f $sub) {
  92                 print STDERR ("$sub  ...                                      \r");
  93                 TidyFile $sub;
  94             }
  95         }
  96         print STDERR ("$cur [Done]                                                                    \n");
  97     }
  98     else {
  99         print STDERR ("$cur [Ingored]\n");
 100     }
 101 }
 102
 103