urlrule_new

   1 #!/usr/bin/perl -w
   2 ###APPNAME:     urlrule_new
   3 ###APPAUTHOR:   xiaoranzzz
   4 ###APPDATE:     Sat Mar 22 22:56:33 2008
   5 ###APPVER:      0.1
   6 ###APPDESC:     create new url rule
   7 ###APPUSAGE:    URL [0-5]
   8 ###APPEXAMPLE:  urlrule_new http://www.shuku.net/novels/gulong/bianlang/bianlang.html 2
   9 ###APPOPTION:
  10 use strict;
  11 use lib $ENV{XR_PERL_MODULE_DIR};
  12 use MyPlace::Script::Usage qw/help_even_empty/;
  13
  14 exit 0 if(help_even_empty($0,@ARGV));
  15
  16 use MyPlace::ColorPrint;
  17 use MyPlace::URLRule qw/parse_rule/;
  18
  19 my %rule = %{parse_rule(@ARGV)};
  20 my $url=$rule{url};
  21 my $level=$rule{level};
  22 my $source=$rule{source};
  23
  24
  25 if(-f $source) {
  26     app_message("Rule for $url exists,opened for editing instead\n");
  27     system("r-vim \"$source\"&");
  28     exit 0;
  29 }
  30
  31 my $date=localtime();
  32 open VIMED,"|-","r-vim -c 'file $source' - ";
  33 print VIMED <<"HEAD";
  34 #!/usr/bin/perl -w
  35 #$url
  36 #$date
  37 use strict;
  38
  39
  40 HEAD
  41
  42 print VIMED <<'FUNC';
  43
  44 =method1
  45 sub apply_rule {
  46  return (
  47        '#use quick parse'=>1,
  48        'data_exp'=>undef,
  49        'data_map'=>undef,
  50        'pass_exp'=>undef,
  51        'pass_map'=>undef,
  52        'pass_name_map'=>undef,
  53        'pages_exp'=>undef,
  54        'pages_map'=>undef,
  55        'pages_pre'=>undef,
  56        'pages_suf'=>undef,
  57        'pages_start'=>undef,
  58        'title_exp'=>undef,
  59        'title_map'=>undef,
  60        'charset'=>undef
  61  );
  62 }
  63 =cut
  64
  65 =method2
  66 use MyPlace::Curl;
  67 #use MyPlace::HTML;
  68
  69 sub _process {
  70     my ($url,$rule,$html) = @_;
  71     my $title = undef;
  72     my @data;
  73     my @pass_data;
  74     #my @html = split(/\n/,$html);
  75     return (
  76         count=>scalar(@data),
  77         data=>[@data],
  78         pass_count=>scalar(@pass_data),
  79         pass_data=>[@pass_data],
  80         base=>$url,
  81         no_subdir=>1,
  82         work_dir=>$title,
  83     );
  84 }
  85
  86 sub apply_rule {
  87     my $url = shift(@_);
  88     my $rule = shift(@_);
  89     my $http = MyPlace::Curl->new();
  90     my (undef,$html) = $http->get($url);
  91     return &_process($url,$rule,$html,@_);
  92 }
  93 =cut
  94
  95 1;
  96 FUNC
  97 print VIMED <<'USAGE';
  98
  99 __END__
 100
 101 #================================================================
 102 # apply_rule will be called with ($RuleBase,%Rule),the result returned have these meaning:
 103 # $result{base}          : Base url to build the full url
 104 # $result{work_dir}      : Working directory (will be created if not exists)
 105 # $result{data}          : Data array extracted from url,will be passed to $result{action}(see followed)
 106 # $result{action}        : Command which the $result{data} will be piped to,can be overrided
 107 # $result{pipeto}        : Same as action,Command which the $result{data} will be piped to,can be overrided
 108 # $result{hook}          : Hook action,function process_data will be called
 109 # $result{no_subdir}     : Do not create sub directories
 110 # $result{pass_data}     : Data array which will be passed to next level of urlrule
 111 # $result{pass_name}     : Names of each $result{pass_data}
 112 # $result{pass_arg}      : Additional arguments to be passed to next level of urlrule
 113 # urlrule_quick_parse($url,$rule,$data_exp,$data_map,$pass_exp,$pass_map,$charset)
 114 # urlrule_parse_data($url,$rule,$data_exp,$data_map,$charset)
 115 # urlrule_parse_pass_data($url,$rule,$pass_exp,$pass_map,$charset)
 116 #
 117 # $result{same_level}    : Keep same rule level for passing down
 118 # $result{level}         : Specify rule level for passing down
 119 #================================================================
 120
 121 #       vim:filetype=perl
 122 USAGE
 123
 124
 125 close VIMED;
 126
 127 if(-f "$source" and ! -x "$source") {
 128     chmod 0775,$source;
 129 }
 130 exit 0;
 131