Perl LWP 股票資料收集

這是一支跑去Yahoo股市查列出股票的股利政策的程式，自動計算、分析是最後的結果。

# 抓取網頁中的 link
use LWP;
use HTML::TreeBuilder;
use Text::Iconv;

# 取得命令列中的網頁位址，放入 $url 中
#my $url=shift || die "您沒有輸入 url 網址!\n";
my $stock=2357;
if(defined $ARGV[0]){
 $stock=$ARGV[0];
}
my $url="http://tw.stock.yahoo.com/d/s/dividend_".$stock.".html"; 

my $ua=LWP::UserAgent->new;
my $res=$ua->get($url);
die "Can't get $url ", $res->status_line unless $res->is_success;
my $html=$res->content;

$converter = Text::Iconv->new("big5", "utf8");
$html = $converter->convert("$html");
#print "$html\r\n";

my $root=HTML::TreeBuilder->new_from_content($html);

my @names=$root->look_down(_tag=>'TITLE');

foreach my $line(@names){
 my $name=$line->as_trimmed_text;
 my @items=split(/\-/,$name);
 print $items[0],",",$items[1],",\r\n";
}

my @links=$root->look_down(
   _tag=>'table',
);
for($j=0; $j<=$#links; $j++){
 if($j == 9){
  #print "\t$j=", $links[$j]->attr('href'),' ', $links[$j]->as_trimmed_text, "\r\n";
  my @items=$links[$j]->look_down(_tag=>'tr');
  #print "Item $j\r\n";
  foreach my $item(@items){
   my @datas=$item->look_down(_tag=>'td');
   foreach my $data(@datas){
    my $tmp=$data->as_trimmed_text;
    $tmp =~ s/\ //ge;
    $tmp =~ s/\　//ge;
    print $tmp,",";
   }
   print "\r\n";
   #print "$j\t", $item->as_trimmed_text,"\r\n";
  }
  #print "$j=", $links[$j]->as_trimmed_text, "\r\n";
 }
}
$root->delete;

這個程式是先跑去抓某個網頁的股票代碼列表(有點舊)然後再去呼叫上面那一支程式把歷史股利政策全抓下來。

# 抓取網頁中的 link
use LWP;
use HTML::TreeBuilder;
use Text::Iconv;

# 取得命令列中的網頁位址，放入 $url 中
#my $url=shift || die "您沒有輸入 url 網址!\n";

my $url="http://www.tej.com.tw/webtej/doc/uid.htm"; 

my $ua=LWP::UserAgent->new;
my $res=$ua->get($url);
die "Can't get $url ", $res->status_line unless $res->is_success;
my $html=$res->content;

$converter = Text::Iconv->new("big5", "utf8");
$html = $converter->convert("$html");
#print "$html\r\n";

my $root=HTML::TreeBuilder->new_from_content($html);

my @links=$root->look_down(
   _tag=>'td'
);
#my @links=$root->query('tr');

for($j=0; $j<=$#links; $j++){
 my $tmp=$links[$j]->as_trimmed_text;
 #my $tmp=$links[$j]->attr('str');
 $tmp=~s/\D//ge;
 if(($tmp < 9999)&&($tmp ne "")){
  #print $tmp, "\r\n";
  my $var=`perl yahooStock01.pl $tmp`;
  print "$var\n";
 }
}
$root->delete;