$BASE=$ARGV[0];
}
+# Somehow Perl modules started reencoding windows-1250 -> utf-8.
+
if ($ENV{"GATEWAY_INTERFACE"}) {
my $future=HTTP::Date::time2str(2000000000);
my $past =HTTP::Date::time2str(1000000000);
print <<"EOH";
-Content-type: text/html; charset=windows-1250
+Content-type: text/html; charset=utf-8
Cache-Control: public
Expires: $future
Last-Modified: $past
print <<"EOH";
<html><head>
<title>iDNES foto@{[ (!$BASE ? "" : ": $BASE") ]}</title>
-<meta http-equiv="Content-Type" content="text/html; charset=windows-1250">
+<meta http-equiv="Content-Type" content="text/html; charset=utf-8">
</head><body>
<form action="idnes-foto" method="get">
<input type="text" name="url" size="50" value="$BASE">
<input type="submit">
</form>
EOH
-$BASE=~m{^http://\w+\Q.idnes.cz/foto.asp?\E(?:r=[-\w]+&c=[-\w]+|c=\w+&r=\w+)$} or $BASE="";
+$BASE=~m{^http://\w+\Q.idnes.cz/foto.asp?\E(?:r=[-_\w]+&c=[-_\w]+|c=[-_\w]+&r=[-_\w]+)$} or $BASE="";
if ($BASE) {
my $pageno=1;
my %seen;
last PAGES if $seen{$base}++;
my $infourl=$BASE.'&styl=zoom&foto='.$base;
my $info=LWP::Simple::get($infourl) or die $infourl;
- $info=~m{<img src="(http://(?:\Qi.idnes.cz\E/\d{2}/\d{3}|\Qimgs.idnes.cz\E/\w+)(?:/\w+)?/[-.\w]+.jpg)" }i
+ $info=~m{<img\s+id="fotka"\s+src="([^<>"]+)" }is
or die "No image found: $infourl";
my $img_src=$1;
print <<"EOH";
<hr>
<img src="$img_src" border="0">
EOH
- my $text_last=keys(%seen_p_text);;
+ my $hit;
for my $text (
+ $info=~m{<span>(.*?)</span>}s,
$info=~m{<p>([^<>]*)</p>},
$info=~m{<div class="text"><!--google_ad_section_start--><h4>([^<>]*)</h4><p>([^<>]*)<!--google_ad_section_end--></p><p>Autor:\s+(.*?)(?:, <a target="_blank" href="http://www.idnes.cz">iDNES.cz</a>)?</p></div>}s,
) {
next if !$text;
+ $text=~s/^\s+//s;
+ $text=~s/\s+$//s;
+ next if !$text;
+ $hit++;
print <<"EOH" if !$seen_p_text{$text}++;
<p>$text</p>
EOH
}
- warn "No text found: $infourl" if $text_last==keys(%seen_p_text);
+ warn "No text found: $infourl" if !$hit;
}
die $pageurl if !$did;
$pageno++;