From c8f94771b5de8eba10e174a63a8d5c778dff065e Mon Sep 17 00:00:00 2001
From: lace <>
Date: Wed, 14 Jul 2010 11:13:07 +0000
Subject: [PATCH] idnes-foto-old1

---
 public_html/cgi-bin/idnes-foto | 39 +++++++++++++++++++++++++--------------
 1 file changed, 25 insertions(+), 14 deletions(-)

diff --git a/public_html/cgi-bin/idnes-foto b/public_html/cgi-bin/idnes-foto
index eaa15c9..dfdbd96 100755
--- a/public_html/cgi-bin/idnes-foto
+++ b/public_html/cgi-bin/idnes-foto
@@ -2,6 +2,7 @@
 use strict;
 use warnings;
 require LWP::Simple;
+require HTTP::Date;
 require URI::Escape;
 
 $|=1;
@@ -18,8 +19,13 @@ else {
 	}
 
 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
+Cache-Control: public
+Expires: $future
+Last-Modified: $past
 
 EOH
 	}
@@ -33,7 +39,7 @@ print <<"EOH";
 	<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;
@@ -42,25 +48,30 @@ PAGES:	for (;;) {
 		my $pageurl=$BASE.'&strana='.$pageno;
 		my $page=LWP::Simple::get($pageurl) or die $pageurl;
 		my $did=0;
-		while ($page=~m{<img src="(http://(?:\Qi.idnes.cz\E/\d{2}/\d{3}|\Qimgs.idnes.cz\E/\w+))(/midi)?/([-\w]+.jpg)" }gi) {
+		while ($page=~m{<img src="http://(?:\Qi.idnes.cz\E/\d{2}/\d{3}|\Qimgs.idnes.cz\E/\w+)(?:/\w+)?/([-.\w]+.jpg)" }gi) {
 			$did=1;
-			my($start,$mid,$base)=($1,$2,$3);
+			my($base)=($1);
 			last PAGES if $seen{$base}++;
-			# FIXME: Sometimes "/maxi" does not exist.
-			$mid&&="/maxi";
-			$base=~s/_1M.JPG$/_V.JPG/;
-			my $maxi=$start.($mid||"")."/".$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
+					or die "No image found: $infourl";
+			my $img_src=$1;
 			print <<"EOH";
 <hr>
-<img src="$maxi" border="0">
+<img src="$img_src" border="0">
 EOH
-			my $infourl=$BASE.'&styl=zoom&foto='.$base;
-			my $info=LWP::Simple::get($infourl) or die $infourl;
-			$info=~m{<p>[^<]*</p>} or die "No text found: $infourl";
-			my $p_text=$&;
-			print <<"EOH" if !$seen_p_text{$p_text}++;
-$p_text
+			my $text_last=keys(%seen_p_text);;
+			for my $text (
+					$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;
+				print <<"EOH" if !$seen_p_text{$text}++;
+<p>$text</p>
 EOH
+				}
+			warn "No text found: $infourl" if $text_last==keys(%seen_p_text);
 			}
 		die $pageurl if !$did;
 		$pageno++;
-- 
1.8.3.1