--- trunk/spider/swishspider	2003/03/24 09:57:44	30
+++ trunk/spider/swishspider	2003/07/29 10:40:58	42
@@ -18,7 +18,9 @@
 
 my $no_parent_url;
 if ($url =~ m/\s/) {
-	($url,$no_parent_url) = split(/\s/,$url,2);
+	($no_parent_url,$url) = split(/\s/,$url,2);
+	# old scheme had URL, no parent and new is reverse
+	($url,$no_parent_url) = ($no_parent_url,$url) if ($no_parent_url =~ m/\Q$url\E/);
 }
 
 my $request = new HTTP::Request( "GET", $url );
@@ -39,7 +41,7 @@
 	if ($no_parent_url) {
 		if ($link =~ m/$no_parent_url/) {
 			# if this URL is below parent URL o.k....
-			print RESP "$link $no_parent_url\n";
+			print RESP "$no_parent_url $link\n";
 		} else {
 			# if not, crawl just this page!
 			print RESP "$link $link\n";
@@ -59,20 +61,28 @@
     my $contents = $response->content();
 
     open( CONTENTS, ">$localpath.contents" ) || die( "Could not open contents file $localpath.contents\n" );
-    # if you don't want content to be indexed, include it in
-    # <noindex> foobar </noindex> tags or surround it with comments
-    # <!-- noindex --> foobar <!-- /noindex -->
-    $contents =~ s,<noindex>.+?</noindex>,,isg;
-    $contents =~ s,<!--\s*noindex\s*-->.+?<!--\s*/noindex\s*-->,,isg;
-    # this will remove all script from indexing content
-    $contents =~ s,<script>.+?</script>,,isg;
-    # remap Windows charset to ISO-8859-2
-    $contents =~ tr/氿炶鎶袔绕/桂捐姗挟绕/;	# 1250 -> iso8859-2
-    # this will fix badly formatted html in form:
-    # <head><title>some text</title
-    # ></head>
-    # which will confuse indexer (or libxml2?)
-    $contents =~ s/[\n\r]^(>)/$1\n/msg;
+    # fixup just HTML files
+    if ($response->header("content-type") =~ "text/html") {
+	# if you don't want content to be indexed, include it in
+	# <noindex> foobar </noindex> tags or surround it with comments
+	# <!-- noindex --> foobar <!-- /noindex -->
+	# <!-- noindex --> foobar <!-- index --> (also supported by swish)
+	$contents =~ s,<noindex>.+?</noindex>,,isg;
+	$contents =~ s,<!--\s*noindex\s*-->.+?<!--\s*/noindex\s*-->,,isg;
+	$contents =~ s,<!--\s*noindex\s*-->.+?<!--\s*index\s*-->,,isg;
+	# this will remove all script from indexing content
+	$contents =~ s,<script>.+?</script>,,isg;
+	# remap Windows charset to ISO-8859-2
+	$contents =~ tr/氿炶鎶袔绕/桂捐姗挟绕/;	# 1250 -> iso8859-2
+	# this will fix badly formatted html in form:
+	# <head><title>some text</title
+	# ></head>
+	# which will confuse indexer (or libxml2?)
+	$contents =~ s/[\n\r]^(>)/$1\n/msg;
+	# remove comments between <html> and <head> texi2html inserts them
+	# there and swish can't find document title then (libxml or swish bug?)
+	while ($contents =~ s/(<html>.*)<!--.*?-->(.*<head>)/$1$2/msi) { };
+    }
     print CONTENTS $contents;
     close( CONTENTS );
 
@@ -88,8 +98,8 @@
 
 sub linkcb {
     my($tag, %links) = @_;
-    if (($tag eq "a" || $tag eq "area") && ($links{"href"})) {
-	my $link = $links{"href"};
+    if (($tag eq "a" || $tag eq "area") && ($links{"href"}) || ($tag eq "frame" && $links{"src"})) {
+	my $link = $links{"href"} || $links{"src"};
 
 	#
 	# Remove fragments
@@ -113,7 +123,7 @@
 
 	if ($no_parent_url) {
 			if ($link =~ m/$no_parent_url/) {
-				print LINKS "$link $no_parent_url\n";
+				print LINKS "$no_parent_url $link\n";
 #				print STDERR "using $link\n";
 #			} else {
 #				print STDERR "skipping $link\n";
