--- trunk/perl/scripts/est-spider	2005/09/17 23:07:52	27
+++ trunk/perl/scripts/est-spider	2005/09/17 23:43:20	28
@@ -17,12 +17,13 @@
 my $exclude;
 
 #$verbose = 1;
+my $debug = 0;
 
 my $result = GetOptions(
 	"collection=s" => \$collection,
 	"path=s" => \$path_add,
 	"verbose!" => \$verbose,
-	"debug!" => \$verbose,
+	"debug!" => \$debug,
 	"exclude=s" => \$exclude,
 	"node=s" => \$node_url,
 );
@@ -133,12 +134,25 @@
 	my $path = $_;
 	my $contents;
 
-	return if (-l $path || $path =~ m#/.svn#);
+	return if (-l $path || $path =~ m#/.svn# || $path =~ m/(~|.bak)$/);
+
+	my $mtime = (stat($path))[9];
+	my $mtime_db = $db->get_doc_attr_by_uri("file:///$path", '@mtime') || -2;
+
+	if ($mtime == $mtime_db) {
+		print STDERR "# same: $path $mtime\n" if ($verbose);
+		return;
+	} else {
+		print STDERR "# changed: $path $mtime != $mtime_db\n" if ($debug);
+	}
+
+	# skip files on which File::MMagic::XS croaks
+	return if ($path =~ m#\.au$#);
 
 	my $type = $mm->checktype_filename($path);
 	$type =~ s/\s+/ /gs;
 
-	print STDERR "# $path $type\n" if ($verbose);
+	print STDERR "# $path $type\n" if ($debug);
 
 	if ($pdftotext && -f $path && $type =~ m/pdf/i) {
 
@@ -175,17 +189,16 @@
 			print STDERR " $page_nr" if ($verbose);
 			my $pre_tmp = $pre_html;
 			$pre_tmp =~ s/##page_nr##/$page_nr<\/title>/s;
-			dump_contents($db, $pre_tmp . $page . $post_html,time(), $path) if ($page !~ m/^\s*$/s);
+			dump_contents($db, $pre_tmp . $page . $post_html, $mtime, $path) if ($page !~ m/^\s*$/s);
 			$page_nr++;
 		}
 
 	} else {
 
 #		return if (! -f $path || ! m/\.(html*|php|pl|txt|info|log|text)$/i);
-		return if (! -f $path || (
-			$type !~ m/html/ ||
-			($type !~ m#text/plain# && m/\.(php|pl|txt|info|log|text)$/)
-		));
+		return unless (-f $path && $type =~ m/html/ ||
+			($type =~ m#text# && $path =~ m/\.(php|pl|txt|info|log|text)$/io)
+		);
 
 		# skip index files
 		return if (m/index_[a-z]\.html*/i || m/index_symbol\.html*/i);
@@ -202,7 +215,7 @@
 		# add optional components to path
 		$path .= " $path_add" if ($path_add);
 
-		dump_contents($db, $contents,time(), $path);
+		dump_contents($db, $contents, $mtime, $path);
 	}
 
 	print STDERR "\n" if ($verbose);