X-Git-Url: https://git.deb.at/?a=blobdiff_plain;f=bin%2Fparse-packages;h=399a62ca060656a82314f4f80000a7462899b938;hb=27d33ebb54f354d5d0d1fb68c94ff6a5682b54c3;hp=7de51b6e89d0780a0d2186051976046cfb62528c;hpb=dc9512b1309f4c8c6bd1171b543183e8ac8b2115;p=deb%2Fpackages.git

diff --git a/bin/parse-packages b/bin/parse-packages
index 7de51b6..399a62c 100755
--- a/bin/parse-packages
+++ b/bin/parse-packages
@@ -22,6 +22,8 @@
 use strict;
 use lib './lib';
 
+$| = 1;
+
 my $what = $ARGV[0] ? "non-free" : "*";
 # max. distinct results for a given package postfix
 my $MAX_PACKAGE_POSTFIXES = 100;
@@ -130,7 +132,7 @@ for my $archive (@ARCHIVES) {
 
 	foreach (keys %virtual_packages) {
 	    my @provided_by = keys %{$virtual_packages{$_}};
-	    $packages_small{$_} .= "$archive $suite virtual - - - - @provided_by\000";
+	    $packages_small{$_} .= "$archive $suite virtual v v v v @provided_by\000";
 	}
 
 	untie %packages_all_db;
@@ -184,8 +186,13 @@ tie %descriptions_db, "DB_File", "$DBDIR/descriptions.db.new",
 open DESCR, ">", "$DBDIR/descriptions.txt" or die "Error creating descriptions textfile";
 for (my $i=1; $i<= $#descriptions; $i++) {
 	my $plain_description = $descriptions[$i];
-	$plain_description =~ s/\n .\n/ /og;
-	$plain_description =~ s/[\n \t]+/ /og;
+# WARNING: This needs to correspond with what happens in
+# Packages/Search.pm:do_fulltext_search
+	$plain_description =~ tr [A-Z] [a-z];
+	# ensure one space on both ends
+	$plain_description = " $plain_description ";
+	$plain_description =~ s/[(),.-]+//og;
+	$plain_description =~ s#[^a-z0-9_/+]+# #og;
 	print DESCR "$plain_description\n";
 	$descriptions_db{$i} = $descriptions[$i];
 }