X-Git-Url: https://git.deb.at/?a=blobdiff_plain;ds=sidebyside;f=bin%2Fparse-packages;h=399a62ca060656a82314f4f80000a7462899b938;hb=fcd2a8e03040589f3048e91737d4b2863c8521fd;hp=38c83e988fbfcc8e890d44a30bbe47b5030c5708;hpb=3210a2b7d904e95b48ebbd84afa2bf1875544e8e;p=deb%2Fpackages.git

diff --git a/bin/parse-packages b/bin/parse-packages
index 38c83e9..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 $MAX_PACKAGE_POSTFIXES = 100;
@@ -184,8 +186,13 @@ tie %descriptions_db, "DB_File", "$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];
 }