(get_filesize_from_synthesis) read back real filesizes from synthesis

(filter_existing_hdlist) readd back real filesizes read from synthesis

else, since hdlist no more have filesizes since 2008, URPM will put a
bogus estimation in newly generated synthesis

2 files changed, 28 insertions, 4 deletions

diff --git a/NEWS b/NEWS
index fcf7aa6..46adfbc 100644
--- a/NEWS
+++ b/NEWS
@@ -1,3 +1,5 @@
+- keep real filesizes in synthesis
+
 Version 6.1 - 09 September 2011, by Thierry Vignaud
 
 - force locale to be C (mga#95, Pascal Terjan)
diff --git a/genhdlist2 b/genhdlist2
index 325456f..a77e510 100644
--- a/genhdlist2
+++ b/genhdlist2
@@ -115,7 +115,7 @@ sub do_it {
     }
     my @xml_media_info = $options{xml_info} ? ('info', 'files', 'changelog') : ();
 
-    build($urpm, \%rpms_todo, $media_info_dir, $rpms_dir, \@xml_media_info, $options{no_incremental}, $options{no_hdlist}, $xml_info_suffix, $xml_info_filter);
+    build($urpm, \%rpms_todo, $media_info_dir, $rpms_dir, \@xml_media_info, $options{no_incremental}, $options{no_hdlist}, $xml_info_suffix, $xml_info_filter, $synthesis);
     build_synthesis($urpm, "$synthesis.tmp", $synthesis_filter);
 
     if (1) {
@@ -162,7 +162,7 @@ sub read_file_deps {
 }
 
 sub build {
-    my ($urpm, $rpms_todo, $media_info_dir, $rpms_dir, $xml_media_info, $b_no_incremental, $b_no_hdlist, $xml_info_suffix, $xml_info_filter) = @_;
+    my ($urpm, $rpms_todo, $media_info_dir, $rpms_dir, $xml_media_info, $b_no_incremental, $b_no_hdlist, $xml_info_suffix, $xml_info_filter, $synthesis) = @_;
 
     my $hdlist = "$media_info_dir/hdlist.cz";
 
@@ -181,9 +181,13 @@ sub build {
 	map { $_ => open_xml_filter("$media_info_dir/$_.xml${xml_info_suffix}.tmp", $xml_info_filter) } @$xml_media_info
     };
 
+    # keep real filesizes from synthesis since hdlist no more have filesizes since 2008
+    # else URPM will put a bogus estimation:
+    my $sizes = get_filesize_from_synthesis($urpm, $rpms_todo, $synthesis);
+
     if (-e $hdlist && !$b_no_incremental) {
 	print "filtering $hdlist into hdlist.cz.tmp\n" if $verbose >= 0;
-	filter_existing_hdlist($urpm, $rpms_todo, $hdlist, $out);
+	filter_existing_hdlist($urpm, $rpms_todo, $hdlist, $out, $sizes);
     }
 
     add_new_rpms_to_hdlist($urpm, $rpms_todo, $out, $rpms_dir);
@@ -191,8 +195,23 @@ sub build {
     close_xml($out->{$_}) foreach @$xml_media_info;
 }
 
+sub get_filesize_from_synthesis {
+    my ($urpm, $rpms_todo, $synthesis) = @_;
+
+    return {} if !-r $synthesis;
+
+    my %sizes;
+    $urpm->parse_synthesis($synthesis, packing => 1, callback => sub {
+	my (undef, $pkg) = @_;
+	my $fullname = $pkg->fullname;
+	$sizes{$fullname} = $pkg->filesize if $rpms_todo->{$fullname};
+	0; # don't keep in memory
+    });
+    \%sizes;
+}
+
 sub filter_existing_hdlist {
-    my ($urpm, $rpms_todo, $in_hdlist, $out) = @_;
+    my ($urpm, $rpms_todo, $in_hdlist, $out, $sizes) = @_;
 
     if ($urpm->parse_hdlist($in_hdlist, packing => 1, callback => sub {
 	my (undef, $pkg) = @_;
@@ -200,6 +219,9 @@ sub filter_existing_hdlist {
 	if (delete $rpms_todo->{$fullname}) {
 	    print "keeping $fullname\n" if $verbose > 1;
 	    add_pkg($out, $pkg);
+	    # readd back real filesize from synthesis since hdlist no more have filesizes since 2008
+	    # else URPM will put a bogus estimation:
+	    $pkg->set_filesize($sizes->{$fullname}) if $sizes->{$fullname};
 	    1; # do keep in memory
 	} else {
 	    print "removing $fullname\n" if $verbose > 0;