X-Git-Url: https://code.th-h.de/?p=usenet%2Fnewsstats.git;a=blobdiff_plain;f=NewsStats.pm;h=cc3dd83d85629a70a41f9702b151646279483912;hp=35e0c90054fb136c153faafecd6c97a6c85ee96d;hb=d3b6810d3d09c7a1dffaef75731624fd9712d914;hpb=78389b28e945178cad3f43a63e5d1091334e8dab;ds=sidebyside

diff --git a/NewsStats.pm b/NewsStats.pm
index 35e0c90..cc3dd83 100644
--- a/NewsStats.pm
+++ b/NewsStats.pm
@@ -31,15 +31,16 @@ require Exporter;
   SplitPeriod
   ListMonth
   ListNewsgroups
+  ReadGroupList
   OutputData
   FormatOutput
   SQLHierarchies
   SQLGroupList
-  GetMaxLenght
+  GetMaxLength
 );
 %EXPORT_TAGS = ( TimePeriods => [qw(GetTimePeriod LastMonth CheckMonth SplitPeriod ListMonth)],
                  Output      => [qw(OutputData FormatOutput)],
-                 SQLHelper   => [qw(SQLHierarchies SQLGroupList GetMaxLenght)]);
+                 SQLHelper   => [qw(SQLHierarchies SQLGroupList GetMaxLength)]);
 $VERSION = '0.01';
 our $PackageVersion = '0.01';
 
@@ -154,15 +155,26 @@ sub ListNewsgroups {
 ### explode a (scalar) list of newsgroup names to a list of newsgroup and
 ### hierarchy names where every newsgroup and hierarchy appears only once:
 ### de.alt.test,de.alt.admin -> de.ALL, de.alt.ALL, de.alt.test, de.alt.admin
-### IN : $Newsgroups: a list of newsgroups (content of Newsgroups: header)
-### OUT: %Newsgroups: hash containing all newsgroup and hierarchy names as keys
-  my ($Newsgroups) = @_;
+### IN : $Newsgroups  : a list of newsgroups (content of Newsgroups: header)
+###      $TLH         : top level hierarchy (all other newsgroups are ignored)
+###      $ValidGroupsR: reference to a hash containing all valid newsgroups
+###                     as keys
+### OUT: %Newsgroups  : hash containing all newsgroup and hierarchy names as keys
+  my ($Newsgroups,$TLH,$ValidGroupsR) = @_;
+  my %ValidGroups = %{$ValidGroupsR} if $ValidGroupsR;
   my %Newsgroups;
   chomp($Newsgroups);
   # remove whitespace from contents of Newsgroups:
   $Newsgroups =~ s/\s//;
   # call &HierarchyCount for each newsgroup in $Newsgroups:
   for (split /,/, $Newsgroups) {
+    # don't count newsgroup/hierarchy in wrong TLH
+    next if($TLH and !/^$TLH/);
+    # don't count invalid newsgroups
+    if(%ValidGroups and !defined($ValidGroups{$_})) {
+      warn (sprintf("DROPPED: %s\n",$_));
+      next;
+    }
     # add original newsgroup to %Newsgroups
     $Newsgroups{$_} = 1;
     # add all hierarchy elements to %Newsgroups, amended by '.ALL',
@@ -194,6 +206,26 @@ sub ParseHierarchies {
   return @Hierarchies;
 };
 
+################################################################################
+sub ReadGroupList {
+################################################################################
+### read a list of valid newsgroups from file (each group on one line,
+### ignoring everything after the first whitespace and so accepting files
+### in checkgroups format as well as (parts of) an INN active file)
+### IN : $Filename    : file to read
+### OUT: \%ValidGroups: hash containing all valid newsgroups
+  my ($Filename) = @_;
+  my %ValidGroups;
+  open (my $LIST,"<$Filename") or die "$MySelf: E: Cannot read $Filename: $!\n";
+  while (<$LIST>) {
+    s/^(\S+).*$/$1/;
+    chomp;
+    $ValidGroups{$_} = '1';
+  };
+  close $LIST;
+  return \%ValidGroups;
+};
+
 ################################################################################
 
 #####----------------------------- TimePeriods ----------------------------#####
@@ -380,7 +412,7 @@ sub SQLHierarchies {
 };
 
 ################################################################################
-sub GetMaxLenght {
+sub GetMaxLength {
 ################################################################################
 ### get length of longest field in future query result
 ### IN : $DBHandle   : database handel