gatherstats.pl: Move TLH check to NewsStats.pm.

[usenet/newsstats.git] / NewsStats.pm

diff --git a/NewsStats.pm b/NewsStats.pm
index 67b0dbc..781272d 100644 (file)
--- a/NewsStats.pm
+++ b/NewsStats.pm
@@ -31,6 +31,7 @@ require Exporter;
   SplitPeriod
   ListMonth
   ListNewsgroups
   SplitPeriod
   ListMonth
   ListNewsgroups

+  ReadGroupList

   OutputData
   FormatOutput
   SQLHierarchies
   OutputData
   FormatOutput
   SQLHierarchies


@@ -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
 ### 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 /,/, $Newsgroups) {
   my %Newsgroups;
   chomp($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',
     # add original newsgroup to %Newsgroups
     $Newsgroups{$_} = 1;
     # add all hierarchy elements to %Newsgroups, amended by '.ALL',


@@ -194,6 +206,26 @@ sub ParseHierarchies {
   return @Hierarchies;
 };
 
   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 ----------------------------#####
 ################################################################################
 
 #####----------------------------- TimePeriods ----------------------------#####


@@ -303,14 +335,27 @@ sub ListMonth {
 sub OutputData {
 ################################################################################
 ### read database query results from DBHandle and print results with formatting
 sub OutputData {
 ################################################################################
 ### read database query results from DBHandle and print results with formatting

-### IN : $Format : format specifier
-###      $DBQuery: database query handle with executed query,
-###                containing $Month, $Key, $Value
+### IN : $Format  : format specifier
+###      $FileName: file name template (-f): filename-YYYY-MM
+###      $DBQuery : database query handle with executed query,
+###                 containing $Month, $Key, $Value

 ###      $PadGroup: padding length for newsgroups field (optional) for 'pretty'
 ###      $PadGroup: padding length for newsgroups field (optional) for 'pretty'

-  my ($Format, $DBQuery,$PadGroup) = @_;
+  my ($Format, $FileName, $DBQuery, $PadGroup) = @_;
+  my ($Handle, $OUT);
+  our $LastIteration;

   while (my ($Month, $Key, $Value) = $DBQuery->fetchrow_array) {
   while (my ($Month, $Key, $Value) = $DBQuery->fetchrow_array) {

-    print &FormatOutput($Format, $Month, $Key, $Value, $PadGroup);
+    # set output file handle
+    if (!$FileName) {
+      $Handle = *STDOUT{IO}; # set $Handle to a reference to STDOUT
+    } elsif (!defined($LastIteration) or $LastIteration ne $Month) {
+      close $OUT if ($LastIteration);
+      open ($OUT,sprintf('>%s-%s',$FileName,$Month)) or die sprintf("$MySelf: E: Cannot open output file '%s-%s': $!\n",$FileName,$Month);
+      $Handle = $OUT;
+    };
+    print $Handle &FormatOutput($Format, $Month, $Key, $Value, $PadGroup);
+    $LastIteration = $Month;

   };
   };

+  close $OUT if ($FileName);

 };
 
 ################################################################################
 };
 
 ################################################################################