Merge branch 'gatherstats' into next

NewsStats.pm

@@ -184,7 +184,7 @@ sub ListNewsgroups {
     next if($TLH and !/^$TLH/);
     # don't count invalid newsgroups
     if(%ValidGroups and !defined($ValidGroups{$_})) {
-      &Bleat(1,sprintf("DROPPED: %s",$_));
+      warn (sprintf("DROPPED: %s\n",$_));
       next;
     }
     # add original newsgroup to %Newsgroups
@@ -230,8 +230,9 @@ sub ReadGroupList {
   my %ValidGroups;
   open (my $LIST,"<$Filename") or &Bleat(2,"Cannot read $Filename: $!");
   while (<$LIST>) {
-    s/^(\S+).*$/$1/;
+    s/^\s*(\S+).*$/$1/;
     chomp;
+    next if /^$/;
     $ValidGroups{$_} = '1';
   };
   close $LIST;

gatherstats.pl

@@ -78,28 +78,30 @@ my $TLH;
 if ($Conf{'TLH'}) {
   # $Conf{'TLH'} is parsed as an array by Config::Auto;
   # make a flat list again, separated by :
-  if (ref($TLH) eq 'ARRAY') {
+  if (ref($Conf{'TLH'}) eq 'ARRAY') {
     $TLH = join(':',@{$Conf{'TLH'}});
   } else {
     $TLH  = $Conf{'TLH'};
   }
   # strip whitespace
   $TLH =~ s/\s//g;
+  # add trailing dots if none are present yet
+  # (using negative look-behind assertions)
+  $TLH =~ s/(?<!\.):/.:/g;
+  $TLH =~ s/(?<!\.)$/./;
   # check for illegal characters
   &Bleat(2,'Config error - illegal characters in TLH definition!')
-    if ($TLH !~ /^[a-zA-Z0-9:]+$/);
+    if ($TLH !~ /^[a-zA-Z0-9:+.-]+$/);
+  # escape dots
+  $TLH =~ s/\./\\./g;
   if ($TLH =~ /:/) {
     # reformat $TLH from a:b to (a)|(b),
-    # e.g. replace '.' by '|'
+    # e.g. replace ':' by ')|('
     $TLH =~ s/:/)|(/g;
     $TLH = '(' . $TLH . ')';
   };
 };
 
-# read list of newsgroups from --checkgroups
-# into a hash
-my %ValidGroups = %{ReadGroupList($OptCheckgroupsFile)} if $OptCheckgroupsFile;
-
 ### init database
 my $DBHandle = InitDB(\%Conf,1);
 
@@ -110,6 +112,11 @@ foreach my $Month (&ListMonth($Period)) {
   print "---------- $Month ----------\n" if $OptDebug;
 
   if ($OptStatsType eq 'all' or $OptStatsType eq 'groups') {
+    # read list of newsgroups from --checkgroups
+    # into a hash
+    my %ValidGroups = %{ReadGroupList(sprintf('%s-%s',$OptCheckgroupsFile,$Month))}
+      if $OptCheckgroupsFile;
+
     ### ----------------------------------------------
     ### get groups data (number of postings per group)
     # get groups data from raw table for given month
@@ -125,7 +132,7 @@ foreach my $Month (&ListMonth($Period)) {
     # count postings per group
     my %Postings;
     while (($_) = $DBQuery->fetchrow_array) {
-      # get list oft newsgroups and hierarchies from Newsgroups:
+      # get list of newsgroups and hierarchies from Newsgroups:
       my %Newsgroups = ListNewsgroups($_,$TLH,
                                       $OptCheckgroupsFile ? \%ValidGroups : '');
       # count each newsgroup and hierarchy once
@@ -138,8 +145,15 @@ foreach my $Month (&ListMonth($Period)) {
     if (%ValidGroups) {
       foreach (sort keys %ValidGroups) {
         if (!defined($Postings{$_})) {
-          $Postings{$_} = 0 ;
+          # expand newsgroup with hierarchies
+          my @Newsgroups = ParseHierarchies($_);
+          # add each empty newsgroup and empty hierarchies, too, as needed
+          foreach (@Newsgroups) {
+            if (!defined($Postings{$_})) {
+              $Postings{$_} = 0;
               warn (sprintf("ADDED: %s as empty group\n",$_));
+            };
+          };
         }
       };
     };
@@ -189,7 +203,7 @@ gatherstats - process statistical data from a raw source
 
 =head1 SYNOPSIS
 
-B<gatherstats> [B<-Vhdt>] [B<-m> I<YYYY-MM> | I<YYYY-MM:YYYY-MM>] [B<-s> I<stats] [B<-c> I<checkgroups file>]] [B<--hierarchy> I<TLH>] [B<--rawdb> I<database table>] [B<-groupsdb> I<database table>] [B<--clientsdb> I<database table>] [B<--hostsdb> I<database table>]
+B<gatherstats> [B<-Vhdt>] [B<-m> I<YYYY-MM> | I<YYYY-MM:YYYY-MM>] [B<-s> I<stats] [B<-c> I<filename template>]] [B<--hierarchy> I<TLH>] [B<--rawdb> I<database table>] [B<-groupsdb> I<database table>] [B<--clientsdb> I<database table>] [B<--hostsdb> I<database table>]
 
 =head1 REQUIREMENTS
 
@@ -283,15 +297,23 @@ Set processing type to one of I<all> and I<groups>. Defaults to all
 (and is currently rather pointless as only I<groups> has been
 implemented).
 
-=item B<-c>, B<--checkgroups> I<filename>
+=item B<-c>, B<--checkgroups> I<filename template>
 
-Check each group against a list of valid newsgroups read from
+Check each group against a list of valid newsgroups read from a file,
-I<filename>, one group on each line and ignoring everything after the
+one group on each line and ignoring everything after the first
-first whitespace (so you can use a file in checkgroups format or (part
+whitespace (so you can use a file in checkgroups format or (part of)
-of) your INN active file).
+your INN active file).
 
-Newsgroups not found in I<filename> will be dropped (and logged to
+The filename is taken from I<filename template>, amended by each B<--
-STDERR), and newsgroups found in I<filename> but having no postings
+month> B<gatherstats> is processing, so that
+
+    gatherstats -m 2010-01:2010-12 -c checkgroups
+
+will check against F<checkgroups-2010-01> for January 2010, against
+F<checkgroups-2010-02> for February 2010 and so on.
+
+Newsgroups not found in the checkgroups file will be dropped (and
+logged to STDERR), and newsgroups found there but having no postings
 will be added with a count of 0 (and logged to STDERR).
 
 =item B<--hierarchy> I<TLH> (newsgroup hierarchy)
@@ -335,9 +357,9 @@ Process all types of information for January of 2010:
     gatherstats --month 2010-01
 
 Process only number of postings for the year of 2010,
-checking against checkgroups-2010.txt:
+checking against checkgroups-*:
 
-    gatherstats -m 2010-01:2010-12 -s groups -c checkgroups-2010.txt
+    gatherstats -m 2010-01:2010-12 -s groups -c checkgroups
 
 =head1 FILES