X-Git-Url: https://code.th-h.de/?p=usenet%2Fnewsstats.git;a=blobdiff_plain;f=groupstats.pl;h=e21aeb78f5380b455ede631a4ff7319508d019a0;hp=a09c632890e36296796b01a663eb6ed23bd9028f;hb=f2ddfd8a923d3276c00daf243a0abf1cbdd8b35b;hpb=d8695b1c044f9a99f06a7943f88fe2191e6cc52a

diff --git a/groupstats.pl b/groupstats.pl
index a09c632..e21aeb7 100755
--- a/groupstats.pl
+++ b/groupstats.pl
@@ -26,7 +26,7 @@ use DBI;
 ################################# Main program #################################
 
 ### read commandline options
-my %Options = &ReadOptions('m:p:n:o:t:l:b:iscqdg:');
+my %Options = &ReadOptions('m:p:an:o:t:l:b:iscqdg:');
 
 ### read configuration
 my %Conf = %{ReadConfig('newsstats.conf')};
@@ -71,51 +71,72 @@ if ($Options{'l'}) {
   warn ("$MySelf: W: Output type forced to '-o pretty' due to usage of '-l'.\n");
 };
 
-### get time period
-my ($StartMonth,$EndMonth) = &GetTimePeriod($Options{'m'},$Options{'p'});
-# reset to one month for 'dump' output type
-if ($Options{'o'} eq 'dump' and $Options{'p'}) {
-  warn ("$MySelf: W: You cannot combine time periods (-p) with '-o dump', changing output type to '-o pretty'.\n");
-  $Options{'o'} = 'pretty';
-};
-
 ### init database
 my $DBHandle = InitDB(\%Conf,1);
 
+### get time period
+my ($StartMonth,$EndMonth);
+# if '-a' is set, set start/end month from database
+# FIXME - it doesn't make that much sense to get first/last month from database to query it
+#         with a time period that equals no time period ...
+if ($Options{'a'}) {
+  undef($Options{'m'});
+  undef($Options{'p'});
+  my $DBQuery = $DBHandle->prepare(sprintf("SELECT MIN(month),MAX(month) FROM %s.%s",$Conf{'DBDatabase'},$Conf{'DBTableGrps'}));
+  $DBQuery->execute or die sprintf("$MySelf: E: Can't get MIN/MAX month from %s.%s: %s\n",$Conf{'DBDatabase'},$Conf{'DBTableGrps'},$DBI::errstr);
+  ($StartMonth,$EndMonth) = $DBQuery->fetchrow_array;
+} else {
+  ($StartMonth,$EndMonth) = &GetTimePeriod($Options{'m'},$Options{'p'});
+};
+# if time period is more than one month: set output type to '-o pretty' or '-o dumpgroup'
+if ($Options{'o'} eq 'dump' and ($Options{'p'} or $Options{'a'})) {
+  if (defined($Options{'n'}) and $Options{'n'} !~ /:|\*/) {
+   warn ("$MySelf: W: You cannot combine time periods (-p) with '-o dump', changing output type to '-o dumpgroup'.\n");
+   $Options{'o'} = 'dumpgroup';
+  } else {
+   warn ("$MySelf: W: You cannot combine time periods (-p) with '-o dump', changing output type to '-o pretty'.\n");
+   $Options{'o'} = 'pretty';
+  }
+};
+
 ### create report
 # get list of newsgroups (-n)
-my ($QueryPart,@GroupList);
+my ($QueryGroupList,$QueryThreshold,@GroupList,@Params);
 my $Newsgroups = $Options{'n'};
 if ($Newsgroups) {
   # explode list of newsgroups for WHERE clause
-  ($QueryPart,@GroupList) = &SQLGroupList($Newsgroups);
+  ($QueryGroupList,@GroupList) = &SQLGroupList($Newsgroups);
 } else {
   # set to dummy value (always true)
-  $QueryPart = 1;
+  $QueryGroupList = 1;
 };
 
 # manage thresholds
 if (defined($Options{'t'})) {
   if ($Options{'i'}) {
     # -i: list groups below threshold
-    $QueryPart .= ' AND postings < ?';
+    $QueryThreshold .= ' postings < ?';
   } else {
     # default: list groups above threshold
-    $QueryPart .= ' AND postings > ?';
+    $QueryThreshold .= ' postings > ?';
   };
-  # push threshold to GroupList to match number of binding vars for DBQuery->execute
-  push @GroupList,$Options{'t'};
+  # push threshold to Params
+  push @Params,$Options{'t'};
+} else {
+  # set to dummy value (always true)
+  $QueryThreshold = 1;  
 }
 
 # construct WHERE clause
-# $QueryPart is "list of newsgroup" (or 1),
+# $QueryGroupList is "list of newsgroup" (or 1),
+# $QueryThreshold is threshold definition (or 1),
 # &SQLHierarchies() takes care of the exclusion of hierarchy levels (.ALL)
 # according to setting of -s
-my $WhereClause = sprintf('month BETWEEN ? AND ? AND %s %s',$QueryPart,&SQLHierarchies($Options{'s'}));
+my $WhereClause = sprintf('month BETWEEN ? AND ? AND %s AND %s %s',$QueryGroupList,$QueryThreshold,&SQLHierarchies($Options{'s'}));
 
-# get lenght of longest newsgroup delivered by query for formatting purposes
+# get length of longest newsgroup delivered by query for formatting purposes
 # FIXME
-my $MaxLength = &GetMaxLenght($DBHandle,$Conf{'DBTableGrps'},'newsgroup',$WhereClause,$StartMonth,$EndMonth,@GroupList);
+my $MaxLength = &GetMaxLenght($DBHandle,$Conf{'DBTableGrps'},'newsgroup',$WhereClause,$StartMonth,$EndMonth,(@GroupList,@Params));
 
 my ($OrderClause,$DBQuery);
 # -b (best of / top list) defined?
@@ -130,7 +151,7 @@ if (!defined($Options{'b'}) and !defined($Options{'l'})) {
   $DBQuery = $DBHandle->prepare(sprintf("SELECT month,newsgroup,postings FROM %s.%s WHERE %s ORDER BY month,%s",$Conf{'DBDatabase'},$Conf{'DBTableGrps'},$WhereClause,$OrderClause));
 } elsif ($Options{'b'}) {
   # -b is set (then -l can't be!)
-  # set sorting order (-i)
+  # set sorting order (-i): top or flop list?
   if ($Options{'i'}) {
     $OrderClause = 'postings';
   } else {
@@ -138,35 +159,45 @@ if (!defined($Options{'b'}) and !defined($Options{'l'})) {
   };
   # set -b to 10 if < 1 (Top 10)
   $Options{'b'} = 10 if $Options{'b'} !~ /^\d*$/ or $Options{'b'} < 1;
-  # push LIMIT to GroupList to match number of binding vars for DBQuery->execute
-  push @GroupList,$Options{'b'};
+  # push LIMIT to Params
+  push @Params,$Options{'b'};
   # prepare query: get sum of postings per group from groups table for given months and newsgroups with LIMIT
   $DBQuery = $DBHandle->prepare(sprintf("SELECT newsgroup,SUM(postings) AS postings FROM %s.%s WHERE %s GROUP BY newsgroup ORDER BY %s,newsgroup LIMIT ?",$Conf{'DBDatabase'},$Conf{'DBTableGrps'},$WhereClause,$OrderClause));
 } else {
   # -l must be set now, as all other cases have been taken care of
-  # set sorting order (-i)
+  # which kind of level (-i): more than -l x or less than -l x?
+  my ($Level);
   if ($Options{'i'}) {
-    $OrderClause = '<';
+    $Level = '<';
   } else {
-    $OrderClause = '>';
+    $Level = '>';
+  };
+  # prepare and execute query: get list of newsgroups meeting level condition
+  $DBQuery = $DBHandle->prepare(sprintf("SELECT newsgroup FROM %s.%s WHERE %s GROUP BY newsgroup HAVING MAX(postings) %s ?",$Conf{'DBDatabase'},$Conf{'DBTableGrps'},$WhereClause,$Level));
+  $DBQuery->execute($StartMonth,$EndMonth,@GroupList,$Options{'l'})
+    or die sprintf("$MySelf: E: Can't get groups data for %s to %s from %s.%s: %s\n",$StartMonth,$EndMonth,$Conf{'DBDatabase'},$Conf{'DBTableGrps'},$DBI::errstr);
+  # add newsgroups to a comma-seperated list ready for IN(...) query
+  my $GroupList;
+  while (my ($Newsgroup) = $DBQuery->fetchrow_array) {
+    $GroupList .= ',' if (defined($GroupList) and $GroupList ne '');
+    $GroupList .= "'$Newsgroup'";
   };
-  # push level and $StartMonth,$EndMonth - again - to GroupList to match number of binding vars for DBQuery->execute
-  # FIXME -- together with the query (see below)
-  push @GroupList,$Options{'l'};
-  push @GroupList,$StartMonth,$EndMonth;
-  # prepare query: get number of postings per group from groups table for given months and 
-  # FIXME -- this query is ... in dire need of impromevent
-  $DBQuery = $DBHandle->prepare(sprintf("SELECT month,newsgroup,postings FROM %s.%s WHERE newsgroup IN (SELECT newsgroup FROM %s.%s WHERE %s GROUP BY newsgroup HAVING MAX(postings) %s ?) AND %s ORDER BY newsgroup,month",$Conf{'DBDatabase'},$Conf{'DBTableGrps'},$Conf{'DBDatabase'},$Conf{'DBTableGrps'},$WhereClause,$OrderClause,$WhereClause));
+  $DBQuery = $DBHandle->prepare(sprintf("SELECT month,newsgroup,postings FROM %s.%s WHERE newsgroup IN (%s) AND %s ORDER BY newsgroup,month",$Conf{'DBDatabase'},$Conf{'DBTableGrps'},$GroupList,$WhereClause));
 };
 
 # execute query
-$DBQuery->execute($StartMonth,$EndMonth,@GroupList)
+$DBQuery->execute($StartMonth,$EndMonth,@GroupList,@Params)
   or die sprintf("$MySelf: E: Can't get groups data for %s to %s from %s.%s: %s\n",$StartMonth,$EndMonth,$Conf{'DBDatabase'},$Conf{'DBTableGrps'},$DBI::errstr);
 
 # output results
 # print caption (-c) with time period if -m or -p is set
-# FIXME - month or period should handled differently
-printf ("----- Report from %s to %s\n",$StartMonth,$EndMonth) if $Options{'c'} and ($Options{'m'} or $Options{'p'});
+if ($Options{'c'}) {
+  if ($Options{'p'}) {
+    printf ("----- Report from %s to %s\n",$StartMonth,$EndMonth);
+  } elsif ($Options{'m'}) {
+    printf ("----- Report for %s\n",$StartMonth);
+  };
+};
 # print caption (-c) with newsgroup list if -n is set
 printf ("----- Newsgroups: %s\n",join(',',split(/:/,$Newsgroups))) if $Options{'c'} and $Options{'n'};
 # print caption (-c) with threshold if -t is set, taking -i in account
@@ -178,16 +209,16 @@ if (!defined($Options{'b'})  and !defined($Options{'l'})) {
   # -b is set (then -l can't be!)
   # we have to read in the query results ourselves, as they do not have standard layout
   while (my ($Newsgroup,$Postings) = $DBQuery->fetchrow_array) {
-    # we just assign "top x" or "bottom x" instead of a month for the caption
-    # FIXME
-    print &FormatOutput($Options{'o'}, ($Options{'i'} ? 'Bottom ' : 'Top ').$Options{'b'}, $Newsgroup, $Postings, $MaxLength);
+    # we just assign "top x" or "bottom x" instead of a month for the caption and force an output type of pretty
+    print &FormatOutput('pretty', ($Options{'i'} ? 'Bottom ' : 'Top ').$Options{'b'}, $Newsgroup, $Postings, $MaxLength);
   };
 } else {
   # -l must be set now, as all other cases have been taken care of
+  # print caption (-c) with level, taking -i in account
+  printf ("----- Newsgroups with %s than %u postings over the whole time period\n",$Options{'i'} ? 'less' : 'more',$Options{'l'}) if $Options{'c'};
   # we have to read in the query results ourselves, as they do not have standard layout
   while (my ($Month,$Newsgroup,$Postings) = $DBQuery->fetchrow_array) {
     # we just switch $Newsgroups and $Month for output generation
-    # FIXME
     print &FormatOutput($Options{'o'}, $Newsgroup, $Month, $Postings, 7);
   };
 };
@@ -205,7 +236,7 @@ groupstats - create reports on newsgroup usage
 
 =head1 SYNOPSIS
 
-B<groupstats> [B<-Vhiscqd>] [B<-m> I<YYYY-MM>] [B<-p> I<YYYY-MM:YYYY-MM>] [B<-n> I<newsgroup(s)>] [B<-t> I<threshold>] [B<-l> I<level>] [B<-b> I<number>] [B<-o> I<output type>] [B<-g> I<database table>]
+B<groupstats> [B<-Vhiscqd>] [B<-m> I<YYYY-MM> | B<-p> I<YYYY-MM:YYYY-MM> | B<-a>] [B<-n> I<newsgroup(s)>] [B<-t> I<threshold>] [B<-l> I<level>] [B<-b> I<number>] [B<-o> I<output type>] [B<-g> I<database table>]
 
 =head1 REQUIREMENTS
 
@@ -263,7 +294,7 @@ below). Captions can be added by setting the B<-c> switch.
 
 =head2 Configuration
 
-F<groupstats.pl> will read its configuration from F<newsstats.conf>
+B<groupstats> will read its configuration from F<newsstats.conf>
 which should be present in the same directory via Config::Auto.
 
 See doc/INSTALL for an overview of possible configuration options.
@@ -285,12 +316,18 @@ Print this man page and exit.
 =item B<-m> I<YYYY-MM> (month)
 
 Set processing period to a month in YYYY-MM format. Ignored if B<-p>
-is set.
+or B<-a> is set.
 
 =item B<-p> I<YYYY-MM:YYYY-MM> (period)
 
 Set processing period to a time period between two month, each in
-YYYY-MM format, separated by a colon. Overrides B<-m>.
+YYYY-MM format, separated by a colon. Overrides B<-m>. Ignored if
+B<-a> is set.
+
+=item B<-a> (all)
+
+Set no processing period (process whole database). Overrides B<-m>
+and B<-p>.
 
 =item B<-n> I<newsgroup(s)> (newsgroups)