groupstats.pl: Add handling of incompatible command line options.

[usenet/newsstats.git] / groupstats.pl

#! /usr/bin/perl -W\r
#\r
# groupstats.pl\r
#\r
# This script will get statistical data on newgroup usage\r
# form a database.\r
# \r
# It is part of the NewsStats package.\r
#\r
# Copyright (c) 2010 Thomas Hochstein <thh@inter.net>\r
#\r
# It can be redistributed and/or modified under the same terms under \r
# which Perl itself is published.\r
\r
BEGIN {\r
  our $VERSION = "0.01";\r
  use File::Basename;\r
  push(@INC, dirname($0));\r
}\r
use strict;\r
\r
use NewsStats qw(:DEFAULT :TimePeriods :Output :SQLHelper);\r
\r
use DBI;\r
\r
################################# Definitions ##################################\r
\r
# ...\r
\r
################################# Main program #################################\r
\r
### read commandline options\r
my %Options = &ReadOptions('m:p:n:o:t:l:b:iscqdg:');\r
\r
### read configuration\r
my %Conf = %{ReadConfig('newsstats.conf')};\r
\r
### override configuration via commandline options\r
my %ConfOverride;\r
$ConfOverride{'DBTableGrps'}  = $Options{'g'} if $Options{'g'};\r
&OverrideConfig(\%Conf,\%ConfOverride);\r
\r
### check for incompatible command line options\r
# you can't mix '-t', '-b' and '-l'\r
# -b/-l take preference over -t, and -b takes preference over -l\r
if ($Options{'b'} or $Options{'l'}) {\r
  if ($Options{'t'}) {\r
    # drop -t\r
    warn ("$MySelf: W: You cannot combine thresholds (-t) and top lists (-b) or levels (-l). Threshold '-t $Options{'t'}' was ignored.\n");\r
    undef($Options{'t'});\r
  };\r
  if ($Options{'b'} and $Options{'l'}) {\r
    # drop -l\r
    warn ("$MySelf: W: You cannot combine top lists (-b) and levels (-l). Level '-l $Options{'l'}' was ignored.\n");\r
    undef($Options{'l'});\r
  };\r
  # -q/-d don't work with -b or -l\r
  warn ("$MySelf: W: Sorting by number of postings (-q) ignored due to top list mode (-b) / levels (-l).\n") if $Options{'q'};\r
  warn ("$MySelf: W: Reverse sorting (-d) ignored due to top list mode (-b) / levels (-l).\n") if $Options{'d'};\r
};\r
\r
### check output type\r
# default output type to 'dump'\r
$Options{'o'} = 'dump' if !$Options{'o'};\r
# fail if more than one newsgroup is combined with 'dumpgroup' type\r
die ("$MySelf: E: You cannot combine newsgroup lists (-n) with more than one group with '-o dumpgroup'!\n") if ($Options{'o'} eq 'dumpgroup' and defined($Options{'n'}) and $Options{'n'} =~ /:|\*/);\r
# accept 'dumpgroup' only with -n\r
if ($Options{'o'} eq 'dumpgroup' and !defined($Options{'n'})) {\r
  $Options{'o'} = 'dump';\r
  warn ("$MySelf: W: You must submit exactly one newsgroup ('-n news.group') for '-o dumpgroup'. Output type was set to 'dump'.\n");\r
};\r
# set output type to 'pretty' for -l\r
if ($Options{'l'}) {\r
  $Options{'o'} = 'pretty';\r
  warn ("$MySelf: W: Output type forced to '-o pretty' due to usage of '-l'.\n");\r
};\r
\r
### get query type, default to 'postings'\r
#die "$MySelf: E: Unknown query type -q $Options{'q'}!\n" if ($Options{'q'} and !exists($LegalTypes{$Options{'q'}}));\r
#die "$MySelf: E: You must submit a threshold ('-t') for query type '-q $Options{'q'}'!\n" if ($Options{'q'} and !$Options{'t'});\r
\r
### get time period\r
my ($StartMonth,$EndMonth) = &GetTimePeriod($Options{'m'},$Options{'p'});\r
# reset to one month for 'dump' type\r
if ($Options{'o'} eq 'dump' and $Options{'p'}) {\r
  $StartMonth = $EndMonth;\r
  warn ("$MySelf: W: You cannot combine time periods (-p) with '-o dump'. Month was set to $StartMonth.\n");\r
};\r
\r
### init database\r
my $DBHandle = InitDB(\%Conf,1);\r
\r
### get data\r
# get list of newsgroups (-n)\r
my ($QueryPart,@GroupList);\r
my $Newsgroups = $Options{'n'};\r
if ($Newsgroups) {\r
  ($QueryPart,@GroupList) = &SQLGroupList($Newsgroups);\r
} else {\r
  $QueryPart = 1;\r
};\r
\r
# manage thresholds\r
if (defined($Options{'t'})) {\r
  if ($Options{'i'}) {\r
    $QueryPart .= ' AND postings < ?';\r
  } else {\r
    $QueryPart .= ' AND postings > ?';\r
  };\r
  push @GroupList,$Options{'t'};\r
}\r
\r
# construct WHERE clause\r
my $WhereClause = sprintf('month BETWEEN ? AND ? AND %s %s',$QueryPart,&SQLHierarchies($Options{'s'}));\r
\r
# get lenght of longest newsgroup delivered by query for formatting purposes\r
my $MaxLength = &GetMaxLenght($DBHandle,$Conf{'DBTableGrps'},'newsgroup',$WhereClause,$StartMonth,$EndMonth,@GroupList);\r
\r
my ($OrderClause,$DBQuery);\r
# -b (best of) defined?\r
if (!defined($Options{'b'}) and !defined($Options{'l'})) {\r
  $OrderClause = 'newsgroup';\r
  $OrderClause = 'postings' if $Options{'q'};\r
  $OrderClause .= ' DESC' if $Options{'d'};\r
  # do query: get number of postings per group from groups table for given months and newsgroups\r
  $DBQuery = $DBHandle->prepare(sprintf("SELECT month,newsgroup,postings FROM %s.%s WHERE %s ORDER BY month,%s",$Conf{'DBDatabase'},$Conf{'DBTableGrps'},$WhereClause,$OrderClause));\r
} elsif ($Options{'b'}) {\r
  # set sorting order (-i)\r
  if ($Options{'i'}) {\r
    $OrderClause = 'postings';\r
  } else {\r
    $OrderClause = 'postings DESC';\r
  };\r
  # push LIMIT to GroupList to match number of binding vars\r
  push @GroupList,$Options{'b'};\r
  # do query: get sum of postings per group from groups table for given months and newsgroups with LIMIT\r
  $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,$OrderClause));\r
} else { # -l\r
  # set sorting order (-i)\r
  if ($Options{'i'}) {\r
    $OrderClause = '<';\r
  } else {\r
    $OrderClause = '>';\r
  };\r
  # push level and $StartMonth,$EndMonth - again - to GroupList to match number of binding vars\r
  push @GroupList,$Options{'l'};\r
  push @GroupList,$StartMonth,$EndMonth;\r
  # do query: get number of postings per group from groups table for given months and \r
  $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));\r
};\r
\r
# execute query\r
$DBQuery->execute($StartMonth,$EndMonth,@GroupList) or die sprintf("$MySelf: E: Can't get groups data for %s to %s from %s.%s: %s\n",$StartMonth,$EndMonth,$Conf{'DBDatabase'},$Conf{'DBTableGrps'},$DBI::errstr);\r
\r
# output result\r
printf ("----- Report from %s to %s\n",$StartMonth,$EndMonth) if $Options{'c'} and ($Options{'m'} or $Options{'p'});\r
printf ("----- Newsgroups: %s\n",join(',',split(/:/,$Newsgroups))) if $Options{'c'} and $Options{'n'};\r
printf ("----- Threshold: %s %u\n",$Options{'i'} ? '<' : '>',$Options{'t'}) if $Options{'c'} and $Options{'t'};\r
if (!defined($Options{'b'})  and !defined($Options{'l'})) {\r
   &OutputData($Options{'o'},$DBQuery,$MaxLength);\r
} elsif ($Options{'b'}) {\r
   while (my ($Newsgroup,$Postings) = $DBQuery->fetchrow_array) {\r
    print &FormatOutput($Options{'o'}, ($Options{'i'} ? 'Bottom ' : 'Top ').$Options{'b'}, $Newsgroup, $Postings, $MaxLength);\r
  };\r
} else { # -l\r
   while (my ($Month,$Newsgroup,$Postings) = $DBQuery->fetchrow_array) {\r
    print &FormatOutput($Options{'o'}, $Newsgroup, $Month, $Postings, 7);\r
  };\r
};\r
\r
### close handles\r
$DBHandle->disconnect;\r
\r