[usenet/newsstats.git] / feedlog.pl

#! /usr/bin/perl -W
#
# feedlog.pl
#
# This script will log headers and other data to a database
# for further analysis by parsing a feed from INN.
# 
# It is part of the NewsStats package.
#
# Copyright (c) 2010-2012 Thomas Hochstein <thh@inter.net>
#
# It can be redistributed and/or modified under the same terms under 
# which Perl itself is published.

BEGIN {
  our $VERSION = "0.01";
  use File::Basename;
  push(@INC, dirname($0));
}
use strict;

use NewsStats;

use Sys::Syslog qw(:standard :macros);

use Date::Format;
use DBI;
use Getopt::Long qw(GetOptions);
Getopt::Long::config ('bundling');

################################# Subroutines ##################################

sub PrepareDB {
### initialise database connection, prepare statement
### and catch errors
### IN : \%Conf   : reference to configuration hash
### OUT: $DBHandle: database handle
###      $DBQuery : prepared statement
  our ($DBHandle, $DBQuery, $OptQuiet);
  my ($ConfigR) = @_;
  my %Conf = %$ConfigR;
  # drop current database connection - hard, if necessary
  if ($DBHandle) {
    $DBHandle->disconnect;
    undef $DBHandle;
  };
  # connect to database; try again every 5 seconds
  while (!$DBHandle) {
    $DBHandle = InitDB($ConfigR,0);
    if (!$DBHandle) {
      syslog(LOG_CRIT, 'Database connection failed: %s', $DBI::errstr);
      sleep(5);
    } else {;
      syslog(LOG_NOTICE, "Database connection (re-)established successfully.") if !$OptQuiet;
    }
  };
  $DBQuery = $DBHandle->prepare(sprintf("INSERT INTO %s.%s (day,date,mid,
                                         timestamp,token,size,peer,path,
                                         newsgroups,headers)
                                         VALUES (?,?,?,?,?,?,?,?,?,?)",
                                         $Conf{'DBDatabase'},
                                         $Conf{'DBTableRaw'}));
  return ($DBHandle,$DBQuery);
}


################################# Main program #################################

### read commandline options
my ($OptDebug,$OptQuiet);
GetOptions ('d|debug!'        => \$OptDebug,
            'q|test!'         => \$OptQuiet,
            'h|help'          => \&ShowPOD,
            'V|version'       => \&ShowVersion) or exit 1;

### read configuration
my %Conf = %{ReadConfig($HomePath.'/newsstats.conf')};

### init syslog
openlog($0, 'nofatal,pid', LOG_NEWS);
syslog(LOG_NOTICE, "$MyVersion starting up.") if !$OptQuiet;

### init database
my ($DBHandle,$DBQuery) = PrepareDB(\%Conf);

### main loop
while (<>) {
  chomp;
  # catch empty lines trailing or leading
  if ($_ eq '') {
    next;
  }
  # first line contains: mid, timestamp, token, size, peer, Path, Newsgroups
  my ($Mid, $Timestamp, $Token, $Size, $Peer, $Path, $Newsgroups) = split;
  # remaining lines contain headers
  my $Headers = "";
  while (<>) {
    chomp;
    # empty line terminates this article
    if ($_ eq '') {
      last;
    }
    # collect headers
    $Headers .= $_."\n" ;
  }

  # parse timestamp to day (YYYY-MM-DD) and to MySQL timestamp
  my $Day  = time2str("%Y-%m-%d", $Timestamp);
  my $Date = time2str("%Y-%m-%d %H:%M:%S", $Timestamp);

  # write to database
  if (!$DBQuery->execute($Day, $Date, $Mid, $Timestamp, $Token, $Size, $Peer,
                         $Path, $Newsgroups, $Headers)) {
    syslog(LOG_ERR, 'Database error %s while processing %s: %s',
           $DBI::err, $Mid, $DBI::errstr);
    # if "MySQL server has gone away", try to recover
    if ($DBI::err == 2006) {
      # try to reconnect to database
      ($DBHandle,$DBQuery) = PrepareDB(\%Conf);
      # try to repeat the write attempt as before
      if (!$DBQuery->execute($Day, $Date, $Mid, $Timestamp, $Token, $Size, $Peer,
                             $Path, $Newsgroups, $Headers)) {
        syslog(LOG_ERR, '%s was dropped and lost.',$Mid);
      };
    # otherwise log missing posting
    } else {
      syslog(LOG_ERR, '%s was dropped and lost.',$Mid);
    };
  };
  $DBQuery->finish;
  
  warn sprintf("-----\nDay: %s\nDate: %s\nMID: %s\nTS: %s\nToken: %s\n".
               "Size: %s\nPeer: %s\nPath: %s\nNewsgroups: %s\nHeaders: %s\n",
               $Day, $Date, $Mid, $Timestamp, $Token, $Size, $Peer, $Path,
               $Newsgroups, $Headers) if $OptDebug;
}

### close handles
$DBHandle->disconnect;
syslog(LOG_NOTICE, "$0 closing down.") if !$OptQuiet;
closelog();

__END__

################################ Documentation #################################

=head1 NAME

feedlog - log data from an INN feed to a database

=head1 SYNOPSIS

B<feedlog> [B<-Vhdq>]

=head1 REQUIREMENTS

See L<doc/README>.

=head1 DESCRIPTION

This script will log overview data and complete headers to a database
table for further examination by parsing a feed from INN. It will
parse that information and write it to a mysql database table in real
time.

All reporting is done to I<syslog> via I<news> facility. If B<feedlog>
fails to initiate a database connection at startup, it will log to
I<syslog> with I<CRIT> priority and go in an endless loop, as
terminating would only result in a rapid respawn.

=head2 Configuration

B<feedlog> will read its configuration from F<newsstats.conf> which
should be present in the same directory via Config::Auto.

See L<doc/INSTALL> for an overview of possible configuration options.

=head1 OPTIONS

=over 3

=item B<-V>, B<--version>

Print out version and copyright information and exit.

=item B<-h>, B<--help>

Print this man page and exit.

=item B<-d>, B<--debug>

Output debugging information to STDERR while parsing STDIN. You'll
find that information most probably in your B<INN> F<errlog> file.

=item B<-q>, B<--quiet>

Suppress logging to syslog.

=back

=head1 INSTALLATION

See L<doc/INSTALL.>

=head1 EXAMPLES

Set up a feed like that in your B<INN> F<newsfeeds> file:

    ## gather statistics for NewsStats
    newsstats!
            :!*,de.*
            :Tc,WmtfbsPNH,Ac:/path/to/feedlog.pl

See L<doc/INSTALL> for further information.

=head1 FILES

=over 4

=item F<feedlog.pl>

The script itself.

=item F<NewsStats.pm>

Library functions for the NewsStats package.

=item F<newsstats.conf>

Runtime configuration file.

=back

=head1 BUGS

Please report any bugs or feature requests to the author or use the
bug tracker at L<http://bugs.th-h.de/>!

=head1 SEE ALSO

=over 2

=item -

L<doc/README>

=item -

L<doc/INSTALL>

=back

This script is part of the B<NewsStats> package.

=head1 AUTHOR

Thomas Hochstein <thh@inter.net>

=head1 COPYRIGHT AND LICENSE

Copyright (c) 2010-2012 Thomas Hochstein <thh@inter.net>

This program is free software; you may redistribute it and/or modify it
under the same terms as Perl itself.

=cut
Commit	Line	Data
	1	#! /usr/bin/perl -W
	2	#
	3	# feedlog.pl
	4	#
	5	# This script will log headers and other data to a database
	6	# for further analysis by parsing a feed from INN.
	7	#
	8	# It is part of the NewsStats package.
	9	#
	10	# Copyright (c) 2010-2012 Thomas Hochstein <thh@inter.net>
	11	#
	12	# It can be redistributed and/or modified under the same terms under
	13	# which Perl itself is published.
	14
	15	BEGIN {
	16	our $VERSION = "0.01";
	17	use File::Basename;
	18	push(@INC, dirname($0));
	19	}
	20	use strict;
	21
	22	use NewsStats;
	23
	24	use Sys::Syslog qw(:standard :macros);
	25
	26	use Date::Format;
	27	use DBI;
	28	use Getopt::Long qw(GetOptions);
	29	Getopt::Long::config ('bundling');
	30
	31	################################# Subroutines ##################################
	32
	33	sub PrepareDB {
	34	### initialise database connection, prepare statement
	35	### and catch errors
	36	### IN : \%Conf : reference to configuration hash
	37	### OUT: $DBHandle: database handle
	38	### $DBQuery : prepared statement
	39	our ($DBHandle, $DBQuery, $OptQuiet);
	40	my ($ConfigR) = @_;
	41	my %Conf = %$ConfigR;
	42	# drop current database connection - hard, if necessary
	43	if ($DBHandle) {
	44	$DBHandle->disconnect;
	45	undef $DBHandle;
	46	};
	47	# connect to database; try again every 5 seconds
	48	while (!$DBHandle) {
	49	$DBHandle = InitDB($ConfigR,0);
	50	if (!$DBHandle) {
	51	syslog(LOG_CRIT, 'Database connection failed: %s', $DBI::errstr);
	52	sleep(5);
	53	} else {;
	54	syslog(LOG_NOTICE, "Database connection (re-)established successfully.") if !$OptQuiet;
	55	}
	56	};
	57	$DBQuery = $DBHandle->prepare(sprintf("INSERT INTO %s.%s (day,date,mid,
	58	timestamp,token,size,peer,path,
	59	newsgroups,headers)
	60	VALUES (?,?,?,?,?,?,?,?,?,?)",
	61	$Conf{'DBDatabase'},
	62	$Conf{'DBTableRaw'}));
	63	return ($DBHandle,$DBQuery);
	64	}
	65
	66
	67	################################# Main program #################################
	68
	69	### read commandline options
	70	my ($OptDebug,$OptQuiet);
	71	GetOptions ('d\|debug!' => \$OptDebug,
	72	'q\|test!' => \$OptQuiet,
	73	'h\|help' => \&ShowPOD,
	74	'V\|version' => \&ShowVersion) or exit 1;
	75
	76	### read configuration
	77	my %Conf = %{ReadConfig($HomePath.'/newsstats.conf')};
	78
	79	### init syslog
	80	openlog($0, 'nofatal,pid', LOG_NEWS);
	81	syslog(LOG_NOTICE, "$MyVersion starting up.") if !$OptQuiet;
	82
	83	### init database
	84	my ($DBHandle,$DBQuery) = PrepareDB(\%Conf);
	85
	86	### main loop
	87	while (<>) {
	88	chomp;
	89	# catch empty lines trailing or leading
	90	if ($_ eq '') {
	91	next;
	92	}
	93	# first line contains: mid, timestamp, token, size, peer, Path, Newsgroups
	94	my ($Mid, $Timestamp, $Token, $Size, $Peer, $Path, $Newsgroups) = split;
	95	# remaining lines contain headers
	96	my $Headers = "";
	97	while (<>) {
	98	chomp;
	99	# empty line terminates this article
	100	if ($_ eq '') {
	101	last;
	102	}
	103	# collect headers
	104	$Headers .= $_."\n" ;
	105	}
	106
	107	# parse timestamp to day (YYYY-MM-DD) and to MySQL timestamp
	108	my $Day = time2str("%Y-%m-%d", $Timestamp);
	109	my $Date = time2str("%Y-%m-%d %H:%M:%S", $Timestamp);
	110
	111	# write to database
	112	if (!$DBQuery->execute($Day, $Date, $Mid, $Timestamp, $Token, $Size, $Peer,
	113	$Path, $Newsgroups, $Headers)) {
	114	syslog(LOG_ERR, 'Database error %s while processing %s: %s',
	115	$DBI::err, $Mid, $DBI::errstr);
	116	# if "MySQL server has gone away", try to recover
	117	if ($DBI::err == 2006) {
	118	# try to reconnect to database
	119	($DBHandle,$DBQuery) = PrepareDB(\%Conf);
	120	# try to repeat the write attempt as before
	121	if (!$DBQuery->execute($Day, $Date, $Mid, $Timestamp, $Token, $Size, $Peer,
	122	$Path, $Newsgroups, $Headers)) {
	123	syslog(LOG_ERR, '%s was dropped and lost.',$Mid);
	124	};
	125	# otherwise log missing posting
	126	} else {
	127	syslog(LOG_ERR, '%s was dropped and lost.',$Mid);
	128	};
	129	};
	130	$DBQuery->finish;
	131
	132	warn sprintf("-----\nDay: %s\nDate: %s\nMID: %s\nTS: %s\nToken: %s\n".
	133	"Size: %s\nPeer: %s\nPath: %s\nNewsgroups: %s\nHeaders: %s\n",
	134	$Day, $Date, $Mid, $Timestamp, $Token, $Size, $Peer, $Path,
	135	$Newsgroups, $Headers) if $OptDebug;
	136	}
	137
	138	### close handles
	139	$DBHandle->disconnect;
	140	syslog(LOG_NOTICE, "$0 closing down.") if !$OptQuiet;
	141	closelog();
	142
	143	__END__
	144
	145	################################ Documentation #################################
	146
	147	=head1 NAME
	148
	149	feedlog - log data from an INN feed to a database
	150
	151	=head1 SYNOPSIS
	152
	153	B<feedlog> [B<-Vhdq>]
	154
	155	=head1 REQUIREMENTS
	156
	157	See L<doc/README>.
	158
	159	=head1 DESCRIPTION
	160
	161	This script will log overview data and complete headers to a database
	162	table for further examination by parsing a feed from INN. It will
	163	parse that information and write it to a mysql database table in real
	164	time.
	165
	166	All reporting is done to I<syslog> via I<news> facility. If B<feedlog>
	167	fails to initiate a database connection at startup, it will log to
	168	I<syslog> with I<CRIT> priority and go in an endless loop, as
	169	terminating would only result in a rapid respawn.
	170
	171	=head2 Configuration
	172
	173	B<feedlog> will read its configuration from F<newsstats.conf> which
	174	should be present in the same directory via Config::Auto.
	175
	176	See L<doc/INSTALL> for an overview of possible configuration options.
	177
	178	=head1 OPTIONS
	179
	180	=over 3
	181
	182	=item B<-V>, B<--version>
	183
	184	Print out version and copyright information and exit.
	185
	186	=item B<-h>, B<--help>
	187
	188	Print this man page and exit.
	189
	190	=item B<-d>, B<--debug>
	191
	192	Output debugging information to STDERR while parsing STDIN. You'll
	193	find that information most probably in your B<INN> F<errlog> file.
	194
	195	=item B<-q>, B<--quiet>
	196
	197	Suppress logging to syslog.
	198
	199	=back
	200
	201	=head1 INSTALLATION
	202
	203	See L<doc/INSTALL.>
	204
	205	=head1 EXAMPLES
	206
	207	Set up a feed like that in your B<INN> F<newsfeeds> file:
	208
	209	## gather statistics for NewsStats
	210	newsstats!
	211	:!,de.
	212	:Tc,WmtfbsPNH,Ac:/path/to/feedlog.pl
	213
	214	See L<doc/INSTALL> for further information.
	215
	216	=head1 FILES
	217
	218	=over 4
	219
	220	=item F<feedlog.pl>
	221
	222	The script itself.
	223
	224	=item F<NewsStats.pm>
	225
	226	Library functions for the NewsStats package.
	227
	228	=item F<newsstats.conf>
	229
	230	Runtime configuration file.
	231
	232	=back
	233
	234	=head1 BUGS
	235
	236	Please report any bugs or feature requests to the author or use the
	237	bug tracker at L<http://bugs.th-h.de/>!
	238
	239	=head1 SEE ALSO
	240
	241	=over 2
	242
	243	=item -
	244
	245	L<doc/README>
	246
	247	=item -
	248
	249	L<doc/INSTALL>
	250
	251	=back
	252
	253	This script is part of the B<NewsStats> package.
	254
	255	=head1 AUTHOR
	256
	257	Thomas Hochstein <thh@inter.net>
	258
	259	=head1 COPYRIGHT AND LICENSE
	260
	261	Copyright (c) 2010-2012 Thomas Hochstein <thh@inter.net>
	262
	263	This program is free software; you may redistribute it and/or modify it
	264	under the same terms as Perl itself.
	265
	266	=cut