Commit | Line | Data |
---|---|---|
3f817eb4 | 1 | #! /usr/bin/perl |
2832c235 TH |
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 | # | |
f05ec4ae | 10 | # Copyright (c) 2010-2012 Thomas Hochstein <thh@inter.net> |
2832c235 TH |
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; | |
3f817eb4 | 21 | use warnings; |
2832c235 TH |
22 | |
23 | use NewsStats; | |
24 | ||
25 | use Sys::Syslog qw(:standard :macros); | |
26 | ||
27 | use Date::Format; | |
28 | use DBI; | |
f05ec4ae TH |
29 | use Getopt::Long qw(GetOptions); |
30 | Getopt::Long::config ('bundling'); | |
2832c235 | 31 | |
98563c61 TH |
32 | ################################# Subroutines ################################## |
33 | ||
34 | sub PrepareDB { | |
35 | ### initialise database connection, prepare statement | |
36 | ### and catch errors | |
37 | ### IN : \%Conf : reference to configuration hash | |
38 | ### OUT: $DBHandle: database handle | |
39 | ### $DBQuery : prepared statement | |
40 | our ($DBHandle, $DBQuery, $OptQuiet); | |
41 | my ($ConfigR) = @_; | |
42 | my %Conf = %$ConfigR; | |
43 | # drop current database connection - hard, if necessary | |
44 | if ($DBHandle) { | |
45 | $DBHandle->disconnect; | |
46 | undef $DBHandle; | |
47 | }; | |
48 | # connect to database; try again every 5 seconds | |
49 | while (!$DBHandle) { | |
50 | $DBHandle = InitDB($ConfigR,0); | |
51 | if (!$DBHandle) { | |
52 | syslog(LOG_CRIT, 'Database connection failed: %s', $DBI::errstr); | |
53 | sleep(5); | |
54 | } else {; | |
55 | syslog(LOG_NOTICE, "Database connection (re-)established successfully.") if !$OptQuiet; | |
56 | } | |
57 | }; | |
58 | $DBQuery = $DBHandle->prepare(sprintf("INSERT INTO %s.%s (day,date,mid, | |
59 | timestamp,token,size,peer,path, | |
60 | newsgroups,headers) | |
61 | VALUES (?,?,?,?,?,?,?,?,?,?)", | |
62 | $Conf{'DBDatabase'}, | |
63 | $Conf{'DBTableRaw'})); | |
64 | return ($DBHandle,$DBQuery); | |
65 | } | |
66 | ||
67 | ||
2832c235 TH |
68 | ################################# Main program ################################# |
69 | ||
70 | ### read commandline options | |
f05ec4ae TH |
71 | my ($OptDebug,$OptQuiet); |
72 | GetOptions ('d|debug!' => \$OptDebug, | |
73 | 'q|test!' => \$OptQuiet, | |
74 | 'h|help' => \&ShowPOD, | |
75 | 'V|version' => \&ShowVersion) or exit 1; | |
2832c235 TH |
76 | |
77 | ### read configuration | |
f05ec4ae | 78 | my %Conf = %{ReadConfig($HomePath.'/newsstats.conf')}; |
2832c235 TH |
79 | |
80 | ### init syslog | |
f05ec4ae TH |
81 | openlog($0, 'nofatal,pid', LOG_NEWS); |
82 | syslog(LOG_NOTICE, "$MyVersion starting up.") if !$OptQuiet; | |
2832c235 TH |
83 | |
84 | ### init database | |
98563c61 | 85 | my ($DBHandle,$DBQuery) = PrepareDB(\%Conf); |
2832c235 TH |
86 | |
87 | ### main loop | |
88 | while (<>) { | |
89 | chomp; | |
90 | # catch empty lines trailing or leading | |
91 | if ($_ eq '') { | |
92 | next; | |
93 | } | |
94 | # first line contains: mid, timestamp, token, size, peer, Path, Newsgroups | |
95 | my ($Mid, $Timestamp, $Token, $Size, $Peer, $Path, $Newsgroups) = split; | |
96 | # remaining lines contain headers | |
97 | my $Headers = ""; | |
98 | while (<>) { | |
99 | chomp; | |
100 | # empty line terminates this article | |
101 | if ($_ eq '') { | |
102 | last; | |
103 | } | |
104 | # collect headers | |
105 | $Headers .= $_."\n" ; | |
106 | } | |
107 | ||
108 | # parse timestamp to day (YYYY-MM-DD) and to MySQL timestamp | |
109 | my $Day = time2str("%Y-%m-%d", $Timestamp); | |
110 | my $Date = time2str("%Y-%m-%d %H:%M:%S", $Timestamp); | |
111 | ||
112 | # write to database | |
f05ec4ae TH |
113 | if (!$DBQuery->execute($Day, $Date, $Mid, $Timestamp, $Token, $Size, $Peer, |
114 | $Path, $Newsgroups, $Headers)) { | |
98563c61 TH |
115 | syslog(LOG_ERR, 'Database error %s while processing %s: %s', |
116 | $DBI::err, $Mid, $DBI::errstr); | |
117 | # if "MySQL server has gone away", try to recover | |
118 | if ($DBI::err == 2006) { | |
119 | # try to reconnect to database | |
120 | ($DBHandle,$DBQuery) = PrepareDB(\%Conf); | |
121 | # try to repeat the write attempt as before | |
122 | if (!$DBQuery->execute($Day, $Date, $Mid, $Timestamp, $Token, $Size, $Peer, | |
123 | $Path, $Newsgroups, $Headers)) { | |
124 | syslog(LOG_ERR, '%s was dropped and lost.',$Mid); | |
125 | }; | |
126 | # otherwise log missing posting | |
127 | } else { | |
128 | syslog(LOG_ERR, '%s was dropped and lost.',$Mid); | |
129 | }; | |
2832c235 TH |
130 | }; |
131 | $DBQuery->finish; | |
132 | ||
f05ec4ae TH |
133 | warn sprintf("-----\nDay: %s\nDate: %s\nMID: %s\nTS: %s\nToken: %s\n". |
134 | "Size: %s\nPeer: %s\nPath: %s\nNewsgroups: %s\nHeaders: %s\n", | |
135 | $Day, $Date, $Mid, $Timestamp, $Token, $Size, $Peer, $Path, | |
136 | $Newsgroups, $Headers) if $OptDebug; | |
2832c235 TH |
137 | } |
138 | ||
139 | ### close handles | |
140 | $DBHandle->disconnect; | |
f05ec4ae | 141 | syslog(LOG_NOTICE, "$0 closing down.") if !$OptQuiet; |
2832c235 TH |
142 | closelog(); |
143 | ||
144 | __END__ | |
145 | ||
146 | ################################ Documentation ################################# | |
147 | ||
148 | =head1 NAME | |
149 | ||
150 | feedlog - log data from an INN feed to a database | |
151 | ||
152 | =head1 SYNOPSIS | |
153 | ||
154 | B<feedlog> [B<-Vhdq>] | |
155 | ||
156 | =head1 REQUIREMENTS | |
157 | ||
f05ec4ae | 158 | See L<doc/README>. |
2832c235 TH |
159 | |
160 | =head1 DESCRIPTION | |
161 | ||
162 | This script will log overview data and complete headers to a database | |
163 | table for further examination by parsing a feed from INN. It will | |
164 | parse that information and write it to a mysql database table in real | |
165 | time. | |
166 | ||
167 | All reporting is done to I<syslog> via I<news> facility. If B<feedlog> | |
168 | fails to initiate a database connection at startup, it will log to | |
169 | I<syslog> with I<CRIT> priority and go in an endless loop, as | |
170 | terminating would only result in a rapid respawn. | |
171 | ||
172 | =head2 Configuration | |
173 | ||
f05ec4ae | 174 | B<feedlog> will read its configuration from F<newsstats.conf> which |
2832c235 TH |
175 | should be present in the same directory via Config::Auto. |
176 | ||
f05ec4ae | 177 | See L<doc/INSTALL> for an overview of possible configuration options. |
2832c235 TH |
178 | |
179 | =head1 OPTIONS | |
180 | ||
181 | =over 3 | |
182 | ||
f05ec4ae | 183 | =item B<-V>, B<--version> |
2832c235 | 184 | |
f05ec4ae | 185 | Print out version and copyright information and exit. |
2832c235 | 186 | |
f05ec4ae | 187 | =item B<-h>, B<--help> |
2832c235 TH |
188 | |
189 | Print this man page and exit. | |
190 | ||
f05ec4ae | 191 | =item B<-d>, B<--debug> |
2832c235 TH |
192 | |
193 | Output debugging information to STDERR while parsing STDIN. You'll | |
194 | find that information most probably in your B<INN> F<errlog> file. | |
195 | ||
f05ec4ae | 196 | =item B<-q>, B<--quiet> |
2832c235 TH |
197 | |
198 | Suppress logging to syslog. | |
199 | ||
200 | =back | |
201 | ||
202 | =head1 INSTALLATION | |
203 | ||
f05ec4ae | 204 | See L<doc/INSTALL.> |
2832c235 TH |
205 | |
206 | =head1 EXAMPLES | |
207 | ||
208 | Set up a feed like that in your B<INN> F<newsfeeds> file: | |
209 | ||
210 | ## gather statistics for NewsStats | |
211 | newsstats! | |
212 | :!*,de.* | |
213 | :Tc,WmtfbsPNH,Ac:/path/to/feedlog.pl | |
214 | ||
f05ec4ae | 215 | See L<doc/INSTALL> for further information. |
2832c235 TH |
216 | |
217 | =head1 FILES | |
218 | ||
219 | =over 4 | |
220 | ||
221 | =item F<feedlog.pl> | |
222 | ||
223 | The script itself. | |
224 | ||
225 | =item F<NewsStats.pm> | |
226 | ||
227 | Library functions for the NewsStats package. | |
228 | ||
229 | =item F<newsstats.conf> | |
230 | ||
f05ec4ae | 231 | Runtime configuration file. |
2832c235 TH |
232 | |
233 | =back | |
234 | ||
235 | =head1 BUGS | |
236 | ||
237 | Please report any bugs or feature requests to the author or use the | |
238 | bug tracker at L<http://bugs.th-h.de/>! | |
239 | ||
240 | =head1 SEE ALSO | |
241 | ||
242 | =over 2 | |
243 | ||
244 | =item - | |
245 | ||
f05ec4ae | 246 | L<doc/README> |
2832c235 TH |
247 | |
248 | =item - | |
249 | ||
f05ec4ae | 250 | L<doc/INSTALL> |
2832c235 TH |
251 | |
252 | =back | |
253 | ||
254 | This script is part of the B<NewsStats> package. | |
255 | ||
256 | =head1 AUTHOR | |
257 | ||
258 | Thomas Hochstein <thh@inter.net> | |
259 | ||
260 | =head1 COPYRIGHT AND LICENSE | |
261 | ||
f05ec4ae | 262 | Copyright (c) 2010-2012 Thomas Hochstein <thh@inter.net> |
2832c235 TH |
263 | |
264 | This program is free software; you may redistribute it and/or modify it | |
265 | under the same terms as Perl itself. | |
266 | ||
267 | =cut |