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. | |
dfc2b81c | 7 | # |
2832c235 TH |
8 | # It is part of the NewsStats package. |
9 | # | |
07c0b258 | 10 | # Copyright (c) 2010-2013 Thomas Hochstein <thh@inter.net> |
2832c235 | 11 | # |
dfc2b81c | 12 | # It can be redistributed and/or modified under the same terms under |
2832c235 TH |
13 | # which Perl itself is published. |
14 | ||
15 | BEGIN { | |
24d2011f | 16 | our $VERSION = "0.02"; |
2832c235 | 17 | use File::Basename; |
2ad99c20 TH |
18 | # we're in .../bin, so our module is in ../lib |
19 | push(@INC, dirname($0).'/../lib'); | |
2832c235 TH |
20 | } |
21 | use strict; | |
3f817eb4 | 22 | use warnings; |
2832c235 TH |
23 | |
24 | use NewsStats; | |
25 | ||
26 | use Sys::Syslog qw(:standard :macros); | |
27 | ||
28 | use Date::Format; | |
29 | use DBI; | |
f05ec4ae TH |
30 | use Getopt::Long qw(GetOptions); |
31 | Getopt::Long::config ('bundling'); | |
2832c235 | 32 | |
98563c61 TH |
33 | ################################# Subroutines ################################## |
34 | ||
35 | sub PrepareDB { | |
36 | ### initialise database connection, prepare statement | |
37 | ### and catch errors | |
38 | ### IN : \%Conf : reference to configuration hash | |
39 | ### OUT: $DBHandle: database handle | |
40 | ### $DBQuery : prepared statement | |
41 | our ($DBHandle, $DBQuery, $OptQuiet); | |
42 | my ($ConfigR) = @_; | |
43 | my %Conf = %$ConfigR; | |
44 | # drop current database connection - hard, if necessary | |
45 | if ($DBHandle) { | |
46 | $DBHandle->disconnect; | |
47 | undef $DBHandle; | |
48 | }; | |
49 | # connect to database; try again every 5 seconds | |
50 | while (!$DBHandle) { | |
51 | $DBHandle = InitDB($ConfigR,0); | |
52 | if (!$DBHandle) { | |
53 | syslog(LOG_CRIT, 'Database connection failed: %s', $DBI::errstr); | |
54 | sleep(5); | |
55 | } else {; | |
56 | syslog(LOG_NOTICE, "Database connection (re-)established successfully.") if !$OptQuiet; | |
57 | } | |
58 | }; | |
59 | $DBQuery = $DBHandle->prepare(sprintf("INSERT INTO %s.%s (day,date,mid, | |
60 | timestamp,token,size,peer,path, | |
61 | newsgroups,headers) | |
62 | VALUES (?,?,?,?,?,?,?,?,?,?)", | |
63 | $Conf{'DBDatabase'}, | |
64 | $Conf{'DBTableRaw'})); | |
65 | return ($DBHandle,$DBQuery); | |
66 | } | |
67 | ||
68 | ||
2832c235 TH |
69 | ################################# Main program ################################# |
70 | ||
71 | ### read commandline options | |
23ab67a0 | 72 | my ($OptDebug,$OptQuiet,$OptConfFile); |
f05ec4ae TH |
73 | GetOptions ('d|debug!' => \$OptDebug, |
74 | 'q|test!' => \$OptQuiet, | |
23ab67a0 | 75 | 'conffile=s' => \$OptConfFile, |
f05ec4ae TH |
76 | 'h|help' => \&ShowPOD, |
77 | 'V|version' => \&ShowVersion) or exit 1; | |
2832c235 TH |
78 | |
79 | ### read configuration | |
23ab67a0 | 80 | my %Conf = %{ReadConfig($OptConfFile)}; |
2832c235 TH |
81 | |
82 | ### init syslog | |
f05ec4ae TH |
83 | openlog($0, 'nofatal,pid', LOG_NEWS); |
84 | syslog(LOG_NOTICE, "$MyVersion starting up.") if !$OptQuiet; | |
2832c235 TH |
85 | |
86 | ### init database | |
98563c61 | 87 | my ($DBHandle,$DBQuery) = PrepareDB(\%Conf); |
2832c235 TH |
88 | |
89 | ### main loop | |
90 | while (<>) { | |
91 | chomp; | |
92 | # catch empty lines trailing or leading | |
93 | if ($_ eq '') { | |
94 | next; | |
95 | } | |
96 | # first line contains: mid, timestamp, token, size, peer, Path, Newsgroups | |
97 | my ($Mid, $Timestamp, $Token, $Size, $Peer, $Path, $Newsgroups) = split; | |
98 | # remaining lines contain headers | |
99 | my $Headers = ""; | |
100 | while (<>) { | |
101 | chomp; | |
102 | # empty line terminates this article | |
103 | if ($_ eq '') { | |
104 | last; | |
105 | } | |
106 | # collect headers | |
107 | $Headers .= $_."\n" ; | |
108 | } | |
109 | ||
110 | # parse timestamp to day (YYYY-MM-DD) and to MySQL timestamp | |
111 | my $Day = time2str("%Y-%m-%d", $Timestamp); | |
112 | my $Date = time2str("%Y-%m-%d %H:%M:%S", $Timestamp); | |
113 | ||
114 | # write to database | |
f05ec4ae TH |
115 | if (!$DBQuery->execute($Day, $Date, $Mid, $Timestamp, $Token, $Size, $Peer, |
116 | $Path, $Newsgroups, $Headers)) { | |
98563c61 TH |
117 | syslog(LOG_ERR, 'Database error %s while processing %s: %s', |
118 | $DBI::err, $Mid, $DBI::errstr); | |
119 | # if "MySQL server has gone away", try to recover | |
120 | if ($DBI::err == 2006) { | |
121 | # try to reconnect to database | |
122 | ($DBHandle,$DBQuery) = PrepareDB(\%Conf); | |
123 | # try to repeat the write attempt as before | |
124 | if (!$DBQuery->execute($Day, $Date, $Mid, $Timestamp, $Token, $Size, $Peer, | |
125 | $Path, $Newsgroups, $Headers)) { | |
126 | syslog(LOG_ERR, '%s was dropped and lost.',$Mid); | |
127 | }; | |
128 | # otherwise log missing posting | |
129 | } else { | |
130 | syslog(LOG_ERR, '%s was dropped and lost.',$Mid); | |
131 | }; | |
2832c235 TH |
132 | }; |
133 | $DBQuery->finish; | |
23ab67a0 | 134 | |
f05ec4ae TH |
135 | warn sprintf("-----\nDay: %s\nDate: %s\nMID: %s\nTS: %s\nToken: %s\n". |
136 | "Size: %s\nPeer: %s\nPath: %s\nNewsgroups: %s\nHeaders: %s\n", | |
137 | $Day, $Date, $Mid, $Timestamp, $Token, $Size, $Peer, $Path, | |
138 | $Newsgroups, $Headers) if $OptDebug; | |
2832c235 TH |
139 | } |
140 | ||
141 | ### close handles | |
142 | $DBHandle->disconnect; | |
f05ec4ae | 143 | syslog(LOG_NOTICE, "$0 closing down.") if !$OptQuiet; |
2832c235 TH |
144 | closelog(); |
145 | ||
146 | __END__ | |
147 | ||
148 | ################################ Documentation ################################# | |
149 | ||
150 | =head1 NAME | |
151 | ||
152 | feedlog - log data from an INN feed to a database | |
153 | ||
154 | =head1 SYNOPSIS | |
155 | ||
e39d4207 | 156 | B<feedlog> [B<-Vhdq>] [B<--conffile> I<filename>] |
2832c235 TH |
157 | |
158 | =head1 REQUIREMENTS | |
159 | ||
f05ec4ae | 160 | See L<doc/README>. |
2832c235 TH |
161 | |
162 | =head1 DESCRIPTION | |
163 | ||
164 | This script will log overview data and complete headers to a database | |
165 | table for further examination by parsing a feed from INN. It will | |
166 | parse that information and write it to a mysql database table in real | |
167 | time. | |
168 | ||
169 | All reporting is done to I<syslog> via I<news> facility. If B<feedlog> | |
170 | fails to initiate a database connection at startup, it will log to | |
171 | I<syslog> with I<CRIT> priority and go in an endless loop, as | |
172 | terminating would only result in a rapid respawn. | |
173 | ||
174 | =head2 Configuration | |
175 | ||
f05ec4ae | 176 | B<feedlog> will read its configuration from F<newsstats.conf> which |
2832c235 TH |
177 | should be present in the same directory via Config::Auto. |
178 | ||
f05ec4ae | 179 | See L<doc/INSTALL> for an overview of possible configuration options. |
2832c235 TH |
180 | |
181 | =head1 OPTIONS | |
182 | ||
183 | =over 3 | |
184 | ||
f05ec4ae | 185 | =item B<-V>, B<--version> |
2832c235 | 186 | |
f05ec4ae | 187 | Print out version and copyright information and exit. |
2832c235 | 188 | |
f05ec4ae | 189 | =item B<-h>, B<--help> |
2832c235 TH |
190 | |
191 | Print this man page and exit. | |
192 | ||
f05ec4ae | 193 | =item B<-d>, B<--debug> |
2832c235 TH |
194 | |
195 | Output debugging information to STDERR while parsing STDIN. You'll | |
196 | find that information most probably in your B<INN> F<errlog> file. | |
197 | ||
f05ec4ae | 198 | =item B<-q>, B<--quiet> |
2832c235 TH |
199 | |
200 | Suppress logging to syslog. | |
201 | ||
23ab67a0 TH |
202 | =item B<--conffile> I<filename> |
203 | ||
204 | Load configuration from I<filename> instead of F<newsstats.conf>. | |
205 | ||
2832c235 TH |
206 | =back |
207 | ||
208 | =head1 INSTALLATION | |
209 | ||
95d9fe2c | 210 | See L<doc/INSTALL>. |
2832c235 TH |
211 | |
212 | =head1 EXAMPLES | |
213 | ||
214 | Set up a feed like that in your B<INN> F<newsfeeds> file: | |
215 | ||
216 | ## gather statistics for NewsStats | |
217 | newsstats! | |
218 | :!*,de.* | |
219 | :Tc,WmtfbsPNH,Ac:/path/to/feedlog.pl | |
220 | ||
f05ec4ae | 221 | See L<doc/INSTALL> for further information. |
2832c235 TH |
222 | |
223 | =head1 FILES | |
224 | ||
225 | =over 4 | |
226 | ||
2ad99c20 | 227 | =item F<bin/feedlog.pl> |
2832c235 TH |
228 | |
229 | The script itself. | |
230 | ||
2ad99c20 | 231 | =item F<lib/NewsStats.pm> |
2832c235 TH |
232 | |
233 | Library functions for the NewsStats package. | |
234 | ||
2ad99c20 | 235 | =item F<etc/newsstats.conf> |
2832c235 | 236 | |
f05ec4ae | 237 | Runtime configuration file. |
2832c235 TH |
238 | |
239 | =back | |
240 | ||
241 | =head1 BUGS | |
242 | ||
243 | Please report any bugs or feature requests to the author or use the | |
244 | bug tracker at L<http://bugs.th-h.de/>! | |
245 | ||
246 | =head1 SEE ALSO | |
247 | ||
248 | =over 2 | |
249 | ||
250 | =item - | |
251 | ||
f05ec4ae | 252 | L<doc/README> |
2832c235 TH |
253 | |
254 | =item - | |
255 | ||
f05ec4ae | 256 | L<doc/INSTALL> |
2832c235 TH |
257 | |
258 | =back | |
259 | ||
260 | This script is part of the B<NewsStats> package. | |
261 | ||
262 | =head1 AUTHOR | |
263 | ||
264 | Thomas Hochstein <thh@inter.net> | |
265 | ||
266 | =head1 COPYRIGHT AND LICENSE | |
267 | ||
28717921 | 268 | Copyright (c) 2010-2013 Thomas Hochstein <thh@inter.net> |
2832c235 TH |
269 | |
270 | This program is free software; you may redistribute it and/or modify it | |
271 | under the same terms as Perl itself. | |
272 | ||
273 | =cut |