z3950 support is coming...
[koha.git] / z3950 / processz3950queue
1 #!/usr/bin/perl
2
3 # $Id$
4
5 use C4::Context;
6 use DBI;
7 use strict;
8 use C4::Biblio;
9 use C4::Output;
10 use C4::Breeding;
11 use Net::Z3950;
12
13
14 if ($< == 0) {
15     # Running as root, switch privs
16     if (-d "/var/run") {
17         open PID, ">/var/run/processz3950queue.pid";
18         print PID $$."\n";
19         close PID;
20     }
21     # Get real apacheuser from koha.conf or reparsing httpd.conf
22     my $apacheuser='paul';
23     my $uid=0;
24     unless ($uid = (getpwnam($apacheuser))[2]) {
25         die "Attempt to run daemon as non-existent or superuser\n";
26     }
27     $>=$uid;
28     $<=$uid;
29 }
30 my $dbh = C4::Context->dbh;
31
32 my $sth=$dbh->prepare("update z3950results set active=0");
33 $sth->execute;
34 $sth->finish;
35 $SIG{CHLD}='reap';
36 $SIG{HUP}='checkqueue';
37
38
39 my $logdir=$ARGV[0];
40
41 open PID, ">$logdir/processz3950queue.pid";
42 print PID $$."\n";
43 close PID;
44
45 my $reapcounter=0;
46 my $forkcounter=0;
47 my $checkqueue=1;
48 my $pid=$$;
49 my $lastrun=0;
50 while (1) {
51         if ((time-$lastrun)>5) {
52                 print "starting loop\n";
53                 if ($checkqueue) { # everytime a SIG{HUP} is recieved
54                         $checkqueue=0;
55                         my $sth=$dbh->prepare("select id,term,type,servers from z3950queue order by id");
56                         $sth->execute;
57                         while (my ($id, $term, $type, $servers) = $sth->fetchrow) {
58                                 if ($forkcounter<12) {
59                                         my $now=time();
60                                         my $stk=$dbh->prepare("select id,server,startdate,enddate,numrecords,active from z3950results where queryid=$id");
61                                         ($stk->execute) || (next);
62                                         my %serverdone;
63                                         unless ($stk->rows) {
64                                                 my $sti=$dbh->prepare("update z3950queue set done=-1,startdate=$now where id=$id");
65                                                 $sti->execute;
66                                         }
67                                         while (my ($r_id, $r_server,$r_startdate,$r_enddate,$r_numrecords,$active) = $stk->fetchrow) {
68                                                 if ($r_enddate >0) {
69                                                         $serverdone{$r_server}=1;
70                                                 } elsif ($active) {
71                                                         $serverdone{$r_server}=1;
72                                                 } else {
73                                                         $serverdone{$r_server}=-1;
74                                                 }
75                                         }
76
77                                         $stk->finish;
78                                         my $attr='';
79                                         if ($type eq 'isbn') {
80                                                 $attr='1=7';
81                                         } elsif ($type eq 'title') {
82                                                 $attr='1=4';
83                                         } elsif ($type eq 'author') {
84                                                 $attr='1=1003';
85                                         } elsif ($type eq 'lccn') {
86                                                 $attr='1=9';
87                                         } elsif ($type eq 'keyword') {
88                                                 $attr='1=1016';
89                                         }
90                                         $term='"'.$term.'"';
91                                         my $query="\@attr $attr $term";
92                                         my $totalrecords=0;
93                                         my $serverinfo;
94                                         my $stillprocessing=0;
95                                         my $globalname;
96                                         foreach $serverinfo (split(/\s+/, $servers)) {
97                                                 (next) if ($serverdone{$serverinfo} == 1);
98                                                 my $stillprocessing=1;
99                                                 if (my $pid=fork()) {
100                                                         $forkcounter++;
101                                                 } else {
102                                                         my $dbi = C4::Context->dbh;
103                                                         my ($name, $server, $database, $user, $password) = split(/\//, $serverinfo, 5);
104                                                         $globalname=$name;
105                                                         $server=~/(.*)\:(\d+)/;
106                                                         my $servername=$1;
107                                                         my $port=$2;
108                                                         print "Processing $type=$term at $name $server $database (".($forkcounter+1)." forks)\n";
109                                                         $now=time();
110                                                         my $q_serverinfo=$dbi->quote($serverinfo);
111                                                         my $resultsid;
112                                                         if ($serverdone{$serverinfo}==-1) {
113                                                                 my $stj=$dbi->prepare("select id from z3950results where server=$q_serverinfo and queryid=$id");
114                                                                 $stj->execute;
115                                                                 ($resultsid) = $stj->fetchrow;
116                                                                 $stj->finish;
117                                                         } else {
118                                                                 my $stj=$dbi->prepare("select id from z3950results where server=$q_serverinfo and queryid=$id");
119                                                                 $stj->execute;
120                                                                 ($resultsid) = $stj->fetchrow;
121                                                                 $stj->finish;
122                                                                 unless ($resultsid) {
123                                                                         $stj=$dbi->prepare("insert into z3950results (server, queryid, startdate) values ($q_serverinfo, $id, $now)");
124                                                                         $stj->execute;
125                                                                         $resultsid=$dbi->{'mysql_insertid'};
126                                                                         $stj->finish;
127                                                                 }
128                                                         }
129                                                         my $stj=$dbh->prepare("update z3950results set active=1 where id=$resultsid");
130                                                         $stj->execute;
131                                                         my $conn;
132                                                         my $noconnection=0;
133                                                         my $error=0;
134                                                         if ($user) {
135                                                                 eval { $conn= new Net::Z3950::Connection($servername, $port, databaseName => $database, user => $user, password => $password); };
136                                                                 if ($@) {
137                                                                         $noconnection=1;
138                                                                 } else {
139                                                                         $error=pe();
140                                                                 }
141                                                         } else {
142                                                                 eval { $conn= new Net::Z3950::Connection($servername, $port, databaseName => $database); };
143                                                                 if ($@) {
144                                                                         $noconnection=1;
145                                                                 } else {
146                                                                         $error=pe();
147                                                                 }
148                                                         }
149                                                         if ($noconnection || $error) {
150                                                                 warn "no connection at $globalname ";
151                                                         } else {
152                                                                 eval { $conn->option(preferredRecordSyntax => Net::Z3950::RecordSyntax::USMARC);};
153                                                                 if ($@) {
154                                                                         print "$globalname ERROR: $@\n";
155                                                                 } else {
156                                                                         print "Q: $query\n";
157                                                                         my $rs=$conn->search($query);
158                                                                         pe();
159                                                                         my $numresults=$rs->size();
160                                                                         pe();
161                                                                         my $i;
162                                                                         my $result='';
163                                                                         my $scantimerstart=time();
164                                                                         for ($i=1; $i<=(($numresults<80) ? ($numresults) : (80)); $i++) {
165                                                                                 my $rec=$rs->record($i);
166                                                                                 my $marcdata=$rec->render();
167                                                                                 my $marcrecord = MARC::File::USMARC::decode($rec->render());
168                                                                                 warn "$globalname ==> ".$marcrecord->as_formatted();
169                                                                                 $result.=$marcdata;
170                                                                                 my ($notmarcrecord,$alreadyindb,$alreadyinfarm,$imported) = ImportBreeding($marcdata,1,"Z3950-$globalname");
171                                                                         }
172                                                                         my $scantimerend=time();
173                                                                         my $numrecords;
174                                                                         ($numresults<80) ? ($numrecords=$numresults) : ($numrecords=80);
175                                                                         my $elapsed=$scantimerend-$scantimerstart;
176                                                                         if ($elapsed) {
177                                                                                 my $speed=int($numresults/$elapsed*100)/100;
178                                                                                 print "$globalname   SPEED: $speed  $server done $numrecords\n";
179                                                                         }
180                                                                         my $q_result=$dbi->quote($result);
181                                                                         ($q_result) || ($q_result='""');
182                                                                         $now=time();
183                                                                         if ($numresults >0) {
184                                                                                 my $task="update z3950results set numrecords=$numresults,numdownloaded=$numrecords,highestseen=0,results=$q_result,enddate=$now where id=$resultsid";
185                                                                                 my $stj=$dbi->prepare($task);
186                                                                                 $stj->execute;
187                                                                         } else { # no results...
188                                                                                 my $task="update z3950results set numrecords=$numresults,numdownloaded=$numrecords,highestseen=0,results='',enddate=$now where id=$resultsid";
189                                                                                 my $stj=$dbi->prepare($task);
190                                                                                 $stj->execute;
191                                                                         }
192                                                                         my $counter=0;
193                                                                         while ($counter<60 && $numrecords<$numresults) {
194                                                                                 $counter++;
195                                                                                 my $stj=$dbi->prepare("select highestseen from z3950results where id=$resultsid");
196                                                                                 $stj->execute;
197                                                                                 my ($highestseen) = $stj->fetchrow;
198                                                                                 if ($highestseen>($numrecords-30)) {
199                                                                                         $counter=0;
200                                                                                         print "   $server rescanning\n";
201                                                                                         my $scantimerstart=time();
202                                                                                         for ($i=$numrecords+1; $i<=(($numresults<($numrecords+40)) ? ($numresults) : ($numrecords+40)); $i++) {
203                                                                                                 my $rec=$rs->record($i);
204                                                                                                 my $marcdata=$rec->rawdata();
205                                                                                                 $result.=$marcdata;
206                                                                                         }
207                                                                                         my $scantimerend=time();
208                                                                                         ($numresults<$numrecords+40) ? ($numrecords=$numresults) : ($numrecords += 40);
209                                                                                         my $elapsed=$scantimerend-$scantimerstart;
210                                                                                         if ($elapsed) {
211                                                                                                 my $speed=int($numresults/$elapsed*100)/100;
212                                                                                                 print "  SPEED: $speed  $server done $numrecords\n";
213                                                                                         }
214
215                                                                                         my $q_result=$dbi->quote($result);
216                                                                                         ($q_result) || ($q_result='""');
217                                                                                         $now=time();
218                                                                                         my $task="update z3950results set numdownloaded=$numrecords,results=$q_result where id=$resultsid";
219                                                                                         my $stj=$dbi->prepare($task);
220                                                                                         $stj->execute;
221                                                                                 }
222                                                                                 sleep 5;
223                                                                         }
224                                                                 }
225                                                         }
226                                                         # FIXME - There's already a $stj in this scope
227                                                         my $stj=$dbi->prepare("update z3950results set active=0 where id=$resultsid");
228                                                         $stj->execute;
229                                                         eval {$stj->finish};
230                                                         print "    $server done.\n";
231                                                         exit;
232 sub pe {
233 #       return 0;
234         my $code=$conn->errcode();
235         my $msg=$conn->errmsg();
236         my $ai=$conn->addinfo();
237         print << "EOF";
238 CODE:  $code
239 MSG:   $msg
240 ADDTL: $ai
241 EOF
242         return 0;
243 }
244                                                 }
245                                         }
246                                         unless ($stillprocessing) {
247                                                 #my $sti=$dbh->prepare("select enddate from z3950queue where id=$id");
248                                                 #$sti->execute;
249                                                 #my ($enddate) = $sti->fetchrow;
250                                                 #unless ($enddate) {
251                                         }
252                                 } else {
253                                 }
254                         }
255                         $lastrun=time();
256                 }
257                 sleep 10;
258         }
259 }
260
261 sub reap {
262     $forkcounter--;
263     wait;
264 }
265
266
267 sub checkqueue {
268     $checkqueue=1;
269 }
270
271