1st draft for z3950 client import.
[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 $$."\n";
19         close PID;
20     }
21     # Get real apacheuser from koha.conf or reparsing httpd.conf
22     my $apacheuser='www-data';
23     my $uid=0;
24     unless ($uid = (getpwnam($apacheuser))[2]) {
25         die "Attempt to run daemon as non-existent or superuser\n";
26     }
27     $>=$uid;
28     $<=$uid;
29 }
30
31
32 my $dbh = C4::Context->dbh;
33
34 my $sth=$dbh->prepare("update z3950results set active=0");
35 $sth->execute;
36 $sth->finish;
37 $SIG{CHLD}='reap';
38 $SIG{HUP}='checkqueue';
39
40
41 my $logdir=$ARGV[0];
42
43 open PID, ">$logdir/processz3950queue.pid";
44 print PID $$."\n";
45 close PID;
46
47 my $reapcounter=0;
48 my $forkcounter=0;
49 my $checkqueue=1;
50 my $pid=$$;
51 my $lastrun=0;
52 while (1) {
53         if ((time-$lastrun)>5) {
54                 print "starting loop\n";
55                 if ($checkqueue) {
56                         print "checkqueue=1\n";
57                         $checkqueue=0;
58                         my $sth=$dbh->prepare("select id,term,type,servers from z3950queue order by id");
59                         $sth->execute;
60                         while (my ($id, $term, $type, $servers) = $sth->fetchrow) {
61                                 if ($forkcounter<12) {
62                                         my $now=time();
63                                         $stk=$dbh->prepare("select id,server,startdate,enddate,numrecords,active from z3950results where queryid=$id");
64                                         ($stk->execute) || (next);
65                                         my %serverdone;
66                                         unless ($stk->rows) {
67                                                 my $sti=$dbh->prepare("update z3950queue set done=-1,startdate=$now where id=$id");
68                                                 $sti->execute;
69                                         }
70                                         while (my ($r_id, $r_server,$r_startdate,$r_enddate,$r_numrecords,$active) = $stk->fetchrow) {
71                                                 if ($r_enddate >0) {
72                                                         $serverdone{$r_server}=1;
73                                                 } elsif ($active) {
74                                                         $serverdone{$r_server}=1;
75                                                 } else {
76                                                         $serverdone{$r_server}=-1;
77                                                 }
78                                         }
79
80                                         $stk->finish;
81                                         my $attr='';
82                                         if ($type eq 'isbn') {
83                                                 $attr='1=7';
84                                         } elsif ($type eq 'title') {
85                                                 $attr='1=4';
86                                         } elsif ($type eq 'author') {
87                                                 $attr='1=1003';
88                                         } elsif ($type eq 'lccn') {
89                                                 $attr='1=9';
90                                         } elsif ($type eq 'keyword') {
91                                                 $attr='1=1016';
92                                         }
93                                         $term='"'.$term.'"';
94                                         $query="\@attr $attr $term";
95                                         my $totalrecords=0;
96                                         my $serverinfo;
97                                         my $stillprocessing=0;
98                                         foreach $serverinfo (split(/\s+/, $servers)) {
99                                                 (next) if ($serverdone{$serverinfo} == 1);
100                                                 my $stillprocessing=1;
101                                                 if (my $pid=fork()) {
102                                                         $forkcounter++;
103                                                 } else {
104                                                         #$sth->finish;
105                                                         #$sti->finish;
106                                                         #$dbh->disconnect;
107                                                         my $dbi = C4::Context->dbh;
108                                                         my ($name, $server, $database, $user, $password) = split(/\//, $serverinfo, 5);
109                                                         $server=~/(.*)\:(\d+)/;
110                                                         my $servername=$1;
111                                                         my $port=$2;
112                                                         print "Processing $type=$term at $name $server $database (".($forkcounter+1)." forks)\n";
113                                                         $now=time();
114                                                         my $q_serverinfo=$dbi->quote($serverinfo);
115                                                         my $resultsid;
116                                                         if ($serverdone{$serverinfo}==-1) {
117                                                                 my $stj=$dbi->prepare("select id from z3950results where server=$q_serverinfo and queryid=$id");
118                                                                 $stj->execute;
119                                                                 ($resultsid) = $stj->fetchrow;
120                                                         } else {
121                                                                 my $stj=$dbi->prepare("select id from z3950results where server=$q_serverinfo and queryid=$id");
122                                                                 $stj->execute;
123                                                                 ($resultsid) = $stj->fetchrow;
124                                                                 unless ($resultsid) {
125                                                                         my $stj=$dbi->prepare("insert into z3950results (server, queryid, startdate) values ($q_serverinfo, $id, $now)");
126                                                                         $stj->execute;
127                                                                         $resultsid=$dbi->{'mysql_insertid'};
128                                                                 }
129                                                         }
130                                                         my $stj=$dbh->prepare("update z3950results set active=1 where id=$resultsid");
131                                                         $stj->execute;
132                                                         my $conn;
133                                                         my $noconnection=0;
134                                                         my $error=0;
135                                                         if ($user) {
136                                                                 eval { $conn= new Net::Z3950::Connection($servername, $port, databaseName => $database, user => $user, password => $password); };
137                                                                 if ($@) {
138                                                                         $noconnection=1;
139                                                                 } else {
140                                                                         $error=pe();
141                                                                 }
142                                                         } else {
143                                                                 eval { $conn= new Net::Z3950::Connection($servername, $port, databaseName => $database); };
144                                                                 if ($@) {
145                                                                         $noconnection=1;
146                                                                 } else {
147                                                                         $error=pe();
148                                                                 }
149                                                         }
150                                                         if ($noconnection || $error) {
151                                                         } else {
152                                                                 eval { $conn->option(preferredRecordSyntax => Net::Z3950::RecordSyntax::USMARC);};
153                                                                 if ($@) {
154                                                                         print "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->rawdata();
167                                                                                 $result.=$marcdata;
168                                                                         }
169                                                                         my $scantimerend=time();
170                                                                         my $numrecords;
171                                                                         ($numresults<80) ? ($numrecords=$numresults) : ($numrecords=80);
172                                                                         my $elapsed=$scantimerend-$scantimerstart;
173                                                                         if ($elapsed) {
174                                                                                 my $speed=int($numresults/$elapsed*100)/100;
175                                                                                 print "  SPEED: $speed  $server done $numrecords\n";
176                                                                         }
177
178                                                                         my $q_result=$dbi->quote($result);
179                                                                         ($q_result) || ($q_result='""');
180                                                                         $now=time();
181                                                                         my $task="update z3950results set numrecords=$numresults,numdownloaded=$numrecords,highestseen=0,results=$q_result,enddate=$now where id=$resultsid";
182                                                                         my $stj=$dbi->prepare($task);
183                                                                         $stj->execute;
184                                                                         ImportBreeding($q_result,1,"Z3950");
185                                                                         my $counter=0;
186                                                                         while ($counter<60 && $numrecords<$numresults) {
187                                                                                 $counter++;
188                                                                                 my $stj=$dbi->prepare("select highestseen from z3950results where id=$resultsid");
189                                                                                 $stj->execute;
190                                                                                 my ($highestseen) = $stj->fetchrow;
191                                                                                 if ($highestseen>($numrecords-30)) {
192                                                                                         $counter=0;
193                                                                                         print "   $server rescanning\n";
194                                                                                         my $scantimerstart=time();
195                                                                                         for ($i=$numrecords+1; $i<=(($numresults<($numrecords+40)) ? ($numresults) : ($numrecords+40)); $i++) {
196                                                                                                 my $rec=$rs->record($i);
197                                                                                                 my $marcdata=$rec->rawdata();
198                                                                                                 $result.=$marcdata;
199                                                                                         }
200                                                                                         my $scantimerend=time();
201                                                                                         ($numresults<$numrecords+40) ? ($numrecords=$numresults) : ($numrecords += 40);
202                                                                                         my $elapsed=$scantimerend-$scantimerstart;
203                                                                                         if ($elapsed) {
204                                                                                                 my $speed=int($numresults/$elapsed*100)/100;
205                                                                                                 print "  SPEED: $speed  $server done $numrecords\n";
206                                                                                         }
207
208                                                                                         my $q_result=$dbi->quote($result);
209                                                                                         ($q_result) || ($q_result='""');
210                                                                                         $now=time();
211                                                                                         my $task="update z3950results set numdownloaded=$numrecords,results=$q_result where id=$resultsid";
212                                                                                         my $stj=$dbi->prepare($task);
213                                                                                         $stj->execute;
214                                                                                 }
215                                                                                 sleep 5;
216                                                                         }
217                                                                 }
218                                                         }
219                                                         # FIXME - There's already a $stj in this scope
220                                                         my $stj=$dbi->prepare("update z3950results set active=0 where id=$resultsid");
221                                                         $stj->execute;
222                                                         eval {$stj->finish};
223                                                         print "    $server done.\n";
224                                                         exit;
225 sub pe {
226         return 0;
227         my $code=$conn->errcode();
228         my $msg=$conn->errmsg();
229         my $ai=$conn->addinfo();
230         print << "EOF";
231 CODE:  $code
232 MSG:   $msg
233 ADDTL: $ai
234 EOF
235                                                         return 0;
236                                                         }
237                                                 }
238                                         }
239                                         unless ($stillprocessing) {
240                                                 #my $sti=$dbh->prepare("select enddate from z3950queue where id=$id");
241                                                 #$sti->execute;
242                                                 #my ($enddate) = $sti->fetchrow;
243                                                 #unless ($enddate) {
244                                         }
245                                 } else {
246                                 }
247                         }
248                         $lastrun=time();
249                 }
250                 sleep 10;
251         }
252 }
253
254 # sub getrecord {
255 #     my $server=shift;
256 #     my $base=shift;
257 #     my $query=shift;
258 #     my $auth=shift;
259 #     my $id=shift;
260 #     open  (M, "|yaz-client -m yaz-$id.mrc >>yaz.out 2>>yaz.err");
261 #     select M;
262 #     $|=1;
263 #     select STDOUT;
264 #     ($auth) && ($auth="authentication $auth\n");
265 #     print M << "EOF";
266 # $auth\open $server
267 # base $base
268 # setnames
269 # $query
270 # s
271 # s
272 # s
273 # s
274 # s
275 # s
276 # s
277 # s
278 # s
279 # s
280 # quit
281 # EOF
282 #     close M;
283 # }
284 sub reap {
285     $forkcounter--;
286     wait;
287 }
288
289
290 sub checkqueue {
291     $checkqueue=1;
292 }
293
294