source: main/trunk/greenstone2/bin/script/activate.pl@ 28088

Last change on this file since 28088 was 28088, checked in by davidb, 11 years ago

Additional print messages added to show the decision making process activate.pl goes through
  • Property svn:executable set to *
File size: 30.6 KB
Line 
1#!/usr/bin/perl -w
2
3###########################################################################
4#
5# activate.pl -- to be called after building a collection to activate it.
6#
7# A component of the Greenstone digital library software
8# from the New Zealand Digital Library Project at the
9# University of Waikato, New Zealand.
10#
11# Copyright (C) 2009 New Zealand Digital Library Project
12#
13# This program is free software; you can redistribute it and/or modify
14# it under the terms of the GNU General Public License as published by
15# the Free Software Foundation; either version 2 of the License, or
16# (at your option) any later version.
17#
18# This program is distributed in the hope that it will be useful,
19# but WITHOUT ANY WARRANTY; without even the implied warranty of
20# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
21# GNU General Public License for more details.
22#
23# You should have received a copy of the GNU General Public License
24# along with this program; if not, write to the Free Software
25# Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
26#
27###########################################################################
28
29
30# This program is designed to support the building process of Greenstone.
31# It deactivates the collection just built, if the web server is running
32# and is a persistent web server (or if the library_URL provided as
33# parameter to this script is of a currently running web server). It then
34# moves building to index, before activating the collection on the GS2 or
35# GS3 web server again if necessary.
36
37use Config;
38
39BEGIN {
40 die "GSDLHOME not set\n" unless defined $ENV{'GSDLHOME'};
41 die "GSDLOS not set\n" unless defined $ENV{'GSDLOS'};
42 unshift (@INC, "$ENV{'GSDLHOME'}/perllib");
43 unshift (@INC, "$ENV{'GSDLHOME'}/perllib/cpan");
44
45 # Adding cpan in, adds in its auto subfolder which conflicts with ActivePerl on Windows
46 # The auto folder has been moved into a perl-5.8 folder, and this will now be included
47 # only if the current version of perl is 5.8 (and not ActivePerl).
48 my $perl_dir;
49
50 # Note: $] encodes the version number of perl
51 if ($]>=5.010) {
52 $perl_dir="perl-5.".substr($],3,2);
53 }
54 elsif ($]>5.008) {
55 # perl 5.8.1 or above
56 $perl_dir = "perl-5.8";
57 }
58 elsif ($]>=5.008) {
59 # perl 5.8.1 or above
60 $perl_dir = "perl-5.8";
61 }
62 elsif ($]<5.008) {
63 # assume perl 5.6
64 $perl_dir = "perl-5.6";
65 }
66 else {
67 print STDERR "Warning: Perl 5.8.0 is not a maintained release.\n";
68 print STDERR " Please upgrade to a newer version of Perl.\n";
69 $perl_dir = "perl-5.8";
70 }
71
72 #if ($ENV{'GSDLOS'} !~ /^windows$/i) {
73 # Use push to put this on the end, so an existing XML::Parser will be used by default
74 if (-d "$ENV{'GSDLHOME'}/perllib/cpan/$perl_dir-mt" && $Config{usethreads}){
75 push (@INC, "$ENV{'GSDLHOME'}/perllib/cpan/$perl_dir-mt");
76 }
77 else{
78 push (@INC, "$ENV{'GSDLHOME'}/perllib/cpan/$perl_dir");
79 }
80 #}
81
82}
83
84
85use strict;
86no strict 'refs'; # allow filehandles to be variables and vice versa
87no strict 'subs'; # allow barewords (eg STDERR) as function arguments
88
89use File::Basename;
90use File::Find;
91
92use HTTP::Response;
93use LWP::Simple qw($ua !head); # import useragent object as $ua from the full LWP to use along with LWP::Simple
94 # don't import LWP::Simple's head function by name since it can conflict with CGI:head())
95#use CGI qw(:standard); # then only CGI.pm defines a head()
96use Net::Ping;
97use URI;
98
99use colcfg;
100use scriptutil;
101use util;
102#use enum;
103
104# enumerations in perl, http://stackoverflow.com/questions/473666/does-perl-have-an-enumeration-type
105# Unfortunately, not part of perl's core
106#use enum qw(LEVEL_NONE LEVEL_ERROR LEVEL_INFO LEVEL_DEBUG); # debugging levels NONE == 0, ERROR=1 INFO=2 DEBUG=3
107
108# global variables
109#my $default_verbosity = LEVEL_ERROR; # by default we display basic error messages
110
111my $default_verbosity = 2; # by default we display basic error and info messages
112
113sub print_task_msg {
114 my ($task_msg, $verbosity_setting) = @_;
115
116 $verbosity_setting = $default_verbosity unless $verbosity_setting;
117 #$verbosity_setting = 1 unless defined $verbosity;
118 if($verbosity_setting >= 1) {
119 print STDERR "\n";
120 print STDERR "************************\n";
121 print STDERR "* $task_msg\n";
122 print STDERR "************************\n";
123 }
124}
125
126# Prints messages if the verbosity is right. Does not add new lines.
127sub print_msg {
128 my ($msg, $min_verbosity, $verbosity_setting) = @_;
129
130 # only display error messages if the current
131 # verbosity setting >= the minimum verbosity level
132 # needed for that message to be displayed.
133
134 $verbosity_setting = $default_verbosity unless defined $verbosity_setting;
135 $min_verbosity = 1 unless defined $min_verbosity;
136 if($verbosity_setting >= $min_verbosity) { # by default display all 1 messages
137 print STDERR "$msg";
138 }
139}
140
141# Method to send a command to a GS2 or GS3 library_URL
142# the commands used in this script can be activate, deactivate, ping,
143# and is-persistent (is-persistent only implemented for GS2).
144sub config {
145 my ($library_url, $command, $check_message_against_regex, $site, $expected_error_code) = @_;
146 # Gatherer.java's configGS3Server doesn't use the site variable
147 # so we don't have to either
148
149 # for GS2, getting the HTTP status isn't enough, we need to read the output
150 # since this is what CollectionManager.config() stipulates.
151 # Using LWP::UserAgent::get($url) for this
152
153 if(!defined $library_url) {
154 return 0;
155 }
156 else {
157 $ua->timeout(5); # set LWP useragent to 5s max timeout for testing the URL
158 # Need to set this, else it takes I don't know how long to timeout
159 # http://www.perlmonks.org/?node_id=618534
160
161 # http://search.cpan.org/~gaas/libwww-perl-6.04/lib/LWP/UserAgent.pm
162 # use LWP::UserAgent's get($url) since it returns an HTTP::Response code
163
164 my $response_obj = $ua->get( $library_url.$command);
165
166 # $response_obj->content stores the content and $response_obj->code the HTTP response code
167 my $response_code = $response_obj->code();
168
169 if(LWP::Simple::is_success($response_code)) {# $response_code eq RC_OK) { # LWP::Simple::is_success($response_code)
170 &print_msg("*** Command $library_url$command\n", 3);
171 &print_msg("*** HTTP Response Status: $response_code - Complete.", 3);
172
173 # check the page content is as expected
174 my $response_content = $response_obj->content;
175 if($response_content =~ m/$check_message_against_regex/) {
176 &print_msg(" Response as expected.\n", 3);
177 return 1;
178 } else {
179 &print_msg("\n\tBUT: command $library_url$command response UNEXPECTED.\n", 3);
180 &print_msg("*** Got message:\n$response_content.\n", 4);
181 return 0; # ping on a collection may "not succeed."
182 }
183 }
184 elsif(LWP::Simple::is_error($response_code)) { # method exported by LWP::Simple, along with HTTP::Status constants
185 # check the page content is as expected
186 if(defined $expected_error_code && $response_code == $expected_error_code) {
187 &print_msg(" Response status $response_code as expected.\n", 3);
188 } else {
189 &print_msg("*** Command $library_url$command\n");
190 &print_msg("*** Unexpected error. HTTP Response Status: $response_code - Failed.\n");
191 }
192 return 0; # return false, since the response_code was an error, expected or not
193 }
194 else {
195 &print_msg("*** Command $library_url$command\n");
196 &print_msg("*** Unexpected error. HTTP Response Status: $response_code - Failed.\n");
197 return 0;
198 }
199 }
200}
201
202sub deactivate_collection {
203 my ($library_url, $gs_mode, $qualified_collection, $site) = @_;
204
205 if($gs_mode eq "gs2") {
206 my $DEACTIVATE_COMMAND = "?a=config&cmd=release-collection&c=";
207 my $check_message_against_regex = q/configured release-collection/;
208 config($library_url, $DEACTIVATE_COMMAND.$qualified_collection, $check_message_against_regex);
209 }
210 elsif ($gs_mode eq "gs3") {
211 my $DEACTIVATE_COMMAND = "?a=s&sa=d&st=collection&sn=";
212 my $check_message_against_regex = "collection: $qualified_collection deactivated";
213 config($library_url, $DEACTIVATE_COMMAND.$qualified_collection, $check_message_against_regex, $site);
214 }
215}
216
217sub activate_collection {
218 my ($library_url, $gs_mode, $qualified_collection, $site) = @_;
219
220 if($gs_mode eq "gs2") {
221 my $ACTIVATE_COMMAND = "?a=config&cmd=add-collection&c=";
222 my $check_message_against_regex = q/configured add-collection/;
223 config($library_url, $ACTIVATE_COMMAND.$qualified_collection, $check_message_against_regex);
224 }
225 elsif ($gs_mode eq "gs3") {
226 my $ACTIVATE_COMMAND = "?a=s&sa=a&st=collection&sn=";
227 my $check_message_against_regex = "collection: $qualified_collection activated";
228 config($library_url, $ACTIVATE_COMMAND.$qualified_collection, $check_message_against_regex, $site);
229 }
230}
231
232sub ping {
233 my ($library_url, $command, $gs_mode, $site) = @_;
234
235 # If the GS server is not running, we *expect* to see a "500" status code.
236 # If the GS server is running, then "Ping" ... "succeeded" is expected on success.
237 # When pinging an inactive collection, it will say it did "not succeed". This is
238 # a message of interest to return.
239 my $check_responsemsg_against_regex = q/(succeeded)/;
240 my $expected_error_code = 500;
241 return config($library_url, $command, $check_responsemsg_against_regex, $site, $expected_error_code);
242}
243
244# send a pingaction to the GS library. General server-level ping.
245sub ping_library {
246 my ($library_url, $gs_mode, $site) = @_;
247
248 my $command = "";
249 if($gs_mode eq "gs2") {
250 $command = "?a=ping";
251 }
252 elsif ($gs_mode eq "gs3") {
253 $command = "?a=s&sa=ping";
254 }
255 return &ping($library_url, $command, $gs_mode, $site);
256}
257
258
259# send a pingaction to a collection in GS library to check if it's active
260sub ping_library_collection {
261 my ($library_url, $gs_mode, $qualified_collection, $site) = @_;
262
263 my $command = "";
264 if($gs_mode eq "gs2") {
265 $command = "?a=ping&c=$qualified_collection";
266 }
267 elsif ($gs_mode eq "gs3") {
268 $command = "?a=s&sa=ping&st=collection&sn=$qualified_collection";
269 }
270 return &ping($library_url, $command, $gs_mode, $site);
271}
272
273# return true if server is persistent, by calling is-persistent on library_url
274# this is only for GS2, since the GS3 server is always persistent
275sub is_persistent {
276 my ($library_url, $gs_mode) = @_;
277
278 if($gs_mode eq "gs3") { # GS3 server is always persistent
279 return 1;
280 }
281
282 my $command = "?a=is-persistent";
283 my $check_responsemsg_against_regex = q/true/; # isPersistent: true versus isPersistent: false
284 return config($library_url, $command, $check_responsemsg_against_regex);
285}
286
287sub get_library_URL {
288 my $gs_mode = shift(@_); # gs3 or gs2
289
290 # If we get here, we are dealing with a server included with GS.
291 # For GS3, we ask ant for the library URL.
292 # For GS2, we derive the URL from the llssite.cfg file.
293
294 my $url = undef;
295
296 if($gs_mode eq "gs2") {
297 my $llssite_cfg = &FileUtils::filenameConcatenate($ENV{'GSDLHOME'}, "llssite.cfg");
298
299 if(-f $llssite_cfg) {
300 # check llssite.cfg for line with url property
301 # for server.exe also need to use portnumber and enterlib properties
302
303 # Read in the entire contents of the file in one hit
304 if (!open (FIN, $llssite_cfg)) {
305 &print_msg("activate.pl::get_library_URL failed to open $llssite_cfg ($!)\n");
306 return undef;
307 }
308
309 my $contents;
310 sysread(FIN, $contents, -s FIN);
311 close(FIN);
312
313 my @lines = split(/[\n\r]+/, $contents); # split on carriage-returns and/or linefeeds
314 my $enterlib = "";
315 my $portnumber = ""; # will remain empty (implicit port 80) unless it's specifically been assigned
316
317 foreach my $line (@lines) {
318 if($line =~ m/^url=(.*)$/) {
319 $url = $1;
320 } elsif($line =~ m/^enterlib=(.*)$/) {
321 $enterlib = $1;
322 } elsif($line =~ m/^portnumber=(.*)$/) {
323 $portnumber = $1;
324 }
325 }
326
327 if(!$url) {
328 return undef;
329 }
330 elsif($url eq "URL_pending") { # library is not running
331 # do not process url=URL_pending in the file, since for server.exe
332 # this just means the Enter Library button hasn't been pressed yet
333 $url = undef;
334 }
335 else {
336 # In the case of server.exe, need to do extra work to get the proper URL
337 # But first, need to know whether we're indeed dealing with server.exe:
338
339 # compare the URL's domain to the full URL
340 # E.g. for http://localhost:8383/greenstone3/cgi-bin, the domain is localhost:8383
341 my $uri = URI->new( $url );
342 my $host = $uri->host;
343 #print STDERR "@@@@@ host: $host\n";
344 if($url =~ m/http:\/\/$host(\/)?$/) {
345 #if($url !~ m/http:\/\/$host:$portnumber(\/)?/ || $url =~ m/http:\/\/$host(\/)?$/) {
346 # (if the URL does not contain the portnumber, OR if the port is implicitly 80 and)
347 # If the domain with http:// prefix is completely the same as the URL, assume server.exe
348 # then the actual URL is the result of suffixing the port and enterlib properties in llssite.cfg
349 $url = $url.":".$portnumber.$enterlib;
350 } # else, apache web server
351
352 }
353 }
354 } elsif($gs_mode eq "gs3") {
355 # Either check build.properties for tomcat.server, tomcat.port and app.name (and default servlet name).
356 # app.name is stored in app.path by build.xml. Need to move app.name in build.properties from build.xml
357
358 # Or, run the new target get-default-servlet-url
359 # the output can look like:
360 #
361 # Buildfile: build.xml
362 # [echo] os.name: Windows Vista
363 #
364 # get-default-servlet-url:
365 # [echo] http://localhost:8383/greenstone3/library
366 # BUILD SUCCESSFUL
367 # Total time: 0 seconds
368
369 #my $output = qx/ant get-default-servlet-url/; # backtick operator, to get STDOUT (else 2>&1)
370 # see http://stackoverflow.com/questions/799968/whats-the-difference-between-perls-backticks-system-and-exec
371
372 # The get-default-servlet-url ant target can be run from anywhere by specifying the
373 # location of GS3's ant build.xml buildfile. Activate.pl can be run from anywhere for GS3
374 # GSDL3SRCHOME will be set for GS3 by gs3-setup.sh, a step that would have been necessary
375 # to run the activate.pl script in the first place
376 my $perl_command = "ant -buildfile \"$ENV{'GSDL3SRCHOME'}/build.xml\" get-default-servlet-url";
377
378 if (open(PIN, "$perl_command |")) {
379 while (defined (my $perl_output_line = <PIN>)) {
380 if($perl_output_line =~ m@http:\/\/(\S*)@) { # grab all the non-whitespace chars
381 $url="http://".$1;
382 }
383 }
384 close(PIN);
385 } else {
386 &print_msg("activate.pl::get_library_URL: Failed to run $perl_command to work out library URL for $gs_mode\n");
387 }
388 }
389
390 # either the url is still undef or it is now set
391 #print STDERR "\n@@@@@ final URL:|$url|\n" if $url;
392 #print STDERR "\n@@@@@ URL still undef\n" if !$url;
393 return $url;
394}
395
396### UNUSED METHODS TO MOVE TO util.pm?
397
398# This method is now unused. Using ping_library instead to send the ping action to a
399# GS2/GS3 server. This method can be used more generally to test whether a URL is alive.
400# http://search.cpan.org/dist/libwww-perl/lib/LWP/Simple.pm
401# and http://www.perlmonks.org/?node_id=618534
402sub is_URL_active {
403 my $url = shift(@_); # gs3 or gs2 URL
404
405 my $status = 0;
406 if(defined $url) {
407 $ua->timeout(5); # set LWP useragent to 5s max timeout for testing the URL
408 # Need to set this, else it takes I don't know how long to timeout
409 # http://www.perlmonks.org/?node_id=618534
410
411 $status = LWP::Simple::head($url); # returns empty list of headers if it fails
412 # LWP::Simple::get($url) is more intensive, so don't need to do that
413 #print STDERR "**** $url is alive.\n" if $status;
414 }
415 return $status;
416}
417
418# Pinging seems to always return true, so this method doesn't work
419sub pingHost {
420 my $url = shift(@_); # gs3 or gs2 URL
421
422 my $status = 0;
423 if(defined $url) {
424 # Get just the domain. "http://localhost/gsdl?uq=332033495" becomes "localhost"
425 # "http://localhost/greenstone/cgi-bin/library.cgi" becomes "localhost" too
426
427 #my $host = $url;
428 #$host =~ s@^http:\/\/(www.)?@@;
429 #$host =~ s@\/.*@@;
430 #print STDERR "**** HOST: $host\n";
431
432 # More robust way
433 # http://stackoverflow.com/questions/827024/how-do-i-extract-the-domain-out-of-an-url
434 my $uri = URI->new( $url );
435 my $host = $uri->host;
436
437 # Ping the host. http://perldoc.perl.org/Net/Ping.html
438 my $p = Net::Ping->new();
439 $status = $p->ping($host); # || 0. Appears to set to undef rather than 0
440 print STDERR "**** $host is alive.\n" if $status; #print "$host is alive.\n" if $p->ping($host);
441 $p->close();
442 }
443 # return whether pinging was a success or failure
444 return $status;
445}
446
447
448# Most of the arguments are familiar from the building scripts like buildcol.pl
449# The special optional argument -library_url is for when we're dealing with a web
450# library server such as an apache that's separate from any included with GS2.
451# In such a case, this script's caller should pass in -library_url <URL>.
452#
453# $site argument must be specified in the cmdline for collectionConfig.xml to get
454# generated which makes $gs_mode=gs3, else collect.cfg gets generated and $gs_mode=gs2
455sub main
456{
457 my ($argc,@argv) = @_;
458
459 if (($argc==0) || (($argc==1) && ($argv[0] =~ m/^--?h(elp)?$/))) {
460 my ($progname) = ($0 =~ m/^.*[\/|\\](.*?)$/);
461
462
463 print STDERR "\n";
464 print STDERR "Usage: $progname [-collectdir c -builddir b -indexdir i -site s -removeold -keepold -verbosity v\n";
465 print STDERR "\t-library_url URL] <[colgroup/]collection>\n";
466 print STDERR "\n";
467
468 exit(-1);
469 }
470
471 # get the collection details
472 my $qualified_collection = pop @argv; # qualified collection
473
474 my $collect_dir = undef; #"collect"; # can't be "collect" when only -site is provided for GS3
475 my $build_dir = undef;
476 my $index_dir = undef;
477 my $site = undef;
478
479 my $removeold = 0;
480 my $keepold = 0;
481 my $incremental = 0; # used by solr
482
483 my $library_url = undef; # to be specified on the cmdline if not using a GS-included web server
484
485 while (my $arg = shift @argv) {
486 if ($arg eq "-collectdir") {
487 $collect_dir = shift @argv;
488 }
489 elsif ($arg eq "-builddir") {
490 $build_dir = shift @argv;
491 }
492 elsif ($arg eq "-indexdir") {
493 $index_dir = shift @argv;
494 }
495 elsif ($arg eq "-site") {
496 $site = shift @argv;
497 }
498 elsif ($arg eq "-removeold") {
499 $removeold = 1;
500 }
501 elsif ($arg eq "-keepold") {
502 $keepold = 1;
503 }
504 elsif ($arg eq "-incremental") {
505 $incremental = 1;
506 }
507 elsif ($arg eq "-library_url") {
508 $library_url = shift @argv;
509 }
510 elsif ($arg eq "-verbosity") {
511 $default_verbosity = shift @argv; # global variable
512
513 # ensure we're working with ints not strings (int context not str context), in case verbosity=0
514 # http://stackoverflow.com/questions/288900/how-can-i-convert-a-string-to-a-number-in-perl
515 $default_verbosity = int($default_verbosity || 0); ### is this the best way?
516 }
517 }
518
519 # work out the building and index dirs
520 my $collection_dir = &util::resolve_collection_dir($collect_dir, $qualified_collection, $site);
521 $build_dir = &FileUtils::filenameConcatenate($collection_dir, "building") unless (defined $build_dir);
522 $index_dir = &FileUtils::filenameConcatenate($collection_dir, "index") unless (defined $index_dir);
523
524 &print_task_msg("Running Collection Activation Stage");
525
526 # get and check the collection name
527 if ((&colcfg::use_collection($site, $qualified_collection, $collect_dir)) eq "") {
528 &print_msg("Unable to use collection \"$qualified_collection\" within \"$collect_dir\"\n");
529 exit -1;
530 }
531
532 # Read in the collection configuration file.
533 # Beware: Only if $site is specified in the cmdline does collectionConfig.xml get
534 # generated and does $gs_mode=gs3, else collect.cfg gets generated and $gs_mode=gs2
535 my $gs_mode = "gs2";
536 if ((defined $site) && ($site ne "")) { # GS3
537 $gs_mode = "gs3";
538 }
539 my $collect_cfg_filename = &colcfg::get_collect_cfg_name(STDERR, $gs_mode);
540 my $collectcfg = &colcfg::read_collection_cfg ($collect_cfg_filename,$gs_mode);
541
542 # look for build.cfg/buildConfig.xml
543 my $build_cfg_filename ="";
544
545 if ($gs_mode eq "gs2") {
546 $build_cfg_filename = &FileUtils::filenameConcatenate($build_dir,"build.cfg");
547 } else {
548 $build_cfg_filename = &FileUtils::filenameConcatenate($build_dir, "buildConfig.xml");
549 # gs_mode is GS3. Set the site now if this was not specified as cmdline argument
550 #$site = "localsite" unless defined $site;
551 }
552
553 # We need to know the buildtype for Solr.
554 # Any change of indexers is already detected and handled by the calling code (buildcol or
555 # full-rebuild), so that at this stage the config file's buildtype reflects the actual buildtype.
556
557 # From buildcol.pl we use searchtype for determining buildtype, but for old versions, use buildtype
558 my $buildtype;
559 if (defined $collectcfg->{'buildtype'}) {
560 $buildtype = $collectcfg->{'buildtype'};
561 } elsif (defined $collectcfg->{'searchtypes'} || defined $collectcfg->{'searchtype'}) {
562 $buildtype = "mgpp";
563 } else {
564 $buildtype = "mg"; #mg is the default
565 }
566
567 # can't do anything without a build directory with something in it to move into index
568 # Except if we're (doing incremental) building for solr, where we want to still
569 # activate and deactivate collections including for the incremental case
570 if(!&FileUtils::directoryExists($build_dir)) {
571 &print_msg("No building folder at $build_dir to move to index.\n");
572 exit -1 unless ($buildtype eq "solr"); #&& $incremental);
573 } elsif (&FileUtils::isDirectoryEmpty($build_dir)) {
574 &print_msg("Nothing in building folder $build_dir to move into index folder.\n");
575 exit -1 unless ($buildtype eq "solr"); #&& $incremental);
576 }
577
578
579 my $solr_server;
580 my @corenames = ();
581 if($buildtype eq "solr") { # start up the jetty server
582 my $solr_ext = $ENV{'GEXT_SOLR'}; # from solr_passes.pl
583 unshift (@INC, "$solr_ext/perllib");
584 require solrserver;
585
586 # Solr cores are named without taking the collection-group name into account, since solr
587 # is used for GS3 and GS3 doesn't use collection groups but has the site concept instead
588 my ($colname, $colgroup) = &util::get_collection_parts($qualified_collection);
589
590 # See solrbuilder.pm to get the indexing levels (document, section) from the collectcfg file
591 # Used to generate core names from them and remove cores by name
592 foreach my $level ( @{$collectcfg->{'levels'}} ){
593 my ($pindex) = $level =~ /^(.)/;
594 my $indexname = $pindex."idx";
595 push(@corenames, "$site-$colname-$indexname"); #"$site-$colname-didx", "$site-$colname-sidx"
596 }
597
598 # If the Solr/Jetty server is not already running, the following starts
599 # it up, and only returns when the server is "reading and listening"
600 $solr_server = new solrserver($build_dir);
601 $solr_server->start();
602
603 # We'll be moving building to index. For solr collection, there's further
604 # special processing to make a corresponding change to the solr.xml
605 # by removing the temporary building cores and (re)creating the index cores
606 }
607
608 # Now the logic in GLI's CollectionManager.java (processComplete()
609 # and installCollection()) and Gatherer.configGS3Server().
610
611 # 1. Get library URL
612
613 # For web servers that are external to a Greenstone installation,
614 # the user can pass in their web server's library URL.
615 # For web servers included with GS (like tomcat for GS3 and server.exe
616 # and apache for GS2), we work out the library URL:
617 if(!$library_url) {
618 $library_url = &get_library_URL($gs_mode); # returns undef if no server is running
619 }
620
621 # CollectionManager's installCollection phase in GLI
622 # 2. Ping the library URL, and if it's a persistent server and running, release the collection
623
624 my $is_persistent_server = undef;
625 if($library_url) { # undef if no valid server URL
626
627 &print_msg("Pinging $library_url\n");
628 if (&ping_library($library_url, $gs_mode, $site)) { # server running
629
630 # server is running, so release the collection if
631 # the server is persistent and the collection is active
632 &print_msg("Checking if Greenstone server is persistent\n");
633 $is_persistent_server = &is_persistent($library_url, $gs_mode);
634
635 if ($is_persistent_server) { # only makes sense to issue activate and deactivate cmds to a persistent server
636
637 &print_msg("Checking if the collection $qualified_collection is already active\n");
638 my $collection_active = &ping_library_collection($library_url, $gs_mode, $qualified_collection, $site);
639
640 if ($collection_active) {
641 &print_msg("De-activating collection $qualified_collection\n");
642 &deactivate_collection($library_url, $gs_mode, $qualified_collection, $site);
643 }
644 else {
645 &print_msg("Collection is not active => No need to deactivate\n");
646 }
647 }
648 else {
649 &print_msg("Server is not persistent => No need to deactivate collection\n");
650 }
651 }
652 else {
653 &print_msg("No response to Ping => Taken to mean server is not running\n");
654 }
655
656 }
657
658 # 3. Do all the moving building to index stuff now
659
660 # If removeold: replace index dir with building dir.
661 # If keepold: move building's contents into index, where only duplicates will get deleted.
662 # removeold and keepold can't both be on at the same time
663 # incremental becomes relevant for solr, though it was irrelevant to what activate.pl does (moving building to index)
664 my $incremental_mode;
665 ($removeold, $keepold, $incremental, $incremental_mode) = &scriptutil::check_removeold_and_keepold($removeold, $keepold,
666 $incremental,
667 $build_dir, # checkdir. Usually archives or export to be deleted. activate.pl deletes building
668 $collectcfg);
669
670 if($removeold) {
671
672 if(&FileUtils::directoryExists($index_dir)) {
673 &print_task_msg("Removing \"index\"");
674
675 if ($buildtype eq "solr") {
676 # if solr, remove any cores that are using the index_dir before deleting this dir
677 foreach my $corename (@corenames) {
678 $solr_server->admin_unload_core($corename);
679 }
680 }
681
682 &FileUtils::removeFilesRecursive($index_dir);
683
684 # Wait for a couple of seconds, just for luck
685 sleep 2;
686
687 if (&FileUtils::directoryExists($index_dir)) {
688 &print_msg("The index directory $index_dir could not be deleted.\n"); # CollectionManager.Index_Not_Deleted
689 }
690 }
691
692 # if remote GS server: gliserver.pl would call activate.pl to activate
693 # the collection at this point since activate.pl lives on the server side
694
695 if ($buildtype eq "solr") {
696 # if solr, remove any cores that are using the building_dir before moving this dir onto index
697 foreach my $corename (@corenames) {
698 $solr_server->admin_unload_core("building-$corename");
699 }
700 }
701
702 # Move the building directory to become the new index directory
703 &print_task_msg("Moving \"building\" -> \"index\"");
704 &FileUtils::moveFiles($build_dir, $index_dir);
705 if(&FileUtils::directoryExists($build_dir) || !&FileUtils::directoryExists($index_dir)) {
706 &print_msg("Could not move $build_dir to $index_dir.\n"); # CollectionManager.Build_Not_Moved
707 }
708 }
709 elsif ($keepold || $incremental) {
710 if ($buildtype eq "solr") {
711 # if solr, remove any cores that may be using the building_dir before moving this dir onto index
712 foreach my $corename (@corenames) {
713 $solr_server->admin_unload_core("building-$corename") if $solr_server->admin_ping_core("building-$corename");
714 }
715 }
716
717 # Copy just the contents of building dir into the index dir, overwriting
718 # existing files, but don't replace index with building.
719 &print_task_msg("Moving \"building\" -> \"index\"");
720 &FileUtils::moveDirectoryContents($build_dir, $index_dir);
721 }
722
723 if ($buildtype eq "solr") {
724 # Call CREATE action to get the old cores pointing to the index folder
725 foreach my $corename (@corenames) {
726 if($removeold) {
727 # Call CREATE action to get all cores pointing to the index folder, since building is now index
728 $solr_server->admin_create_core($corename, $index_dir);
729
730 } elsif ($keepold || $incremental) {
731 # Call RELOAD core. Should already be using the index_dir directory for $keepold and $incremental case
732
733 # Ping to see if corename exists, if it does, reload, else create
734 if ($solr_server->admin_ping_core($corename)) {
735 $solr_server->admin_reload_core($corename);
736 } else {
737 $solr_server->admin_create_core($corename, $index_dir);
738 }
739 }
740 }
741
742 # copy the just updated ext/solr/solr.xml to web/ext/solr/solr.xml
743 $solr_server->copy_solrxml_to_web();
744 }
745
746 # 4. Ping the library URL, and if it's a persistent server and running, activate the collection again
747
748 # Check for success: if building does not exist OR is empty
749 if(!&FileUtils::directoryExists($build_dir) || &FileUtils::isDirectoryEmpty($build_dir)) {
750
751 if($library_url) { # undef if no valid server URL
752
753 &print_msg("Pinging $library_url\n");
754 if (&ping_library($library_url, $gs_mode, $site)) { # server running
755
756 # don't need to work out persistency of server more than once, since the libraryURL hasn't changed
757 if (!defined $is_persistent_server) {
758 &print_msg("Checking if Greenstone server is persistent\n");
759 $is_persistent_server = &is_persistent($library_url, $gs_mode);
760 }
761
762
763 if ($is_persistent_server) { # persistent server, so can try activating collection
764
765 &print_msg("Checking if the collection $qualified_collection is not already active\n");
766 my $collection_active = &ping_library_collection($library_url, $gs_mode, $qualified_collection, $site);
767
768 if (!$collection_active) {
769 &print_msg("Activating collection $qualified_collection\n");
770 &activate_collection($library_url, $gs_mode, $qualified_collection, $site);
771
772 # unless an error occurred, the collection should now be active:
773 $collection_active = &ping_library_collection($library_url, $gs_mode, $qualified_collection, $site);
774 if(!$collection_active) {
775 &print_msg("ERROR: collection $qualified_collection did not get activated\n");
776 }
777 }
778 else {
779 &print_msg("Collection is not active => No need to deactivate\n");
780 }
781 }
782 else {
783 &print_msg("Server is not persistent => No need to deactivate collection\n");
784 }
785 }
786 else {
787 &print_msg("No response to Ping => Taken to mean server is not running\n");
788 }
789 }
790 } else { # installcollection failed
791 #CollectionManager.Preview_Ready_Failed
792 &print_msg("Building directory is not empty or still exists. Failed to properly move $build_dir to $index_dir.\n");
793 }
794
795 &print_msg("\n");
796
797 if($buildtype eq "solr") {
798 if ($solr_server->explicitly_started()) {
799 $solr_server->stop();
800 }
801 }
802}
803
804&main(scalar(@ARGV),@ARGV);
Note: See TracBrowser for help on using the repository browser.