source: main/trunk/greenstone2/bin/script/activate.pl@ 30517

Last change on this file since 30517 was 30517, checked in by ak19, 8 years ago

Fixing incremental-rebuild when the database is gdbm. At this point (see buildcolutils.pm), the code needs to deactivate the collection before calling make_infodatabase(), since otherwise there's a lock on the gdbm database which prevents successful incremental-rebuild and activation.
  • Property svn:executable set to *
File size: 34.9 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, $silent) = @_;
146 # Gatherer.java's configGS3Server doesn't use the site variable
147 # so we don't have to either
148
149 # for GS2, getting the HTTP status isn't enough, we need to read the output
150 # since this is what CollectionManager.config() stipulates.
151 # Using LWP::UserAgent::get($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 my $resultstr = $response_content;
176 $resultstr =~ s@.*gs_content\"\>@@s;
177 $resultstr =~ s@</div>.*@@s;
178
179 if($response_content =~ m/$check_message_against_regex/) {
180 &print_msg(" Response as expected.\n", 3);
181 &print_msg("@@@@@@ Got result:\n$resultstr\n", 4);
182 return 1;
183 } else {
184 # if we expect the collection to be inactive, then we'd be in silent mode: if so,
185 # don't print out the "ping did not succeed" response, but print out any other messages
186
187 # So we only suppress the ping col "did not succeed" response if we're in silent mode
188 # But if any message other than ping "did not succeed" is returned, we always print it
189 if($response_content !~ m/did not succeed/ || !$silent) {
190 &print_msg("\n\tBUT: command $library_url$command response UNEXPECTED.\n", 3);
191 &print_msg("*** Got message:\n$response_content.\n", 4);
192 &print_msg("*** Got result:\n$resultstr\n", 3);
193 }
194 return 0; # ping on a collection may "not succeed."
195 }
196 }
197 elsif(LWP::Simple::is_error($response_code)) { # method exported by LWP::Simple, along with HTTP::Status constants
198 # check the page content is as expected
199 if(defined $expected_error_code && $response_code == $expected_error_code) {
200 &print_msg(" Response status $response_code as expected.\n", 3);
201 } else {
202 &print_msg("*** Command $library_url$command\n");
203 &print_msg("*** Unexpected error. HTTP Response Status: $response_code - Failed.\n");
204 }
205 return 0; # return false, since the response_code was an error, expected or not
206 }
207 else {
208 &print_msg("*** Command $library_url$command\n");
209 &print_msg("*** Unexpected error. HTTP Response Status: $response_code - Failed.\n");
210 return 0;
211 }
212 }
213}
214
215sub deactivate_collection {
216 my ($library_url, $gs_mode, $qualified_collection, $site) = @_;
217
218 if($gs_mode eq "gs2") {
219 my $DEACTIVATE_COMMAND = "?a=config&cmd=release-collection&c=";
220 my $check_message_against_regex = q/configured release-collection/;
221 config($library_url, $DEACTIVATE_COMMAND.$qualified_collection, $check_message_against_regex);
222 }
223 elsif ($gs_mode eq "gs3") {
224 my $DEACTIVATE_COMMAND = "?a=s&sa=d&st=collection&sn=";
225 my $check_message_against_regex = "collection: $qualified_collection deactivated";
226 config($library_url, $DEACTIVATE_COMMAND.$qualified_collection, $check_message_against_regex, $site);
227 }
228}
229
230sub activate_collection {
231 my ($library_url, $gs_mode, $qualified_collection, $site) = @_;
232
233 if($gs_mode eq "gs2") {
234 my $ACTIVATE_COMMAND = "?a=config&cmd=add-collection&c=";
235 my $check_message_against_regex = q/configured add-collection/;
236 config($library_url, $ACTIVATE_COMMAND.$qualified_collection, $check_message_against_regex);
237 }
238 elsif ($gs_mode eq "gs3") {
239 my $ACTIVATE_COMMAND = "?a=s&sa=a&st=collection&sn=";
240 my $check_message_against_regex = "collection: $qualified_collection activated";
241 config($library_url, $ACTIVATE_COMMAND.$qualified_collection, $check_message_against_regex, $site);
242 }
243}
244
245sub ping {
246 my ($library_url, $command, $gs_mode, $site, $silent) = @_;
247
248 # If the GS server is not running, we *expect* to see a "500" status code.
249 # If the GS server is running, then "Ping" ... "succeeded" is expected on success.
250 # When pinging an inactive collection, it will say it did "not succeed". This is
251 # a message of interest to return.
252 my $check_responsemsg_against_regex = q/(succeeded)/;
253 my $expected_error_code = 500;
254
255 &print_msg("*** COMMAND WAS: |$command|***\n", 4);
256
257 return config($library_url, $command, $check_responsemsg_against_regex, $site, $expected_error_code, $silent);
258}
259
260# send a pingaction to the GS library. General server-level ping.
261sub ping_library {
262 my ($library_url, $gs_mode, $site) = @_;
263
264 my $command = "";
265 if($gs_mode eq "gs2") {
266 $command = "?a=ping";
267 }
268 elsif ($gs_mode eq "gs3") {
269 $command = "?a=s&sa=ping";
270 }
271 return &ping($library_url, $command, $gs_mode, $site);
272}
273
274
275# send a pingaction to a collection in GS library to check if it's active
276sub ping_library_collection {
277 my ($library_url, $gs_mode, $qualified_collection, $site, $silent) = @_;
278
279 my $command = "";
280 if($gs_mode eq "gs2") {
281 $command = "?a=ping&c=$qualified_collection";
282 }
283 elsif ($gs_mode eq "gs3") {
284 $command = "?a=s&sa=ping&st=collection&sn=$qualified_collection";
285 }
286 return &ping($library_url, $command, $gs_mode, $site, $silent);
287}
288
289# return true if server is persistent, by calling is-persistent on library_url
290# this is only for GS2, since the GS3 server is always persistent
291sub is_persistent {
292 my ($library_url, $gs_mode) = @_;
293
294 if($gs_mode eq "gs3") { # GS3 server is always persistent
295 return 1;
296 }
297
298 my $command = "?a=is-persistent";
299 my $check_responsemsg_against_regex = q/true/; # isPersistent: true versus isPersistent: false
300 return config($library_url, $command, $check_responsemsg_against_regex);
301}
302
303sub get_library_URL {
304 my ($gs_mode, $lib_name) = @_; # gs_mode can be gs3 or gs2, lib_name is the custom servlet name
305
306 # If we get here, we are dealing with a server included with GS.
307 # For GS3, we ask ant for the library URL.
308 # For GS2, we derive the URL from the llssite.cfg file.
309
310 my $url = undef;
311
312 if($gs_mode eq "gs2") {
313 my $llssite_cfg = &FileUtils::filenameConcatenate($ENV{'GSDLHOME'}, "llssite.cfg");
314
315 if(-f $llssite_cfg) {
316 # check llssite.cfg for line with url property
317 # for server.exe also need to use portnumber and enterlib properties
318
319 # Read in the entire contents of the file in one hit
320 if (!open (FIN, $llssite_cfg)) {
321 &print_msg("activate.pl::get_library_URL failed to open $llssite_cfg ($!)\n");
322 return undef;
323 }
324
325 my $contents;
326 sysread(FIN, $contents, -s FIN);
327 close(FIN);
328
329 my @lines = split(/[\n\r]+/, $contents); # split on carriage-returns and/or linefeeds
330 my $enterlib = "";
331 my $portnumber = ""; # will remain empty (implicit port 80) unless it's specifically been assigned
332
333 foreach my $line (@lines) {
334 if($line =~ m/^url=(.*)$/) {
335 $url = $1;
336 } elsif($line =~ m/^enterlib=(.*)$/) {
337 $enterlib = $1;
338 } elsif($line =~ m/^portnumber=(.*)$/) {
339 $portnumber = $1;
340 }
341 }
342
343 if(!$url) {
344 return undef;
345 }
346 elsif($url eq "URL_pending") { # library is not running
347 # do not process url=URL_pending in the file, since for server.exe
348 # this just means the Enter Library button hasn't been pressed yet
349 $url = undef;
350 }
351 else {
352 # In the case of server.exe, need to do extra work to get the proper URL
353 # But first, need to know whether we're indeed dealing with server.exe:
354
355 # compare the URL's domain to the full URL
356 # E.g. for http://localhost:8383/greenstone3/cgi-bin, the domain is localhost:8383
357 my $uri = URI->new( $url );
358 my $host = $uri->host;
359 #print STDERR "@@@@@ host: $host\n";
360 if($url =~ m/http:\/\/$host(\/)?$/) {
361 #if($url !~ m/http:\/\/$host:$portnumber(\/)?/ || $url =~ m/http:\/\/$host(\/)?$/) {
362 # (if the URL does not contain the portnumber, OR if the port is implicitly 80 and)
363 # If the domain with http:// prefix is completely the same as the URL, assume server.exe
364 # then the actual URL is the result of suffixing the port and enterlib properties in llssite.cfg
365 $url = $url.":".$portnumber.$enterlib;
366 } # else, apache web server
367
368 }
369 }
370 } elsif($gs_mode eq "gs3") {
371 # Either check build.properties for tomcat.server, tomcat.port and app.name (and default servlet name).
372 # app.name is stored in app.path by build.xml. Need to move app.name in build.properties from build.xml
373
374 # Or, run the new target get-default-servlet-url
375 # the output can look like:
376 #
377 # Buildfile: build.xml
378 # [echo] os.name: Windows Vista
379 #
380 # get-default-servlet-url:
381 # [echo] http://localhost:8383/greenstone3/library
382 # BUILD SUCCESSFUL
383 # Total time: 0 seconds
384
385 #my $output = qx/ant get-default-servlet-url/; # backtick operator, to get STDOUT (else 2>&1)
386 # see http://stackoverflow.com/questions/799968/whats-the-difference-between-perls-backticks-system-and-exec
387
388 # The get-default-servlet-url ant target can be run from anywhere by specifying the
389 # location of GS3's ant build.xml buildfile. Activate.pl can be run from anywhere for GS3
390 # GSDL3SRCHOME will be set for GS3 by gs3-setup.sh, a step that would have been necessary
391 # to run the activate.pl script in the first place
392 my $perl_command = "ant -buildfile \"$ENV{'GSDL3SRCHOME'}/build.xml\" get-default-servlet-url";
393
394 if (open(PIN, "$perl_command |")) {
395 while (defined (my $perl_output_line = <PIN>)) {
396 if($perl_output_line =~ m@http:\/\/(\S*)@) { # grab all the non-whitespace chars
397 $url="http://".$1;
398 }
399 }
400 close(PIN);
401 } else {
402 &print_msg("activate.pl::get_library_URL: Failed to run $perl_command to work out library URL for $gs_mode\n");
403 }
404 if(defined $lib_name) {
405 # replace the servlet_name portion of the url found, with the given library_name
406 $url =~ s@/[^/]*$@/$lib_name@;
407 }
408 }
409
410 # either the url is still undef or it is now set
411 #print STDERR "\n@@@@@ final URL:|$url|\n" if $url;
412 #print STDERR "\n@@@@@ URL still undef\n" if !$url;
413 return $url;
414}
415
416### UNUSED METHODS TO MOVE TO util.pm?
417
418# This method is now unused. Using ping_library instead to send the ping action to a
419# GS2/GS3 server. This method can be used more generally to test whether a URL is alive.
420# http://search.cpan.org/dist/libwww-perl/lib/LWP/Simple.pm
421# and http://www.perlmonks.org/?node_id=618534
422sub is_URL_active {
423 my $url = shift(@_); # gs3 or gs2 URL
424
425 my $status = 0;
426 if(defined $url) {
427 $ua->timeout(10); # set LWP useragent to 5s max timeout for testing the URL
428 # Need to set this, else it takes I don't know how long to timeout
429 # http://www.perlmonks.org/?node_id=618534
430
431 $status = LWP::Simple::head($url); # returns empty list of headers if it fails
432 # LWP::Simple::get($url) is more intensive, so don't need to do that
433 #print STDERR "**** $url is alive.\n" if $status;
434 }
435 return $status;
436}
437
438# Pinging seems to always return true, so this method doesn't work
439sub pingHost {
440 my $url = shift(@_); # gs3 or gs2 URL
441
442 my $status = 0;
443 if(defined $url) {
444 # Get just the domain. "http://localhost/gsdl?uq=332033495" becomes "localhost"
445 # "http://localhost/greenstone/cgi-bin/library.cgi" becomes "localhost" too
446
447 #my $host = $url;
448 #$host =~ s@^http:\/\/(www.)?@@;
449 #$host =~ s@\/.*@@;
450 #print STDERR "**** HOST: $host\n";
451
452 # More robust way
453 # http://stackoverflow.com/questions/827024/how-do-i-extract-the-domain-out-of-an-url
454 my $uri = URI->new( $url );
455 my $host = $uri->host;
456
457 # Ping the host. http://perldoc.perl.org/Net/Ping.html
458 my $p = Net::Ping->new();
459 $status = $p->ping($host); # || 0. Appears to set to undef rather than 0
460 print STDERR "**** $host is alive.\n" if $status; #print "$host is alive.\n" if $p->ping($host);
461 $p->close();
462 }
463 # return whether pinging was a success or failure
464 return $status;
465}
466
467sub do_deactivate {
468 my($is_persistent_server, $library_url, $gs_mode, $site, $qualified_collection) = @_;
469
470
471 &print_msg("Pinging $library_url\n");
472 if (&ping_library($library_url, $gs_mode, $site)) { # server running
473
474 # server is running, so release the collection if
475 # the server is persistent and the collection is active
476 &print_msg("Checking if Greenstone server is persistent\n");
477 $is_persistent_server = &is_persistent($library_url, $gs_mode);
478
479 if ($is_persistent_server) { # only makes sense to issue activate and deactivate cmds to a persistent server
480
481 &print_msg("Checking if the collection $qualified_collection is already active\n");
482 my $collection_active = &ping_library_collection($library_url, $gs_mode, $qualified_collection, $site);
483
484 if ($collection_active) {
485 &print_msg("De-activating collection $qualified_collection\n");
486 &deactivate_collection($library_url, $gs_mode, $qualified_collection, $site);
487 }
488 else {
489 &print_msg("Collection is not active => No need to deactivate\n");
490 }
491 }
492 else {
493 &print_msg("Server is not persistent => No need to deactivate collection\n");
494 }
495 }
496 else {
497 &print_msg("No response to Ping => Taken to mean server is not running\n");
498 }
499
500 return $is_persistent_server;
501}
502
503sub do_activate {
504 my($is_persistent_server, $library_url, $gs_mode, $site, $qualified_collection) = @_;
505
506
507 &print_msg("Pinging $library_url\n");
508 if (&ping_library($library_url, $gs_mode, $site)) { # server running
509
510 # don't need to work out persistency of server more than once, since the libraryURL hasn't changed
511 if (!defined $is_persistent_server) {
512 &print_msg("Checking if Greenstone server is persistent\n");
513 $is_persistent_server = &is_persistent($library_url, $gs_mode);
514 }
515
516
517 if ($is_persistent_server) { # persistent server, so can try activating collection
518
519 &print_msg("Checking if the collection $qualified_collection is not already active\n");
520
521 # Since we could have deactivated the collection at this point,
522 # it is likely that it is not yet active. When pinging the collection
523 # a "ping did not succeed" message is expected, therefore tell the ping
524 # to proceed silently
525 my $silent = 1;
526 my $collection_active = &ping_library_collection($library_url, $gs_mode, $qualified_collection, $site, $silent);
527
528 if (!$collection_active) {
529 &print_msg(" Collection is not active.\n");
530 &print_msg("Activating collection $qualified_collection\n");
531 &activate_collection($library_url, $gs_mode, $qualified_collection, $site);
532
533 # unless an error occurred, the collection should now be active:
534 $collection_active = &ping_library_collection($library_url, $gs_mode, $qualified_collection, $site); # not silent if ping did not succeed
535 if(!$collection_active) {
536 &print_msg("ERROR: collection $qualified_collection did not get activated\n");
537 }
538 }
539 else {
540 &print_msg("Collection is already active => No need to activate\n");
541 }
542 }
543 else {
544 &print_msg("Server is not persistent => No need to activate collection\n");
545 }
546 }
547 else {
548 &print_msg("No response to Ping => Taken to mean server is not running\n");
549 }
550
551 return $is_persistent_server;
552}
553
554
555# Most of the arguments are familiar from the building scripts like buildcol.pl
556# The special optional argument -library_url is for when we're dealing with a web
557# library server such as an apache that's separate from any included with GS2.
558# In such a case, this script's caller should pass in -library_url <URL>.
559#
560# $site argument must be specified in the cmdline for collectionConfig.xml to get
561# generated which makes $gs_mode=gs3, else collect.cfg gets generated and $gs_mode=gs2
562sub main
563{
564 my ($argc,@argv) = @_;
565
566 if (($argc==0) || (($argc==1) && ($argv[0] =~ m/^--?h(elp)?$/))) {
567 my ($progname) = ($0 =~ m/^.*[\/|\\](.*?)$/);
568
569
570 print STDERR "\n";
571 print STDERR "Usage: $progname [-collectdir c -builddir b -indexdir i -site s -skipactivation -removeold -keepold -verbosity v\n";
572 print STDERR "\t-library_url URL -library_name n] <[colgroup/]collection>\n";
573 print STDERR "\n";
574
575 exit(-1);
576 }
577
578 # http://stackoverflow.com/questions/6156742/how-can-i-capture-the-complete-commandline-in-perl
579 #print STDERR "@@@@@@@@@ ACTIVATE CMD: " . join " ", $0, @ARGV;
580
581 # get the collection details
582 my $qualified_collection = pop @argv; # qualified collection
583
584 my $collect_dir = undef; #"collect"; # can't be "collect" when only -site is provided for GS3
585 my $build_dir = undef;
586 my $index_dir = undef;
587 my $site = undef;
588
589 # if run from server (java code), it will handle deactivation and activation to prevent open file handles when java launches this script and exits:
590 my $skipactivation = 0;
591 my $removeold = 0;
592 my $keepold = 0;
593 my $incremental = 0; # used by solr
594
595 # some further parameters
596 my $just_activate = 0; # don't do anything else than activating the collection if the server is running
597 my $just_deactivate = 0; # don't do anything else than deactivating the collection if the server is running
598
599
600 my $library_url = $ENV{'GSDL_LIBRARY_URL'} || undef; # to be specified on the cmdline if not using a GS-included web server
601 # the GSDL_LIBRARY_URL env var is useful when running cmdline buildcol.pl in the linux package manager versions of GS3
602
603 my $library_name = undef;
604
605 while (my $arg = shift @argv) {
606 if ($arg eq "-collectdir") {
607 $collect_dir = shift @argv;
608 }
609 elsif ($arg eq "-builddir") {
610 $build_dir = shift @argv;
611 }
612 elsif ($arg eq "-indexdir") {
613 $index_dir = shift @argv;
614 }
615 elsif ($arg eq "-site") {
616 $site = shift @argv;
617 }
618 elsif ($arg eq "-skipactivation") {
619 $skipactivation = 1;
620 }
621 elsif ($arg eq "-removeold") {
622 $removeold = 1;
623 }
624 elsif ($arg eq "-keepold") {
625 $keepold = 1;
626 }
627 elsif ($arg eq "-incremental") {
628 $incremental = 1;
629 }
630 elsif ($arg eq "-justactivate") {
631 $just_activate = 1;
632 }
633 elsif ($arg eq "-justdeactivate") {
634 $just_deactivate = 1;
635 }
636 elsif ($arg eq "-library_url") {
637 $library_url = shift @argv;
638 }
639 elsif ($arg eq "-library_name") {
640 $library_name = shift @argv;
641 }
642 elsif ($arg eq "-verbosity") {
643 $default_verbosity = shift @argv; # global variable
644
645 # ensure we're working with ints not strings (int context not str context), in case verbosity=0
646 # http://stackoverflow.com/questions/288900/how-can-i-convert-a-string-to-a-number-in-perl
647 $default_verbosity = int($default_verbosity || 0); ### is this the best way?
648 }
649 }
650
651 # work out the building and index dirs
652 my $collection_dir = &util::resolve_collection_dir($collect_dir, $qualified_collection, $site);
653 $build_dir = &FileUtils::filenameConcatenate($collection_dir, "building") unless (defined $build_dir);
654 $index_dir = &FileUtils::filenameConcatenate($collection_dir, "index") unless (defined $index_dir);
655
656 &print_task_msg("Running Collection Activation Stage");
657
658 # get and check the collection name
659 if ((&colcfg::use_collection($site, $qualified_collection, $collect_dir)) eq "") {
660 &print_msg("Unable to use collection \"$qualified_collection\" within \"$collect_dir\"\n");
661 exit -1;
662 }
663
664 # Read in the collection configuration file.
665 # Beware: Only if $site is specified in the cmdline does collectionConfig.xml get
666 # generated and does $gs_mode=gs3, else collect.cfg gets generated and $gs_mode=gs2
667 my $gs_mode = "gs2";
668 if ((defined $site) && ($site ne "")) { # GS3
669 $gs_mode = "gs3";
670 }
671 my $collect_cfg_filename = &colcfg::get_collect_cfg_name(STDERR, $gs_mode);
672 my $collectcfg = &colcfg::read_collection_cfg ($collect_cfg_filename,$gs_mode);
673
674 # look for build.cfg/buildConfig.xml
675 my $build_cfg_filename ="";
676
677 if ($gs_mode eq "gs2") {
678 $build_cfg_filename = &FileUtils::filenameConcatenate($build_dir,"build.cfg");
679 } else {
680 $build_cfg_filename = &FileUtils::filenameConcatenate($build_dir, "buildConfig.xml");
681 # gs_mode is GS3. Set the site now if this was not specified as cmdline argument
682 #$site = "localsite" unless defined $site;
683 }
684
685 # We need to know the buildtype for Solr.
686 # Any change of indexers is already detected and handled by the calling code (buildcol or
687 # full-rebuild), so that at this stage the config file's buildtype reflects the actual buildtype.
688
689 # From buildcol.pl we use searchtype for determining buildtype, but for old versions, use buildtype
690 my $buildtype;
691 if (defined $collectcfg->{'buildtype'}) {
692 $buildtype = $collectcfg->{'buildtype'};
693 } elsif (defined $collectcfg->{'searchtypes'} || defined $collectcfg->{'searchtype'}) {
694 $buildtype = "mgpp";
695 } else {
696 $buildtype = "mg"; #mg is the default
697 }
698
699 # can't do anything without a build directory with something in it to move into index
700 # Except if we're (doing incremental) building for solr, where we want to still
701 # activate and deactivate collections including for the incremental case
702 if(!&FileUtils::directoryExists($build_dir)) {
703 &print_msg("No building folder at $build_dir to move to index.\n");
704 exit -1 unless ($buildtype eq "solr"); #&& $incremental);
705 } elsif (&FileUtils::isDirectoryEmpty($build_dir)) {
706 &print_msg("Nothing in building folder $build_dir to move into index folder.\n");
707 exit -1 unless ($buildtype eq "solr"); #&& $incremental);
708 }
709
710 # Now the logic in GLI's CollectionManager.java (processComplete()
711 # and installCollection()) and Gatherer.configGS3Server().
712
713 # 1. Get library URL
714
715 # For web servers that are external to a Greenstone installation,
716 # the user can pass in their web server's library URL.
717 # For web servers included with GS (like tomcat for GS3 and server.exe
718 # and apache for GS2), we work out the library URL:
719 if(!$library_url) {
720 $library_url = &get_library_URL($gs_mode, $library_name); # returns undef if no server is running
721 }
722
723 # CollectionManager's installCollection phase in GLI
724 # 2. Ping the library URL, and if it's a persistent server and running, release the collection
725 my $is_persistent_server = undef;
726 if($library_url && !$skipactivation) { # undef if no valid server URL
727 if($just_deactivate || !$just_activate) { # either $justdeactivate is provided or neither $justactivate and $justdeactivate are provided
728 # and deactivation will go through
729 $is_persistent_server = &do_deactivate($is_persistent_server, $library_url, $gs_mode, $site, $qualified_collection);
730 }
731 }
732
733 # 2b. If we're working with a solr collection, then start up the solrserver now.
734 my $solr_server;
735if(!$just_activate && !$just_deactivate) {
736 my @corenames = ();
737 if($buildtype eq "solr") { # start up the jetty server
738 my $solr_ext = $ENV{'GEXT_SOLR'}; # from solr_passes.pl
739 unshift (@INC, "$solr_ext/perllib");
740 require solrserver;
741
742 # Solr cores are named without taking the collection-group name into account, since solr
743 # is used for GS3 and GS3 doesn't use collection groups but has the site concept instead
744 my ($colname, $colgroup) = &util::get_collection_parts($qualified_collection);
745
746 # See solrbuilder.pm to get the indexing levels (document, section) from the collectcfg file
747 # Used to generate core names from them and remove cores by name
748 foreach my $level ( @{$collectcfg->{'levels'}} ){
749 my ($pindex) = $level =~ /^(.)/;
750 my $indexname = $pindex."idx";
751 push(@corenames, "$site-$colname-$indexname"); #"$site-$colname-didx", "$site-$colname-sidx"
752 }
753
754 # If the Solr/Jetty server is not already running, the following starts
755 # it up, and only returns when the server is "reading and listening"
756 $solr_server = new solrserver($build_dir);
757 $solr_server->start();
758
759 # We'll be moving building to index. For solr collection, there's further
760 # special processing to make a corresponding change to the solr.xml
761 # by removing the temporary building cores and (re)creating the index cores
762 }
763
764
765 # 3. Do all the moving building to index stuff now
766
767 # If removeold: replace index dir with building dir.
768 # If keepold: move building's contents into index, where only duplicates will get deleted.
769 # removeold and keepold can't both be on at the same time
770 # incremental becomes relevant for solr, though it was irrelevant to what activate.pl does (moving building to index)
771 my $incremental_mode;
772 ($removeold, $keepold, $incremental, $incremental_mode) = &scriptutil::check_removeold_and_keepold($removeold, $keepold,
773 $incremental,
774 $build_dir, # checkdir. Usually archives or export to be deleted. activate.pl deletes building
775 $collectcfg);
776
777 if($removeold) {
778
779 if(&FileUtils::directoryExists($index_dir)) {
780 &print_task_msg("Removing \"index\"");
781
782 if ($buildtype eq "solr") {
783 # if solr, remove any cores that are using the index_dir before deleting this dir
784 foreach my $corename (@corenames) {
785 $solr_server->admin_unload_core($corename);
786 }
787 }
788
789 &FileUtils::removeFilesRecursive($index_dir);
790
791 # Wait for a couple of seconds, just for luck
792 sleep 2;
793
794 if (&FileUtils::directoryExists($index_dir)) {
795 &print_msg("The index directory $index_dir could not be deleted.\n"); # CollectionManager.Index_Not_Deleted
796 }
797 }
798
799 # if remote GS server: gliserver.pl would call activate.pl to activate
800 # the collection at this point since activate.pl lives on the server side
801
802 if ($buildtype eq "solr") {
803 # if solr, remove any cores that are using the building_dir before moving this dir onto index
804 foreach my $corename (@corenames) {
805 $solr_server->admin_unload_core("building-$corename");
806 }
807 }
808
809 # Move the building directory to become the new index directory
810 &print_task_msg("Moving \"building\" -> \"index\"");
811 &FileUtils::moveFiles($build_dir, $index_dir);
812 if(&FileUtils::directoryExists($build_dir) || !&FileUtils::directoryExists($index_dir)) {
813 &print_msg("Could not move $build_dir to $index_dir.\n"); # CollectionManager.Build_Not_Moved
814 }
815 }
816 elsif ($keepold || $incremental) {
817 if ($buildtype eq "solr") {
818 # if solr, remove any cores that may be using the building_dir before moving this dir onto index
819 foreach my $corename (@corenames) {
820 $solr_server->admin_unload_core("building-$corename") if $solr_server->admin_ping_core("building-$corename");
821 }
822 }
823
824 if($build_dir eq $index_dir) { # building_dir can have been set to "index" folder, see incremental-buildcol.pl
825 &print_task_msg("building folder is index folder, not moving");
826 } else {
827 # Copy just the contents of building dir into the index dir, overwriting
828 # existing files, but don't replace index with building.
829 &print_task_msg("Moving \"building\" -> \"index\"");
830 &FileUtils::moveDirectoryContents($build_dir, $index_dir);
831 }
832 }
833
834 if ($buildtype eq "solr") {
835 # Call CREATE action to get the old cores pointing to the index folder
836 foreach my $corename (@corenames) {
837 if($removeold) {
838 # Call CREATE action to get all cores pointing to the index folder, since building is now index
839 $solr_server->admin_create_core($corename, $index_dir);
840
841 } elsif ($keepold || $incremental) {
842 # Call RELOAD core. Should already be using the index_dir directory for $keepold and $incremental case
843
844 # Ping to see if corename exists, if it does, reload, else create
845 if ($solr_server->admin_ping_core($corename)) {
846 $solr_server->admin_reload_core($corename);
847 } else {
848 $solr_server->admin_create_core($corename, $index_dir);
849 }
850 }
851 }
852
853 # regenerate the solr.xml.in from solr.xml in case we are working off a dvd.
854 $solr_server->solr_xml_to_solr_xml_in();
855 }
856}#end if(!$just_activate && !$just_deactivate)
857
858 # 4. Ping the library URL, and if it's a persistent server and running, activate the collection again
859
860 # Check for success: if building does not exist OR is empty OR if building is index (in which case there was no move)
861 if($build_dir eq $index_dir || !&FileUtils::directoryExists($build_dir) || &FileUtils::isDirectoryEmpty($build_dir)) {
862
863 if($library_url && (!$skipactivation)) { # undef if no valid server URL
864 if($just_activate || !$just_deactivate) { # either $justactivate is provided or neither $justdeactivate and $justactivate are provided
865 # then activation will go through
866 $is_persistent_server = &do_activate($is_persistent_server, $library_url, $gs_mode, $site, $qualified_collection);
867 }
868 }
869 } else { # installcollection failed
870 #CollectionManager.Preview_Ready_Failed
871 &print_msg("Building directory is not empty or still exists. Failed to properly move $build_dir to $index_dir.\n");
872 }
873
874 &print_msg("\n");
875
876 if(!$just_activate && !$just_deactivate && $buildtype eq "solr") {
877 if ($solr_server->explicitly_started()) {
878 $solr_server->stop();
879 }
880 }
881}
882
883&main(scalar(@ARGV),@ARGV);
Note: See TracBrowser for help on using the repository browser.