root/main/trunk/greenstone2/perllib/g2futil.pm @ 24192

Revision 24192, 13.1 KB (checked in by ak19, 8 years ago)

Sam discovered that using dollar-Config{perlpath} in place of dollar-hat-X is the better way to obtain the path to the perl that is being used. We hope this will not be a relative path on the Mac as dollar-hat-x was on Professor Witten's Mac when we tried it there today.

Line 
1package g2futil;
2
3
4BEGIN
5{
6    if (!defined $ENV{'FEDORA_HOME'}) {
7    print STDERR "Error: Environment variable FEDORA_HOME not set.\n";
8    exit 1;
9    }
10
11    my $fedora_client_bin = &util::filename_cat($ENV{'FEDORA_HOME'},"client","bin");
12    &util::envvar_append("PATH",$fedora_client_bin);
13}
14
15use strict;
16use util;
17use Config; # for getting the perlpath in the recommended way
18
19sub run_cmd_old
20{
21    my ($cmd,$verbosity,$tolerate_error) = @_;
22
23    if (($verbosity == 0)
24    || (defined $tolerate_error && ($tolerate_error eq "tolerate_error"))) {
25    if($ENV{'GSDLOS'} =~ /^windows$/i) {
26        $cmd .= " > nul";
27    } else {
28        $cmd .= " > /dev/null";
29    }
30    }
31   
32    if ($verbosity >= 2) {
33    print "Runing command:\n";
34    print "$cmd\n";
35    }
36
37    my $status = system($cmd);
38
39    if ($verbosity >= 2) {
40    print "Exit status = ", $status/256, "\n";
41    }
42
43    if ((!defined $tolerate_error) || ($tolerate_error ne "tolerate_error")) {
44    if ($status>0) {
45        print STDERR "Error executing:\n$cmd\n";
46        print STDERR "$!\n";
47    }
48    }
49
50    return $status;
51}
52
53
54sub run_cmd
55{
56    my ($prog,$arguments,$verbosity,$tolerate_error) = @_;
57
58    my $cmd_status = undef;
59
60    my $script_ext = ($ENV{'GSDLOS'} =~ m/^windows/) ? ".bat" : ".sh";
61
62    if ($prog =~ m/^fedora-/) {
63    $prog .= $script_ext;
64    }
65    if (($prog =~ m/.pl$/i) && ($ENV{'GSDLOS'} =~ m/^windows/)) {
66    $prog ="\"$Config{perlpath}\" -S $prog";
67    }
68 
69    my $cmd = "$prog $arguments";
70
71###    print "*** cmd = $cmd\n";
72
73    if (open(CMD,"$cmd 2>&1 |"))
74    {
75    my $result = "";
76    my $line;
77    while (defined ($line = <CMD>))
78    {   
79        $result .= $line;       
80
81        if ((!defined $tolerate_error) || ($tolerate_error ne "tolerate_error"))
82        {
83        print $line;
84        }
85
86
87    }
88   
89    close(CMD);
90   
91    $cmd_status = $?;
92
93    if ($cmd_status == 0) {
94        # Check for any lines in result begining 'Error:'
95       
96        if ($result =~ m/^Error\s*:/m) {
97        # Fedora script generated an error, but did not exit
98        # with an error status => artificially raise one
99
100        $cmd_status = -1;
101        }
102    }
103
104    if ($cmd_status != 0) {
105
106        if ((!defined $tolerate_error) || ($tolerate_error ne "tolerate_error"))
107        {
108        print STDERR "Error: processing command failed.  Exit status $cmd_status\n";
109       
110        if ($verbosity >= 2) {
111            print STDERR "  Command was: $cmd\n";
112        }
113        if ($verbosity >= 3) {
114            print STDERR "result: $result\n";
115        }
116
117        }
118    }
119    }
120    else
121    {
122    print STDERR "Error: failed to execute $cmd\n";
123    }
124
125
126    return $cmd_status;
127}
128
129
130sub run_datastore_info
131{
132    my ($pid,$options) = @_;
133
134    my $verbosity = $options->{'verbosity'};
135
136    my $hostname = $options->{'hostname'};
137    my $port     = $options->{'port'};
138    my $username = $options->{'username'};
139    my $password = $options->{'password'};
140    my $protocol = $options->{'protocol'};
141
142    my $prog = "fedora-dsinfo";
143    my $arguments = "$hostname $port $username $password $pid $protocol";
144    my $status = run_cmd($prog,$arguments,$verbosity,"tolerate_error");
145
146    return $status;
147}
148
149sub run_purge
150{
151    my ($pid,$options) = @_;
152
153    my $verbosity = $options->{'verbosity'};
154
155    my $hostname = $options->{'hostname'};
156    my $port     = $options->{'port'};
157    my $username = $options->{'username'};
158    my $password = $options->{'password'};
159    my $protocol = $options->{'protocol'};
160
161    my $server = "$hostname:$port";
162
163    my $prog = "fedora-purge";
164    my $arguments = "$server $username $password $pid $protocol";
165    $arguments .= " \\\n \"Automated_purge_by_g2f_script\"";
166
167    my $status = run_cmd($prog,$arguments,$verbosity);
168
169    return $status;
170}
171
172sub run_ingest
173{
174    my ($docmets_filename,$options) = @_;
175
176    my $verbosity = $options->{'verbosity'};
177
178    my $hostname = $options->{'hostname'};
179    my $port     = $options->{'port'};
180    my $username = $options->{'username'};
181    my $password = $options->{'password'};
182    my $protocol = $options->{'protocol'};
183
184    my $server = "$hostname:$port";
185
186    my $prog = "fedora-ingest";
187
188    my $type = undef;
189   
190    if ($ENV{'FEDORA_VERSION'} =~ m/^2/) { # checking if major version is 2
191        $type = "metslikefedora1";
192    }
193    else {
194    $type = "info:fedora/fedora-system:METSFedoraExt-1.1";
195    }
196
197    my $arguments = "file \"$docmets_filename\" $type $server $username $password $protocol";
198    $arguments .= " \\\n \"Automated_purge_by_g2f_script\"";
199
200    my $status = run_cmd($prog,$arguments,$verbosity);
201
202    return $status;
203}
204
205
206sub rec_get_all_hash_dirs
207{
208    my ($full_dir,$all_dirs) = @_;
209
210    if (opendir(DIR, $full_dir)) {
211    my @sub_dirs = grep { ($_ !~ /^\./) && (-d &util::filename_cat($full_dir,$_)) } readdir(DIR);
212    closedir DIR;
213
214    my @hash_dirs = grep { $_ =~ m/\.dir$/ } @sub_dirs;
215    my @rec_dirs = grep { $_ !~ m/\.dir$/ } @sub_dirs;
216   
217    foreach my $hd (@hash_dirs) {
218        my $full_hash_dir = &util::filename_cat($full_dir,$hd);
219        push(@$all_dirs,$full_hash_dir);
220    }
221
222    foreach my $rd (@rec_dirs) {
223        my $full_rec_dir = &util::filename_cat($full_dir,$rd);
224        rec_get_all_hash_dirs($full_rec_dir,$all_dirs);
225    }       
226    }
227}
228
229sub get_all_hash_dirs
230{
231    my ($start_dir,$maxdocs) = @_;
232   
233    my @all_dirs = ();
234    rec_get_all_hash_dirs($start_dir,\@all_dirs);
235
236    if ((defined $maxdocs) && ($maxdocs ne "")) {
237    my @maxdoc_dirs = ();
238    for (my $i=0; $i<$maxdocs; $i++) {
239        push(@maxdoc_dirs,shift(@all_dirs));
240    }
241    @all_dirs = @maxdoc_dirs;
242    }
243
244    return @all_dirs;
245}
246
247sub get_hash_id
248{
249    my ($hash_dir) = @_;
250
251    my $hash_id = undef;
252
253    my $docmets_filename = &util::filename_cat($hash_dir,"docmets.xml");
254
255    if (open(DIN,"<$docmets_filename"))
256    {
257    while (defined (my $line = <DIN>))
258    {
259        if ($line =~ m/<dc:identifier>(.*?)<\/dc:identifier>/)
260        {
261        $hash_id = $1;
262        last;
263        }
264    }
265   
266    close(DIN);
267    }
268    else
269    {
270    print STDERR "Warning: Unable to open \"$docmets_filename\"\n";
271    }
272
273    return $hash_id;
274
275}
276
277
278# Subroutine to write the gsdl.xml file in FEDORA_HOME/tomcat/conf/Catalina/<host/localhost>/
279# This xml file will tell Fedora where to find the parent folder of the GS collect dir
280# so that it can obtain the FedoraMETS files for ingestion.
281# It depends on the Fedora server being on the same machine as the Greenstone server that
282# this code is part of.
283sub write_gsdl_xml_file
284{
285    my ($fedora_host, $collect_dir, $options) = @_;
286    my $verbosity = $options->{'verbosity'};
287    my $hostname = $options->{'hostname'};
288    my $port     = $options->{'port'};
289    my $protocol = $options->{'protocol'};
290
291    print STDERR "Ensuring that a correct gsdl.xml file exists on the Fedora server end\n";
292    # The top of this file has already made sure that FEDORA_HOME is set, but for GS3
293    # CATALINA_HOME is set to GS' own tomcat. Since we'll be working with fedora, we need
294    # to temporarily set CATALINA_HOME to fedora's tomcat. (Catalina is undefined for GS2.)
295    my $gs_catalina_home = $ENV{'CATALINA_HOME'} if defined $ENV{'CATALINA_HOME'};
296    $ENV{'CATALINA_HOME'} = &util::filename_cat($ENV{'FEDORA_HOME'}, "tomcat");
297   
298    # 1. Find out which folder to write to: fedora_host or localhost
299    # whichever contains fedora.xml is the one we want (if none, exit with error value?)
300    my $fedora_home = $ENV{'FEDORA_HOME'};
301    my $base_path = &util::filename_cat($fedora_home, "tomcat", "conf", "Catalina");
302
303    my $host_path = &util::filename_cat($base_path, $fedora_host);
304    my $xmlFile = &util::filename_cat($host_path, "fedora.xml");
305    if (!-e $xmlFile) {
306    # check if the folder localhost contains fedoraXML
307    $host_path = &util::filename_cat($base_path, "localhost");
308    $xmlFile = &util::filename_cat($host_path, "fedora.xml");
309    if(!-e $xmlFile) {
310        # try putting gsdl in this folder, but still print a warning
311        print STDERR "$host_path does not contain file fedora.xml. Hoping gsdl.xml belongs there anyway\n";
312    }
313    }
314
315    # 2. Construct the string we are going write to the gsdl.xml file
316    # a. get the parent directory of collect_dir by removinbg the word
317    # "collect" from it and any optional OS-type slash at the end.
318    # (Path slash direction does not matter here.)
319    my $collectParentDir = $collect_dir;
320    $collectParentDir =~ s/collect(\/|\\)?//;
321 
322    # b. Use the collectParentDir to create the contents of gsdl.xml
323    my $greenstone_url_prefix = &util::get_greenstone_url_prefix(); # would have the required slash at front
324    my $gsdlXMLcontents = "<?xml version='1.0' encoding='utf-8'?>\n<Context docBase=\"";
325    $gsdlXMLcontents = $gsdlXMLcontents.$collectParentDir."\" path=\"$greenstone_url_prefix\"></Context>";
326   
327    # 3. If there is already a gsdl.xml file in host_path, compare the string we
328    # want to write with what is already in there. If they're the same, we can return
329    $xmlFile = &util::filename_cat($host_path, "gsdl.xml");
330    if(-e $xmlFile) {
331    # such a file exists, so read the contents
332    unless(open(FIN, "<$xmlFile")) {
333        print STDERR "g2f-import.pl: Unable to open existing $xmlFile for comparing...Recoverable. $!\n";
334        # doesn't matter, we'll just overwrite it then
335    }   
336    my $xml_contents;
337    {
338        local $/ = undef;        # Read entire file at once
339        $xml_contents = <FIN>;   # Now file is read in as one single 'line'
340    }
341    close(FIN); # close the file
342    if($xml_contents eq $gsdlXMLcontents) {
343        print STDERR "Fedora links to the FLI import folder through gsdl.xml.\n";
344        # it already contains what we want, we're done
345        return "gsdl.xml";
346    }
347    }
348
349    # 4. If we're here, the contents of gsdl.xml need to be updated:
350    # a. First stop the fedora server
351    my $script_ext = ($ENV{'GSDLOS'} =~ m/^windows/) ? ".bat" : ".sh";
352    my $stop_tomcat = &util::filename_cat($fedora_home, "tomcat", "bin", "shutdown".$script_ext);
353    # execute the command
354    $! = 0; # does this initialise the return value?
355    my $status = system($stop_tomcat);
356    if ($status!=0) { # to get the actual exit value, divide by 256, but not useful here
357    # possible tomcat was already stopped - it's not the end of the world
358    print STDERR "Failed to stop Fedora server. Perhaps it was not running. $!\n";
359    print "Exit status = ", $status/256, "\n";
360    }
361
362    # b. overwrite the file that has outdated contents with the contents we just constructed
363    unless(open(FOUT, ">$xmlFile")) {  # create or overwrite gsdl.xml file
364    die "g2f-import.pl: Unable to open $xmlFile for telling Fedora where the collect dir is...ERROR: $!\n";
365    }
366    # write out the updated contents and close the file
367    print FOUT $gsdlXMLcontents;
368    close(FOUT);
369
370    # c. Restart the fedora server
371    my $start_tomcat = &util::filename_cat($fedora_home, "tomcat", "bin", "startup".$script_ext);
372    $! = 0;
373    $status = system($start_tomcat);
374    if ($status!=0) {
375    print STDERR "Failed to restart the Fedora server... ERROR: $!\n";
376    print "Exit status = ", $status/256, "\n";
377    }
378
379    # reset CATALINA_HOME to GS' Tomcat (it is undefined for GS2 since GS2 has no tomcat):
380    $ENV{'CATALINA_HOME'} = $gs_catalina_home if defined $gs_catalina_home;
381   
382    # Starting up the Fedora server takes a long time. We need to wait for the server to be
383    # ready before import can continue, because g2f-import relies on an up-and-running Fedora
384    # server to purge the collection from it while g2f-build.pl needs a ready Fedora server
385    # in order to make it ingest the FedoraMETS. Sleeping is not sufficient (#sleep 10;) since
386    # the subsequent steps depend on a proper server restart.
387    # Dr Bainbridge's suggestion: test the server is ready with a call to wget.
388   
389    # Wget tries to retrieve the fedora search page (protocol://host:port/fedora/search)
390    # 20 times, waiting 3 seconds between each failed attempt. If it ultimately fails, we
391    # print a message to the user.
392    # The wget --spider option makes it check that the page is merely there rather than
393    # downloading it (see http://www.gnu.org/software/wget/manual/wget.html#Download-Options)
394    # -q is for quiet, --tries for the number of retries, --waitretry is the number of seconds
395    # between each attempt. Usually wget returns the contents of the page, but in our case it
396    # will return 0 for success since we are not downloading.
397
398    print STDERR "Fedora server restarted. Waiting for it to become ready...\n";
399    #print STDERR "****$protocol://$hostname:$port/fedora/search\n";
400    $! = 0;
401    #my $fedoraServerReady = system("wget -q --spider --waitretry=10 --tries=20 $protocol://$hostname:$port/fedora/search");
402
403    # The retries above won't work if the server isn't running:
404    # http://www.gnu.org/software/wget/manual/wget.html
405    #'--tries=number'
406    # Set number of retries to number. Specify 0 or 'inf' for infinite retrying. The default is to retry 20 times,
407    # with the exception of fatal errors like "connection refused" or "not found" (404), which ARE NOT RETRIED.
408
409    # retry fedora server every second for a total of 20 times until the server is ready
410    my $fedoraServerReady = 0;
411    my $count = 0;
412    do {
413    $fedoraServerReady = system("wget -q --spider $protocol://$hostname:$port/fedora/search");
414    if($fedoraServerReady != 0) {
415        sleep(1);
416        $count++;
417        #print STDERR "$count second(s)\n";
418    }
419    } while($fedoraServerReady != 0 && $count < 20);
420
421    if($fedoraServerReady != 0) {
422    print STDERR "Fedora server is still not ready... ERROR: $!\n";
423    print "Exit status = ", $fedoraServerReady/256, "\n";
424    die "Exiting....\n";
425    }
426
427    # return some indication that things went well
428    return "gsdl.xml";
429}
430
431
4321;
Note: See TracBrowser for help on using the browser.