1 | #!/usr/bin/perl -w
|
---|
2 |
|
---|
3 | BEGIN
|
---|
4 | {
|
---|
5 | if (!defined $ENV{'GSDLHOME'}) {
|
---|
6 | print STDERR "Environment variable GSDLHOME not set.\n";
|
---|
7 | print STDERR " Have you sourced Greenstone's 'setup.bash' file?\n";
|
---|
8 | exit 1;
|
---|
9 | }
|
---|
10 |
|
---|
11 | $ENV{'FEDORA_HOSTNAME'} = "localhost" if (!defined $ENV{'FEDORA_HOSTNAME'});
|
---|
12 | $ENV{'FEDORA_SERVER_PORT'} = "8080" if (!defined $ENV{'FEDORA_SERVER_PORT'});
|
---|
13 | $ENV{'FEDORA_USER'} = "fedoraAdmin" if (!defined $ENV{'FEDORA_USER'});
|
---|
14 | $ENV{'FEDORA_PASS'} = "fedoraAdmin" if (!defined $ENV{'FEDORA_PASS'});
|
---|
15 | $ENV{'FEDORA_PROTOCOL'} = "http" if (!defined $ENV{'FEDORA_PROTOCOL'});
|
---|
16 | $ENV{'FEDORA_PID_NAMESPACE'} = "greenstone" if (!defined $ENV{'FEDORA_PID_NAMESPACE'});
|
---|
17 | $ENV{'FEDORA_PREFIX'} = "/fedora" if (!defined $ENV{'FEDORA_PREFIX'});
|
---|
18 |
|
---|
19 | unshift (@INC, "$ENV{'GSDLHOME'}/perllib/");
|
---|
20 | }
|
---|
21 |
|
---|
22 |
|
---|
23 | use strict;
|
---|
24 | no strict 'refs'; # allow filehandles to be variables and vice versa
|
---|
25 | no strict 'subs'; # allow barewords (e.g. STDERR) as function arguments
|
---|
26 |
|
---|
27 | use util;
|
---|
28 | use gsprintf 'gsprintf';
|
---|
29 | use printusage;
|
---|
30 | use parse2;
|
---|
31 |
|
---|
32 |
|
---|
33 | use g2futil;
|
---|
34 |
|
---|
35 |
|
---|
36 | my $arguments =
|
---|
37 | [
|
---|
38 | { 'name' => "verbosity",
|
---|
39 | 'desc' => "{import.verbosity}",
|
---|
40 | 'type' => "string",
|
---|
41 | 'deft' => '1',
|
---|
42 | 'reqd' => "no",
|
---|
43 | 'hiddengli' => "no" },
|
---|
44 | { 'name' => "hostname",
|
---|
45 | 'desc' => "Domain hostname of Fedora server",
|
---|
46 | 'type' => "string",
|
---|
47 | 'deft' => $ENV{'FEDORA_HOSTNAME'},
|
---|
48 | 'reqd' => "no",
|
---|
49 | 'hiddengli' => "no" },
|
---|
50 | { 'name' => "port",
|
---|
51 | 'desc' => "Port that the Fedora server is running on.",
|
---|
52 | 'type' => "string",
|
---|
53 | 'deft' => $ENV{'FEDORA_SERVER_PORT'},
|
---|
54 | 'reqd' => "no",
|
---|
55 | 'hiddengli' => "no" },
|
---|
56 | { 'name' => "username",
|
---|
57 | 'desc' => "Fedora admin username",
|
---|
58 | 'type' => "string",
|
---|
59 | 'deft' => $ENV{'FEDORA_USER'},
|
---|
60 | 'reqd' => "no",
|
---|
61 | 'hiddengli' => "no" },
|
---|
62 | { 'name' => "password",
|
---|
63 | 'desc' => "Fedora admin password",
|
---|
64 | 'type' => "string",
|
---|
65 | 'deft' => $ENV{'FEDORA_PASS'},
|
---|
66 | 'reqd' => "no",
|
---|
67 | 'hiddengli' => "no" },
|
---|
68 | { 'name' => "protocol",
|
---|
69 | 'desc' => "Fedora protocol, e.g. 'http' or 'https'",
|
---|
70 | 'type' => "string",
|
---|
71 | 'deft' => $ENV{'FEDORA_PROTOCOL'},
|
---|
72 | 'reqd' => "no",
|
---|
73 | 'hiddengli' => "no" },
|
---|
74 | { 'name' => "pidnamespace",
|
---|
75 | 'desc' => "Fedora prefix for PIDs",
|
---|
76 | 'type' => "string",
|
---|
77 | 'deft' => $ENV{'FEDORA_PID_NAMESPACE'},
|
---|
78 | 'reqd' => "no",
|
---|
79 | 'hiddengli' => "no" },
|
---|
80 | { 'name' => "maxdocs",
|
---|
81 | 'desc' => "{import.maxdocs}",
|
---|
82 | 'type' => "int",
|
---|
83 | 'reqd' => "no",
|
---|
84 | 'range' => "1,",
|
---|
85 | 'modegli' => "1" },
|
---|
86 | { 'name' => "gli",
|
---|
87 | 'desc' => "",
|
---|
88 | 'type' => "flag",
|
---|
89 | 'reqd' => "no",
|
---|
90 | 'hiddengli' => "yes" },
|
---|
91 | { 'name' => "xml",
|
---|
92 | 'desc' => "{scripts.xml}",
|
---|
93 | 'type' => "flag",
|
---|
94 | 'reqd' => "no",
|
---|
95 | 'hiddengli' => "yes" },
|
---|
96 | { 'name' => "removeold",
|
---|
97 | 'desc' => "{import.removeold}",
|
---|
98 | 'type' => "flag",
|
---|
99 | 'reqd' => "no",
|
---|
100 | 'modegli' => "3" },
|
---|
101 | { 'name' => "language",
|
---|
102 | 'desc' => "{scripts.language}",
|
---|
103 | 'type' => "string",
|
---|
104 | 'reqd' => "no",
|
---|
105 | 'modegli' => "3" },
|
---|
106 | { 'name' => "collectdir",
|
---|
107 | 'desc' => "{import.collectdir}",
|
---|
108 | 'type' => "string",
|
---|
109 | 'deft' => "",
|
---|
110 | 'reqd' => "no",
|
---|
111 | 'hiddengli' => "yes" },
|
---|
112 |
|
---|
113 |
|
---|
114 | ];
|
---|
115 |
|
---|
116 | my $prog_options
|
---|
117 | = { 'name' => "g2fimport.pl",
|
---|
118 | 'desc' => "Generate Fedora METS format using Greenstone document processing capabilities",
|
---|
119 | 'args' => $arguments };
|
---|
120 |
|
---|
121 |
|
---|
122 | sub main
|
---|
123 | {
|
---|
124 | my (@ARGV) = @_;
|
---|
125 |
|
---|
126 | my $GSDLHOME = $ENV{'GSDLHOME'};
|
---|
127 |
|
---|
128 |
|
---|
129 | my $options = {};
|
---|
130 | # general options available to all plugins
|
---|
131 | my $intArgLeftinAfterParsing = parse2::parse(\@ARGV,$arguments,$options,"allow_extra_options");
|
---|
132 |
|
---|
133 | # Something went wrong with parsing
|
---|
134 | if ($intArgLeftinAfterParsing ==-1)
|
---|
135 | {
|
---|
136 | &PrintUsage::print_txt_usage($prog_options, "[options] greenstone-col");
|
---|
137 | die "\n";
|
---|
138 | }
|
---|
139 |
|
---|
140 | my $xml = $options->{'xml'};
|
---|
141 | my $gli = $options->{'gli'};
|
---|
142 |
|
---|
143 | if ($intArgLeftinAfterParsing != 1)
|
---|
144 | {
|
---|
145 | if ($xml) {
|
---|
146 | &PrintUsage::print_xml_usage($prog_options);
|
---|
147 | print "\n";
|
---|
148 | return;
|
---|
149 | }
|
---|
150 | else {
|
---|
151 | &PrintUsage::print_txt_usage($prog_options, "[options] greenstone-col");
|
---|
152 | print "\n";
|
---|
153 | return;
|
---|
154 | }
|
---|
155 |
|
---|
156 | }
|
---|
157 |
|
---|
158 | if ($gli) { # the gli wants strings to be in UTF-8
|
---|
159 | &gsprintf::output_strings_in_UTF8;
|
---|
160 | }
|
---|
161 |
|
---|
162 | my $gs_col = $ARGV[0];
|
---|
163 |
|
---|
164 | my $verbosity = $options->{'verbosity'};
|
---|
165 | my $hostname = $options->{'hostname'};
|
---|
166 | my $port = $options->{'port'};
|
---|
167 | my $username = $options->{'username'};
|
---|
168 | my $password = $options->{'password'};
|
---|
169 | my $protocol = $options->{'protocol'};
|
---|
170 | my $pid_namespace = $options->{'pidnamespace'};
|
---|
171 |
|
---|
172 | # The following are needed in the FedoraMETS plugout
|
---|
173 | $ENV{'FEDORA_HOSTNAME'} = $hostname;
|
---|
174 | $ENV{'FEDORA_SERVER_PORT'} = $port;
|
---|
175 |
|
---|
176 | my $language = $options->{'language'};
|
---|
177 | my $collectdir = $options->{'collectdir'};
|
---|
178 | my $removeold = $options->{'removeold'};
|
---|
179 | my $maxdocs = $options->{'maxdocs'};
|
---|
180 |
|
---|
181 | if (!$collectdir) {
|
---|
182 | # Explicitly set one, depending on whether it's GS2 or GS3
|
---|
183 | if($ENV{'GSDL3HOME'}) {
|
---|
184 | $collectdir = &util::filename_cat($ENV{'GSDL3HOME'},"sites","localsite","collect");
|
---|
185 | } else {
|
---|
186 | $collectdir = &util::filename_cat($ENV{'GSDLHOME'},"collect");
|
---|
187 | }
|
---|
188 | }
|
---|
189 |
|
---|
190 | # if GS3, and if Fedora uses Greenstone 3's tomcat, then we do not need to write out the file gsdl.xml into Fedora's tomcat
|
---|
191 | my $localfedora = &util::filename_cat($ENV{'GSDL3SRCHOME'}, "packages", "tomcat", "conf", "Catalina", "localhost", "fedora.xml");
|
---|
192 | unless($ENV{'GSDL3SRCHOME'} && -e $localfedora) {
|
---|
193 | # method that will tell Fedora where the ultimate output of g2f-import and g2f-build
|
---|
194 | # will be found: Points Fedora to the collect directory where FedoraMETS will be output.
|
---|
195 | &g2futil::write_gsdl_xml_file($hostname, $collectdir, $options);
|
---|
196 | }
|
---|
197 |
|
---|
198 | my $full_gs_col = &util::filename_cat($collectdir,$gs_col);
|
---|
199 |
|
---|
200 | if (!-e $full_gs_col ) {
|
---|
201 | print STDERR "Unable to find Greenstone collection $full_gs_col\n";
|
---|
202 | exit 1;
|
---|
203 | }
|
---|
204 |
|
---|
205 | my $export_dir = &util::filename_cat($full_gs_col,"export");
|
---|
206 |
|
---|
207 | if ( -e $export_dir ) {
|
---|
208 | print "***\n";
|
---|
209 | print "* Updating existing Greenstone $gs_col objects from Fedora $pid_namespace\n";
|
---|
210 | print "***\n";
|
---|
211 |
|
---|
212 | # set up fedoragsearch for updating the index upon ingesting documents
|
---|
213 | my $fedoragsearch_webapp = &g2futil::gsearch_webapp_folder();
|
---|
214 |
|
---|
215 | # need the username and password preset in order to run fedoraGSearch's RESTClient script
|
---|
216 | # this assumes that the fedoragsearch authentication details are the same as for fedora
|
---|
217 | if (defined $fedoragsearch_webapp) {
|
---|
218 | $ENV{'fgsUserName'} = $options->{'username'};
|
---|
219 | $ENV{'fgsPassword'} = $options->{'password'};
|
---|
220 | }
|
---|
221 |
|
---|
222 | # readdir
|
---|
223 | if (opendir(DIR, $export_dir)) {
|
---|
224 | my @xml_files = grep { $_ =~ m/^greenstone-http.*\.xml$/ } readdir(DIR);
|
---|
225 | closedir DIR;
|
---|
226 |
|
---|
227 | # purge all the (URL,hashID) metadata files that we inserted
|
---|
228 | # into fedora at the end of g2f-buildcol.pl
|
---|
229 | # convert the filenames into fedora-pids
|
---|
230 | # filename = greenstone-http=tmpcol-http-__test1-html.xml -> fpid = greenstone-http:tmpcol-http-__test1.html
|
---|
231 | foreach my $file (@xml_files) {
|
---|
232 | my $fedora_pid = $file;
|
---|
233 | $fedora_pid =~ s/\.xml$//;
|
---|
234 | $fedora_pid =~ s/\=/:/;
|
---|
235 | $fedora_pid =~ s/(.*)-(.*)$/$1.$2/;
|
---|
236 |
|
---|
237 | # print STDERR "#### fedora_pid: $fedora_pid\n";
|
---|
238 | &g2futil::run_purge($fedora_pid,$options); # displays error message if first time (nothing to purge)
|
---|
239 | }
|
---|
240 |
|
---|
241 | my @hash_dirs = &g2futil::get_all_hash_dirs($export_dir,$maxdocs);
|
---|
242 |
|
---|
243 | # for each hash dir, purge its respective PID
|
---|
244 | foreach my $full_hd (@hash_dirs) {
|
---|
245 |
|
---|
246 | my $hash_id = &g2futil::get_hash_id($full_hd);
|
---|
247 |
|
---|
248 | next if !defined $hash_id;
|
---|
249 |
|
---|
250 | my $pid = "$pid_namespace:$gs_col-$hash_id";
|
---|
251 |
|
---|
252 | my $dsinfo_status = &g2futil::run_datastore_info($pid,$options);
|
---|
253 |
|
---|
254 | if ($dsinfo_status == 0) {
|
---|
255 | # first remove the doc from the gsearch index before removing it from the fedora repository
|
---|
256 | print " deleting $pid from GSearch index\n";
|
---|
257 | &g2futil::run_delete_from_index($fedoragsearch_webapp,$pid,$options) if defined $fedoragsearch_webapp;
|
---|
258 |
|
---|
259 | print " $pid being updated.\n";
|
---|
260 | &g2futil::run_purge($pid,$options);
|
---|
261 | }
|
---|
262 | else {
|
---|
263 | print " $pid not present.\n";
|
---|
264 | }
|
---|
265 | }
|
---|
266 | }
|
---|
267 | else {
|
---|
268 | print STDERR "Error: Unable to open directory $export_dir: $!\n";
|
---|
269 | exit;
|
---|
270 | }
|
---|
271 | }
|
---|
272 |
|
---|
273 | print "***\n";
|
---|
274 | print "* Processing $gs_col into FedoraMETS format\n";
|
---|
275 | print "***\n";
|
---|
276 |
|
---|
277 | my $gs_export_opts = "-saveas FedoraMETS -fedora_namespace $pid_namespace";
|
---|
278 |
|
---|
279 | my $gs_opts = " -verbosity $verbosity";
|
---|
280 | $gs_opts .= " -gli" if ($gli);
|
---|
281 |
|
---|
282 | $gs_opts .= " -language $language" if ($language);
|
---|
283 | $gs_opts .= " -collectdir \"$collectdir\"" if ($collectdir);
|
---|
284 | $gs_opts .= " -removeold" if ($removeold);
|
---|
285 | $gs_opts .= " -maxdocs $maxdocs" if ($maxdocs);
|
---|
286 |
|
---|
287 | $gs_export_opts .= " $gs_opts -exportdir \"$export_dir\"";
|
---|
288 |
|
---|
289 | my $gs_export_arguments = "$gs_export_opts $gs_col";
|
---|
290 |
|
---|
291 | &g2futil::run_cmd("export.pl", $gs_export_arguments, $options);
|
---|
292 |
|
---|
293 | print STDERR "**** Just for now, also run Greenstone's import.pl\n";
|
---|
294 | # if we have the FedoraMETSPlugIN then we wouldn't have to run import anymore
|
---|
295 | my $gs_import_arguments = "$gs_opts $gs_col";
|
---|
296 |
|
---|
297 | &g2futil::run_cmd("import.pl", $gs_import_arguments, $options);
|
---|
298 | }
|
---|
299 |
|
---|
300 | &main(@ARGV);
|
---|
301 |
|
---|
302 |
|
---|
303 |
|
---|