| 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 | if (!defined $ENV{'JAVA_HOME'}) {
|
|---|
| 12 | print STDERR "Environment variable JAVA_HOME not set.\n";
|
|---|
| 13 | print STDERR "Needed by Fedora command line scripts.\n";
|
|---|
| 14 | exit 1;
|
|---|
| 15 | }
|
|---|
| 16 |
|
|---|
| 17 | $ENV{'FEDORA_HOSTNAME'} = "localhost" if (!defined $ENV{'FEDORA_HOSTNAME'});
|
|---|
| 18 | $ENV{'FEDORA_SERVER_PORT'} = "8080" if (!defined $ENV{'FEDORA_SERVER_PORT'});
|
|---|
| 19 | $ENV{'FEDORA_USER'} = "fedoraAdmin" if (!defined $ENV{'FEDORA_USER'});
|
|---|
| 20 | $ENV{'FEDORA_PASS'} = "fedoraAdmin" if (!defined $ENV{'FEDORA_PASS'});
|
|---|
| 21 | $ENV{'FEDORA_PROTOCOL'} = "http" if (!defined $ENV{'FEDORA_PROTOCOL'});
|
|---|
| 22 | $ENV{'FEDORA_PID_NAMESPACE'} = "greenstone" if (!defined $ENV{'FEDORA_PID_NAMESPACE'});
|
|---|
| 23 | $ENV{'FEDORA_PREFIX'} = "/fedora" if (!defined $ENV{'FEDORA_PREFIX'});
|
|---|
| 24 |
|
|---|
| 25 | unshift (@INC, "$ENV{'GSDLHOME'}/perllib/");
|
|---|
| 26 |
|
|---|
| 27 | }
|
|---|
| 28 |
|
|---|
| 29 |
|
|---|
| 30 | use strict;
|
|---|
| 31 | no strict 'refs'; # allow filehandles to be variables and vice versa
|
|---|
| 32 | no strict 'subs'; # allow barewords (e.g. STDERR) as function arguments
|
|---|
| 33 |
|
|---|
| 34 | use util;
|
|---|
| 35 | use gsprintf 'gsprintf';
|
|---|
| 36 | use printusage;
|
|---|
| 37 | use parse2;
|
|---|
| 38 | use cfgread;
|
|---|
| 39 | use colcfg;
|
|---|
| 40 |
|
|---|
| 41 | use g2futil;
|
|---|
| 42 |
|
|---|
| 43 | use dbutil;
|
|---|
| 44 |
|
|---|
| 45 | my $arguments =
|
|---|
| 46 | [
|
|---|
| 47 | { 'name' => "verbosity",
|
|---|
| 48 | 'desc' => "Level of verbosity generated",
|
|---|
| 49 | 'type' => "string",
|
|---|
| 50 | 'deft' => '1',
|
|---|
| 51 | 'reqd' => "no",
|
|---|
| 52 | 'hiddengli' => "no" },
|
|---|
| 53 | { 'name' => "hostname",
|
|---|
| 54 | 'desc' => "Domain hostname of Fedora server",
|
|---|
| 55 | 'type' => "string",
|
|---|
| 56 | 'deft' => $ENV{'FEDORA_HOSTNAME'},
|
|---|
| 57 | 'reqd' => "no",
|
|---|
| 58 | 'hiddengli' => "no" },
|
|---|
| 59 | { 'name' => "port",
|
|---|
| 60 | 'desc' => "Port that the Fedora server is running on.",
|
|---|
| 61 | 'type' => "string",
|
|---|
| 62 | 'deft' => $ENV{'FEDORA_SERVER_PORT'},
|
|---|
| 63 | 'reqd' => "no",
|
|---|
| 64 | 'hiddengli' => "no" },
|
|---|
| 65 | { 'name' => "username",
|
|---|
| 66 | 'desc' => "Fedora admin username",
|
|---|
| 67 | 'type' => "string",
|
|---|
| 68 | 'deft' => $ENV{'FEDORA_USER'},
|
|---|
| 69 | 'reqd' => "no",
|
|---|
| 70 | 'hiddengli' => "no" },
|
|---|
| 71 | { 'name' => "password",
|
|---|
| 72 | 'desc' => "Fedora admin password",
|
|---|
| 73 | 'type' => "string",
|
|---|
| 74 | 'deft' => $ENV{'FEDORA_PASS'},
|
|---|
| 75 | 'reqd' => "no",
|
|---|
| 76 | 'hiddengli' => "no" },
|
|---|
| 77 | { 'name' => "protocol",
|
|---|
| 78 | 'desc' => "Fedora protocol, e.g. 'http' or 'https'",
|
|---|
| 79 | 'type' => "string",
|
|---|
| 80 | 'deft' => $ENV{'FEDORA_PROTOCOL'},
|
|---|
| 81 | 'reqd' => "no",
|
|---|
| 82 | 'hiddengli' => "no" },
|
|---|
| 83 | { 'name' => "pidnamespace",
|
|---|
| 84 | 'desc' => "Fedora prefix for PIDs",
|
|---|
| 85 | 'type' => "string",
|
|---|
| 86 | 'deft' => $ENV{'FEDORA_PID_NAMESPACE'},
|
|---|
| 87 | 'reqd' => "no",
|
|---|
| 88 | 'hiddengli' => "no" },
|
|---|
| 89 | { 'name' => "gli",
|
|---|
| 90 | 'desc' => "",
|
|---|
| 91 | 'type' => "flag",
|
|---|
| 92 | 'reqd' => "no",
|
|---|
| 93 | 'hiddengli' => "yes" },
|
|---|
| 94 | { 'name' => "xml",
|
|---|
| 95 | 'desc' => "{scripts.xml}",
|
|---|
| 96 | 'type' => "flag",
|
|---|
| 97 | 'reqd' => "no",
|
|---|
| 98 | 'hiddengli' => "yes" },
|
|---|
| 99 | { 'name' => "removeold",
|
|---|
| 100 | 'desc' => "{import.removeold}",
|
|---|
| 101 | 'type' => "flag",
|
|---|
| 102 | 'reqd' => "no",
|
|---|
| 103 | 'modegli' => "3" },
|
|---|
| 104 | { 'name' => "language",
|
|---|
| 105 | 'desc' => "{scripts.language}",
|
|---|
| 106 | 'type' => "string",
|
|---|
| 107 | 'reqd' => "no",
|
|---|
| 108 | 'modegli' => "3" },
|
|---|
| 109 | { 'name' => "collectdir",
|
|---|
| 110 | 'desc' => "{import.collectdir}",
|
|---|
| 111 | 'type' => "string",
|
|---|
| 112 | 'deft' => "",
|
|---|
| 113 | 'reqd' => "no",
|
|---|
| 114 | 'hiddengli' => "yes" }
|
|---|
| 115 | ];
|
|---|
| 116 |
|
|---|
| 117 | my $prog_options
|
|---|
| 118 | = { 'name' => "g2fbuildcol.pl",
|
|---|
| 119 | 'desc' => "Ingest Greenstone directory of FedoraMETS documents into Fedora",
|
|---|
| 120 | 'args' => $arguments };
|
|---|
| 121 |
|
|---|
| 122 |
|
|---|
| 123 | sub main
|
|---|
| 124 | {
|
|---|
| 125 | my (@ARGV) = @_;
|
|---|
| 126 |
|
|---|
| 127 | my $GSDLHOME = $ENV{'GSDLHOME'};
|
|---|
| 128 |
|
|---|
| 129 |
|
|---|
| 130 | my $options = {};
|
|---|
| 131 | # general options available to all plugins
|
|---|
| 132 | my $intArgLeftinAfterParsing = parse2::parse(\@ARGV,$arguments,$options,"allow_extra_options");
|
|---|
| 133 |
|
|---|
| 134 | # Something went wrong with parsing
|
|---|
| 135 | if ($intArgLeftinAfterParsing ==-1)
|
|---|
| 136 | {
|
|---|
| 137 | &PrintUsage::print_txt_usage($prog_options, "[options] greenstone-col");
|
|---|
| 138 | die "\n";
|
|---|
| 139 | }
|
|---|
| 140 |
|
|---|
| 141 | my $xml = $options->{'xml'};
|
|---|
| 142 | my $gli = $options->{'gli'};
|
|---|
| 143 |
|
|---|
| 144 | if ($intArgLeftinAfterParsing != 1)
|
|---|
| 145 | {
|
|---|
| 146 | if ($xml) {
|
|---|
| 147 | &PrintUsage::print_xml_usage($prog_options);
|
|---|
| 148 | print "\n";
|
|---|
| 149 | return;
|
|---|
| 150 | }
|
|---|
| 151 | else {
|
|---|
| 152 | &PrintUsage::print_txt_usage($prog_options, "[options] greenstone-col");
|
|---|
| 153 | print "\n";
|
|---|
| 154 | return;
|
|---|
| 155 | }
|
|---|
| 156 |
|
|---|
| 157 | }
|
|---|
| 158 |
|
|---|
| 159 | my $gs_col = $ARGV[0];
|
|---|
| 160 |
|
|---|
| 161 | my $verbosity = $options->{'verbosity'};
|
|---|
| 162 | my $hostname = $options->{'hostname'};
|
|---|
| 163 | my $port = $options->{'port'};
|
|---|
| 164 | my $username = $options->{'username'};
|
|---|
| 165 | my $password = $options->{'password'};
|
|---|
| 166 | my $protocol = $options->{'protocol'};
|
|---|
| 167 | my $pid_namespace = $options->{'pidnamespace'};
|
|---|
| 168 |
|
|---|
| 169 | # The following are needed in the FedoraMETS plugout
|
|---|
| 170 | $ENV{'FEDORA_HOSTNAME'} = $hostname;
|
|---|
| 171 | $ENV{'FEDORA_SERVER_PORT'} = $port;
|
|---|
| 172 |
|
|---|
| 173 | my $collectdir = $options->{'collectdir'};
|
|---|
| 174 |
|
|---|
| 175 | if (!$collectdir) {
|
|---|
| 176 | if($ENV{'GSDL3HOME'}) {
|
|---|
| 177 | $collectdir = &util::filename_cat($ENV{'GSDL3HOME'},"sites","localsite","collect");
|
|---|
| 178 | } else {
|
|---|
| 179 | $collectdir = util::filename_cat($ENV{'GSDLHOME'},"collect");
|
|---|
| 180 | }
|
|---|
| 181 | }
|
|---|
| 182 |
|
|---|
| 183 | my $full_gs_col = util::filename_cat($collectdir,$gs_col);
|
|---|
| 184 |
|
|---|
| 185 |
|
|---|
| 186 | if (!-e $full_gs_col ) {
|
|---|
| 187 | print STDERR "Unable to find Greenstone collection $full_gs_col\n";
|
|---|
| 188 | exit 1;
|
|---|
| 189 | }
|
|---|
| 190 |
|
|---|
| 191 | ## my $archives_dir = &util::filename_cat($full_gs_col,"archives");
|
|---|
| 192 | my $export_dir = &util::filename_cat($full_gs_col,"export");
|
|---|
| 193 |
|
|---|
| 194 |
|
|---|
| 195 | print "***\n";
|
|---|
| 196 | print "* Ingesting Greenstone processed files into Fedora $pid_namespace\n";
|
|---|
| 197 | print "***\n";
|
|---|
| 198 |
|
|---|
| 199 | # Following falls foul of Schematron rule checking
|
|---|
| 200 | my $fd_add_prog = "fedora-ingest";
|
|---|
| 201 | # my $fd_add_cmd;
|
|---|
| 202 | # $fd_add_args = "dir $export_dir O metslikefedora1 $hostname:$port $username $password \\\n";
|
|---|
| 203 | # $fd_add_args .= " \"Automated_ingest_by_gs2fed.pl\"";
|
|---|
| 204 |
|
|---|
| 205 | # &g2futil::run_cmd($fd_add_prog,$fd_add_args,$options);
|
|---|
| 206 |
|
|---|
| 207 |
|
|---|
| 208 | # => Ingest individually!
|
|---|
| 209 |
|
|---|
| 210 | if (opendir(DIR, $export_dir)) {
|
|---|
| 211 | closedir DIR;
|
|---|
| 212 | ## my @hash_dirs = grep { /\.dir$/ } readdir(DIR);
|
|---|
| 213 | my @hash_dirs = &g2futil::get_all_hash_dirs($export_dir);
|
|---|
| 214 |
|
|---|
| 215 |
|
|---|
| 216 | # for each hash dir, purge its respective PID
|
|---|
| 217 | foreach my $hd (@hash_dirs) {
|
|---|
| 218 |
|
|---|
| 219 | my $hash_id = &g2futil::get_hash_id($hd);
|
|---|
| 220 |
|
|---|
| 221 | if (defined $hash_id) {
|
|---|
| 222 |
|
|---|
| 223 | my $pid = "$pid_namespace:$gs_col-$hash_id";
|
|---|
| 224 |
|
|---|
| 225 |
|
|---|
| 226 | my $dsinfo_status = &g2futil::run_datastore_info($pid,$options);
|
|---|
| 227 |
|
|---|
| 228 | if ($dsinfo_status == 0) {
|
|---|
| 229 | print " $pid being updated.\n";
|
|---|
| 230 | &g2futil::run_purge($pid,$options);
|
|---|
| 231 | }
|
|---|
| 232 | else {
|
|---|
| 233 | print " $pid not present.\n";
|
|---|
| 234 | }
|
|---|
| 235 | }
|
|---|
| 236 |
|
|---|
| 237 | my $docmets_filename
|
|---|
| 238 | = &util::filename_cat($hd,"docmets.xml");
|
|---|
| 239 |
|
|---|
| 240 | print STDERR "<Build>\n" if $gli;
|
|---|
| 241 |
|
|---|
| 242 | print "Ingesting $docmets_filename\n";
|
|---|
| 243 |
|
|---|
| 244 | &g2futil::run_ingest($docmets_filename,$options);
|
|---|
| 245 | print STDERR "</Build>\n" if $gli;
|
|---|
| 246 |
|
|---|
| 247 | }
|
|---|
| 248 | }
|
|---|
| 249 | else {
|
|---|
| 250 | print STDERR "Error: Unable to open directory $export_dir: $!\n";
|
|---|
| 251 | exit 1;
|
|---|
| 252 | }
|
|---|
| 253 |
|
|---|
| 254 |
|
|---|
| 255 | # can possibly use inexport instead of running buildcol.pl through system()
|
|---|
| 256 | print STDERR "**** Just for now, also run Greenstone's buildcol.pl\n";
|
|---|
| 257 |
|
|---|
| 258 | my $gs_opts = " -verbosity $verbosity";
|
|---|
| 259 | $gs_opts .= " -gli" if ($gli);
|
|---|
| 260 | $gs_opts .= " -collectdir \"$collectdir\"" if ($collectdir);
|
|---|
| 261 | $gs_opts .= " -mode infodb";
|
|---|
| 262 |
|
|---|
| 263 | my $gs_buildcol_arguments = "$gs_opts $gs_col";
|
|---|
| 264 |
|
|---|
| 265 | &g2futil::run_cmd("buildcol.pl", $gs_buildcol_arguments, $options);
|
|---|
| 266 |
|
|---|
| 267 | # read in collect cfg file to work out db type
|
|---|
| 268 | my $collectcfg = &util::filename_cat ($collectdir, $gs_col, "etc", "collectionConfig.xml");
|
|---|
| 269 | #print STDERR "**** collectcfg file: $collectcfg\n";
|
|---|
| 270 | unless(open(FIN, "<$collectcfg")) {
|
|---|
| 271 | print STDERR "g2f-buildcol.pl: Unable to open $collectcfg...ERROR: $!\n";
|
|---|
| 272 | exit 1;
|
|---|
| 273 | }
|
|---|
| 274 | close(FIN);
|
|---|
| 275 |
|
|---|
| 276 | # for now we assume GS3, since that's what the following gets implemented for
|
|---|
| 277 | my $collect_cfg = &colcfg::read_collection_cfg ($collectcfg, "gs3");
|
|---|
| 278 | # get the database type for this collection from its configuration file (may be undefined)
|
|---|
| 279 | my $infodbtype = $collect_cfg->{'infodbtype'} || &dbutil::get_default_infodb_type();
|
|---|
| 280 |
|
|---|
| 281 | # open .gdbm database file in building/text/$colname.gdb, using dbutil
|
|---|
| 282 | my $colname = $gs_col;
|
|---|
| 283 | $colname =~ s/(:?\\|\/)(.*)$/$1/; # remove any collect group from collection name to get tailname
|
|---|
| 284 |
|
|---|
| 285 | my $building_txt_dir = &util::filename_cat ($collectdir, $gs_col, "building", "text");
|
|---|
| 286 | my $building_txt_db = &dbutil::get_infodb_file_path($infodbtype, "$colname", $building_txt_dir);
|
|---|
| 287 |
|
|---|
| 288 | # foreach key that matches http://dir1/dir2/....file.xxx
|
|---|
| 289 | my $db_keys = {};
|
|---|
| 290 | &dbutil::read_infodb_keys($infodbtype,$building_txt_db, $db_keys);
|
|---|
| 291 |
|
|---|
| 292 | foreach my $key (keys %$db_keys) {
|
|---|
| 293 | if($key =~ m@^http://@) {
|
|---|
| 294 |
|
|---|
| 295 | # get value for the key
|
|---|
| 296 | my $src_rec_string = &dbutil::read_infodb_entry($infodbtype,$building_txt_db, $key);
|
|---|
| 297 | my $src_rec = &dbutil::convert_infodb_string_to_hash($src_rec_string);
|
|---|
| 298 | my $OID_hash_value = $src_rec->{'section'}->[0];
|
|---|
| 299 | $OID_hash_value = "$pid_namespace:$gs_col-".$OID_hash_value; # convert to fedoraPID
|
|---|
| 300 |
|
|---|
| 301 | # its fedora pid = "greenstone-http:$colname-http:||dir|file.xxx"
|
|---|
| 302 | # except that fedorapids don't like extra colons and don't like |
|
|---|
| 303 | my $fedora_identifier = "$pid_namespace-http:$gs_col-$key";
|
|---|
| 304 | # CAN'T HAVE | OR : (as in "http:||one|two.html") in fedoraPID
|
|---|
| 305 | $key =~ s@/@_@g;
|
|---|
| 306 | $key =~ s@:@-@g;
|
|---|
| 307 | my $fedora_pid = "$pid_namespace-http:$gs_col-$key";
|
|---|
| 308 |
|
|---|
| 309 | # To run fedora ingest on the new file need to have sensible
|
|---|
| 310 | # filenames that won't offend windows
|
|---|
| 311 | my $fedora_key_file_name = "$fedora_pid";
|
|---|
| 312 | $fedora_key_file_name =~ s@\.@-@g;
|
|---|
| 313 | $fedora_key_file_name =~ s/\:/=/g;
|
|---|
| 314 | $fedora_key_file_name .= ".xml";
|
|---|
| 315 | print STDERR "+++++ fpid: $fedora_pid, fedora-key filename: $fedora_key_file_name\n";
|
|---|
| 316 |
|
|---|
| 317 | # write out a FedoraMets File for this key (in /tmp)
|
|---|
| 318 | # -> it has one metadata value, which is 'dc:title' = HASHxxxxxx
|
|---|
| 319 |
|
|---|
| 320 | # The HASHID shouldn't be the title: then will have
|
|---|
| 321 | # duplicate titles and it will be hard to search for
|
|---|
| 322 | # unique ones. What about making the filename the
|
|---|
| 323 | # dc.title and the HASHID the dc.identifier
|
|---|
| 324 |
|
|---|
| 325 | my $contents = "<?xml version=\"1.0\" encoding=\"UTF-8\" standalone=\"no\"?>\n";
|
|---|
| 326 | $contents .= "<mets:mets xmlns:mets=\"http://www.loc.gov/METS/\"\n";
|
|---|
| 327 | $contents .= " xmlns:xsi=\"http://www.w3.org/2001/XMLSchema-instance\"\n";
|
|---|
| 328 | $contents .= " xmlns:gsdl3=\"http://www.greenstone.org/namespace/gsdlmetadata/1.0/\"\n";
|
|---|
| 329 | $contents .= " xmlns:xlink=\"http://www.w3.org/1999/xlink\"\n";
|
|---|
| 330 | $contents .= " xsi:schemaLocation=\"http://www.loc.gov/METS/\n";
|
|---|
| 331 | $contents .= " http://www.loc.gov/standards/mets/mets.xsd\n";
|
|---|
| 332 | $contents .= " http://www.greenstone.org/namespace/gsdlmetadata/1.0/\n";
|
|---|
| 333 | $contents .= " http://www.greenstone.org/namespace/gsdlmetadata/1.0/gsdl_metadata.xsd\"\n";
|
|---|
| 334 | $contents .= " OBJID=\"$fedora_pid\"\n";
|
|---|
| 335 | # $contents .= " OBJID=\"greenstone:$gs_col-HASH1f814d07252c354039ee11\"\n";
|
|---|
| 336 | $contents .= " TYPE=\"FedoraObject\" LABEL=\"$fedora_pid\" EXT_VERSION=\"1.1\">\n";
|
|---|
| 337 | $contents .= "<mets:metsHdr RECORDSTATUS=\"A\"/>\n";
|
|---|
| 338 | $contents .= " <mets:amdSec ID=\"DC\" >\n";
|
|---|
| 339 | $contents .= " <mets:techMD ID=\"DC.0\">\n";
|
|---|
| 340 | $contents .= " <mets:mdWrap LABEL=\"Metadata\" MDTYPE=\"OTHER\" OTHERMDTYPE=\"gsdl3\" ID=\"DCgsdl1\">\n";
|
|---|
| 341 | $contents .= " <mets:xmlData>\n";
|
|---|
| 342 | $contents .= " <oai_dc:dc xmlns:dc=\"http://purl.org/dc/elements/1.1/\" xmlns:oai_dc=\"http://www.openarchives.org/OAI/2.0/oai_dc/\" >\n";
|
|---|
| 343 | $contents .= " <dc:title>$OID_hash_value</dc:title>\n";
|
|---|
| 344 | # $contents .= " <dc:identifier>$fedora_identifier</dc:identifier>\n";
|
|---|
| 345 | $contents .= " </oai_dc:dc>\n";
|
|---|
| 346 | $contents .= " </mets:xmlData>\n";
|
|---|
| 347 | $contents .= " </mets:mdWrap>\n";
|
|---|
| 348 | $contents .= " </mets:techMD>\n";
|
|---|
| 349 | $contents .= " </mets:amdSec>\n";
|
|---|
| 350 | $contents .= "</mets:mets>\n";
|
|---|
| 351 |
|
|---|
| 352 |
|
|---|
| 353 | # write out the file and then run fedora ingest on that file
|
|---|
| 354 | # The file gets purged in g2f-import.pl, so don't remove it from export dir now
|
|---|
| 355 | my $fedora_key_file_path = &util::filename_cat($export_dir, $fedora_key_file_name);
|
|---|
| 356 | unless(open(FOUT, ">$fedora_key_file_path")) {
|
|---|
| 357 | print STDERR "g2f-buildcol.pl: Unable to open $fedora_key_file_path...ERROR: $!\n";
|
|---|
| 358 | exit 1;
|
|---|
| 359 | }
|
|---|
| 360 | print FOUT $contents;
|
|---|
| 361 | close(FOUT);
|
|---|
| 362 |
|
|---|
| 363 | print STDERR "<Build>\n" if $gli;
|
|---|
| 364 | print STDERR "Ingesting $fedora_key_file_name\n";
|
|---|
| 365 | print STDERR "#### ".join(",", %$options)."\n";
|
|---|
| 366 |
|
|---|
| 367 | &g2futil::run_ingest($fedora_key_file_path,$options);
|
|---|
| 368 | print STDERR "</Build>\n" if $gli;
|
|---|
| 369 | }
|
|---|
| 370 |
|
|---|
| 371 | }
|
|---|
| 372 |
|
|---|
| 373 |
|
|---|
| 374 | # If successful!!! Then need to think about:
|
|---|
| 375 | # [CLX] nodes
|
|---|
| 376 | # Doing this with FedoraMETSPlugin
|
|---|
| 377 |
|
|---|
| 378 |
|
|---|
| 379 | }
|
|---|
| 380 |
|
|---|
| 381 | &main(@ARGV);
|
|---|
| 382 |
|
|---|
| 383 |
|
|---|
| 384 |
|
|---|
