| 1 | ##########################################################################
|
|---|
| 2 | #
|
|---|
| 3 | # modmetadataaction.pm --
|
|---|
| 4 | # A component of the Greenstone digital library software
|
|---|
| 5 | # from the New Zealand Digital Library Project at the
|
|---|
| 6 | # University of Waikato, New Zealand.
|
|---|
| 7 | #
|
|---|
| 8 | # Copyright (C) 2009 New Zealand Digital Library Project
|
|---|
| 9 | #
|
|---|
| 10 | # This program is free software; you can redistribute it and/or modify
|
|---|
| 11 | # it under the terms of the GNU General Public License as published by
|
|---|
| 12 | # the Free Software Foundation; either version 2 of the License, or
|
|---|
| 13 | # (at your option) any later version.
|
|---|
| 14 | #
|
|---|
| 15 | # This program is distributed in the hope that it will be useful,
|
|---|
| 16 | # but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|---|
| 17 | # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|---|
| 18 | # GNU General Public License for more details.
|
|---|
| 19 | #
|
|---|
| 20 | # You should have received a copy of the GNU General Public License
|
|---|
| 21 | # along with this program; if not, write to the Free Software
|
|---|
| 22 | # Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
|
|---|
| 23 | #
|
|---|
| 24 | ###########################################################################
|
|---|
| 25 |
|
|---|
| 26 | # This is a submodule of metadataaction.pm. It is loaded conditionally
|
|---|
| 27 | # upon having the permissions to modify (not just get) metadata.
|
|---|
| 28 | # With sufficient permissions, these additional subroutines should be made
|
|---|
| 29 | # available to metadataaction.pm
|
|---|
| 30 |
|
|---|
| 31 | # See http://www.perlmonks.org/?node_id=881761 for splitting module into multiple files
|
|---|
| 32 | # and how variables declared with 'our' are used there.
|
|---|
| 33 |
|
|---|
| 34 | package metadataaction;
|
|---|
| 35 |
|
|---|
| 36 | use strict;
|
|---|
| 37 |
|
|---|
| 38 | use cgiactions::baseaction;
|
|---|
| 39 | use cgiactions::metadataactioncommon;
|
|---|
| 40 |
|
|---|
| 41 | use dbutil;
|
|---|
| 42 | use ghtml;
|
|---|
| 43 |
|
|---|
| 44 | use JSON;
|
|---|
| 45 |
|
|---|
| 46 | # convenient to have shorter version around
|
|---|
| 47 | my $FLDV_HISTORY_DIR = $metadataactioncommon::FLDV_HISTORY_DIR;
|
|---|
| 48 |
|
|---|
| 49 | $metadataaction::modmeta_action_table = #OR: our $modmeta_action_table =
|
|---|
| 50 | {
|
|---|
| 51 | #SET METHODS
|
|---|
| 52 | "set-live-metadata" => {
|
|---|
| 53 | 'compulsory-args' => [ "d", "metaname", "metavalue" ],
|
|---|
| 54 | 'optional-args' => [ ] },
|
|---|
| 55 |
|
|---|
| 56 | "set-metadata" => { # generic set-meta function. The 'where' param can be set to any combination of index|archives|import|live. docid d is still compulsory for setting index, archives and live meta
|
|---|
| 57 | 'compulsory-args' => [ "metaname", "metavalue" ],
|
|---|
| 58 | 'optional-args' => [ "where", "metapos", "metamode", "prevmetavalue", "d", "dv", "f" ] },
|
|---|
| 59 |
|
|---|
| 60 | "set-index-metadata" => {
|
|---|
| 61 | 'compulsory-args' => [ "d", "metaname", "metavalue" ],
|
|---|
| 62 | 'optional-args' => [ "metapos", "metamode" ] },
|
|---|
| 63 |
|
|---|
| 64 | "set-archives-metadata" => {
|
|---|
| 65 | 'compulsory-args' => [ "d", "metaname", "metavalue" ],
|
|---|
| 66 | 'optional-args' => [ "dv", "metapos", "metamode", "prevmetavalue" ] }, # metamode can be "accumulate", "override",
|
|---|
| 67 |
|
|---|
| 68 | ###### !!!!!
|
|---|
| 69 | # Should there not be a "set-archives-text" also ???!!!!??
|
|---|
| 70 | ###### !!!!!
|
|---|
| 71 |
|
|---|
| 72 | "set-archives-assocfile" => {
|
|---|
| 73 | 'compulsory-args' => [ "d", "assocname" ],
|
|---|
| 74 | 'optional-args' => [ "dv", "fileupload", "filedata" ] ,
|
|---|
| 75 | 'help-string' => [ "Because this action uploads a file, this action needs to be POSTed. The 'fileupload' parameter/field is the uploaded file content; 'assocname' specifies the filename within the archives's document's area that it will be saved as" ]
|
|---|
| 76 | },
|
|---|
| 77 |
|
|---|
| 78 | "set-import-metadata" => {
|
|---|
| 79 | 'compulsory-args' => [ "metaname", "metavalue" ],
|
|---|
| 80 | 'optional-args' => [ "d", "f", "metamode", "metapos", "prevmetavalue" ] }, # metamode can be "accumulate", "override", or "unique-id". Also need to add the ability to specify a previous metadata value to overwrite (because we can't use metapos). Metapos now supported, but assumes you are working with a Simple (instead of Complex) collection
|
|---|
| 81 |
|
|---|
| 82 | #SET METHODS (ARRAY)
|
|---|
| 83 | "set-metadata-array" => {
|
|---|
| 84 | 'compulsory-args' => [ "where", "json" ],
|
|---|
| 85 | 'optional-args' => [ ],
|
|---|
| 86 | 'help-string' => [
|
|---|
| 87 | 'A simple example: metadata-server.pl?a=set-metadata-array&where=archives|index|import&c=demo&json=[{"docid":"HASHc5bce2d6d3e5b04e470ec9","metaname":"Title","metavalue":"Tralalala","metamode":"accumulate"},{"docid":"HASHbe483fa4df4e096335d1c8","metaname":"Title","metavalue":"Lala was here","metapos":0, "metamode":"override"}]',
|
|---|
| 88 |
|
|---|
| 89 | 'A more complex example: metadata-server.pl?a=set-metadata-array&where=archives|index&c=demo&json=[{"docid":"HASHc5bce2d6d3e5b04e470ec9.1","metatable":[{"metaname":"Title","metavals":["Transformers","Robots in disguise","Autobots"]}],"metamode":"override"},{"docid":"HASHbe483fa4df4e096335d1c8.2","metaname":"Title","metavalue":"Pinky was here","metamode":"accumulate"}]' ] },
|
|---|
| 90 |
|
|---|
| 91 | # The same examples rewritten for when running the metadata-server.pl script from the commandline:
|
|---|
| 92 |
|
|---|
| 93 | # the simple example: metadata-server.pl a="set-metadata-array" where="archives|index|import" c="demo" json="[{\"docid\":\"HASHc5bce2d6d3e5b04e470ec9\",\"metaname\":\"Title\",\"metavalue\":\"Tralalala\",\"metamode\":\"accumulate\"},{\"docid\":\"HASHbe483fa4df4e096335d1c8\",\"metaname\":\"Title\",\"metavalue\":\"Lala was here\",\"metapos\":0, \"metamode\":\"override\"}]",
|
|---|
| 94 |
|
|---|
| 95 | # the more complex example: metadata-server.pl a="set-metadata-array" where="archives|index" c="demo" json="[{\"docid\":\"HASHc5bce2d6d3e5b04e470ec9.1\",\"metatable\":[{\"metaname\":\"Title\",\"metavals\":[\"Transformers\",\"Robots in disguise\",\"Autobots\"]}],\"metamode\":\"override\"},{\"docid\":\"HASHbe483fa4df4e096335d1c8.2\",\"metaname\":\"Title\",\"metavalue\":\"Pinky was here\",\"metamode\":\"accumulate\"}]"
|
|---|
| 96 |
|
|---|
| 97 | "set-archives-metadata-array" => {
|
|---|
| 98 | 'compulsory-args' => [ "json" ],
|
|---|
| 99 | 'optional-args' => [ ] },
|
|---|
| 100 |
|
|---|
| 101 | "set-import-metadata-array" => {
|
|---|
| 102 | 'compulsory-args' => [ "json" ],
|
|---|
| 103 | 'optional-args' => [ ] },
|
|---|
| 104 |
|
|---|
| 105 | "set-index-metadata-array" => {
|
|---|
| 106 | 'compulsory-args' => [ "json" ],
|
|---|
| 107 | 'optional-args' => [ ] },
|
|---|
| 108 |
|
|---|
| 109 | "set-live-metadata-array" => {
|
|---|
| 110 | 'compulsory-args' => [ "json" ],
|
|---|
| 111 | 'optional-args' => [ ] },
|
|---|
| 112 |
|
|---|
| 113 | #REMOVE METHODS
|
|---|
| 114 | "remove-import-metadata" => {
|
|---|
| 115 | 'compulsory-args' => [ "d", "metaname" ], #TODO: add f argument
|
|---|
| 116 | 'optional-args' => [ "metapos", "metavalue", "metamode" ] }, # only provide metapos arg for SIMPLE collections.
|
|---|
| 117 | # Metavalue is now an optional arg for remove_import_metadata() based on what the implementation did, which allowed metavalue to be undefined, and if so, used metapos.
|
|---|
| 118 |
|
|---|
| 119 | "remove-archives-metadata" => {
|
|---|
| 120 | 'compulsory-args' => [ "d", "metaname" ], #TODO: add f argument
|
|---|
| 121 | 'optional-args' => [ "dv", "metapos", "metavalue", "metamode" ] },
|
|---|
| 122 |
|
|---|
| 123 | "remove-live-metadata" => {
|
|---|
| 124 | 'compulsory-args' => [ "d", "metaname" ],
|
|---|
| 125 | 'optional-args' => [ ] },
|
|---|
| 126 |
|
|---|
| 127 | "remove-index-metadata" => {
|
|---|
| 128 | 'compulsory-args' => [ "d", "metaname" ],
|
|---|
| 129 | 'optional-args' => [ "metapos", "metavalue" ] },
|
|---|
| 130 |
|
|---|
| 131 | "remove-metadata" => { # generic remove-meta function. The 'where' param can be set to any combination of index|archives|import|live. docid d is still compulsory for setting index, archives and live meta
|
|---|
| 132 | 'compulsory-args' => [ "d", "metaname" ],
|
|---|
| 133 | 'optional-args' => [ "where", "dv", "metapos", "metavalue", "metamode" ] }, # metamode is optional since remove-metadata can call any of remove_import_meta and remove_archives_meta, remove_index_meta, of which the first two accept metamode as an optional param
|
|---|
| 134 |
|
|---|
| 135 |
|
|---|
| 136 | #REMOVE MULTIPLE (REMOVE ALL MATCHING METANAME AT DOCID)
|
|---|
| 137 | "erase-import-metadata" => {
|
|---|
| 138 | 'compulsory-args' => [ "d", "metaname" ] },
|
|---|
| 139 |
|
|---|
| 140 | "erase-archives-metadata" => {
|
|---|
| 141 | 'compulsory-args' => [ "d", "metaname" ],
|
|---|
| 142 | 'optional-args' => [ "dv" ] },
|
|---|
| 143 |
|
|---|
| 144 | "erase-index-metadata" => {
|
|---|
| 145 | 'compulsory-args' => [ "d", "metaname" ] },
|
|---|
| 146 |
|
|---|
| 147 | "erase-live-metadata" => {
|
|---|
| 148 | 'compulsory-args' => [ "d", "metaname" ] },
|
|---|
| 149 |
|
|---|
| 150 | "erase-metadata" => { # generic erase-meta function. The 'where' param can be set to any combination of index|archives|import|live. docid d is still compulsory for setting index, archives and live meta
|
|---|
| 151 | 'compulsory-args' => [ "d", "metaname" ],
|
|---|
| 152 | 'optional-args' => [ "where" ] },
|
|---|
| 153 |
|
|---|
| 154 |
|
|---|
| 155 | #INSERT METHODS
|
|---|
| 156 | "insert-metadata" => {
|
|---|
| 157 | 'compulsory-args' => [ "d", "metaname", "metavalue" ],
|
|---|
| 158 | 'optional-args' => [ ] }
|
|---|
| 159 | };
|
|---|
| 160 |
|
|---|
| 161 | sub _set_live_metadata
|
|---|
| 162 | {
|
|---|
| 163 | my $self = shift @_;
|
|---|
| 164 |
|
|---|
| 165 | my $collect = $self->{'collect'};
|
|---|
| 166 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
|---|
| 167 | my $infodbtype = $self->{'infodbtype'};
|
|---|
| 168 |
|
|---|
| 169 | # Obtain the collect dir
|
|---|
| 170 | my $site = $self->{'site'};
|
|---|
| 171 | my $collect_dir = $gsdl_cgi->get_collection_dir($site);
|
|---|
| 172 |
|
|---|
| 173 |
|
|---|
| 174 | # look up additional args
|
|---|
| 175 | my $docid = $self->{'d'};
|
|---|
| 176 | if ((!defined $docid) || ($docid =~ m/^\s*$/)) {
|
|---|
| 177 | $gsdl_cgi->generate_error("No docid (d=...) specified."); # generates error and dies
|
|---|
| 178 | }
|
|---|
| 179 | my $metavalue = $self->{'metavalue'};
|
|---|
| 180 |
|
|---|
| 181 | # Generate the dbkey
|
|---|
| 182 | my $metaname = $self->{'metaname'};
|
|---|
| 183 | my $dbkey = "$docid.$metaname";
|
|---|
| 184 |
|
|---|
| 185 | # To people who know $collect_tail please add some comments
|
|---|
| 186 | # Obtain path to the database
|
|---|
| 187 | my $collect_tail = $collect;
|
|---|
| 188 | $collect_tail =~ s/^.*[\/|\\]//;
|
|---|
| 189 | my $index_text_directory = &util::filename_cat($collect_dir,$collect,"index","text");
|
|---|
| 190 | my $infodb_file_path = &dbutil::get_infodb_file_path($infodbtype, "live-$collect_tail", $index_text_directory);
|
|---|
| 191 |
|
|---|
| 192 | # Set the new value
|
|---|
| 193 | my $cmd = "gdbmset \"$infodb_file_path\" \"$dbkey\" \"$metavalue\"";
|
|---|
| 194 | my $status = system($cmd);
|
|---|
| 195 | if ($status != 0) {
|
|---|
| 196 | # Catch error if gdbmget failed
|
|---|
| 197 | my $mess = "Failed to set metadata key: $dbkey\n";
|
|---|
| 198 |
|
|---|
| 199 | $mess .= "PATH: $ENV{'PATH'}\n";
|
|---|
| 200 | $mess .= "cmd = $cmd\n";
|
|---|
| 201 | $mess .= "Exit status: $status\n";
|
|---|
| 202 | $mess .= "System Error Message: $!\n";
|
|---|
| 203 |
|
|---|
| 204 | $gsdl_cgi->generate_error($mess);
|
|---|
| 205 | }
|
|---|
| 206 | else {
|
|---|
| 207 | $gsdl_cgi->generate_ok_message("set-live-metadata successful: Key[$metaname]=$metavalue");
|
|---|
| 208 | }
|
|---|
| 209 |
|
|---|
| 210 | #return $status; # in case calling functions have any further use for this
|
|---|
| 211 | }
|
|---|
| 212 |
|
|---|
| 213 | sub set_live_metadata
|
|---|
| 214 | {
|
|---|
| 215 | my $self = shift @_;
|
|---|
| 216 |
|
|---|
| 217 | my $username = $self->{'username'};
|
|---|
| 218 | my $collect = $self->{'collect'};
|
|---|
| 219 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
|---|
| 220 |
|
|---|
| 221 | if ($baseaction::authentication_enabled) {
|
|---|
| 222 | # Ensure the user is allowed to edit this collection
|
|---|
| 223 | $self->authenticate_user($username, $collect); #&authenticate_user($gsdl_cgi, $username, $collect);
|
|---|
| 224 | }
|
|---|
| 225 |
|
|---|
| 226 | # Make sure the collection isn't locked by someone else
|
|---|
| 227 | $self->lock_collection($username, $collect);
|
|---|
| 228 |
|
|---|
| 229 | $self->_set_live_metadata(@_);
|
|---|
| 230 |
|
|---|
| 231 | # Release the lock once it is done
|
|---|
| 232 | $self->unlock_collection($username, $collect);
|
|---|
| 233 | }
|
|---|
| 234 |
|
|---|
| 235 | sub set_index_metadata_entry
|
|---|
| 236 | {
|
|---|
| 237 | my $self = shift @_;
|
|---|
| 238 | my ($collect_dir,$collect,$infodbtype,$docid,$metaname,$metapos,$metavalue,$metamode,$prevmetavalue) = @_;
|
|---|
| 239 |
|
|---|
| 240 | # To people who know $collect_tail please add some comments
|
|---|
| 241 | # Obtain path to the database
|
|---|
| 242 | my $collect_tail = $collect;
|
|---|
| 243 | $collect_tail =~ s/^.*[\/|\\]//;
|
|---|
| 244 | my $index_text_directory = &util::filename_cat($collect_dir,$collect,"index","text");
|
|---|
| 245 | my $infodb_file_path = &dbutil::get_infodb_file_path($infodbtype, $collect_tail, $index_text_directory);
|
|---|
| 246 |
|
|---|
| 247 | # print STDERR "**** infodb file path = $infodb_file_path\n";
|
|---|
| 248 | # print STDERR "***** infodb type = $infodbtype\n";
|
|---|
| 249 |
|
|---|
| 250 | # Read the docid entry
|
|---|
| 251 | my $doc_rec = &dbutil::read_infodb_entry($infodbtype, $infodb_file_path, $docid);
|
|---|
| 252 |
|
|---|
| 253 | # Set the metadata value
|
|---|
| 254 | if (defined $metapos) {
|
|---|
| 255 | # if metamode=accumulate AND metapos, warn user and then use metapos
|
|---|
| 256 | if (defined $metamode && $metamode eq "accumulate") {
|
|---|
| 257 | print STDERR "**** Warning: metamode is set to accumulate yet metapos is also provided for $docid\n";
|
|---|
| 258 | print STDERR "**** Proceeding by using metapos\n";
|
|---|
| 259 | }
|
|---|
| 260 | $doc_rec->{$metaname}->[$metapos] = $metavalue;
|
|---|
| 261 | }
|
|---|
| 262 | elsif (defined $prevmetavalue) {
|
|---|
| 263 | my $array = $doc_rec->{$metaname};
|
|---|
| 264 | my $length = @$array;
|
|---|
| 265 |
|
|---|
| 266 | my $found = 0;
|
|---|
| 267 | for (my $i = 0; $i < $length; $i++){
|
|---|
| 268 | if(defined $doc_rec->{$metaname}->[$i] && $doc_rec->{$metaname}->[$i] eq $prevmetavalue){
|
|---|
| 269 | $doc_rec->{$metaname}->[$i] = $metavalue;
|
|---|
| 270 | $found = 1;
|
|---|
| 271 | last;
|
|---|
| 272 | }
|
|---|
| 273 | }
|
|---|
| 274 |
|
|---|
| 275 | if($found == 0){
|
|---|
| 276 | $doc_rec->{$metaname} = [ $metavalue ];
|
|---|
| 277 | }
|
|---|
| 278 | }
|
|---|
| 279 | elsif (defined $metamode && $metamode eq "override") {
|
|---|
| 280 | $doc_rec->{$metaname} = [ $metavalue ];
|
|---|
| 281 | }
|
|---|
| 282 | else { # default for index was to override, but because accumulate is less destructive,
|
|---|
| 283 | # and because accumulate is the default for archives and import, that's the new default for index too
|
|---|
| 284 | if(defined $doc_rec->{$metaname}) {
|
|---|
| 285 | push(@{$doc_rec->{$metaname}}, $metavalue); # accumulate the value for that metaname
|
|---|
| 286 | } else {
|
|---|
| 287 | $doc_rec->{$metaname} = [ $metavalue ];
|
|---|
| 288 | }
|
|---|
| 289 | }
|
|---|
| 290 |
|
|---|
| 291 | my $status = &dbutil::set_infodb_entry($infodbtype, $infodb_file_path,$docid,$doc_rec);
|
|---|
| 292 |
|
|---|
| 293 | return $status;
|
|---|
| 294 |
|
|---|
| 295 | }
|
|---|
| 296 |
|
|---|
| 297 | sub _set_import_metadata
|
|---|
| 298 | {
|
|---|
| 299 | my $self = shift @_;
|
|---|
| 300 |
|
|---|
| 301 | my $collect = $self->{'collect'};
|
|---|
| 302 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
|---|
| 303 | my $infodbtype = $self->{'infodbtype'};
|
|---|
| 304 |
|
|---|
| 305 | # Obtain the collect and archive dir
|
|---|
| 306 | my $site = $self->{'site'};
|
|---|
| 307 | my $collect_dir = $gsdl_cgi->get_collection_dir($site);
|
|---|
| 308 | my $archive_dir = &util::filename_cat($collect_dir,$collect,"archives");
|
|---|
| 309 |
|
|---|
| 310 | # look up additional args
|
|---|
| 311 | # want either d= or f=
|
|---|
| 312 | my $docid = $self->{'d'};
|
|---|
| 313 | my ($docid_root,$docid_secnum);
|
|---|
| 314 | if(defined $docid) {
|
|---|
| 315 | ($docid_root,$docid_secnum) = ($docid =~ m/^(.*?)(\..*)?$/);
|
|---|
| 316 | # as yet no support for setting subsection metadata in metadata.xml
|
|---|
| 317 | if ((defined $docid_secnum) && ($docid_secnum !~ m/^\s*$/)) {
|
|---|
| 318 | $gsdl_cgi->generate_message("*** No support yet for setting import metadata at subsections level.\n");
|
|---|
| 319 | return;
|
|---|
| 320 | }
|
|---|
| 321 | }
|
|---|
| 322 |
|
|---|
| 323 | my $import_file = $self->{'f'};
|
|---|
| 324 | if ((!defined $docid || $docid =~ m/^\s*$/) && (!defined $import_file || $import_file =~ m/^\s*$/)) {
|
|---|
| 325 | $gsdl_cgi->generate_error("No docid (d=...) or import file (f=) specified."); # at least d or f must be specified
|
|---|
| 326 | }
|
|---|
| 327 |
|
|---|
| 328 | # Get the parameters and set default mode to "accumulate"
|
|---|
| 329 | my $metaname = $self->{'metaname'};
|
|---|
| 330 | my $metavalue = $self->{'metavalue'};
|
|---|
| 331 | ## $metavalue =~ s/&lt;(.*?)&gt;/<$1>/g;
|
|---|
| 332 | $metavalue =~ s/<(.*?)>/<$1>/g;
|
|---|
| 333 |
|
|---|
| 334 | my $metamode = $self->{'metamode'};
|
|---|
| 335 | if ((!defined $metamode) || ($metamode =~ m/^\s*$/)) {
|
|---|
| 336 | # make "accumulate" the default (less destructive, as it won't actually
|
|---|
| 337 | # delete any existing values)
|
|---|
| 338 | $metamode = "accumulate";
|
|---|
| 339 | }
|
|---|
| 340 |
|
|---|
| 341 | # adding metapos and prevmetavalue support to import_metadata subroutines
|
|---|
| 342 | my $metapos = $self->{'metapos'}; # don't force undef to 0. Undef has meaning when metamode=override
|
|---|
| 343 | my $prevmetavalue = $self->{'prevmetavalue'};
|
|---|
| 344 | $metapos = undef if(defined $metapos && ($metapos =~ m/^\s*$/));
|
|---|
| 345 | $prevmetavalue = undef if(defined $prevmetavalue && ($prevmetavalue =~ m/^\s*$/));
|
|---|
| 346 |
|
|---|
| 347 | my $arcinfo_doc_filename = &dbutil::get_infodb_file_path($infodbtype, "archiveinf-doc", $archive_dir);
|
|---|
| 348 | my $metadata_xml_filename = $self->set_import_metadata_entry($gsdl_cgi, $arcinfo_doc_filename, $infodbtype, $docid_root, $metaname, $metapos,$metavalue, $metamode,$prevmetavalue, $collect, $collect_dir); # at this point, docid_root = docid
|
|---|
| 349 |
|
|---|
| 350 | my $mess = "set-import-metadata successful: Key[$docid] -> $metadata_xml_filename\n";
|
|---|
| 351 | $mess .= " $metaname";
|
|---|
| 352 | $mess .= " = $metavalue";
|
|---|
| 353 | $mess .= " ($metamode)\n";
|
|---|
| 354 |
|
|---|
| 355 | $gsdl_cgi->generate_ok_message($mess);
|
|---|
| 356 |
|
|---|
| 357 | #return $status; # in case calling functions have any further use for this
|
|---|
| 358 | }
|
|---|
| 359 |
|
|---|
| 360 | # the version of set_archives_meta that doesn't do authentication
|
|---|
| 361 | sub _set_archives_metadata
|
|---|
| 362 | {
|
|---|
| 363 | my $self = shift @_;
|
|---|
| 364 |
|
|---|
| 365 | my $collect = $self->{'collect'};
|
|---|
| 366 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
|---|
| 367 | my $infodbtype = $self->{'infodbtype'};
|
|---|
| 368 |
|
|---|
| 369 | # Obtain the collect and archive dir
|
|---|
| 370 | my $site = $self->{'site'};
|
|---|
| 371 | my $collect_dir = $gsdl_cgi->get_collection_dir($site);
|
|---|
| 372 | my $archive_dir = &util::filename_cat($collect_dir,$collect,"archives");
|
|---|
| 373 |
|
|---|
| 374 | # look up additional args
|
|---|
| 375 | my $docid = $self->{'d'};
|
|---|
| 376 | my $dv = $self->{'dv'};
|
|---|
| 377 |
|
|---|
| 378 | my $metaname = $self->{'metaname'};
|
|---|
| 379 | my $metavalue = $self->{'metavalue'};
|
|---|
| 380 | my $prevmetavalue = $self->{'prevmetavalue'};
|
|---|
| 381 |
|
|---|
| 382 | my $metapos = $self->{'metapos'}; # don't force undef to 0. Undef has meaning when metamode=override
|
|---|
| 383 | # Don't append "|| undef", since if metapos=0 it will then be set to undef
|
|---|
| 384 |
|
|---|
| 385 | $metapos = undef if(defined $metapos && ($metapos =~ m/^\s*$/));
|
|---|
| 386 | $prevmetavalue = undef if(defined $prevmetavalue && ($prevmetavalue =~ m/^\s*$/));
|
|---|
| 387 |
|
|---|
| 388 | my $metamode = $self->{'metamode'};
|
|---|
| 389 | if ((!defined $metamode) || ($metamode =~ m/^\s*$/)) {
|
|---|
| 390 | # make "accumulate" the default (less destructive, as it won't actually
|
|---|
| 391 | # delete any existing values)
|
|---|
| 392 | $metamode = "accumulate";
|
|---|
| 393 | }
|
|---|
| 394 |
|
|---|
| 395 | my $status = $self->set_archives_metadata_entry($gsdl_cgi,$archive_dir, $collect_dir,$collect, $infodbtype,
|
|---|
| 396 | $docid,$dv, $metaname,$metapos,$metavalue,$metamode,$prevmetavalue);
|
|---|
| 397 |
|
|---|
| 398 | if ($status == 0) {
|
|---|
| 399 | my $mess = "set-archives-metadata successful: Key[$docid]\n";
|
|---|
| 400 | $mess .= " $metaname";
|
|---|
| 401 | $mess .= "->[$metapos]" if (defined $metapos);
|
|---|
| 402 | $mess .= " = $metavalue";
|
|---|
| 403 | $mess .= " ($metamode)\n";
|
|---|
| 404 |
|
|---|
| 405 | $gsdl_cgi->generate_ok_message($mess);
|
|---|
| 406 | }
|
|---|
| 407 | else {
|
|---|
| 408 | my $mess .= "Failed to set archives metadata key: $docid\n";
|
|---|
| 409 | $mess .= "Exit status: $status\n";
|
|---|
| 410 | if(defined $self->{'error_msg'}) {
|
|---|
| 411 | $mess .= "Error Message: $self->{'error_msg'}\n";
|
|---|
| 412 | } else {
|
|---|
| 413 | $mess .= "System Error Message: $!\n";
|
|---|
| 414 | }
|
|---|
| 415 | $mess .= "-" x 20 . "\n";
|
|---|
| 416 |
|
|---|
| 417 | $gsdl_cgi->generate_error($mess);
|
|---|
| 418 | }
|
|---|
| 419 |
|
|---|
| 420 | #return $status; # in case calling functions have any further use for this
|
|---|
| 421 | }
|
|---|
| 422 |
|
|---|
| 423 |
|
|---|
| 424 | # the version of set_index_meta that doesn't do authentication
|
|---|
| 425 | sub _set_index_metadata
|
|---|
| 426 | {
|
|---|
| 427 | print STDERR "START SET INDEX METADATA\n";
|
|---|
| 428 | my $self = shift @_;
|
|---|
| 429 |
|
|---|
| 430 | my $collect = $self->{'collect'};
|
|---|
| 431 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
|---|
| 432 |
|
|---|
| 433 | # Obtain the collect dir
|
|---|
| 434 | my $site = $self->{'site'};
|
|---|
| 435 | my $collect_dir = $gsdl_cgi->get_collection_dir($site);
|
|---|
| 436 |
|
|---|
| 437 | # look up additional args
|
|---|
| 438 | my $docid = $self->{'d'};
|
|---|
| 439 | my $metaname = $self->{'metaname'};
|
|---|
| 440 | my $metapos = $self->{'metapos'}; # undef has meaning
|
|---|
| 441 | my $metavalue = $self->{'metavalue'};
|
|---|
| 442 | my $infodbtype = $self->{'infodbtype'};
|
|---|
| 443 | my $metamode = $self->{'metamode'};
|
|---|
| 444 | my $prevmetavalue = $self->{'prevmetavalue'};
|
|---|
| 445 |
|
|---|
| 446 | $metapos = undef if(defined $metapos && ($metapos =~ m/^\s*$/));
|
|---|
| 447 | $prevmetavalue = undef if(defined $prevmetavalue && ($prevmetavalue =~ m/^\s*$/));
|
|---|
| 448 |
|
|---|
| 449 | print STDERR "SETTING INDEX METADATA ENTRY\n";
|
|---|
| 450 | my $status = $self->set_index_metadata_entry($collect_dir,$collect,$infodbtype,$docid,$metaname,$metapos,$metavalue,$metamode,$prevmetavalue);
|
|---|
| 451 | print STDERR "DONE SETTING INDEX METADATA ENTRY\n";
|
|---|
| 452 | if ($status != 0) {
|
|---|
| 453 | # Catch error if set infodb entry failed
|
|---|
| 454 | my $mess = "Failed to set metadata key: $docid\n";
|
|---|
| 455 |
|
|---|
| 456 | $mess .= "PATH: $ENV{'PATH'}\n";
|
|---|
| 457 | $mess .= "Exit status: $status\n";
|
|---|
| 458 | $mess .= "System Error Message: $!\n";
|
|---|
| 459 |
|
|---|
| 460 | $gsdl_cgi->generate_error($mess);
|
|---|
| 461 | }
|
|---|
| 462 | else {
|
|---|
| 463 | my $mess = "set-index-metadata successful: Key[$docid]\n";
|
|---|
| 464 | $mess .= " $metaname";
|
|---|
| 465 | $mess .= "->[$metapos]" if (defined $metapos);
|
|---|
| 466 | $mess .= " = $metavalue\n";
|
|---|
| 467 |
|
|---|
| 468 | $gsdl_cgi->generate_ok_message($mess);
|
|---|
| 469 | }
|
|---|
| 470 |
|
|---|
| 471 | print STDERR "END SET INDEX METADATA\n";
|
|---|
| 472 | #return $status; # in case calling functions have any further use for this
|
|---|
| 473 | }
|
|---|
| 474 |
|
|---|
| 475 | sub set_index_metadata
|
|---|
| 476 | {
|
|---|
| 477 | my $self = shift @_;
|
|---|
| 478 |
|
|---|
| 479 | my $username = $self->{'username'};
|
|---|
| 480 | my $collect = $self->{'collect'};
|
|---|
| 481 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
|---|
| 482 |
|
|---|
| 483 | if ($baseaction::authentication_enabled) {
|
|---|
| 484 | # Ensure the user is allowed to edit this collection
|
|---|
| 485 | $self->authenticate_user($username, $collect); #&authenticate_user($gsdl_cgi, $username, $collect);
|
|---|
| 486 | }
|
|---|
| 487 |
|
|---|
| 488 | # Obtain the collect dir
|
|---|
| 489 | my $site = $self->{'site'};
|
|---|
| 490 | my $collect_dir = $gsdl_cgi->get_collection_dir($site);
|
|---|
| 491 |
|
|---|
| 492 | $gsdl_cgi->checked_chdir($collect_dir);
|
|---|
| 493 |
|
|---|
| 494 | # Make sure the collection isn't locked by someone else
|
|---|
| 495 | $self->lock_collection($username, $collect);
|
|---|
| 496 |
|
|---|
| 497 | $self->_set_index_metadata(@_);
|
|---|
| 498 |
|
|---|
| 499 | # Release the lock once it is done
|
|---|
| 500 | $self->unlock_collection($username, $collect);
|
|---|
| 501 | }
|
|---|
| 502 |
|
|---|
| 503 | # call this to set the metadata for a combination of dirs archives, import or index, or live
|
|---|
| 504 | # if none specified, defaults to index which was the original behaviour of set_metadata.
|
|---|
| 505 | sub set_metadata
|
|---|
| 506 | {
|
|---|
| 507 | my $self = shift @_;
|
|---|
| 508 |
|
|---|
| 509 | # Testing that not defining a variable, setting it to "" or to " " all return false
|
|---|
| 510 | # >perl -e 'my $whichdirs=""; if($whichdirs) {print "$whichdirs\n"};'
|
|---|
| 511 |
|
|---|
| 512 | my $where = $self->{'where'};
|
|---|
| 513 | if(!$where || ($where =~ m/^\s*$/)) {
|
|---|
| 514 | $self->set_index_metadata(@_); # call the full version of set_index_meta for the default behaviour
|
|---|
| 515 | return;
|
|---|
| 516 | }
|
|---|
| 517 |
|
|---|
| 518 | # authenticate and lock collection once, even if processing multiple dirs
|
|---|
| 519 | my $username = $self->{'username'};
|
|---|
| 520 | my $collect = $self->{'collect'};
|
|---|
| 521 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
|---|
| 522 |
|
|---|
| 523 | if ($baseaction::authentication_enabled) {
|
|---|
| 524 | # Ensure the user is allowed to edit this collection
|
|---|
| 525 | $self->authenticate_user($username, $collect); #&authenticate_user($gsdl_cgi, $username, $collect);
|
|---|
| 526 | }
|
|---|
| 527 |
|
|---|
| 528 | if($where =~ m/index/) {
|
|---|
| 529 | my $site = $self->{'site'};
|
|---|
| 530 | my $collect_dir = $gsdl_cgi->get_collection_dir($site);
|
|---|
| 531 | $gsdl_cgi->checked_chdir($collect_dir);
|
|---|
| 532 | }
|
|---|
| 533 |
|
|---|
| 534 | # Make sure the collection isn't locked by someone else
|
|---|
| 535 | $self->lock_collection($username, $collect);
|
|---|
| 536 |
|
|---|
| 537 |
|
|---|
| 538 | # now at last can set the metadata. $where can specify multiple
|
|---|
| 539 | # $where is of the form: import|archives|index, or a subset thereof
|
|---|
| 540 |
|
|---|
| 541 | #my @whichdirs = split('\|', $where);
|
|---|
| 542 |
|
|---|
| 543 | # just check whether $where contains import/archives/index/live in turn, and
|
|---|
| 544 | # for each case, process it accordingly
|
|---|
| 545 | if($where =~ m/import/) {
|
|---|
| 546 | $self->_set_import_metadata(@_);
|
|---|
| 547 | }
|
|---|
| 548 |
|
|---|
| 549 | if($where =~ m/archives/) {
|
|---|
| 550 |
|
|---|
| 551 | # look up docID arg which is optional to set_metadata because it's optional
|
|---|
| 552 | # to set_import, but which is compulsory to set_archives_metadata
|
|---|
| 553 | my $docid = $self->{'d'};
|
|---|
| 554 | if ((!defined $docid) || ($docid =~ m/^\s*$/)) {
|
|---|
| 555 | $gsdl_cgi->generate_error("No docid (d=...) specified."); # generates error and dies
|
|---|
| 556 | }
|
|---|
| 557 | # we have a docid, so can set archives meta
|
|---|
| 558 | $self->_set_archives_metadata(@_);
|
|---|
| 559 | }
|
|---|
| 560 |
|
|---|
| 561 | if($where =~ m/index/) {
|
|---|
| 562 |
|
|---|
| 563 | # look up docID arg which is optional to set_metadata because it's optional
|
|---|
| 564 | # to set_import, but which is compulsory to set_archives_metadata
|
|---|
| 565 | my $docid = $self->{'d'};
|
|---|
| 566 | if ((!defined $docid) || ($docid =~ m/^\s*$/)) {
|
|---|
| 567 | $gsdl_cgi->generate_error("No docid (d=...) specified.");
|
|---|
| 568 | }
|
|---|
| 569 | # we have a docid, so can set index meta
|
|---|
| 570 | $self->_set_index_metadata(@_);
|
|---|
| 571 | }
|
|---|
| 572 |
|
|---|
| 573 | if($where =~ m/live/) {
|
|---|
| 574 | $self->_set_live_metadata(@_); # docid param, d, is compulsory, but is checked for in subroutine
|
|---|
| 575 | }
|
|---|
| 576 |
|
|---|
| 577 | # Release the lock once it is done
|
|---|
| 578 | $self->unlock_collection($username, $collect);
|
|---|
| 579 | }
|
|---|
| 580 |
|
|---|
| 581 | sub set_metadata_array
|
|---|
| 582 | {
|
|---|
| 583 | my $self = shift @_;
|
|---|
| 584 |
|
|---|
| 585 | my $where = $self->{'where'};
|
|---|
| 586 | if(!$where || ($where =~ m/^\s*$/)) {
|
|---|
| 587 | $self->set_index_metadata_array(@_); # default behaviour is the full version of set_index_meta_array
|
|---|
| 588 | return;
|
|---|
| 589 | }
|
|---|
| 590 |
|
|---|
| 591 | my $username = $self->{'username'};
|
|---|
| 592 | my $collect = $self->{'collect'};
|
|---|
| 593 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
|---|
| 594 |
|
|---|
| 595 | if ($baseaction::authentication_enabled) {
|
|---|
| 596 | # Ensure the user is allowed to edit this collection
|
|---|
| 597 | $self->authenticate_user($username, $collect); #&authenticate_user($gsdl_cgi, $username, $collect);
|
|---|
| 598 | }
|
|---|
| 599 |
|
|---|
| 600 | # Not sure if the checked_chdir is necessary, since lock_collection also does a chdir
|
|---|
| 601 | # But including the stmt during this code reorganisation to preserve as-is what used to happen
|
|---|
| 602 | my $site = $self->{'site'};
|
|---|
| 603 | my $collect_dir = $gsdl_cgi->get_collection_dir($site);
|
|---|
| 604 | $gsdl_cgi->checked_chdir($collect_dir);
|
|---|
| 605 |
|
|---|
| 606 | # Make sure the collection isn't locked by someone else
|
|---|
| 607 | $self->lock_collection($username, $collect);
|
|---|
| 608 |
|
|---|
| 609 | if($where =~ m/import/) {
|
|---|
| 610 | $self->_set_import_metadata_array(@_);
|
|---|
| 611 | }
|
|---|
| 612 | if($where =~ m/archives/) {
|
|---|
| 613 | $self->_set_archives_metadata_array(@_);
|
|---|
| 614 | }
|
|---|
| 615 | if($where =~ m/index/) {
|
|---|
| 616 | $self->_set_index_metadata_array(@_);
|
|---|
| 617 | }
|
|---|
| 618 | if($where =~ m/live/) {
|
|---|
| 619 | $self->_set_live_metadata_array(@_);
|
|---|
| 620 | }
|
|---|
| 621 |
|
|---|
| 622 | # Release the lock once it is done
|
|---|
| 623 | $self->unlock_collection($username, $collect);
|
|---|
| 624 | }
|
|---|
| 625 |
|
|---|
| 626 | sub _set_index_metadata_array
|
|---|
| 627 | {
|
|---|
| 628 | my $self = shift @_;
|
|---|
| 629 |
|
|---|
| 630 | my $collect = $self->{'collect'};
|
|---|
| 631 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
|---|
| 632 |
|
|---|
| 633 | # Obtain the collect dir
|
|---|
| 634 | my $site = $self->{'site'};
|
|---|
| 635 | my $collect_dir = $gsdl_cgi->get_collection_dir($site);
|
|---|
| 636 |
|
|---|
| 637 | # look up additional args
|
|---|
| 638 |
|
|---|
| 639 | my $infodbtype = $self->{'infodbtype'};
|
|---|
| 640 |
|
|---|
| 641 | my $json_str = $self->{'json'};
|
|---|
| 642 | my $doc_array = decode_json $json_str;
|
|---|
| 643 |
|
|---|
| 644 |
|
|---|
| 645 | my $global_status = 0;
|
|---|
| 646 | my $global_mess = "";
|
|---|
| 647 |
|
|---|
| 648 | my @all_docids = ();
|
|---|
| 649 |
|
|---|
| 650 | foreach my $doc_array_rec ( @$doc_array ) {
|
|---|
| 651 |
|
|---|
| 652 | my $status = -1;
|
|---|
| 653 | my $docid = $doc_array_rec->{'docid'};
|
|---|
| 654 |
|
|---|
| 655 | push(@all_docids,$docid);
|
|---|
| 656 |
|
|---|
| 657 | my $metaname = $doc_array_rec->{'metaname'};
|
|---|
| 658 | if(defined $metaname) {
|
|---|
| 659 | my $metapos = $doc_array_rec->{'metapos'}; # can legitimately be undef
|
|---|
| 660 | my $metavalue = $doc_array_rec->{'metavalue'};
|
|---|
| 661 | my $metamode = $doc_array_rec->{'metamode'} || $self->{'metamode'};
|
|---|
| 662 |
|
|---|
| 663 | $status = $self->set_index_metadata_entry($collect_dir,$collect,$infodbtype,$docid,$metaname,$metapos,$metavalue,$metamode);
|
|---|
| 664 | } elsif (defined $doc_array_rec->{'metatable'}) { # if no metaname, we expect a metatable
|
|---|
| 665 | my $metatable = $doc_array_rec->{'metatable'}; # a subarray, or need to generate an error saying JSON structure is wrong
|
|---|
| 666 |
|
|---|
| 667 | foreach my $metatable_rec ( @$metatable ) { # the subarray metatable is an array of hashmaps
|
|---|
| 668 | $metaname = $metatable_rec->{'metaname'};
|
|---|
| 669 | my $metamode = $metatable_rec->{'metamode'} || $doc_array_rec->{'metamode'} || $self->{'metamode'};
|
|---|
| 670 | my $metapos = undef;
|
|---|
| 671 | my $metavals = $metatable_rec->{'metavals'}; # a sub-subarray
|
|---|
| 672 |
|
|---|
| 673 | foreach my $metavalue ( @$metavals ) { # metavals is an array
|
|---|
| 674 | $status = $self->set_index_metadata_entry($collect_dir,$collect,$infodbtype,$docid,$metaname,$metapos,$metavalue,$metamode); # how do we use metamode in set_meta_entry?
|
|---|
| 675 | if($metamode eq "override") { # now, having overridden the metavalue for the first,
|
|---|
| 676 | # need to accumulate subsequent metavals for this metaname, else the just-assigned
|
|---|
| 677 | # metavalue for this metaname will be lost
|
|---|
| 678 | $metamode = "accumulate";
|
|---|
| 679 | }
|
|---|
| 680 | }
|
|---|
| 681 | }
|
|---|
| 682 | }
|
|---|
| 683 |
|
|---|
| 684 | if ($status != 0) {
|
|---|
| 685 | # Catch error if set infodb entry failed
|
|---|
| 686 | $global_status = $status;
|
|---|
| 687 | $global_mess .= "Failed to set metadata key: $docid\n";
|
|---|
| 688 | $global_mess .= "Exit status: $status\n";
|
|---|
| 689 | $global_mess .= "System Error Message: $!\n";
|
|---|
| 690 | $global_mess .= "-" x 20;
|
|---|
| 691 | }
|
|---|
| 692 | }
|
|---|
| 693 |
|
|---|
| 694 | if ($global_status != 0) {
|
|---|
| 695 | $global_mess .= "PATH: $ENV{'PATH'}\n";
|
|---|
| 696 | $gsdl_cgi->generate_error($global_mess);
|
|---|
| 697 | }
|
|---|
| 698 | else {
|
|---|
| 699 | my $mess = "set-metadata-array successful: Keys[ ".join(", ",@all_docids)."]\n";
|
|---|
| 700 | $gsdl_cgi->generate_ok_message($mess);
|
|---|
| 701 | }
|
|---|
| 702 | }
|
|---|
| 703 |
|
|---|
| 704 | sub set_index_metadata_array
|
|---|
| 705 | {
|
|---|
| 706 | my $self = shift @_;
|
|---|
| 707 |
|
|---|
| 708 | my $username = $self->{'username'};
|
|---|
| 709 | my $collect = $self->{'collect'};
|
|---|
| 710 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
|---|
| 711 |
|
|---|
| 712 | if ($baseaction::authentication_enabled) {
|
|---|
| 713 | # Ensure the user is allowed to edit this collection
|
|---|
| 714 | $self->authenticate_user($username, $collect); #&authenticate_user($gsdl_cgi, $username, $collect);
|
|---|
| 715 | }
|
|---|
| 716 |
|
|---|
| 717 | # Obtain the collect dir
|
|---|
| 718 | my $site = $self->{'site'};
|
|---|
| 719 | my $collect_dir = $gsdl_cgi->get_collection_dir($site);
|
|---|
| 720 |
|
|---|
| 721 | $gsdl_cgi->checked_chdir($collect_dir);
|
|---|
| 722 |
|
|---|
| 723 |
|
|---|
| 724 | # Make sure the collection isn't locked by someone else
|
|---|
| 725 | $self->lock_collection($username, $collect);
|
|---|
| 726 |
|
|---|
| 727 | $self->_set_index_metadata_array(@_);
|
|---|
| 728 |
|
|---|
| 729 | # Release the lock once it is done
|
|---|
| 730 | $self->unlock_collection($username, $collect);
|
|---|
| 731 | }
|
|---|
| 732 |
|
|---|
| 733 | # experimental, newly added in and untested
|
|---|
| 734 | sub _set_live_metadata_array
|
|---|
| 735 | {
|
|---|
| 736 | my $self = shift @_;
|
|---|
| 737 |
|
|---|
| 738 | my $collect = $self->{'collect'};
|
|---|
| 739 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
|---|
| 740 |
|
|---|
| 741 | # Obtain the collect dir
|
|---|
| 742 | my $site = $self->{'site'};
|
|---|
| 743 | my $collect_dir = $gsdl_cgi->get_collection_dir($site);
|
|---|
| 744 |
|
|---|
| 745 | # look up additional args
|
|---|
| 746 | my $infodbtype = $self->{'infodbtype'};
|
|---|
| 747 | # To people who know $collect_tail please add some comments
|
|---|
| 748 | # Obtain path to the database
|
|---|
| 749 | my $collect_tail = $collect;
|
|---|
| 750 | $collect_tail =~ s/^.*[\/|\\]//;
|
|---|
| 751 | my $index_text_directory = &util::filename_cat($collect_dir,$collect,"index","text");
|
|---|
| 752 | my $infodb_file_path = &dbutil::get_infodb_file_path($infodbtype, "live-$collect_tail", $index_text_directory);
|
|---|
| 753 |
|
|---|
| 754 |
|
|---|
| 755 | my $json_str = $self->{'json'};
|
|---|
| 756 | my $doc_array = decode_json $json_str;
|
|---|
| 757 |
|
|---|
| 758 |
|
|---|
| 759 | my $global_status = 0;
|
|---|
| 760 | my $global_mess = "";
|
|---|
| 761 |
|
|---|
| 762 | my @all_docids = ();
|
|---|
| 763 |
|
|---|
| 764 |
|
|---|
| 765 | foreach my $doc_array_rec ( @$doc_array ) {
|
|---|
| 766 |
|
|---|
| 767 | my $status = -1;
|
|---|
| 768 | my $docid = $doc_array_rec->{'docid'};
|
|---|
| 769 |
|
|---|
| 770 | push(@all_docids,$docid);
|
|---|
| 771 |
|
|---|
| 772 | my $metaname = $doc_array_rec->{'metaname'};
|
|---|
| 773 | if(defined $metaname) {
|
|---|
| 774 | my $dbkey = "$docid.$metaname";
|
|---|
| 775 | my $metavalue = $doc_array_rec->{'metavalue'};
|
|---|
| 776 |
|
|---|
| 777 | # Set the new value
|
|---|
| 778 | my $cmd = "gdbmset \"$infodb_file_path\" \"$dbkey\" \"$metavalue\"";
|
|---|
| 779 | $status = system($cmd);
|
|---|
| 780 |
|
|---|
| 781 | } elsif (defined $doc_array_rec->{'metatable'}) { # if no metaname, we expect a metatable
|
|---|
| 782 | my $metatable = $doc_array_rec->{'metatable'}; # a subarray, or need to generate an error saying JSON structure is wrong
|
|---|
| 783 | foreach my $metatable_rec ( @$metatable ) {
|
|---|
| 784 | $metaname = $metatable_rec->{'metaname'};
|
|---|
| 785 | my $dbkey = "$docid.$metaname";
|
|---|
| 786 |
|
|---|
| 787 | my $metavals = $metatable_rec->{'metavals'}; # a sub-subarray
|
|---|
| 788 | foreach my $metavalue ( @$metavals ) {
|
|---|
| 789 | my $cmd = "gdbmset \"$infodb_file_path\" \"$dbkey\" \"$metavalue\"";
|
|---|
| 790 | $status = system($cmd);
|
|---|
| 791 | }
|
|---|
| 792 | }
|
|---|
| 793 |
|
|---|
| 794 | }
|
|---|
| 795 |
|
|---|
| 796 | if ($status != 0) {
|
|---|
| 797 | # Catch error if gdbmget failed
|
|---|
| 798 | $global_status = $status;
|
|---|
| 799 | $global_mess .= "Failed to set metadata key: $docid\n"; # $dbkey
|
|---|
| 800 | $global_mess .= "Exit status: $status\n";
|
|---|
| 801 | $global_mess .= "System Error Message: $!\n";
|
|---|
| 802 | $global_mess .= "-" x 20;
|
|---|
| 803 | }
|
|---|
| 804 | }
|
|---|
| 805 |
|
|---|
| 806 | if ($global_status != 0) {
|
|---|
| 807 | $global_mess .= "PATH: $ENV{'PATH'}\n";
|
|---|
| 808 | $gsdl_cgi->generate_error($global_mess);
|
|---|
| 809 | }
|
|---|
| 810 | else {
|
|---|
| 811 | my $mess = "set-live-metadata-array successful: Keys[ ".join(", ",@all_docids)."]\n";
|
|---|
| 812 | $gsdl_cgi->generate_ok_message($mess);
|
|---|
| 813 | }
|
|---|
| 814 | }
|
|---|
| 815 |
|
|---|
| 816 | sub set_live_metadata_array
|
|---|
| 817 | {
|
|---|
| 818 | my $self = shift @_;
|
|---|
| 819 |
|
|---|
| 820 | my $username = $self->{'username'};
|
|---|
| 821 | my $collect = $self->{'collect'};
|
|---|
| 822 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
|---|
| 823 |
|
|---|
| 824 | if ($baseaction::authentication_enabled) {
|
|---|
| 825 | # Ensure the user is allowed to edit this collection
|
|---|
| 826 | $self->authenticate_user($username, $collect); #&authenticate_user($gsdl_cgi, $username, $collect);
|
|---|
| 827 | }
|
|---|
| 828 |
|
|---|
| 829 | # Obtain the collect dir
|
|---|
| 830 | my $site = $self->{'site'};
|
|---|
| 831 | my $collect_dir = $gsdl_cgi->get_collection_dir($site);
|
|---|
| 832 |
|
|---|
| 833 | $gsdl_cgi->checked_chdir($collect_dir);
|
|---|
| 834 |
|
|---|
| 835 | # Make sure the collection isn't locked by someone else
|
|---|
| 836 | $self->lock_collection($username, $collect);
|
|---|
| 837 |
|
|---|
| 838 | $self->_set_live_metadata_array(@_);
|
|---|
| 839 |
|
|---|
| 840 | # Release the lock once it is done
|
|---|
| 841 | $self->unlock_collection($username, $collect);
|
|---|
| 842 | }
|
|---|
| 843 |
|
|---|
| 844 |
|
|---|
| 845 | sub dxml_metadata
|
|---|
| 846 | {
|
|---|
| 847 | my ($tagname, $attrHash, $contextArray, $parentDataArray, $parser) = @_;
|
|---|
| 848 | my $metaname = $parser->{'parameters'}->{'metaname'};
|
|---|
| 849 | my $metamode = $parser->{'parameters'}->{'metamode'};
|
|---|
| 850 |
|
|---|
| 851 | ###!!! print STDERR "**** Processing closing </Metadata> tag\n";
|
|---|
| 852 |
|
|---|
| 853 | my $opt_doc_secnum = $parser->{'parameters'}->{'secnum'};
|
|---|
| 854 |
|
|---|
| 855 | # Find the right metadata tag and checks if we are going to
|
|---|
| 856 | # override it
|
|---|
| 857 | #
|
|---|
| 858 | # Note: This over writes the first metadata block it
|
|---|
| 859 | # encountered. If there are multiple Sections in the doc.xml, it
|
|---|
| 860 | # might not behave as you would expect
|
|---|
| 861 |
|
|---|
| 862 | my $curr_secnum = $parser->{'parameters'}->{'curr_section_num'};
|
|---|
| 863 | ## print STDERR "**** checking $opt_doc_secnum <=> $curr_secnum\n" if (defined $opt_doc_secnum);
|
|---|
| 864 | ## print STDERR "**** metamode = $metamode\n";
|
|---|
| 865 |
|
|---|
| 866 | if ((!defined $opt_doc_secnum) || ($opt_doc_secnum eq $curr_secnum))
|
|---|
| 867 | {
|
|---|
| 868 | my $name_attr = $attrHash->{'name'};
|
|---|
| 869 | # print STDOUT "*** testing: $name_attr eq $metaname ? and $metamode eq override ?\n";
|
|---|
| 870 | if (($name_attr eq $metaname) && ($metamode eq "override"))
|
|---|
| 871 | {
|
|---|
| 872 | if (!defined $parser->{'parameters'}->{'poscount'})
|
|---|
| 873 | {
|
|---|
| 874 | $parser->{'parameters'}->{'poscount'} = 0;
|
|---|
| 875 | }
|
|---|
| 876 | else
|
|---|
| 877 | {
|
|---|
| 878 | $parser->{'parameters'}->{'poscount'}++;
|
|---|
| 879 | }
|
|---|
| 880 |
|
|---|
| 881 | if (defined $parser->{'parameters'}->{'metapos'}) { # if pos defined, ignore any prevmeta if also defined, and just wait for pos match
|
|---|
| 882 |
|
|---|
| 883 | if ($parser->{'parameters'}->{'poscount'} == $parser->{'parameters'}->{'metapos'}) {
|
|---|
| 884 | ###print STDERR "#### got pos match!!\n";
|
|---|
| 885 | # Get the value and override the current value
|
|---|
| 886 | my $metavalue = $parser->{'parameters'}->{'metavalue'};
|
|---|
| 887 | $attrHash->{'_content'} = $metavalue;
|
|---|
| 888 |
|
|---|
| 889 | # until we have clearly determined the role of metamode in doc.xml files (we know what it means in import/metadata.xml files)
|
|---|
| 890 | # we're not setting metamode to override here.
|
|---|
| 891 |
|
|---|
| 892 | # Don't want it to wipe out any other pieces of metadata
|
|---|
| 893 | $parser->{'parameters'}->{'metamode'} = "done";
|
|---|
| 894 | }
|
|---|
| 895 | }
|
|---|
| 896 | elsif (defined $parser->{'parameters'}->{'prevmetavalue'}) { # if no pos defined but prevmeta defined, then wait for prevmeta match
|
|---|
| 897 | if ($parser->{'parameters'}->{'prevmetavalue'} eq $attrHash->{'_content'}) {
|
|---|
| 898 | ###print STDERR "### prev meta value matches\n";
|
|---|
| 899 | my $metavalue = $parser->{'parameters'}->{'metavalue'};
|
|---|
| 900 | $attrHash->{'_content'} = $metavalue;
|
|---|
| 901 |
|
|---|
| 902 | # until we have clearly determined the role of metamode in doc.xml files (we know what it means in import/metadata.xml files)
|
|---|
| 903 | # we're not setting metamode to override here.
|
|---|
| 904 |
|
|---|
| 905 | $parser->{'parameters'}->{'metamode'} = "done";
|
|---|
| 906 |
|
|---|
| 907 | }
|
|---|
| 908 | }
|
|---|
| 909 | # Note that before commit rev 32076, if neither metapos nor prevmeta defined, then code tried to set pos 0 for that metaname to the new value
|
|---|
| 910 | # However that is not how the code works now (and couldn't get it to work that way yet after rev 32076).
|
|---|
| 911 | # What it does now, and what Dr Bainbridge approved of is that:
|
|---|
| 912 | # If no pos or prevmeta defined, then on override, the metavalue set for the specificed metaname becomes the ONLY instance of that metaname
|
|---|
| 913 | # So any earlier instances of that metaname is removed and replaced by the single new instance.
|
|---|
| 914 | }
|
|---|
| 915 | }
|
|---|
| 916 |
|
|---|
| 917 | # RAW is [$tagname => $attrHash] not $tagname => $attrHash!!
|
|---|
| 918 | return [$tagname => $attrHash];
|
|---|
| 919 | }
|
|---|
| 920 |
|
|---|
| 921 | # This method exists purely for catching invalid section numbers that the client
|
|---|
| 922 | # requested to edit. Once the parser has reached the end (the final </Archive> tag),
|
|---|
| 923 | # we've seen all the Sections in the doc.xml, and none of their section nums matched
|
|---|
| 924 | # if the metamode has not been set to 'done' by then.
|
|---|
| 925 | sub dxml_archive
|
|---|
| 926 | {
|
|---|
| 927 | my ($tagname, $attrHash, $contextArray, $parentDataArray, $parser) = @_;
|
|---|
| 928 | my $metamode = $parser->{'parameters'}->{'metamode'};
|
|---|
| 929 |
|
|---|
| 930 | my $curr_secnum = $parser->{'parameters'}->{'curr_section_num'};
|
|---|
| 931 | my $opt_doc_secnum = $parser->{'parameters'}->{'secnum'};
|
|---|
| 932 |
|
|---|
| 933 | # print STDERR "@@@ $tagname Processing a closing </Archive> tag [$curr_secnum|$opt_doc_secnum]\n";
|
|---|
| 934 |
|
|---|
| 935 | if ($metamode ne "done" && $curr_secnum ne $opt_doc_secnum) {
|
|---|
| 936 | #print STDERR "@@@ $tagname Finished processing FINAL Section.\n";
|
|---|
| 937 |
|
|---|
| 938 | my $metaname = $parser->{'parameters'}->{'metaname'};
|
|---|
| 939 | my $metavalue = $parser->{'parameters'}->{'metavalue'};
|
|---|
| 940 |
|
|---|
| 941 | print STDERR "@@@ Requested section number $opt_doc_secnum not found.\n";
|
|---|
| 942 | print STDERR "\t(last seen section number in document was $curr_secnum)\n";
|
|---|
| 943 | print STDERR "\tDiscarded metadata value '$metavalue' for meta '$metaname'\n";
|
|---|
| 944 | print STDERR "\tin section $opt_doc_secnum.\n";
|
|---|
| 945 | $parser->{'custom_err_msg'} = "Requested section number $opt_doc_secnum not found.";
|
|---|
| 946 | }
|
|---|
| 947 |
|
|---|
| 948 | # RAW is [$tagname => $attrHash] not $tagname => $attrHash!!
|
|---|
| 949 | return [$tagname => $attrHash];
|
|---|
| 950 | }
|
|---|
| 951 |
|
|---|
| 952 | sub dxml_description
|
|---|
| 953 | {
|
|---|
| 954 | my ($tagname, $attrHash, $contextArray, $parentDataArray, $parser) = @_;
|
|---|
| 955 | my $metamode = $parser->{'parameters'}->{'metamode'};
|
|---|
| 956 |
|
|---|
| 957 | my $curr_secnum = $parser->{'parameters'}->{'curr_section_num'};
|
|---|
| 958 | my $opt_doc_secnum = $parser->{'parameters'}->{'secnum'} || "";
|
|---|
| 959 |
|
|---|
| 960 | ###!!! print STDERR "**** Processing a closing </Description> tag \n";
|
|---|
| 961 | # print STDERR "@@@ $tagname Processing a closing </Description> tag [$curr_secnum|$opt_doc_secnum]\n";
|
|---|
| 962 |
|
|---|
| 963 | # Accumulate the metadata
|
|---|
| 964 |
|
|---|
| 965 | # We'll be accumulating metadata at this point if we haven't found and therefore
|
|---|
| 966 | # haven't processed the metadata yet.
|
|---|
| 967 | # For subsections, this means that if we're at a matching subsection, but haven't
|
|---|
| 968 | # found the correct metaname to override in that subsection, we accumulate it as new
|
|---|
| 969 | # meta in the subsection by adding it to the current description.
|
|---|
| 970 | # If there's no subsection info for the metadata, it will accumulate at the top level
|
|---|
| 971 | # section description if we hadn't found a matching metaname to override at this point.
|
|---|
| 972 |
|
|---|
| 973 | # Both curr_secnum and opt_doc_secnum can be "". In the former case, it means we're now
|
|---|
| 974 | # at the toplevel section. In the latter case, it means we want to process meta in the
|
|---|
| 975 | # toplevel section. So the eq check between the values below will work in all cases.
|
|---|
| 976 |
|
|---|
| 977 | # The only time this won't work is if an opt_doc_secnum beyond the section numbers of
|
|---|
| 978 | # this document has been provided. In that case, the metadata for that opt_doc_secnum
|
|---|
| 979 | # won't get attached/accumulated to any part of the doc, not even its top-level section.
|
|---|
| 980 |
|
|---|
| 981 | if ($curr_secnum eq $opt_doc_secnum
|
|---|
| 982 | && ($metamode eq "accumulate" || $metamode eq "override")) {
|
|---|
| 983 | # note, metamode would have been set to done if a metadata operation had already succeeded.
|
|---|
| 984 | # If we get here, we still have the metadata to append.
|
|---|
| 985 | if ($metamode eq "override") {
|
|---|
| 986 | print "Got to end of <Description> block. No metadata value to override. Switching 'metamode' to accumulate\n";
|
|---|
| 987 | }
|
|---|
| 988 |
|
|---|
| 989 | # If we get to here and metamode is override, this means there
|
|---|
| 990 | # was no existing value to overide => treat as an append operation
|
|---|
| 991 |
|
|---|
| 992 | # Tack a new metadata tag on to the end of the <Metadata>+ block
|
|---|
| 993 | my $metaname = $parser->{'parameters'}->{'metaname'};
|
|---|
| 994 | my $metavalue = $parser->{'parameters'}->{'metavalue'};
|
|---|
| 995 |
|
|---|
| 996 | my $metadata_attr = {
|
|---|
| 997 | '_content' => $metavalue,
|
|---|
| 998 | 'name' => $metaname,
|
|---|
| 999 | 'mode' => "accumulate"
|
|---|
| 1000 | };
|
|---|
| 1001 |
|
|---|
| 1002 | my $append_metadata = [ "Metadata" => $metadata_attr ];
|
|---|
| 1003 | my $description_content = $attrHash->{'_content'};
|
|---|
| 1004 | print "Appending metadata to doc.xml\n";
|
|---|
| 1005 |
|
|---|
| 1006 | if (ref($description_content)) {
|
|---|
| 1007 | # got some existing interesting nested content
|
|---|
| 1008 | push(@$description_content, " ", $append_metadata, "\n ");
|
|---|
| 1009 | }
|
|---|
| 1010 | else {
|
|---|
| 1011 | #description_content is most likely a string such as "\n"
|
|---|
| 1012 | $attrHash->{'_content'} = [$description_content, " ", $append_metadata ,"\n" ];
|
|---|
| 1013 | }
|
|---|
| 1014 |
|
|---|
| 1015 | $parser->{'parameters'}->{'metamode'} = "done";
|
|---|
| 1016 | }
|
|---|
| 1017 | else {
|
|---|
| 1018 | # metamode most likely "done" signifying that it has already found a position to add the metadata to.
|
|---|
| 1019 | ## print STDERR "**** NOT ACCUMULATE?!? \n";
|
|---|
| 1020 | }
|
|---|
| 1021 | # RAW is [$tagname => $attrHash] not $tagname => $attrHash!!
|
|---|
| 1022 | return [$tagname => $attrHash];
|
|---|
| 1023 | }
|
|---|
| 1024 |
|
|---|
| 1025 | sub edit_xml_file
|
|---|
| 1026 | {
|
|---|
| 1027 | my $self = shift @_;
|
|---|
| 1028 | my ($gsdl_cgi, $filename, $start_rules, $rules, $options) = @_;
|
|---|
| 1029 |
|
|---|
| 1030 | # use XML::Rules to add it in (read in and out again)
|
|---|
| 1031 | my $parser = XML::Rules->new(start_rules => $start_rules,
|
|---|
| 1032 | rules => $rules,
|
|---|
| 1033 | style => 'filter',
|
|---|
| 1034 | output_encoding => 'utf8' );
|
|---|
| 1035 |
|
|---|
| 1036 | my $xml_in = "";
|
|---|
| 1037 | if (!open(MIN,"<$filename")) {
|
|---|
| 1038 | $gsdl_cgi->generate_error("Unable to read in $filename: $!");
|
|---|
| 1039 | }
|
|---|
| 1040 | else {
|
|---|
| 1041 | # Read all the text in
|
|---|
| 1042 | my $line;
|
|---|
| 1043 | while (defined ($line=<MIN>)) {
|
|---|
| 1044 | $xml_in .= $line if($line !~ m/^\s*$/); # preserve all but empty lines
|
|---|
| 1045 | }
|
|---|
| 1046 | close(MIN);
|
|---|
| 1047 |
|
|---|
| 1048 | my $MOUT;
|
|---|
| 1049 | if (!open($MOUT,">$filename")) {
|
|---|
| 1050 | $gsdl_cgi->generate_error("Unable to write out to $filename: $!");
|
|---|
| 1051 | }
|
|---|
| 1052 | else {
|
|---|
| 1053 | # Matched lines will get handled by the call backs
|
|---|
| 1054 | ## my $xml_out = "";
|
|---|
| 1055 |
|
|---|
| 1056 | binmode($MOUT,":utf8");
|
|---|
| 1057 | $parser->filter($xml_in,$MOUT, $options);
|
|---|
| 1058 |
|
|---|
| 1059 | # binmode(MOUT,":utf8");
|
|---|
| 1060 | # print MOUT $xml_out;
|
|---|
| 1061 | close($MOUT);
|
|---|
| 1062 | }
|
|---|
| 1063 | }
|
|---|
| 1064 |
|
|---|
| 1065 | # copy across any custom error information that was stored during parsing
|
|---|
| 1066 | $self->{'error_msg'} = $parser->{'custom_err_msg'} if(defined $parser->{'custom_err_msg'});
|
|---|
| 1067 | }
|
|---|
| 1068 |
|
|---|
| 1069 | sub edit_doc_xml
|
|---|
| 1070 | {
|
|---|
| 1071 | my $self = shift @_;
|
|---|
| 1072 | my ($gsdl_cgi, $doc_xml_filename, $metaname, $metavalue, $metapos, $metamode, $opt_secnum, $prevmetavalue) = @_;
|
|---|
| 1073 |
|
|---|
| 1074 | my $info_mess = <<RAWEND;
|
|---|
| 1075 | ****************************
|
|---|
| 1076 | edit_doc_xml()
|
|---|
| 1077 | ****************************
|
|---|
| 1078 | RAWEND
|
|---|
| 1079 |
|
|---|
| 1080 | $info_mess .= " doc_xml_filename = $doc_xml_filename\n" if defined($doc_xml_filename);
|
|---|
| 1081 | $info_mess .= " metaname = $metaname\n" if defined($metaname);
|
|---|
| 1082 | $info_mess .= " metapos = $metapos\n" if defined($metapos);
|
|---|
| 1083 | $info_mess .= " metavalue = $metavalue\n" if defined($metavalue);
|
|---|
| 1084 | $info_mess .= " metamode = $metamode\n" if defined($metamode);
|
|---|
| 1085 | $info_mess .= " opt_secnum = $opt_secnum\n" if defined($opt_secnum);
|
|---|
| 1086 | $info_mess .= " prevmetaval = $prevmetavalue\n" if defined($prevmetavalue);
|
|---|
| 1087 |
|
|---|
| 1088 | $info_mess .= "****************************\n";
|
|---|
| 1089 |
|
|---|
| 1090 | $gsdl_cgi->generate_message($info_mess);
|
|---|
| 1091 |
|
|---|
| 1092 | # To monitor which section/subsection number we are in
|
|---|
| 1093 | my @start_rules =
|
|---|
| 1094 | ( 'Section' => \&dxml_start_section );
|
|---|
| 1095 |
|
|---|
| 1096 | # use XML::Rules to add it in (read in and out again)
|
|---|
| 1097 | # Set the call back functions
|
|---|
| 1098 | my @rules =
|
|---|
| 1099 | ( _default => 'raw',
|
|---|
| 1100 | 'Metadata' => \&dxml_metadata,
|
|---|
| 1101 | 'Description' => \&dxml_description,
|
|---|
| 1102 | 'Archive' => \&dxml_archive); # just for catching errors at end
|
|---|
| 1103 |
|
|---|
| 1104 | # Sets the parameters
|
|---|
| 1105 | my $options = { 'metaname' => $metaname,
|
|---|
| 1106 | 'metapos' => $metapos,
|
|---|
| 1107 | 'metavalue' => $metavalue,
|
|---|
| 1108 | 'metamode' => $metamode,
|
|---|
| 1109 | 'prevmetavalue' => $prevmetavalue };
|
|---|
| 1110 |
|
|---|
| 1111 | if (defined $opt_secnum) {
|
|---|
| 1112 | $options->{'secnum'} = $opt_secnum;
|
|---|
| 1113 | }
|
|---|
| 1114 |
|
|---|
| 1115 | $self->edit_xml_file($gsdl_cgi,$doc_xml_filename,\@start_rules,\@rules,$options);
|
|---|
| 1116 | }
|
|---|
| 1117 |
|
|---|
| 1118 | sub set_archives_metadata_entry
|
|---|
| 1119 | {
|
|---|
| 1120 | my $self = shift @_;
|
|---|
| 1121 | my ($gsdl_cgi, $archive_dir, $collect_dir, $collect, $infodbtype, $docid, $dv, $metaname, $metapos, $metavalue, $metamode, $prevmetavalue) = @_;
|
|---|
| 1122 |
|
|---|
| 1123 | my $info_mess = <<RAWEND;
|
|---|
| 1124 | ****************************
|
|---|
| 1125 | set_archives_metadata_entry()
|
|---|
| 1126 | ****************************
|
|---|
| 1127 | RAWEND
|
|---|
| 1128 |
|
|---|
| 1129 | $info_mess .= " archive_dir = $archive_dir\n" if defined($archive_dir);
|
|---|
| 1130 | $info_mess .= " collect_dir = $collect_dir\n" if defined($collect_dir);
|
|---|
| 1131 | $info_mess .= " collect = $collect\n" if defined($collect);
|
|---|
| 1132 | $info_mess .= " infodbtype = $infodbtype\n" if defined($infodbtype);
|
|---|
| 1133 | $info_mess .= " docid = $docid\n" if defined($docid);
|
|---|
| 1134 | $info_mess .= " metaname = $metaname\n" if defined($metaname);
|
|---|
| 1135 | $info_mess .= " metapos = $metapos\n" if defined($metapos);
|
|---|
| 1136 | $info_mess .= " metavalue = $metavalue\n" if defined($metavalue);
|
|---|
| 1137 | $info_mess .= " metamode = $metamode\n" if defined($metamode);
|
|---|
| 1138 | $info_mess .= " prevmetaval = $prevmetavalue\n" if defined($prevmetavalue);
|
|---|
| 1139 |
|
|---|
| 1140 | $info_mess .= "****************************\n";
|
|---|
| 1141 |
|
|---|
| 1142 | $gsdl_cgi->generate_message($info_mess);
|
|---|
| 1143 |
|
|---|
| 1144 | # Obtain the doc.xml path for the specified docID
|
|---|
| 1145 | my ($docid_root,$docid_secnum) = ($docid =~ m/^(.*?)(\..*)?$/);
|
|---|
| 1146 |
|
|---|
| 1147 | my $arcinfo_doc_filename = &dbutil::get_infodb_file_path($infodbtype, "archiveinf-doc", $archive_dir);
|
|---|
| 1148 | my $doc_rec = &dbutil::read_infodb_entry($infodbtype, $arcinfo_doc_filename, $docid_root);
|
|---|
| 1149 | my $doc_xml_file = $doc_rec->{'doc-file'}->[0];
|
|---|
| 1150 |
|
|---|
| 1151 | # check if request if for file-level doc-version history 'nminus-<n>' version
|
|---|
| 1152 | if (defined $dv && ($dv ne "")) {
|
|---|
| 1153 | # Need to insert '_fldv_history/nminus-<n>' into doc_filename
|
|---|
| 1154 |
|
|---|
| 1155 | my ($doc_xml_tailname, $doc_xml_dirname) = File::Basename::fileparse($doc_xml_file);
|
|---|
| 1156 | $doc_xml_file = &util::filename_cat($doc_xml_dirname,$FLDV_HISTORY_DIR,$dv,$doc_xml_tailname);
|
|---|
| 1157 | }
|
|---|
| 1158 |
|
|---|
| 1159 | # The $doc_xml_file is relative to the archives, and now let's get the full path
|
|---|
| 1160 | my $archives_dir = &util::filename_cat($collect_dir,$collect,"archives");
|
|---|
| 1161 |
|
|---|
| 1162 | my $doc_xml_filename = &util::filename_cat($archives_dir,$doc_xml_file);
|
|---|
| 1163 |
|
|---|
| 1164 | # If we're overriding everything, then $metamode=override combined with $metapos=undefined and $prevmetavalue=undefined
|
|---|
| 1165 | # in which case, we need to remove all metavalues for the metaname at the given (sub)section
|
|---|
| 1166 | # Thereafter, we will finally be setting the overriding metavalue for this metaname
|
|---|
| 1167 | if (!defined $prevmetavalue && !defined $metapos && $metamode eq "override") {
|
|---|
| 1168 | print "override mode with no pos or prev value set. Removing all existing values for $metaname, and setting mode to accumulate for the new value.\n";
|
|---|
| 1169 | # remove all values of $metaname metadata
|
|---|
| 1170 | $self->remove_from_doc_xml($gsdl_cgi, &util::filename_cat($archive_dir, $doc_xml_file), $metaname, undef, undef, $docid_secnum, $metamode);
|
|---|
| 1171 | $metamode = "accumulate";
|
|---|
| 1172 | }
|
|---|
| 1173 | # Edit the doc.xml file with the specified metadata name, value and position.
|
|---|
| 1174 | # TODO: there is a potential problem here as this edit_doc_xml function
|
|---|
| 1175 | # is assuming the simple doc.xml situation where there is only one Section and no SubSections.
|
|---|
| 1176 | # Running import.pl -groupsize will cause this to have multiple sections in one doc.xml
|
|---|
| 1177 |
|
|---|
| 1178 | # dxml_metadata method ignores metapos if metamode anything other than override
|
|---|
| 1179 | $self->edit_doc_xml($gsdl_cgi,$doc_xml_filename,
|
|---|
| 1180 | $metaname,$metavalue,$metapos,$metamode,$docid_secnum,$prevmetavalue);
|
|---|
| 1181 |
|
|---|
| 1182 | # return 0; # return 0 for now to indicate no error
|
|---|
| 1183 | return (defined $self->{'error_msg'}) ? 1 : 0;
|
|---|
| 1184 | }
|
|---|
| 1185 |
|
|---|
| 1186 |
|
|---|
| 1187 | sub set_archives_metadata
|
|---|
| 1188 | {
|
|---|
| 1189 | my $self = shift @_;
|
|---|
| 1190 |
|
|---|
| 1191 | my $username = $self->{'username'};
|
|---|
| 1192 | my $collect = $self->{'collect'};
|
|---|
| 1193 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
|---|
| 1194 |
|
|---|
| 1195 | if ($baseaction::authentication_enabled) {
|
|---|
| 1196 | # Ensure the user is allowed to edit this collection
|
|---|
| 1197 | $self->authenticate_user($username, $collect);
|
|---|
| 1198 | }
|
|---|
| 1199 |
|
|---|
| 1200 | # Make sure the collection isn't locked by someone else
|
|---|
| 1201 | $self->lock_collection($username, $collect);
|
|---|
| 1202 |
|
|---|
| 1203 | $self->_set_archives_metadata(@_);
|
|---|
| 1204 |
|
|---|
| 1205 | # Release the lock once it is done
|
|---|
| 1206 | $self->unlock_collection($username, $collect);
|
|---|
| 1207 | }
|
|---|
| 1208 |
|
|---|
| 1209 | sub _set_archives_metadata_array
|
|---|
| 1210 | {
|
|---|
| 1211 | my $self = shift @_;
|
|---|
| 1212 |
|
|---|
| 1213 | my $collect = $self->{'collect'};
|
|---|
| 1214 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
|---|
| 1215 |
|
|---|
| 1216 | # Obtain the collect dir
|
|---|
| 1217 | my $site = $self->{'site'};
|
|---|
| 1218 | my $collect_dir = $gsdl_cgi->get_collection_dir($site);
|
|---|
| 1219 |
|
|---|
| 1220 | # look up additional args
|
|---|
| 1221 |
|
|---|
| 1222 | my $infodbtype = $self->{'infodbtype'};
|
|---|
| 1223 |
|
|---|
| 1224 | my $archive_dir = &util::filename_cat($collect_dir,$collect,"archives");
|
|---|
| 1225 |
|
|---|
| 1226 | my $json_str = $self->{'json'};
|
|---|
| 1227 | my $doc_array = decode_json $json_str;
|
|---|
| 1228 |
|
|---|
| 1229 |
|
|---|
| 1230 | my $global_status = 0;
|
|---|
| 1231 | my $global_mess = "";
|
|---|
| 1232 |
|
|---|
| 1233 | my @all_docids = ();
|
|---|
| 1234 |
|
|---|
| 1235 | foreach my $doc_array_rec ( @$doc_array ) {
|
|---|
| 1236 | my $status = -1;
|
|---|
| 1237 | my $docid = $doc_array_rec->{'docid'};
|
|---|
| 1238 | my $dv = $doc_array_rec->{'dv'};
|
|---|
| 1239 |
|
|---|
| 1240 | push(@all_docids,$docid);
|
|---|
| 1241 |
|
|---|
| 1242 | my $metaname = $doc_array_rec->{'metaname'};
|
|---|
| 1243 | if(defined $metaname) {
|
|---|
| 1244 |
|
|---|
| 1245 | my $metapos = $doc_array_rec->{'metapos'}; # don't force undef to 0. Undef has meaning when metamode=override
|
|---|
| 1246 |
|
|---|
| 1247 | my $metamode = $doc_array_rec->{'metamode'} || $self->{'metamode'};
|
|---|
| 1248 | my $metavalue = $doc_array_rec->{'metavalue'};
|
|---|
| 1249 | my $prevmetavalue = $self->{'prevmetavalue'}; # to make this sub behave as _set_archives_metadata
|
|---|
| 1250 |
|
|---|
| 1251 |
|
|---|
| 1252 | if ((!defined $metamode) || ($metamode =~ m/^\s*$/)) {
|
|---|
| 1253 | # make "accumulate" the default (less destructive, as it won't actually
|
|---|
| 1254 | # delete any existing values)
|
|---|
| 1255 | $metamode = "accumulate";
|
|---|
| 1256 | }
|
|---|
| 1257 |
|
|---|
| 1258 | $status = $self->set_archives_metadata_entry($gsdl_cgi,$archive_dir, $collect_dir,$collect, $infodbtype,
|
|---|
| 1259 | $docid,$dv, $metaname,$metapos,$metavalue,$metamode,$prevmetavalue);
|
|---|
| 1260 | } elsif (defined $doc_array_rec->{'metatable'}) { # if no metaname, we expect a metatable
|
|---|
| 1261 | my $metatable = $doc_array_rec->{'metatable'}; # a subarray, or need to generate an error saying JSON structure is wrong
|
|---|
| 1262 |
|
|---|
| 1263 | foreach my $metatable_rec ( @$metatable ) {
|
|---|
| 1264 | $metaname = $metatable_rec->{'metaname'};
|
|---|
| 1265 | my $metamode = $metatable_rec->{'metamode'} || $doc_array_rec->{'metamode'} || $self->{'metamode'};
|
|---|
| 1266 | my $metapos = undef;
|
|---|
| 1267 | my $prevmetavalue = undef;
|
|---|
| 1268 | my $metavals = $metatable_rec->{'metavals'}; # a sub-subarray
|
|---|
| 1269 |
|
|---|
| 1270 | foreach my $metavalue ( @$metavals ) {
|
|---|
| 1271 | $status = $self->set_archives_metadata_entry($gsdl_cgi,$archive_dir, $collect_dir,$collect,$infodbtype,
|
|---|
| 1272 | $docid,$dv, $metaname,$metapos,$metavalue,$metamode,$prevmetavalue);
|
|---|
| 1273 |
|
|---|
| 1274 | if($metamode eq "override") { # now, having overridden the metavalue for the first,
|
|---|
| 1275 | # need to accumulate subsequent metavals for this metaname, else the just-assigned
|
|---|
| 1276 | # metavalue for this metaname will be lost
|
|---|
| 1277 | $metamode = "accumulate";
|
|---|
| 1278 | }
|
|---|
| 1279 | }
|
|---|
| 1280 | }
|
|---|
| 1281 | }
|
|---|
| 1282 |
|
|---|
| 1283 | if ($status != 0) {
|
|---|
| 1284 | # Catch error if set infodb entry failed
|
|---|
| 1285 | $global_status = $status;
|
|---|
| 1286 | $global_mess .= "Failed to set metadata key: $docid\n";
|
|---|
| 1287 | $global_mess .= "Exit status: $status\n";
|
|---|
| 1288 | $global_mess .= "System Error Message: $!\n";
|
|---|
| 1289 | $global_mess .= "-" x 20 . "\n";
|
|---|
| 1290 | }
|
|---|
| 1291 | }
|
|---|
| 1292 |
|
|---|
| 1293 | if ($global_status != 0) {
|
|---|
| 1294 | $global_mess .= "PATH: $ENV{'PATH'}\n";
|
|---|
| 1295 | $gsdl_cgi->generate_error($global_mess);
|
|---|
| 1296 | }
|
|---|
| 1297 | else {
|
|---|
| 1298 | my $mess = "set-archives-metadata-array successful: Keys[ ".join(", ",@all_docids)."]\n";
|
|---|
| 1299 | $gsdl_cgi->generate_ok_message($mess);
|
|---|
| 1300 | }
|
|---|
| 1301 | }
|
|---|
| 1302 |
|
|---|
| 1303 | sub set_archives_metadata_array
|
|---|
| 1304 | {
|
|---|
| 1305 | my $self = shift @_;
|
|---|
| 1306 |
|
|---|
| 1307 | my $username = $self->{'username'};
|
|---|
| 1308 | my $collect = $self->{'collect'};
|
|---|
| 1309 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
|---|
| 1310 |
|
|---|
| 1311 | if ($baseaction::authentication_enabled) {
|
|---|
| 1312 | # Ensure the user is allowed to edit this collection
|
|---|
| 1313 | $self->authenticate_user($username, $collect); #&authenticate_user($gsdl_cgi, $username, $collect);
|
|---|
| 1314 | }
|
|---|
| 1315 |
|
|---|
| 1316 | # Obtain the collect dir
|
|---|
| 1317 | my $site = $self->{'site'};
|
|---|
| 1318 | my $collect_dir = $gsdl_cgi->get_collection_dir($site);
|
|---|
| 1319 |
|
|---|
| 1320 | $gsdl_cgi->checked_chdir($collect_dir);
|
|---|
| 1321 |
|
|---|
| 1322 | # Make sure the collection isn't locked by someone else
|
|---|
| 1323 | $self->lock_collection($username, $collect);
|
|---|
| 1324 |
|
|---|
| 1325 | $self->_set_archives_metadata_array(@_);
|
|---|
| 1326 |
|
|---|
| 1327 | # Release the lock once it is done
|
|---|
| 1328 | $self->unlock_collection($username, $collect);
|
|---|
| 1329 | }
|
|---|
| 1330 |
|
|---|
| 1331 |
|
|---|
| 1332 |
|
|---|
| 1333 | # the version of set_archives_meta that doesn't do authentication
|
|---|
| 1334 | sub _set_archives_assocfile
|
|---|
| 1335 | {
|
|---|
| 1336 | my $self = shift @_;
|
|---|
| 1337 |
|
|---|
| 1338 | my $collect = $self->{'collect'};
|
|---|
| 1339 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
|---|
| 1340 | my $infodbtype = $self->{'infodbtype'};
|
|---|
| 1341 |
|
|---|
| 1342 | # Obtain the collect and archive dir
|
|---|
| 1343 | my $site = $self->{'site'};
|
|---|
| 1344 | my $collect_dir = $gsdl_cgi->get_collection_dir($site);
|
|---|
| 1345 | my $archive_dir = &util::filename_cat($collect_dir,$collect,"archives");
|
|---|
| 1346 |
|
|---|
| 1347 | # look up additional args
|
|---|
| 1348 | my $docid = $self->{'d'};
|
|---|
| 1349 |
|
|---|
| 1350 | my $assocname = $self->{'assocname'};
|
|---|
| 1351 |
|
|---|
| 1352 | my $arcinfo_doc_filename = &dbutil::get_infodb_file_path($infodbtype, "archiveinf-doc", $archive_dir);
|
|---|
| 1353 | my $doc_rec = &dbutil::read_infodb_entry($infodbtype, $arcinfo_doc_filename, $docid);
|
|---|
| 1354 |
|
|---|
| 1355 | my $doc_file = $doc_rec->{'doc-file'}->[0];
|
|---|
| 1356 |
|
|---|
| 1357 | # check if request if for file-level doc-version history 'nminus-<n>' version
|
|---|
| 1358 | my $dv = $self->{'dv'};
|
|---|
| 1359 |
|
|---|
| 1360 | my ($unused_tailname, $doc_dirname) = File::Basename::fileparse($doc_file);
|
|---|
| 1361 |
|
|---|
| 1362 | my $output_file = $assocname;
|
|---|
| 1363 |
|
|---|
| 1364 | if (defined $dv && ($dv ne "")) {
|
|---|
| 1365 | # Need to insert '_fldv_history/nminus-<n>' into output_file
|
|---|
| 1366 |
|
|---|
| 1367 | $output_file = &util::filename_cat($doc_dirname,$FLDV_HISTORY_DIR,$dv,$output_file);
|
|---|
| 1368 | }
|
|---|
| 1369 | else {
|
|---|
| 1370 | $output_file = &util::filename_cat($doc_dirname,$output_file);
|
|---|
| 1371 | }
|
|---|
| 1372 |
|
|---|
| 1373 | my $output_filename = &util::filename_cat($archive_dir, $output_file);
|
|---|
| 1374 |
|
|---|
| 1375 | my $filedata = $self->{'filedata'};
|
|---|
| 1376 | my $FIN = $self->{'fileupload'};
|
|---|
| 1377 |
|
|---|
| 1378 | if ((!defined $filedata) && (!defined $FIN)) {
|
|---|
| 1379 | $gsdl_cgi->generate_ok_message("set-archives-assocfile: either 'fileupload' or 'filedata' must be set");
|
|---|
| 1380 | }
|
|---|
| 1381 | else {
|
|---|
| 1382 |
|
|---|
| 1383 | my $had_error = 0;
|
|---|
| 1384 | my $had_error_mess = undef;
|
|---|
| 1385 |
|
|---|
| 1386 | if (defined $filedata) {
|
|---|
| 1387 | if (open(FOUT,">$output_filename")) {
|
|---|
| 1388 | print FOUT $filedata;
|
|---|
| 1389 | close(FOUT);
|
|---|
| 1390 | }
|
|---|
| 1391 | else {
|
|---|
| 1392 | $had_error = 1;
|
|---|
| 1393 | $had_error_mess = "Failed to save filedata to: $output_filename";
|
|---|
| 1394 | print STDERR "Error - $had_error_mess:\n$!\n";
|
|---|
| 1395 | }
|
|---|
| 1396 | }
|
|---|
| 1397 | else {
|
|---|
| 1398 | my $FIN = $self->{'fileupload'};
|
|---|
| 1399 | if (open(FOUT,">$output_filename")) {
|
|---|
| 1400 | binmode(FOUT, ":raw");
|
|---|
| 1401 |
|
|---|
| 1402 | while (1) {
|
|---|
| 1403 | my $buffer = "";
|
|---|
| 1404 | my $bytes_read = read($FIN, $buffer, 1024);
|
|---|
| 1405 |
|
|---|
| 1406 | if (defined $bytes_read) {
|
|---|
| 1407 | if ($bytes_read>0) {
|
|---|
| 1408 | print FOUT $buffer;
|
|---|
| 1409 | }
|
|---|
| 1410 | last if $bytes_read < 1024;
|
|---|
| 1411 | }
|
|---|
| 1412 | else {
|
|---|
| 1413 | $had_error = 1;
|
|---|
| 1414 | $had_error_mess = "set-archives-assocfile: Failed to open uploaded file for reading: $!\n";
|
|---|
| 1415 | print STDERR "Error - $had_error_mess";
|
|---|
| 1416 | last;
|
|---|
| 1417 | }
|
|---|
| 1418 | }
|
|---|
| 1419 |
|
|---|
| 1420 | close($FIN);
|
|---|
| 1421 | close(FOUT);
|
|---|
| 1422 | }
|
|---|
| 1423 | else {
|
|---|
| 1424 | $had_error = 1;
|
|---|
| 1425 | $had_error_mess = "Failed to save file uploaded file as: $output_filename";
|
|---|
| 1426 | print STDERR "Error - $had_error_mess:\n$!\n";
|
|---|
| 1427 | }
|
|---|
| 1428 | }
|
|---|
| 1429 |
|
|---|
| 1430 | if (!$had_error) {
|
|---|
| 1431 | my $mess = "set-archives-assocfile successful save uploaded content into 'archives' as: $output_file\n";
|
|---|
| 1432 | $gsdl_cgi->generate_ok_message($mess);
|
|---|
| 1433 | }
|
|---|
| 1434 | else {
|
|---|
| 1435 | $gsdl_cgi->generate_error($had_error_mess);
|
|---|
| 1436 | }
|
|---|
| 1437 | }
|
|---|
| 1438 | }
|
|---|
| 1439 |
|
|---|
| 1440 |
|
|---|
| 1441 | sub set_archives_assocfile
|
|---|
| 1442 | {
|
|---|
| 1443 | my $self = shift @_;
|
|---|
| 1444 |
|
|---|
| 1445 | my $username = $self->{'username'};
|
|---|
| 1446 | my $collect = $self->{'collect'};
|
|---|
| 1447 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
|---|
| 1448 |
|
|---|
| 1449 | if ($baseaction::authentication_enabled) {
|
|---|
| 1450 | # Ensure the user is allowed to edit this collection
|
|---|
| 1451 | $self->authenticate_user($username, $collect);
|
|---|
| 1452 | }
|
|---|
| 1453 |
|
|---|
| 1454 |
|
|---|
| 1455 | # Make sure the collection isn't locked by someone else
|
|---|
| 1456 | $self->lock_collection($username, $collect);
|
|---|
| 1457 |
|
|---|
| 1458 | $self->_set_archives_assocfile(@_);
|
|---|
| 1459 |
|
|---|
| 1460 | # Release the lock once it is done
|
|---|
| 1461 | $self->unlock_collection($username, $collect);
|
|---|
| 1462 | }
|
|---|
| 1463 |
|
|---|
| 1464 |
|
|---|
| 1465 | sub _remove_archives_metadata
|
|---|
| 1466 | {
|
|---|
| 1467 | my $self = shift @_;
|
|---|
| 1468 |
|
|---|
| 1469 | my $collect = $self->{'collect'};
|
|---|
| 1470 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
|---|
| 1471 | my $infodbtype = $self->{'infodbtype'};
|
|---|
| 1472 |
|
|---|
| 1473 |
|
|---|
| 1474 | # Obtain the collect and archive dir
|
|---|
| 1475 | my $site = $self->{'site'};
|
|---|
| 1476 | my $collect_dir = $gsdl_cgi->get_collection_dir($site);
|
|---|
| 1477 |
|
|---|
| 1478 | my $archive_dir = &util::filename_cat($collect_dir,$collect,"archives");
|
|---|
| 1479 |
|
|---|
| 1480 | # look up additional args
|
|---|
| 1481 | my ($docid, $docid_secnum) = ($self->{'d'} =~ m/^(.*?)(\..*)?$/);
|
|---|
| 1482 |
|
|---|
| 1483 | my $metaname = $self->{'metaname'};
|
|---|
| 1484 | my $metapos = $self->{'metapos'};
|
|---|
| 1485 | my $metavalue = $self->{'metavalue'};
|
|---|
| 1486 |
|
|---|
| 1487 | $metapos = undef if(defined $metapos && ($metapos =~ m/^\s*$/));
|
|---|
| 1488 | $metavalue = undef if(defined $metavalue && ($metavalue =~ m/^\s*$/)); # necessary to force fallback to undef here
|
|---|
| 1489 |
|
|---|
| 1490 | # if the user hasn't told us what to delete, not having given a metavalue or metapos,
|
|---|
| 1491 | # default to deleting the first metavalue for the given metaname
|
|---|
| 1492 | # Beware that if both metapos AND metavalue are defined, both matches (if any)
|
|---|
| 1493 | # seem to get deleted in one single remove_archives_meta action invocation.
|
|---|
| 1494 | # Similarly, if 2 identical metavalues for a metaname exist and that metavalue is being
|
|---|
| 1495 | # deleted, both get deleted.
|
|---|
| 1496 | if(!defined $metapos && !defined $metavalue) {
|
|---|
| 1497 | $metapos = 0;
|
|---|
| 1498 | }
|
|---|
| 1499 |
|
|---|
| 1500 | my $metamode = $self->{'metamode'};
|
|---|
| 1501 | $metamode = undef if(defined $metamode && ($metamode =~ m/^\s*$/));
|
|---|
| 1502 |
|
|---|
| 1503 | my $arcinfo_doc_filename = &dbutil::get_infodb_file_path($infodbtype, "archiveinf-doc", $archive_dir);
|
|---|
| 1504 | my $doc_rec = &dbutil::read_infodb_entry($infodbtype, $arcinfo_doc_filename, $docid);
|
|---|
| 1505 |
|
|---|
| 1506 | # This now stores the full pathname
|
|---|
| 1507 | my $doc_file = $doc_rec->{'doc-file'}->[0];
|
|---|
| 1508 |
|
|---|
| 1509 | # check if request if for file-level doc-version history 'nminus-<n>' version
|
|---|
| 1510 | my $dv = $self->{'dv'};
|
|---|
| 1511 | if (defined $dv && ($dv ne "")) {
|
|---|
| 1512 | # Need to insert '_fldv_history/nminus-<n>' into doc_filename
|
|---|
| 1513 |
|
|---|
| 1514 | my ($doc_tailname, $doc_dirname) = File::Basename::fileparse($doc_file);
|
|---|
| 1515 | $doc_file = &util::filename_cat($doc_dirname,$FLDV_HISTORY_DIR,$dv,$doc_tailname);
|
|---|
| 1516 | }
|
|---|
| 1517 |
|
|---|
| 1518 | my $doc_filename = &util::filename_cat($archive_dir, $doc_file);
|
|---|
| 1519 |
|
|---|
| 1520 | my $status = $self->remove_from_doc_xml($gsdl_cgi, $doc_filename, $metaname, $metapos, $metavalue, $docid_secnum, $metamode);
|
|---|
| 1521 | # my $status = $self->remove_from_doc_xml($gsdl_cgi, $doc_filename, $metaname, $metapos, undef, $docid_secnum);
|
|---|
| 1522 |
|
|---|
| 1523 | if ($status == 0)
|
|---|
| 1524 | {
|
|---|
| 1525 | my $mess = "\nremove-archives-metadata successful: \nKey[$docid]\n";
|
|---|
| 1526 | $mess .= " $metaname";
|
|---|
| 1527 | $mess .= "->[$metapos]" if (defined $metapos);
|
|---|
| 1528 | $mess .= " ($metavalue)" if (defined $metavalue);
|
|---|
| 1529 | $gsdl_cgi->generate_ok_message($mess);
|
|---|
| 1530 | }
|
|---|
| 1531 | else
|
|---|
| 1532 | {
|
|---|
| 1533 | my $mess .= "Failed to remove archives metadata key: $docid\n";
|
|---|
| 1534 | $mess .= "Exit status: $status\n";
|
|---|
| 1535 | $mess .= "System Error Message: $!\n";
|
|---|
| 1536 | $mess .= "-" x 20 . "\n";
|
|---|
| 1537 |
|
|---|
| 1538 | $gsdl_cgi->generate_error($mess);
|
|---|
| 1539 | }
|
|---|
| 1540 |
|
|---|
| 1541 | #return $status; # in case calling functions have a use for this
|
|---|
| 1542 | }
|
|---|
| 1543 |
|
|---|
| 1544 | sub remove_archives_metadata
|
|---|
| 1545 | {
|
|---|
| 1546 | my $self = shift @_;
|
|---|
| 1547 |
|
|---|
| 1548 | my $username = $self->{'username'};
|
|---|
| 1549 | my $collect = $self->{'collect'};
|
|---|
| 1550 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
|---|
| 1551 |
|
|---|
| 1552 | if ($baseaction::authentication_enabled)
|
|---|
| 1553 | {
|
|---|
| 1554 | # Ensure the user is allowed to edit this collection
|
|---|
| 1555 | $self->authenticate_user($username, $collect); #&authenticate_user($gsdl_cgi, $username, $collect);
|
|---|
| 1556 | }
|
|---|
| 1557 |
|
|---|
| 1558 | # Make sure the collection isn't locked by someone else
|
|---|
| 1559 | $self->lock_collection($username, $collect);
|
|---|
| 1560 |
|
|---|
| 1561 | $self->_remove_archives_metadata(@_);
|
|---|
| 1562 |
|
|---|
| 1563 | # Release the lock once it is done
|
|---|
| 1564 | $self->unlock_collection($username, $collect);
|
|---|
| 1565 | }
|
|---|
| 1566 |
|
|---|
| 1567 | sub remove_from_doc_xml
|
|---|
| 1568 | {
|
|---|
| 1569 | my $self = shift @_;
|
|---|
| 1570 | my ($gsdl_cgi, $doc_xml_filename, $metaname, $metapos, $metavalue, $secid, $metamode) = @_;
|
|---|
| 1571 |
|
|---|
| 1572 | my @start_rules = ('Section' => \&dxml_start_section);
|
|---|
| 1573 |
|
|---|
| 1574 | # Set the call-back functions for the metadata tags
|
|---|
| 1575 | my @rules =
|
|---|
| 1576 | (
|
|---|
| 1577 | _default => 'raw',
|
|---|
| 1578 | 'Metadata' => \&rfdxml_metadata
|
|---|
| 1579 | );
|
|---|
| 1580 |
|
|---|
| 1581 | my $parser = XML::Rules->new
|
|---|
| 1582 | (
|
|---|
| 1583 | start_rules => \@start_rules,
|
|---|
| 1584 | rules => \@rules,
|
|---|
| 1585 | style => 'filter',
|
|---|
| 1586 | output_encoding => 'utf8',
|
|---|
| 1587 | # normalisespaces => 1, # http://search.cpan.org/~jenda/XML-Rules-1.16/lib/XML/Rules.pm
|
|---|
| 1588 | stripspaces => 2|0|0 # ineffectual
|
|---|
| 1589 | );
|
|---|
| 1590 |
|
|---|
| 1591 | my $status = 0;
|
|---|
| 1592 | my $xml_in = "";
|
|---|
| 1593 | if (!open(MIN,"<$doc_xml_filename"))
|
|---|
| 1594 | {
|
|---|
| 1595 | $gsdl_cgi->generate_error("Unable to read in $doc_xml_filename: $!");
|
|---|
| 1596 | $status = 1;
|
|---|
| 1597 | }
|
|---|
| 1598 | else
|
|---|
| 1599 | {
|
|---|
| 1600 | # Read them in
|
|---|
| 1601 | my $line;
|
|---|
| 1602 | while (defined ($line=<MIN>)) {
|
|---|
| 1603 | $xml_in .= $line if($line !~ m/^\s*$/); # preserve all but empty lines
|
|---|
| 1604 | }
|
|---|
| 1605 | close(MIN);
|
|---|
| 1606 |
|
|---|
| 1607 | # Filter with the call-back functions
|
|---|
| 1608 | my $xml_out = "";
|
|---|
| 1609 |
|
|---|
| 1610 | my $MOUT;
|
|---|
| 1611 | if (!open($MOUT,">$doc_xml_filename")) {
|
|---|
| 1612 | $gsdl_cgi->generate_error("Unable to write out to $doc_xml_filename: $!");
|
|---|
| 1613 | $status = 1;
|
|---|
| 1614 | }
|
|---|
| 1615 | else {
|
|---|
| 1616 | binmode($MOUT,":utf8");
|
|---|
| 1617 | $parser->filter($xml_in, $MOUT, {metaname => $metaname, metapos => $metapos, metavalue => $metavalue, secid => $secid, metamode => $metamode});
|
|---|
| 1618 | close($MOUT);
|
|---|
| 1619 | }
|
|---|
| 1620 | }
|
|---|
| 1621 | return $status;
|
|---|
| 1622 | }
|
|---|
| 1623 |
|
|---|
| 1624 | sub rfdxml_metadata
|
|---|
| 1625 | {
|
|---|
| 1626 | my ($tagname, $attrHash, $contextArray, $parentDataArray, $parser) = @_;
|
|---|
| 1627 |
|
|---|
| 1628 | # For comparisons, toplevel section is indicated by ""
|
|---|
| 1629 | my $curr_sec_num = $parser->{'parameters'}->{'curr_section_num'} || "";
|
|---|
| 1630 | my $secid = $parser->{'parameters'}->{'secid'} || "";
|
|---|
| 1631 |
|
|---|
| 1632 | if (!($secid eq $curr_sec_num))
|
|---|
| 1633 | {
|
|---|
| 1634 | # RAW is [$tagname => $attrHash] not $tagname => $attrHash!!
|
|---|
| 1635 | return [$tagname => $attrHash];
|
|---|
| 1636 | }
|
|---|
| 1637 |
|
|---|
| 1638 | if ($parser->{'parameters'}->{'metaname'} eq $attrHash->{'name'})
|
|---|
| 1639 | {
|
|---|
| 1640 | if (!defined $parser->{'parameters'}->{'poscount'})
|
|---|
| 1641 | {
|
|---|
| 1642 | $parser->{'parameters'}->{'poscount'} = 0;
|
|---|
| 1643 | }
|
|---|
| 1644 | else
|
|---|
| 1645 | {
|
|---|
| 1646 | $parser->{'parameters'}->{'poscount'}++;
|
|---|
| 1647 | }
|
|---|
| 1648 |
|
|---|
| 1649 | # if overriding (for set-meta) but no metapos, then clear all the meta for this metaname
|
|---|
| 1650 | if ((defined $parser->{'parameters'}->{'metamode'}) && ($parser->{'parameters'}->{'metamode'} eq "override") && (!defined $parser->{'parameters'}->{'metapos'}) &&(!defined $parser->{'parameters'}->{'metavalue'})) {
|
|---|
| 1651 | return [];
|
|---|
| 1652 | }
|
|---|
| 1653 |
|
|---|
| 1654 | if ((defined $parser->{'parameters'}->{'metapos'}) && ($parser->{'parameters'}->{'poscount'} == $parser->{'parameters'}->{'metapos'}))
|
|---|
| 1655 | {
|
|---|
| 1656 | return [];
|
|---|
| 1657 | }
|
|---|
| 1658 |
|
|---|
| 1659 | if ((defined $parser->{'parameters'}->{'metavalue'}) && ($parser->{'parameters'}->{'metavalue'} eq $attrHash->{'_content'}))
|
|---|
| 1660 | {
|
|---|
| 1661 | return [];
|
|---|
| 1662 | }
|
|---|
| 1663 | }
|
|---|
| 1664 |
|
|---|
| 1665 | # RAW is [$tagname => $attrHash] not $tagname => $attrHash!!
|
|---|
| 1666 | return [$tagname => $attrHash];
|
|---|
| 1667 | }
|
|---|
| 1668 |
|
|---|
| 1669 | sub mxml_metadata
|
|---|
| 1670 | {
|
|---|
| 1671 | my ($tagname, $attrHash, $contextArray, $parentDataArray, $parser) = @_;
|
|---|
| 1672 | my $metaname = $parser->{'parameters'}->{'metaname'};
|
|---|
| 1673 | my $metamode = $parser->{'parameters'}->{'metamode'};
|
|---|
| 1674 |
|
|---|
| 1675 | # Report error if we don't see FileName tag before this
|
|---|
| 1676 | die "Fatal Error: Unexpected metadata.xml structure. Undefined current_file, possibly encountered Description before FileName" if (!defined($parser->{'parameters'}->{'current_file'}));
|
|---|
| 1677 |
|
|---|
| 1678 | # Don't do anything if we are not in the right FileSet
|
|---|
| 1679 | my $file_regexp = $parser->{'parameters'}->{'current_file'};
|
|---|
| 1680 | if ($file_regexp =~ /\.\*/) {
|
|---|
| 1681 | # Only interested in a file_regexp if it specifies precisely one
|
|---|
| 1682 | # file.
|
|---|
| 1683 | # So, skip anything with a .* in it as it is too general
|
|---|
| 1684 | ## print STDERR "@@@@ Skipping entry in metadata.xml where FileName=.* as it is too general\n";
|
|---|
| 1685 | return [$tagname => $attrHash];
|
|---|
| 1686 | }
|
|---|
| 1687 | my $src_file = $parser->{'parameters'}->{'src_file'};
|
|---|
| 1688 | if (!($src_file =~ /$file_regexp/)) {
|
|---|
| 1689 | return [$tagname => $attrHash];
|
|---|
| 1690 | }
|
|---|
| 1691 | ## print STDERR "*** mxl metamode = $metamode\n";
|
|---|
| 1692 |
|
|---|
| 1693 | # Find the right metadata tag and checks if we are going to override it
|
|---|
| 1694 | my $name_attr = $attrHash->{'name'};
|
|---|
| 1695 | if (($name_attr eq $metaname) && ($metamode eq "override")) {
|
|---|
| 1696 |
|
|---|
| 1697 | # now metadata.xml functions need to keep track of metapos
|
|---|
| 1698 | if (!defined $parser->{'parameters'}->{'poscount'})
|
|---|
| 1699 | {
|
|---|
| 1700 | $parser->{'parameters'}->{'poscount'} = 0;
|
|---|
| 1701 | }
|
|---|
| 1702 | else
|
|---|
| 1703 | {
|
|---|
| 1704 | $parser->{'parameters'}->{'poscount'}++;
|
|---|
| 1705 | }
|
|---|
| 1706 |
|
|---|
| 1707 | # If either the metapos or prevmetavalue is set,
|
|---|
| 1708 | # get the value and override the current value
|
|---|
| 1709 | my $metavalue = $parser->{'parameters'}->{'metavalue'};
|
|---|
| 1710 |
|
|---|
| 1711 | if(defined $parser->{'parameters'}->{'prevmetavalue'} && $parser->{'parameters'}->{'prevmetavalue'} eq $attrHash->{'_content'})
|
|---|
| 1712 | {
|
|---|
| 1713 | $attrHash->{'_content'} = $metavalue;
|
|---|
| 1714 |
|
|---|
| 1715 | ## print STDERR "**** overriding metadata.xml\n";
|
|---|
| 1716 |
|
|---|
| 1717 | # Don't want it to wipe out any other pieces of metadata
|
|---|
| 1718 | $parser->{'parameters'}->{'metamode'} = "done";
|
|---|
| 1719 | }
|
|---|
| 1720 | elsif(defined $parser->{'parameters'}->{'metapos'} && $parser->{'parameters'}->{'poscount'} == $parser->{'parameters'}->{'metapos'})
|
|---|
| 1721 | {
|
|---|
| 1722 | $attrHash->{'_content'} = $metavalue;
|
|---|
| 1723 | $parser->{'parameters'}->{'metamode'} = "done";
|
|---|
| 1724 | }
|
|---|
| 1725 | }
|
|---|
| 1726 |
|
|---|
| 1727 | # mxml_description will process the metadata if metadata is accumulate,
|
|---|
| 1728 | # or if we haven't found the metadata to override
|
|---|
| 1729 |
|
|---|
| 1730 | # RAW is [$tagname => $attrHash] not $tagname => $attrHash!!
|
|---|
| 1731 | return [$tagname => $attrHash];
|
|---|
| 1732 | }
|
|---|
| 1733 |
|
|---|
| 1734 |
|
|---|
| 1735 | sub mxml_description
|
|---|
| 1736 | {
|
|---|
| 1737 | my ($tagname, $attrHash, $contextArray, $parentDataArray, $parser) = @_;
|
|---|
| 1738 | my $metamode = $parser->{'parameters'}->{'metamode'};
|
|---|
| 1739 |
|
|---|
| 1740 | # Failed... Report error if we don't see FileName tag before this
|
|---|
| 1741 | die "Fatal Error: Unexpected metadata.xml structure. Undefind current_file, possiblely encountered Description before FileName" if (!defined($parser->{'parameters'}->{'current_file'}));
|
|---|
| 1742 |
|
|---|
| 1743 | # Don't do anything if we are not in the right FileSet
|
|---|
| 1744 | my $file_regexp = $parser->{'parameters'}->{'current_file'};
|
|---|
| 1745 | if ($file_regexp =~ m/\.\*/) {
|
|---|
| 1746 | # Only interested in a file_regexp if it specifies precisely one
|
|---|
| 1747 | # file.
|
|---|
| 1748 | # So, skip anything with a .* in it as it is too general
|
|---|
| 1749 | return [$tagname => $attrHash];
|
|---|
| 1750 | }
|
|---|
| 1751 | my $src_file = $parser->{'parameters'}->{'src_file'};
|
|---|
| 1752 |
|
|---|
| 1753 | if (!($src_file =~ m/$file_regexp/)) {
|
|---|
| 1754 | return [$tagname => $attrHash];
|
|---|
| 1755 | }
|
|---|
| 1756 |
|
|---|
| 1757 | # Accumulate the metadata block to the end of the description block
|
|---|
| 1758 | # Note: This adds metadata block to all description blocks, so if there are
|
|---|
| 1759 | # multiple FileSets, it will add to all of them
|
|---|
| 1760 | if (($metamode eq "accumulate") || ($metamode eq "override")) {
|
|---|
| 1761 |
|
|---|
| 1762 | # if metamode was "override" but get to here then it failed to
|
|---|
| 1763 | # find an item to override, in which case it should append its
|
|---|
| 1764 | # value to the end, just like the "accumulate" mode
|
|---|
| 1765 |
|
|---|
| 1766 | if ($metamode eq "override") {
|
|---|
| 1767 | print "No metadata value to override. Switching 'metamode' to accumulate\n";
|
|---|
| 1768 | }
|
|---|
| 1769 |
|
|---|
| 1770 | # tack a new metadata tag on to the end of the <Metadata>+ block
|
|---|
| 1771 | my $metaname = $parser->{'parameters'}->{'metaname'};
|
|---|
| 1772 | my $metavalue = $parser->{'parameters'}->{'metavalue'};
|
|---|
| 1773 |
|
|---|
| 1774 | my $metadata_attr = { '_content' => $metavalue,
|
|---|
| 1775 | 'name' => $metaname,
|
|---|
| 1776 | 'mode' => "accumulate" };
|
|---|
| 1777 |
|
|---|
| 1778 | my $append_metadata = [ "Metadata" => $metadata_attr ];
|
|---|
| 1779 | my $description_content = $attrHash->{'_content'};
|
|---|
| 1780 |
|
|---|
| 1781 | ## print STDERR "*** appending to metadata.xml\n";
|
|---|
| 1782 |
|
|---|
| 1783 | # append the new metadata element to the end of the current
|
|---|
| 1784 | # content contained inside this tag
|
|---|
| 1785 | if (ref($description_content) eq "") {
|
|---|
| 1786 | # => string or numeric literal
|
|---|
| 1787 | # this is caused by a <Description> block has no <Metadata> child elements
|
|---|
| 1788 | # => set up an empty array in '_content'
|
|---|
| 1789 | $attrHash->{'_content'} = [ "\n" ];
|
|---|
| 1790 | $description_content = $attrHash->{'_content'};
|
|---|
| 1791 | }
|
|---|
| 1792 |
|
|---|
| 1793 | push(@$description_content, " ", $append_metadata, "\n ");
|
|---|
| 1794 | $parser->{'parameters'}->{'metamode'} = "done";
|
|---|
| 1795 | }
|
|---|
| 1796 |
|
|---|
| 1797 | # RAW is [$tagname => $attrHash] not $tagname => $attrHash!!
|
|---|
| 1798 | return [$tagname => $attrHash];
|
|---|
| 1799 | }
|
|---|
| 1800 |
|
|---|
| 1801 |
|
|---|
| 1802 | sub mxml_fileset
|
|---|
| 1803 | {
|
|---|
| 1804 | my ($tagname, $attrHash, $contextArray, $parentDataArray, $parser) = @_;
|
|---|
| 1805 |
|
|---|
| 1806 | # Initilise the current_file
|
|---|
| 1807 | # Note: According to http://greenstone.org/dtd/DirectoryMetadata/1.0/DirectoryMetadata.dtd
|
|---|
| 1808 | # FileName tag must come before Description tag
|
|---|
| 1809 | $parser->{'parameters'}->{'current_file'} = "";
|
|---|
| 1810 |
|
|---|
| 1811 | # RAW is [$tagname => $attrHash] not $tagname => $attrHash!!
|
|---|
| 1812 | return [$tagname => $attrHash];
|
|---|
| 1813 | }
|
|---|
| 1814 |
|
|---|
| 1815 | sub mxml_directorymetadata
|
|---|
| 1816 | {
|
|---|
| 1817 | my ($tagname, $attrHash, $contextArray, $parentDataArray, $parser) = @_;
|
|---|
| 1818 |
|
|---|
| 1819 | # if we haven't processed the metadata when we reach the end of metadata.xml
|
|---|
| 1820 | # it's because there's no particular FileSet element whose FileName matched
|
|---|
| 1821 | # In which case, add a new FileSet for this FileName
|
|---|
| 1822 | my $metamode = $parser->{'parameters'}->{'metamode'};
|
|---|
| 1823 | if($metamode ne "done") {
|
|---|
| 1824 |
|
|---|
| 1825 | if ($metamode eq "override") {
|
|---|
| 1826 | print "No metadata value to override. Switching 'metamode' to accumulate\n";
|
|---|
| 1827 | }
|
|---|
| 1828 |
|
|---|
| 1829 | # If we get to here and metamode is override, this means there
|
|---|
| 1830 | # was no existing value to overide => treat as an append operation
|
|---|
| 1831 |
|
|---|
| 1832 | # Create a new FileSet element and append to DirectoryMetadata
|
|---|
| 1833 | # <FileSet>
|
|---|
| 1834 | # <FileName>src_file</FileName>
|
|---|
| 1835 | # <Description>
|
|---|
| 1836 | # <Metadata mode="" name="">metavalue</Metadata>
|
|---|
| 1837 | # </Description>
|
|---|
| 1838 | # </FileSet>
|
|---|
| 1839 | my $src_file = $parser->{'parameters'}->{'src_file'};
|
|---|
| 1840 | my $metaname = $parser->{'parameters'}->{'metaname'};
|
|---|
| 1841 | my $metavalue = $parser->{'parameters'}->{'metavalue'};
|
|---|
| 1842 | my $metadata_attr = {
|
|---|
| 1843 | '_content' => $metavalue,
|
|---|
| 1844 | 'name' => $metaname,
|
|---|
| 1845 | 'mode' => "accumulate"
|
|---|
| 1846 | };
|
|---|
| 1847 | my $append_metadata = [ "Metadata" => $metadata_attr ];
|
|---|
| 1848 | my $description_attr = { '_content' => [ "\n\t\t ", $append_metadata, "\n\t\t"] };
|
|---|
| 1849 | my $description_element = [ "Description" => $description_attr ];
|
|---|
| 1850 |
|
|---|
| 1851 | #_content is not an attribute, it's special and holds the children of this element
|
|---|
| 1852 | # including the textnode value embedded in this element if any.
|
|---|
| 1853 | my $filename_attr = {'_content' => $src_file};
|
|---|
| 1854 | my $filename_element = [ "FileName" => $filename_attr ];
|
|---|
| 1855 |
|
|---|
| 1856 | my $fileset_attr = {};
|
|---|
| 1857 | $fileset_attr->{'_content'} = [ "\n\t\t", $filename_element,"\n\t\t",$description_element ,"\n\t" ];
|
|---|
| 1858 | my $fileset = [ "FileSet" => $fileset_attr ]; #my $fileset = [ "FileSet" => {} ];
|
|---|
| 1859 |
|
|---|
| 1860 |
|
|---|
| 1861 | # get children of dirmeta, and push the new FileSet element onto it
|
|---|
| 1862 | print "Appending metadata to metadata.xml\n";
|
|---|
| 1863 | my $dirmeta_content = $attrHash->{'_content'};
|
|---|
| 1864 | if (ref($dirmeta_content)) {
|
|---|
| 1865 | # got some existing interesting nested content
|
|---|
| 1866 | #push(@$dirmeta_content, " ", $fileset ,"\n ");
|
|---|
| 1867 | push(@$dirmeta_content, "\t", $fileset ,"\n");
|
|---|
| 1868 | }
|
|---|
| 1869 | else {
|
|---|
| 1870 | #description_content is most likely a string such as "\n"
|
|---|
| 1871 | #$attrHash->{'_content'} = [$dirmeta_content, " ", $fileset ,"\n" ];
|
|---|
| 1872 | $attrHash->{'_content'} = [$dirmeta_content, "\t", $fileset ,"\n" ];
|
|---|
| 1873 | }
|
|---|
| 1874 |
|
|---|
| 1875 | $parser->{'parameters'}->{'metamode'} = "done";
|
|---|
| 1876 | }
|
|---|
| 1877 | # RAW is [$tagname => $attrHash] not $tagname => $attrHash!!
|
|---|
| 1878 | return [$tagname => $attrHash];
|
|---|
| 1879 | }
|
|---|
| 1880 |
|
|---|
| 1881 |
|
|---|
| 1882 | sub edit_metadata_xml
|
|---|
| 1883 | {
|
|---|
| 1884 | my $self = shift @_;
|
|---|
| 1885 | my ($gsdl_cgi, $metadata_xml_filename, $metaname, $metapos, $metavalue, $metamode, $src_file, $prevmetavalue) = @_;
|
|---|
| 1886 |
|
|---|
| 1887 | # Set the call-back functions for the metadata tags
|
|---|
| 1888 | my @rules =
|
|---|
| 1889 | ( _default => 'raw',
|
|---|
| 1890 | 'FileName' => \&mxml_filename,
|
|---|
| 1891 | 'Metadata' => \&mxml_metadata,
|
|---|
| 1892 | 'Description' => \&mxml_description,
|
|---|
| 1893 | 'FileSet' => \&mxml_fileset,
|
|---|
| 1894 | 'DirectoryMetadata' => \&mxml_directorymetadata);
|
|---|
| 1895 |
|
|---|
| 1896 | # use XML::Rules to add it in (read in and out again)
|
|---|
| 1897 | my $parser = XML::Rules->new(rules => \@rules,
|
|---|
| 1898 | style => 'filter',
|
|---|
| 1899 | output_encoding => 'utf8',
|
|---|
| 1900 | stripspaces => 2|0|0); # http://search.cpan.org/~jenda/XML-Rules-1.16/lib/XML/Rules.pm
|
|---|
| 1901 |
|
|---|
| 1902 | if (!-e $metadata_xml_filename) {
|
|---|
| 1903 |
|
|---|
| 1904 | if (open(MOUT,">$metadata_xml_filename")) {
|
|---|
| 1905 |
|
|---|
| 1906 | my $src_file_re = &util::filename_to_regex($src_file);
|
|---|
| 1907 | # shouldn't the following also be in the above utility routine??
|
|---|
| 1908 | # $src_file_re =~ s/\./\\./g;
|
|---|
| 1909 |
|
|---|
| 1910 | print MOUT "<?xml version=\"1.0\"?>\n";
|
|---|
| 1911 | print MOUT "<DirectoryMetadata>\n";
|
|---|
| 1912 | print MOUT " <FileSet>\n";
|
|---|
| 1913 | print MOUT " <FileName>$src_file_re</FileName>\n";
|
|---|
| 1914 | print MOUT " <Description>\n";
|
|---|
| 1915 | print MOUT " </Description>\n";
|
|---|
| 1916 | print MOUT " </FileSet>\n";
|
|---|
| 1917 | print MOUT "</DirectoryMetadata>\n";
|
|---|
| 1918 |
|
|---|
| 1919 | close(MOUT);
|
|---|
| 1920 | }
|
|---|
| 1921 | else {
|
|---|
| 1922 | $gsdl_cgi->generate_error("Unable to create $metadata_xml_filename: $!");
|
|---|
| 1923 | }
|
|---|
| 1924 | }
|
|---|
| 1925 |
|
|---|
| 1926 |
|
|---|
| 1927 | my $xml_in = "";
|
|---|
| 1928 | if (!open(MIN,"<$metadata_xml_filename")) {
|
|---|
| 1929 | $gsdl_cgi->generate_error("Unable to read in $metadata_xml_filename: $!");
|
|---|
| 1930 | }
|
|---|
| 1931 | else {
|
|---|
| 1932 | # Read them in
|
|---|
| 1933 | my $line;
|
|---|
| 1934 | while (defined ($line=<MIN>)) {
|
|---|
| 1935 | $xml_in .= $line if($line !~ m/^\s*$/); # preserve all but empty lines
|
|---|
| 1936 | }
|
|---|
| 1937 | close(MIN);
|
|---|
| 1938 |
|
|---|
| 1939 | # Filter with the call-back functions
|
|---|
| 1940 | my $xml_out = "";
|
|---|
| 1941 |
|
|---|
| 1942 | my $MOUT;
|
|---|
| 1943 | if (!open($MOUT,">$metadata_xml_filename")) {
|
|---|
| 1944 | $gsdl_cgi->generate_error("Unable to write out to $metadata_xml_filename: $!");
|
|---|
| 1945 | }
|
|---|
| 1946 | else {
|
|---|
| 1947 | binmode($MOUT,":utf8");
|
|---|
| 1948 |
|
|---|
| 1949 | # Some wise person please find out how to keep the DTD and encode lines in after it gets filtered by this XML::Rules
|
|---|
| 1950 | # At the moment, I will just hack it!
|
|---|
| 1951 | #my $header_with_utf8_dtd = "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n";
|
|---|
| 1952 | #$header_with_utf8_dtd .= "<!DOCTYPE DirectoryMetadata SYSTEM \"http://greenstone.org/dtd/DirectoryMetadata/1.0/DirectoryMetadata.dtd\">";
|
|---|
| 1953 | #$xml_out =~ s/\<\?xml\sversion\=\"1.0\"\?\>/$header_with_utf8_dtd/;
|
|---|
| 1954 | #print MOUT $xml_out;
|
|---|
| 1955 |
|
|---|
| 1956 | $parser->filter($xml_in, $MOUT, { metaname => $metaname,
|
|---|
| 1957 | metapos => $metapos,
|
|---|
| 1958 | metavalue => $metavalue,
|
|---|
| 1959 | metamode => $metamode,
|
|---|
| 1960 | src_file => $src_file,
|
|---|
| 1961 | prevmetavalue => $prevmetavalue,
|
|---|
| 1962 | current_file => undef} );
|
|---|
| 1963 | close($MOUT);
|
|---|
| 1964 | }
|
|---|
| 1965 | }
|
|---|
| 1966 | }
|
|---|
| 1967 |
|
|---|
| 1968 |
|
|---|
| 1969 | sub set_import_metadata
|
|---|
| 1970 | {
|
|---|
| 1971 | my $self = shift @_;
|
|---|
| 1972 |
|
|---|
| 1973 | my $username = $self->{'username'};
|
|---|
| 1974 | my $collect = $self->{'collect'};
|
|---|
| 1975 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
|---|
| 1976 |
|
|---|
| 1977 | if ($baseaction::authentication_enabled) {
|
|---|
| 1978 | # Ensure the user is allowed to edit this collection
|
|---|
| 1979 | $self->authenticate_user($username, $collect);
|
|---|
| 1980 | }
|
|---|
| 1981 |
|
|---|
| 1982 | # Make sure the collection isn't locked by someone else
|
|---|
| 1983 | $self->lock_collection($username, $collect);
|
|---|
| 1984 |
|
|---|
| 1985 | $self->_set_import_metadata(@_);
|
|---|
| 1986 |
|
|---|
| 1987 | # Release the lock once it is done
|
|---|
| 1988 | $self->unlock_collection($username, $collect);
|
|---|
| 1989 |
|
|---|
| 1990 | }
|
|---|
| 1991 |
|
|---|
| 1992 | sub set_import_metadata_array
|
|---|
| 1993 | {
|
|---|
| 1994 | my $self = shift @_;
|
|---|
| 1995 |
|
|---|
| 1996 | my $username = $self->{'username'};
|
|---|
| 1997 | my $collect = $self->{'collect'};
|
|---|
| 1998 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
|---|
| 1999 |
|
|---|
| 2000 | if ($baseaction::authentication_enabled) {
|
|---|
| 2001 | # Ensure the user is allowed to edit this collection
|
|---|
| 2002 | $self->authenticate_user($username, $collect); #&authenticate_user($gsdl_cgi, $username, $collect);
|
|---|
| 2003 | }
|
|---|
| 2004 |
|
|---|
| 2005 | # Obtain the collect dir
|
|---|
| 2006 | my $site = $self->{'site'};
|
|---|
| 2007 | my $collect_dir = $gsdl_cgi->get_collection_dir($site);
|
|---|
| 2008 |
|
|---|
| 2009 | $gsdl_cgi->checked_chdir($collect_dir);
|
|---|
| 2010 |
|
|---|
| 2011 | # Make sure the collection isn't locked by someone else
|
|---|
| 2012 | $self->lock_collection($username, $collect);
|
|---|
| 2013 |
|
|---|
| 2014 | $self->_set_import_metadata_array(@_);
|
|---|
| 2015 |
|
|---|
| 2016 | # Release the lock once it is done
|
|---|
| 2017 | $self->unlock_collection($username, $collect);
|
|---|
| 2018 |
|
|---|
| 2019 | }
|
|---|
| 2020 |
|
|---|
| 2021 |
|
|---|
| 2022 | sub _set_import_metadata_array
|
|---|
| 2023 | {
|
|---|
| 2024 | my $self = shift @_;
|
|---|
| 2025 |
|
|---|
| 2026 | my $collect = $self->{'collect'};
|
|---|
| 2027 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
|---|
| 2028 |
|
|---|
| 2029 | # Obtain the collect dir
|
|---|
| 2030 | my $site = $self->{'site'};
|
|---|
| 2031 | my $collect_dir = $gsdl_cgi->get_collection_dir($site);
|
|---|
| 2032 |
|
|---|
| 2033 | # look up additional args
|
|---|
| 2034 |
|
|---|
| 2035 | my $infodbtype = $self->{'infodbtype'};
|
|---|
| 2036 |
|
|---|
| 2037 | my $archive_dir = &util::filename_cat($collect_dir,$collect,"archives");
|
|---|
| 2038 | my $arcinfo_doc_filename = &dbutil::get_infodb_file_path($infodbtype, "archiveinf-doc", $archive_dir);
|
|---|
| 2039 |
|
|---|
| 2040 | my $json_str = $self->{'json'};
|
|---|
| 2041 | my $doc_array = decode_json $json_str;
|
|---|
| 2042 |
|
|---|
| 2043 | my $global_status = 0;
|
|---|
| 2044 | my $global_mess = "";
|
|---|
| 2045 |
|
|---|
| 2046 | my @all_docids = ();
|
|---|
| 2047 |
|
|---|
| 2048 | foreach my $doc_array_rec ( @$doc_array )
|
|---|
| 2049 | {
|
|---|
| 2050 | my $status = -1;
|
|---|
| 2051 | my $docid = $doc_array_rec->{'docid'};
|
|---|
| 2052 |
|
|---|
| 2053 | my ($docid_root,$docid_secnum);
|
|---|
| 2054 | if(defined $docid) {
|
|---|
| 2055 | ($docid_root,$docid_secnum) = ($docid =~ m/^(.*?)(\..*)?$/);
|
|---|
| 2056 | # as yet no support for setting subsection metadata in metadata.xml
|
|---|
| 2057 | if ((defined $docid_secnum) && ($docid_secnum !~ m/^\s*$/)) {
|
|---|
| 2058 | $gsdl_cgi->generate_message("*** docid: $docid. No support yet for setting import metadata at subsections level.\n");
|
|---|
| 2059 | next; # skip this docid in for loop
|
|---|
| 2060 | }
|
|---|
| 2061 | }
|
|---|
| 2062 |
|
|---|
| 2063 | push(@all_docids,$docid); # docid_root rather
|
|---|
| 2064 |
|
|---|
| 2065 | my $metaname = $doc_array_rec->{'metaname'};
|
|---|
| 2066 | if (defined $metaname) {
|
|---|
| 2067 | my $metamode = $doc_array_rec->{'metamode'} || $self->{'metamode'};
|
|---|
| 2068 | my $metavalue = $doc_array_rec->{'metavalue'};
|
|---|
| 2069 | $metavalue =~ s/<(.*?)>/<$1>/g;
|
|---|
| 2070 |
|
|---|
| 2071 | if ((!defined $metamode) || ($metamode =~ m/^\s*$/)) {
|
|---|
| 2072 | # make "accumulate" the default (less destructive, as it won't actually
|
|---|
| 2073 | # delete any existing values)
|
|---|
| 2074 | $metamode = "accumulate";
|
|---|
| 2075 | }
|
|---|
| 2076 |
|
|---|
| 2077 | # adding metapos and prevmetavalue support to import_metadata subroutines
|
|---|
| 2078 | my $metapos = $doc_array_rec->{'metapos'}; # don't force undef to 0. Undef has meaning when metamode=override
|
|---|
| 2079 | my $prevmetavalue = $self->{'prevmetavalue'};
|
|---|
| 2080 |
|
|---|
| 2081 | $self->set_import_metadata_entry($gsdl_cgi, $arcinfo_doc_filename, $infodbtype, $docid_root, $metaname, $metapos, $metavalue, $metamode, $prevmetavalue, $collect, $collect_dir); # at this point, docid_root = docid
|
|---|
| 2082 |
|
|---|
| 2083 | } elsif (defined $doc_array_rec->{'metatable'}) { # if no metaname, we expect a metatable
|
|---|
| 2084 | my $metatable = $doc_array_rec->{'metatable'}; # a subarray, or need to generate an error saying JSON structure is wrong
|
|---|
| 2085 |
|
|---|
| 2086 | foreach my $metatable_rec ( @$metatable ) {
|
|---|
| 2087 | $metaname = $metatable_rec->{'metaname'};
|
|---|
| 2088 | my $metamode = $metatable_rec->{'metamode'} || $doc_array_rec->{'metamode'} || $self->{'metamode'};
|
|---|
| 2089 | if ((!defined $metamode) || ($metamode =~ m/^\s*$/)) {
|
|---|
| 2090 | # make "accumulate" the default (less destructive, as it won't actually
|
|---|
| 2091 | # delete any existing values)
|
|---|
| 2092 | $metamode = "accumulate";
|
|---|
| 2093 | }
|
|---|
| 2094 |
|
|---|
| 2095 | # No support for metapos and prevmetavalue in the JSON metatable substructure
|
|---|
| 2096 | my $metapos = undef;
|
|---|
| 2097 | my $prevmetavalue = undef;
|
|---|
| 2098 | my $metavals = $metatable_rec->{'metavals'}; # a sub-subarray
|
|---|
| 2099 |
|
|---|
| 2100 | foreach my $metavalue ( @$metavals ) {
|
|---|
| 2101 | $metavalue =~ s/<(.*?)>/<$1>/g;
|
|---|
| 2102 |
|
|---|
| 2103 | $self->set_import_metadata_entry($gsdl_cgi, $arcinfo_doc_filename, $infodbtype, $docid_root, $metaname, $metapos, $metavalue, $metamode, $prevmetavalue, $collect, $collect_dir); # at this point, docid_root = docid
|
|---|
| 2104 | if($metamode eq "override") { # now, having overridden the first metavalue of the metaname,
|
|---|
| 2105 | # need to accumulate subsequent metavals for this metaname, else the just-assigned
|
|---|
| 2106 | # metavalue for this metaname will be lost
|
|---|
| 2107 | $metamode = "accumulate";
|
|---|
| 2108 | }
|
|---|
| 2109 | }
|
|---|
| 2110 | }
|
|---|
| 2111 | }
|
|---|
| 2112 | }
|
|---|
| 2113 |
|
|---|
| 2114 | # always a success message
|
|---|
| 2115 | my $mess = "set-archives-metadata-array successful: Keys[ ".join(", ",@all_docids)."]\n";
|
|---|
| 2116 | $gsdl_cgi->generate_ok_message($mess);
|
|---|
| 2117 | }
|
|---|
| 2118 |
|
|---|
| 2119 | # always returns true (1)
|
|---|
| 2120 | sub set_import_metadata_entry
|
|---|
| 2121 | {
|
|---|
| 2122 | my $self = shift @_;
|
|---|
| 2123 | my ($gsdl_cgi, $arcinfo_doc_filename, $infodbtype, $docid, $metaname, $metapos, $metavalue, $metamode, $prevmetavalue, $collect, $collect_dir) = @_;
|
|---|
| 2124 |
|
|---|
| 2125 | my $info_mess = <<RAWEND;
|
|---|
| 2126 | ****************************
|
|---|
| 2127 | set_import_metadata_entry()
|
|---|
| 2128 | ****************************
|
|---|
| 2129 | RAWEND
|
|---|
| 2130 |
|
|---|
| 2131 | $info_mess .= " collect_dir = $collect_dir\n" if defined($collect_dir);
|
|---|
| 2132 | $info_mess .= " collect = $collect\n" if defined($collect);
|
|---|
| 2133 | $info_mess .= " infodbtype = $infodbtype\n" if defined($infodbtype);
|
|---|
| 2134 | $info_mess .= " arcinfo_doc_filename = $arcinfo_doc_filename\n" if defined($arcinfo_doc_filename);
|
|---|
| 2135 | $info_mess .= " docid = $docid\n" if defined($docid);
|
|---|
| 2136 | $info_mess .= " metaname = $metaname\n" if defined($metaname);
|
|---|
| 2137 | $info_mess .= " metapos = $metapos\n" if defined($metapos);
|
|---|
| 2138 | $info_mess .= " metavalue = $metavalue\n" if defined($metavalue);
|
|---|
| 2139 | $info_mess .= " metamode = $metamode\n" if defined($metamode);
|
|---|
| 2140 | $info_mess .= " prevmetaval = $prevmetavalue\n" if defined($prevmetavalue);
|
|---|
| 2141 |
|
|---|
| 2142 | $info_mess .= "****************************\n";
|
|---|
| 2143 |
|
|---|
| 2144 | $gsdl_cgi->generate_message($info_mess);
|
|---|
| 2145 |
|
|---|
| 2146 | # import works with metadata.xml which can have inherited metadata
|
|---|
| 2147 | # so setting or removing at a metapos can have unintended effects for a COMPLEX collection
|
|---|
| 2148 | # (a collection that has or can have inherited metadata). Metapos has expected behaviour for
|
|---|
| 2149 | # a SIMPLE collection, which is one that doesn't have inherited metadata. Assume caller knows
|
|---|
| 2150 | # what they're doing if they provide a metapos.
|
|---|
| 2151 | if(defined $metapos) {
|
|---|
| 2152 | print STDERR "@@@@ WARNING: metapos defined.\n";
|
|---|
| 2153 | print STDERR "@@@@ Assuming SIMPLE collection and proceeding to modify the import meta at $metapos.\n";
|
|---|
| 2154 | }
|
|---|
| 2155 |
|
|---|
| 2156 | # Obtain where the metadata.xml is from the archiveinfo-doc.gdb file
|
|---|
| 2157 | # If the doc oid is not specified, we assume the metadata.xml is next to the specified "f"
|
|---|
| 2158 | my $metadata_xml_file;
|
|---|
| 2159 | my $import_filename = undef;
|
|---|
| 2160 |
|
|---|
| 2161 | if (defined $docid) {
|
|---|
| 2162 | # my $arcinfo_doc_filename = &dbutil::get_infodb_file_path($infodbtype, "archiveinf-doc", $archive_dir);
|
|---|
| 2163 | my $doc_rec = &dbutil::read_infodb_entry($infodbtype, $arcinfo_doc_filename, $docid);
|
|---|
| 2164 |
|
|---|
| 2165 | # This now stores the full pathname
|
|---|
| 2166 | $import_filename = $doc_rec->{'src-file'}->[0];
|
|---|
| 2167 | $import_filename = &util::placeholders_to_abspath($import_filename);
|
|---|
| 2168 |
|
|---|
| 2169 | } else { # only for set_import_meta, not the case when calling method is set_import_metadata_array
|
|---|
| 2170 | # as the array version of the method doesn't support the -f parameter yet
|
|---|
| 2171 | my $import_file = $self->{'f'};
|
|---|
| 2172 | $import_filename = &util::filename_cat($collect_dir,$collect,$import_file);
|
|---|
| 2173 | }
|
|---|
| 2174 |
|
|---|
| 2175 | # figure out correct metadata.xml file [?]
|
|---|
| 2176 | # Assuming the metadata.xml file is next to the source file
|
|---|
| 2177 | # Note: This will not work if it is using the inherited metadata from the parent folder
|
|---|
| 2178 | my ($import_tailname, $import_dirname) = File::Basename::fileparse($import_filename);
|
|---|
| 2179 | my $metadata_xml_filename = &util::filename_cat($import_dirname,"metadata.xml");
|
|---|
| 2180 |
|
|---|
| 2181 | # If we're overriding everything, then $prevmetavalue=undefined and
|
|---|
| 2182 | # $metamode=override combined with $metapos=undefined
|
|---|
| 2183 | # in which case we need to remove all metavalues for the metaname at the given (sub)section
|
|---|
| 2184 | # Thereafter, we will finally be able to set the overriding metavalue for this metaname
|
|---|
| 2185 | if(!defined $prevmetavalue && !defined $metapos && $metamode eq "override") {
|
|---|
| 2186 | ## print STDERR "@@@ REMOVING all import metadata for $metaname\n";
|
|---|
| 2187 | $self->remove_from_metadata_xml($gsdl_cgi, $metadata_xml_filename, $metaname, $metapos, undef, $import_tailname, $metamode); # we're removing all values, so metavalue=undef
|
|---|
| 2188 |
|
|---|
| 2189 | }
|
|---|
| 2190 |
|
|---|
| 2191 | # Edit the metadata.xml
|
|---|
| 2192 | # Modified by Jeffrey from DL Consulting
|
|---|
| 2193 | # Handle the case where there is one metadata.xml file for multiple FileSets
|
|---|
| 2194 | # The XML filter needs to know whether it is in the right FileSet
|
|---|
| 2195 | # TODO: This doesn't fix the problem where the metadata.xml is not next to the src file.
|
|---|
| 2196 | # TODO: This doesn't handle the common metadata (where FileName doesn't point to a single file)
|
|---|
| 2197 | $self->edit_metadata_xml($gsdl_cgi, $metadata_xml_filename, $metaname,
|
|---|
| 2198 | $metapos, $metavalue, $metamode, $import_tailname, $prevmetavalue);
|
|---|
| 2199 | #return 0;
|
|---|
| 2200 | return $metadata_xml_filename;
|
|---|
| 2201 | }
|
|---|
| 2202 |
|
|---|
| 2203 | sub _remove_import_metadata
|
|---|
| 2204 | {
|
|---|
| 2205 | my $self = shift @_;
|
|---|
| 2206 |
|
|---|
| 2207 | my $collect = $self->{'collect'};
|
|---|
| 2208 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
|---|
| 2209 | my $infodbtype = $self->{'infodbtype'};
|
|---|
| 2210 |
|
|---|
| 2211 | # Obtain the collect and archive dir
|
|---|
| 2212 | my $site = $self->{'site'};
|
|---|
| 2213 | my $collect_dir = $gsdl_cgi->get_collection_dir($site);
|
|---|
| 2214 |
|
|---|
| 2215 | my $archive_dir = &util::filename_cat($collect_dir,$collect,"archives");
|
|---|
| 2216 |
|
|---|
| 2217 | # look up additional args
|
|---|
| 2218 | my $docid = $self->{'d'};
|
|---|
| 2219 | if ((!defined $docid) || ($docid =~ m/^\s*$/))
|
|---|
| 2220 | {
|
|---|
| 2221 | $gsdl_cgi->generate_error("No docid (d=...) specified.\n");
|
|---|
| 2222 | }
|
|---|
| 2223 |
|
|---|
| 2224 | my $metaname = $self->{'metaname'};
|
|---|
| 2225 | my $metapos = $self->{'metapos'};
|
|---|
| 2226 | my $metavalue = $self->{'metavalue'};
|
|---|
| 2227 |
|
|---|
| 2228 | $metapos = undef if(defined $metapos && ($metapos =~ m/^\s*$/));
|
|---|
| 2229 | $metavalue = undef if(defined $metavalue && ($metavalue =~ m/^\s*$/));
|
|---|
| 2230 |
|
|---|
| 2231 | if(defined $metavalue) { # metavalue is now a compulsory arg for remove_import_metadata()
|
|---|
| 2232 | $metavalue =~ s/<(.*?)>/<$1>/g;
|
|---|
| 2233 | } elsif (!defined $metapos) { # if given no metavalue or metapos to delete, default to deleting the 1st
|
|---|
| 2234 | $metapos = 0;
|
|---|
| 2235 | }
|
|---|
| 2236 | my $metamode = $self->{'metamode'};
|
|---|
| 2237 | $metamode = undef if(defined $metamode && ($metamode =~ m/^\s*$/));
|
|---|
| 2238 |
|
|---|
| 2239 | # import works with metadata.xml which can have inherited metadata
|
|---|
| 2240 | # so setting or removing at a metapos can have unintended effects for a COMPLEX collection
|
|---|
| 2241 | # (a collection that has or can have inherited metadata). Metapos has expected behaviour for
|
|---|
| 2242 | # a SIMPLE collection, which is one that doesn't have inherited metadata. Assume caller knows
|
|---|
| 2243 | # what they're doing if they provide a metapos.
|
|---|
| 2244 | if(defined $metapos) {
|
|---|
| 2245 | print STDERR "@@@@ WARNING: metapos defined.\n";
|
|---|
| 2246 | print STDERR "@@@@ Assuming SIMPLE collection and proceeding to modify the import meta at $metapos.\n";
|
|---|
| 2247 | }
|
|---|
| 2248 |
|
|---|
| 2249 | # Obtain where the metadata.xml is from the archiveinfo-doc.gdb file
|
|---|
| 2250 | # If the doc oid is not specified, we assume the metadata.xml is next to the specified "f"
|
|---|
| 2251 | my $metadata_xml_file;
|
|---|
| 2252 | my $import_filename = undef;
|
|---|
| 2253 | if (defined $docid)
|
|---|
| 2254 | {
|
|---|
| 2255 | my $arcinfo_doc_filename = &dbutil::get_infodb_file_path($infodbtype, "archiveinf-doc", $archive_dir);
|
|---|
| 2256 | my $doc_rec = &dbutil::read_infodb_entry($infodbtype, $arcinfo_doc_filename, $docid);
|
|---|
| 2257 |
|
|---|
| 2258 | # This now stores the full pathname
|
|---|
| 2259 | $import_filename = $doc_rec->{'src-file'}->[0];
|
|---|
| 2260 | $import_filename = &util::placeholders_to_abspath($import_filename);
|
|---|
| 2261 | }
|
|---|
| 2262 |
|
|---|
| 2263 | if((!defined $import_filename) || ($import_filename =~ m/^\s*$/))
|
|---|
| 2264 | {
|
|---|
| 2265 | $gsdl_cgi->generate_error("There is no metadata\n");
|
|---|
| 2266 | }
|
|---|
| 2267 |
|
|---|
| 2268 | # figure out correct metadata.xml file [?]
|
|---|
| 2269 | # Assuming the metadata.xml file is next to the source file
|
|---|
| 2270 | # Note: This will not work if it is using the inherited metadata from the parent folder
|
|---|
| 2271 | my ($import_tailname, $import_dirname) = File::Basename::fileparse($import_filename);
|
|---|
| 2272 | my $metadata_xml_filename = &util::filename_cat($import_dirname,"metadata.xml");
|
|---|
| 2273 |
|
|---|
| 2274 | $self->remove_from_metadata_xml($gsdl_cgi, $metadata_xml_filename, $metaname, $metapos, $metavalue, $import_tailname, $metamode); # metamode has no meaning for removing meta, but is used by set_meta when overriding All
|
|---|
| 2275 |
|
|---|
| 2276 | my $mess = "remove-import-metadata successful: Key[$docid] -> $metadata_xml_filename\n";
|
|---|
| 2277 | $mess .= " $metaname";
|
|---|
| 2278 | $mess .= " = $metavalue\n";
|
|---|
| 2279 |
|
|---|
| 2280 | $gsdl_cgi->generate_ok_message($mess);
|
|---|
| 2281 |
|
|---|
| 2282 | #return $status; # in case calling functions have a use for this
|
|---|
| 2283 | }
|
|---|
| 2284 |
|
|---|
| 2285 | sub remove_import_metadata
|
|---|
| 2286 | {
|
|---|
| 2287 | my $self = shift @_;
|
|---|
| 2288 |
|
|---|
| 2289 | my $username = $self->{'username'};
|
|---|
| 2290 | my $collect = $self->{'collect'};
|
|---|
| 2291 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
|---|
| 2292 |
|
|---|
| 2293 | if ($baseaction::authentication_enabled) {
|
|---|
| 2294 | # Ensure the user is allowed to edit this collection
|
|---|
| 2295 | $self->authenticate_user($username, $collect); #&authenticate_user($gsdl_cgi, $username, $collect);
|
|---|
| 2296 | }
|
|---|
| 2297 |
|
|---|
| 2298 | # Make sure the collection isn't locked by someone else
|
|---|
| 2299 | $self->lock_collection($username, $collect);
|
|---|
| 2300 |
|
|---|
| 2301 | $self->_remove_import_metadata(@_);
|
|---|
| 2302 |
|
|---|
| 2303 | # Release the lock once it is done
|
|---|
| 2304 | $self->unlock_collection($username, $collect);
|
|---|
| 2305 |
|
|---|
| 2306 | }
|
|---|
| 2307 |
|
|---|
| 2308 | sub remove_from_metadata_xml
|
|---|
| 2309 | {
|
|---|
| 2310 | my $self = shift @_;
|
|---|
| 2311 | my ($gsdl_cgi, $metadata_xml_filename, $metaname, $metapos, $metavalue, $src_file, $metamode) = @_;
|
|---|
| 2312 | # metamode generally has no meaning for removing meta, but is used by set_meta
|
|---|
| 2313 | # when overriding all metavals for a metaname, in which case remove_meta is called with metamode
|
|---|
| 2314 |
|
|---|
| 2315 | # Set the call-back functions for the metadata tags
|
|---|
| 2316 | my @rules =
|
|---|
| 2317 | (
|
|---|
| 2318 | _default => 'raw',
|
|---|
| 2319 | 'Metadata' => \&rfmxml_metadata,
|
|---|
| 2320 | 'FileName' => \&mxml_filename
|
|---|
| 2321 | );
|
|---|
| 2322 |
|
|---|
| 2323 | my $parser = XML::Rules->new
|
|---|
| 2324 | (
|
|---|
| 2325 | rules => \@rules,
|
|---|
| 2326 | style => 'filter',
|
|---|
| 2327 | output_encoding => 'utf8',
|
|---|
| 2328 | #normalisespaces => 1,
|
|---|
| 2329 | stripspaces => 2|0|0 # ineffectual
|
|---|
| 2330 | );
|
|---|
| 2331 |
|
|---|
| 2332 | my $xml_in = "";
|
|---|
| 2333 | if (!open(MIN,"<$metadata_xml_filename"))
|
|---|
| 2334 | {
|
|---|
| 2335 | $gsdl_cgi->generate_error("Unable to read in $metadata_xml_filename: $!");
|
|---|
| 2336 | }
|
|---|
| 2337 | else
|
|---|
| 2338 | {
|
|---|
| 2339 | # Read them in
|
|---|
| 2340 | my $line;
|
|---|
| 2341 | while (defined ($line=<MIN>)) {
|
|---|
| 2342 | $xml_in .= $line if($line !~ m/^\s*$/); # preserve all but empty lines
|
|---|
| 2343 | }
|
|---|
| 2344 | close(MIN);
|
|---|
| 2345 |
|
|---|
| 2346 | # Filter with the call-back functions
|
|---|
| 2347 | my $xml_out = "";
|
|---|
| 2348 |
|
|---|
| 2349 | my $MOUT;
|
|---|
| 2350 | if (!open($MOUT,">$metadata_xml_filename")) {
|
|---|
| 2351 | $gsdl_cgi->generate_error("Unable to write out to $metadata_xml_filename: $!");
|
|---|
| 2352 | }
|
|---|
| 2353 | else {
|
|---|
| 2354 | binmode($MOUT,":utf8");
|
|---|
| 2355 | $parser->filter($xml_in, $MOUT, {metaname => $metaname, metapos => $metapos, metavalue => $metavalue, src_file => $src_file, metamode => $metamode, current_file => undef});
|
|---|
| 2356 | close($MOUT);
|
|---|
| 2357 | }
|
|---|
| 2358 | }
|
|---|
| 2359 | }
|
|---|
| 2360 |
|
|---|
| 2361 | sub rfmxml_metadata
|
|---|
| 2362 | {
|
|---|
| 2363 | my ($tagname, $attrHash, $contextArray, $parentDataArray, $parser) = @_;
|
|---|
| 2364 |
|
|---|
| 2365 | # metadata.xml does not handle subsections
|
|---|
| 2366 |
|
|---|
| 2367 | # since metadata.xml now has to deal with metapos, we keep track of the metadata position
|
|---|
| 2368 | if (($parser->{'parameters'}->{'src_file'} eq $parser->{'parameters'}->{'current_file'})
|
|---|
| 2369 | && $parser->{'parameters'}->{'metaname'} eq $attrHash->{'name'})
|
|---|
| 2370 | {
|
|---|
| 2371 | if (!defined $parser->{'parameters'}->{'poscount'})
|
|---|
| 2372 | {
|
|---|
| 2373 | $parser->{'parameters'}->{'poscount'} = 0;
|
|---|
| 2374 | }
|
|---|
| 2375 | else
|
|---|
| 2376 | {
|
|---|
| 2377 | $parser->{'parameters'}->{'poscount'}++;
|
|---|
| 2378 | }
|
|---|
| 2379 |
|
|---|
| 2380 | # if overriding but no metapos, then clear all the meta for this metaname
|
|---|
| 2381 | if ((defined $parser->{'parameters'}->{'metamode'}) && ($parser->{'parameters'}->{'metamode'} eq "override") && (!defined $parser->{'parameters'}->{'metapos'}) && (!defined $parser->{'parameters'}->{'metavalue'})) {
|
|---|
| 2382 | return [];
|
|---|
| 2383 | }
|
|---|
| 2384 |
|
|---|
| 2385 | if ((defined $parser->{'parameters'}->{'metapos'}) && ($parser->{'parameters'}->{'poscount'} == $parser->{'parameters'}->{'metapos'}))
|
|---|
| 2386 | {
|
|---|
| 2387 | return [];
|
|---|
| 2388 | }
|
|---|
| 2389 | if ((defined $parser->{'parameters'}->{'metavalue'}) && ($attrHash->{'_content'} eq $parser->{'parameters'}->{'metavalue'}))
|
|---|
| 2390 | {
|
|---|
| 2391 | return [];
|
|---|
| 2392 | }
|
|---|
| 2393 | }
|
|---|
| 2394 |
|
|---|
| 2395 | # RAW is [$tagname => $attrHash] not $tagname => $attrHash!!
|
|---|
| 2396 | return [$tagname => $attrHash];
|
|---|
| 2397 | }
|
|---|
| 2398 |
|
|---|
| 2399 | sub _remove_live_metadata
|
|---|
| 2400 | {
|
|---|
| 2401 | my $self = shift @_;
|
|---|
| 2402 |
|
|---|
| 2403 | my $collect = $self->{'collect'};
|
|---|
| 2404 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
|---|
| 2405 | my $infodbtype = $self->{'infodbtype'};
|
|---|
| 2406 |
|
|---|
| 2407 | # Obtain the collect dir
|
|---|
| 2408 | my $site = $self->{'site'};
|
|---|
| 2409 | my $collect_dir = $gsdl_cgi->get_collection_dir($site);
|
|---|
| 2410 |
|
|---|
| 2411 | # look up additional args
|
|---|
| 2412 | my $docid = $self->{'d'};
|
|---|
| 2413 | if ((!defined $docid) || ($docid =~ m/^\s*$/)) {
|
|---|
| 2414 | $gsdl_cgi->generate_error("No docid (d=...) specified.");
|
|---|
| 2415 | }
|
|---|
| 2416 |
|
|---|
| 2417 | # Generate the dbkey
|
|---|
| 2418 | my $metaname = $self->{'metaname'};
|
|---|
| 2419 | my $dbkey = "$docid.$metaname";
|
|---|
| 2420 |
|
|---|
| 2421 | # To people who know $collect_tail please add some comments
|
|---|
| 2422 | # Obtain the live gdbm_db path
|
|---|
| 2423 | my $collect_tail = $collect;
|
|---|
| 2424 | $collect_tail =~ s/^.*[\/|\\]//;
|
|---|
| 2425 | my $index_text_directory = &util::filename_cat($collect_dir,$collect,"index","text");
|
|---|
| 2426 | my $infodb_file_path = &dbutil::get_infodb_file_path($infodbtype, "live-$collect_tail", $index_text_directory);
|
|---|
| 2427 |
|
|---|
| 2428 | # Remove the key
|
|---|
| 2429 | my $cmd = "gdbmdel \"$infodb_file_path\" \"$dbkey\"";
|
|---|
| 2430 | my $status = system($cmd);
|
|---|
| 2431 | if ($status != 0) {
|
|---|
| 2432 | # Catch error if gdbmdel failed
|
|---|
| 2433 | my $mess = "Failed to delete metadata key: $dbkey\n";
|
|---|
| 2434 |
|
|---|
| 2435 | $mess .= "PATH: $ENV{'PATH'}\n";
|
|---|
| 2436 | $mess .= "cmd = $cmd\n";
|
|---|
| 2437 | $mess .= "Exit status: $status\n";
|
|---|
| 2438 | $mess .= "System Error Message: $!\n";
|
|---|
| 2439 |
|
|---|
| 2440 | $gsdl_cgi->generate_error($mess);
|
|---|
| 2441 | }
|
|---|
| 2442 | else {
|
|---|
| 2443 | $gsdl_cgi->generate_ok_message("DB remove successful: Key[$metaname]");
|
|---|
| 2444 | }
|
|---|
| 2445 |
|
|---|
| 2446 | }
|
|---|
| 2447 |
|
|---|
| 2448 | sub remove_live_metadata
|
|---|
| 2449 | {
|
|---|
| 2450 | my $self = shift @_;
|
|---|
| 2451 |
|
|---|
| 2452 | my $username = $self->{'username'};
|
|---|
| 2453 | my $collect = $self->{'collect'};
|
|---|
| 2454 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
|---|
| 2455 |
|
|---|
| 2456 | if ($baseaction::authentication_enabled) {
|
|---|
| 2457 | # Ensure the user is allowed to edit this collection
|
|---|
| 2458 | $self->authenticate_user($username, $collect); #&authenticate_user($gsdl_cgi, $username, $collect);
|
|---|
| 2459 | }
|
|---|
| 2460 |
|
|---|
| 2461 | # Make sure the collection isn't locked by someone else
|
|---|
| 2462 | $self->lock_collection($username, $collect);
|
|---|
| 2463 |
|
|---|
| 2464 | $self->_remove_live_metadata(@_);
|
|---|
| 2465 |
|
|---|
| 2466 | $self->unlock_collection($username, $collect);
|
|---|
| 2467 | }
|
|---|
| 2468 |
|
|---|
| 2469 | sub remove_metadata
|
|---|
| 2470 | {
|
|---|
| 2471 | my $self = shift @_;
|
|---|
| 2472 |
|
|---|
| 2473 | my $where = $self->{'where'};
|
|---|
| 2474 | if(!$where || ($where =~ m/^\s*$/)) {
|
|---|
| 2475 | $self->remove_index_metadata(@_); # call the full version of set_index_meta for the default behaviour
|
|---|
| 2476 | return;
|
|---|
| 2477 | }
|
|---|
| 2478 |
|
|---|
| 2479 | my $username = $self->{'username'};
|
|---|
| 2480 | my $collect = $self->{'collect'};
|
|---|
| 2481 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
|---|
| 2482 |
|
|---|
| 2483 | if ($baseaction::authentication_enabled) {
|
|---|
| 2484 | # Ensure the user is allowed to edit this collection
|
|---|
| 2485 | $self->authenticate_user($username, $collect); #&authenticate_user($gsdl_cgi, $username, $collect);
|
|---|
| 2486 | }
|
|---|
| 2487 |
|
|---|
| 2488 | # Make sure the collection isn't locked by someone else
|
|---|
| 2489 | $self->lock_collection($username, $collect);
|
|---|
| 2490 |
|
|---|
| 2491 | # check which directories need to be processed, specified in $where as
|
|---|
| 2492 | # any combination of import|archives|index|live
|
|---|
| 2493 | if($where =~ m/import/) {
|
|---|
| 2494 | $self->_remove_import_metadata(@_);
|
|---|
| 2495 | }
|
|---|
| 2496 | if($where =~ m/archives/) {
|
|---|
| 2497 | $self->_remove_archives_metadata(@_);
|
|---|
| 2498 | }
|
|---|
| 2499 | if($where =~ m/index/) {
|
|---|
| 2500 | $self->_remove_index_metadata(@_);
|
|---|
| 2501 | }
|
|---|
| 2502 |
|
|---|
| 2503 | # Release the lock once it is done
|
|---|
| 2504 | $self->unlock_collection($username, $collect);
|
|---|
| 2505 | }
|
|---|
| 2506 |
|
|---|
| 2507 | # the internal version, without authentication
|
|---|
| 2508 | sub _remove_index_metadata
|
|---|
| 2509 | {
|
|---|
| 2510 | my $self = shift @_;
|
|---|
| 2511 |
|
|---|
| 2512 | my $collect = $self->{'collect'};
|
|---|
| 2513 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
|---|
| 2514 | my $infodbtype = $self->{'infodbtype'};
|
|---|
| 2515 |
|
|---|
| 2516 | # Obtain the collect dir
|
|---|
| 2517 | my $site = $self->{'site'};
|
|---|
| 2518 | my $collect_dir = $gsdl_cgi->get_collection_dir($site);
|
|---|
| 2519 |
|
|---|
| 2520 | # look up additional args
|
|---|
| 2521 | my $docid = $self->{'d'};
|
|---|
| 2522 | if ((!defined $docid) || ($docid =~ m/^\s*$/)) {
|
|---|
| 2523 | $gsdl_cgi->generate_error("No docid (d=...) specified.");
|
|---|
| 2524 | }
|
|---|
| 2525 | my $metaname = $self->{'metaname'};
|
|---|
| 2526 | my $metapos = $self->{'metapos'};
|
|---|
| 2527 | my $metavalue = $self->{'metavalue'};
|
|---|
| 2528 |
|
|---|
| 2529 | $metapos = undef if(defined $metapos && ($metapos =~ m/^\s*$/));
|
|---|
| 2530 | $metavalue = undef if(defined $metavalue && ($metavalue =~ m/^\s*$/)); # necessary to force fallback to undef here
|
|---|
| 2531 |
|
|---|
| 2532 | # To people who know $collect_tail please add some comments
|
|---|
| 2533 | # -> In collection groups, I think collect_tailname is the subcollection name,
|
|---|
| 2534 | # e.g. colgroup-name/col-tail-name
|
|---|
| 2535 | # Obtain the path to the database
|
|---|
| 2536 | my $collect_tail = $collect;
|
|---|
| 2537 | $collect_tail =~ s/^.*[\/|\\]//;
|
|---|
| 2538 | my $index_text_directory = &util::filename_cat($collect_dir,$collect,"index","text");
|
|---|
| 2539 | my $infodb_file_path = &dbutil::get_infodb_file_path($infodbtype, $collect_tail, $index_text_directory);
|
|---|
| 2540 |
|
|---|
| 2541 | # Read the docid entry
|
|---|
| 2542 | my $doc_rec = &dbutil::read_infodb_entry($infodbtype, $infodb_file_path, $docid);
|
|---|
| 2543 |
|
|---|
| 2544 | # Check to make sure the key does exist
|
|---|
| 2545 | if (!defined ($doc_rec->{$metaname})) {
|
|---|
| 2546 | $gsdl_cgi->generate_error("No metadata field \"" . $metaname . "\" in the specified document: [" . $docid . "]");
|
|---|
| 2547 | }
|
|---|
| 2548 |
|
|---|
| 2549 | # Obtain the specified metadata pos
|
|---|
| 2550 | # if no metavalue or metapos to delete, default to deleting the 1st value for the metaname
|
|---|
| 2551 | if(!defined $metapos && !defined $metavalue) {
|
|---|
| 2552 | $metapos = 0;
|
|---|
| 2553 | }
|
|---|
| 2554 |
|
|---|
| 2555 |
|
|---|
| 2556 | # consider check key is defined before deleting?
|
|---|
| 2557 | # Loop through the metadata array and ignore the specified position
|
|---|
| 2558 | my $filtered_metadata = [];
|
|---|
| 2559 | my $num_metadata_vals = scalar(@{$doc_rec->{$metaname}});
|
|---|
| 2560 | for (my $i=0; $i<$num_metadata_vals; $i++) {
|
|---|
| 2561 | my $metaval = shift(@{$doc_rec->{$metaname}});
|
|---|
| 2562 |
|
|---|
| 2563 | if (!defined $metavalue && $i != $metapos) {
|
|---|
| 2564 | push(@$filtered_metadata,$metaval);
|
|---|
| 2565 | }
|
|---|
| 2566 |
|
|---|
| 2567 | if(defined $metavalue && !($metavalue eq $metaval))
|
|---|
| 2568 | {
|
|---|
| 2569 | push(@$filtered_metadata,$metaval);
|
|---|
| 2570 | }
|
|---|
| 2571 | }
|
|---|
| 2572 | $doc_rec->{$metaname} = $filtered_metadata;
|
|---|
| 2573 |
|
|---|
| 2574 | ## Use the dbutil set_entry method instead of assuming the database is gdbm
|
|---|
| 2575 | my $status = &dbutil::set_infodb_entry($infodbtype, $infodb_file_path, $docid, $doc_rec);
|
|---|
| 2576 |
|
|---|
| 2577 | if ($status != 0) {
|
|---|
| 2578 | my $mess = "Failed to set metadata key: $docid\n";
|
|---|
| 2579 |
|
|---|
| 2580 | $mess .= "PATH: $ENV{'PATH'}\n";
|
|---|
| 2581 | $mess .= "Exit status: $status\n";
|
|---|
| 2582 | $mess .= "System Error Message: $!\n";
|
|---|
| 2583 |
|
|---|
| 2584 | $gsdl_cgi->generate_error($mess);
|
|---|
| 2585 | }
|
|---|
| 2586 | else {
|
|---|
| 2587 | my $mess = "DB set (with item deleted) successful: Key[$docid]\n";
|
|---|
| 2588 | $mess .= " $metaname";
|
|---|
| 2589 | $mess .= "->[$metapos]" if (defined $metapos);
|
|---|
| 2590 | $mess .= " ($metavalue)" if (defined $metavalue);
|
|---|
| 2591 |
|
|---|
| 2592 | $gsdl_cgi->generate_ok_message($mess);
|
|---|
| 2593 | }
|
|---|
| 2594 |
|
|---|
| 2595 | #return $status; # in case calling functions have a use for this
|
|---|
| 2596 | }
|
|---|
| 2597 |
|
|---|
| 2598 | sub remove_index_metadata
|
|---|
| 2599 | {
|
|---|
| 2600 | my $self = shift @_;
|
|---|
| 2601 |
|
|---|
| 2602 | my $username = $self->{'username'};
|
|---|
| 2603 | my $collect = $self->{'collect'};
|
|---|
| 2604 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
|---|
| 2605 |
|
|---|
| 2606 | if ($baseaction::authentication_enabled) {
|
|---|
| 2607 | # Ensure the user is allowed to edit this collection
|
|---|
| 2608 | $self->authenticate_user($username, $collect); #&authenticate_user($gsdl_cgi, $username, $collect);
|
|---|
| 2609 | }
|
|---|
| 2610 |
|
|---|
| 2611 | # Obtain the collect dir
|
|---|
| 2612 | my $site = $self->{'site'};
|
|---|
| 2613 | my $collect_dir = $gsdl_cgi->get_collection_dir($site);
|
|---|
| 2614 |
|
|---|
| 2615 | # Make sure the collection isn't locked by someone else
|
|---|
| 2616 | $self->lock_collection($username, $collect);
|
|---|
| 2617 |
|
|---|
| 2618 | $self->_remove_index_metadata(@_);
|
|---|
| 2619 |
|
|---|
| 2620 | # Release the lock once it is done
|
|---|
| 2621 | $self->unlock_collection($username, $collect);
|
|---|
| 2622 | }
|
|---|
| 2623 |
|
|---|
| 2624 | #################################################################################
|
|---|
| 2625 | # ERASE META METHODS: ERASE ALL VALUES FOR MATCHING METANAME AT SPECIFIED DOCID #
|
|---|
| 2626 | #################################################################################
|
|---|
| 2627 |
|
|---|
| 2628 | sub erase_metadata
|
|---|
| 2629 | {
|
|---|
| 2630 | my $self = shift @_;
|
|---|
| 2631 |
|
|---|
| 2632 | my $where = $self->{'where'};
|
|---|
| 2633 | # when $where is unspecified, following behaviour of sub remove_metadata by defaulting to erasing metadata from index
|
|---|
| 2634 | if(!$where || ($where =~ m/^\s*$/)) {
|
|---|
| 2635 | $self->erase_index_metadata(@_); # call the full version of set_index_meta for the default behaviour
|
|---|
| 2636 | return;
|
|---|
| 2637 | }
|
|---|
| 2638 |
|
|---|
| 2639 | my $username = $self->{'username'};
|
|---|
| 2640 | my $collect = $self->{'collect'};
|
|---|
| 2641 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
|---|
| 2642 |
|
|---|
| 2643 | if ($baseaction::authentication_enabled) {
|
|---|
| 2644 | # Ensure the user is allowed to edit this collection
|
|---|
| 2645 | $self->authenticate_user($username, $collect); #&authenticate_user($gsdl_cgi, $username, $collect);
|
|---|
| 2646 | }
|
|---|
| 2647 |
|
|---|
| 2648 | # Make sure the collection isn't locked by someone else
|
|---|
| 2649 | $self->lock_collection($username, $collect);
|
|---|
| 2650 |
|
|---|
| 2651 | # check which directories need to be processed, specified in $where as
|
|---|
| 2652 | # any combination of import|archives|index|live
|
|---|
| 2653 | if($where =~ m/import/) {
|
|---|
| 2654 | $self->_erase_import_metadata(@_);
|
|---|
| 2655 | }
|
|---|
| 2656 | if($where =~ m/archives/) {
|
|---|
| 2657 | $self->_erase_archives_metadata(@_);
|
|---|
| 2658 | }
|
|---|
| 2659 | if($where =~ m/index/) {
|
|---|
| 2660 | $self->_erase_index_metadata(@_);
|
|---|
| 2661 | }
|
|---|
| 2662 |
|
|---|
| 2663 | # Release the lock once it is done
|
|---|
| 2664 | $self->unlock_collection($username, $collect);
|
|---|
| 2665 | }
|
|---|
| 2666 |
|
|---|
| 2667 | # the internal version, without authentication
|
|---|
| 2668 | sub _erase_index_metadata
|
|---|
| 2669 | {
|
|---|
| 2670 | my $self = shift @_;
|
|---|
| 2671 |
|
|---|
| 2672 | my $collect = $self->{'collect'};
|
|---|
| 2673 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
|---|
| 2674 | my $infodbtype = $self->{'infodbtype'};
|
|---|
| 2675 |
|
|---|
| 2676 | # Obtain the collect dir
|
|---|
| 2677 | my $site = $self->{'site'};
|
|---|
| 2678 | my $collect_dir = $gsdl_cgi->get_collection_dir($site);
|
|---|
| 2679 |
|
|---|
| 2680 | # look up additional args
|
|---|
| 2681 | my $docid = $self->{'d'};
|
|---|
| 2682 | if ((!defined $docid) || ($docid =~ m/^\s*$/)) {
|
|---|
| 2683 | $gsdl_cgi->generate_error("No docid (d=...) specified.");
|
|---|
| 2684 | }
|
|---|
| 2685 | my $metaname = $self->{'metaname'};
|
|---|
| 2686 | my $metapos = undef;
|
|---|
| 2687 | my $metavalue = undef;
|
|---|
| 2688 |
|
|---|
| 2689 | # To people who know $collect_tail please add some comments
|
|---|
| 2690 | # -> In collection groups, I think collect_tailname is the subcollection name,
|
|---|
| 2691 | # e.g. colgroup-name/col-tail-name
|
|---|
| 2692 | # Obtain the path to the database
|
|---|
| 2693 | my $collect_tail = $collect;
|
|---|
| 2694 | $collect_tail =~ s/^.*[\/|\\]//;
|
|---|
| 2695 | my $index_text_directory = &util::filename_cat($collect_dir,$collect,"index","text");
|
|---|
| 2696 | my $infodb_file_path = &dbutil::get_infodb_file_path($infodbtype, $collect_tail, $index_text_directory);
|
|---|
| 2697 |
|
|---|
| 2698 | # Read the docid entry
|
|---|
| 2699 | my $doc_rec = &dbutil::read_infodb_entry($infodbtype, $infodb_file_path, $docid);
|
|---|
| 2700 |
|
|---|
| 2701 | # Check to make sure the key does exist
|
|---|
| 2702 | if (defined ($doc_rec->{$metaname})) {
|
|---|
| 2703 |
|
|---|
| 2704 | $doc_rec->{$metaname} = [];
|
|---|
| 2705 |
|
|---|
| 2706 | ## Use the dbutil set_entry method instead of assuming the database is gdbm
|
|---|
| 2707 | my $status = &dbutil::set_infodb_entry($infodbtype, $infodb_file_path, $docid, $doc_rec);
|
|---|
| 2708 |
|
|---|
| 2709 | if ($status != 0) {
|
|---|
| 2710 | my $mess = "Failed to erase metadata key: $docid\n";
|
|---|
| 2711 |
|
|---|
| 2712 | $mess .= "PATH: $ENV{'PATH'}\n";
|
|---|
| 2713 | $mess .= "Exit status: $status\n";
|
|---|
| 2714 | $mess .= "System Error Message: $!\n";
|
|---|
| 2715 |
|
|---|
| 2716 | $gsdl_cgi->generate_error($mess);
|
|---|
| 2717 | }
|
|---|
| 2718 | else {
|
|---|
| 2719 | my $mess = "DB set (with item deleted) successful: Key[$docid]\n";
|
|---|
| 2720 | $mess .= " $metaname";
|
|---|
| 2721 | $mess .= "->[$metapos]" if (defined $metapos);
|
|---|
| 2722 | $mess .= " ($metavalue)" if (defined $metavalue);
|
|---|
| 2723 |
|
|---|
| 2724 | $gsdl_cgi->generate_ok_message($mess);
|
|---|
| 2725 | }
|
|---|
| 2726 | }
|
|---|
| 2727 | else { # no such metaname. Is it really an error?
|
|---|
| 2728 | $gsdl_cgi->generate_ok_message("Can't erase. No metadata field \"" . $metaname . "\" in the specified document: [" . $docid . "]");
|
|---|
| 2729 | }
|
|---|
| 2730 | #return $status; # in case calling functions have a use for this
|
|---|
| 2731 | }
|
|---|
| 2732 |
|
|---|
| 2733 | sub erase_index_metadata
|
|---|
| 2734 | {
|
|---|
| 2735 | my $self = shift @_;
|
|---|
| 2736 |
|
|---|
| 2737 | my $username = $self->{'username'};
|
|---|
| 2738 | my $collect = $self->{'collect'};
|
|---|
| 2739 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
|---|
| 2740 |
|
|---|
| 2741 | if ($baseaction::authentication_enabled) {
|
|---|
| 2742 | # Ensure the user is allowed to edit this collection
|
|---|
| 2743 | $self->authenticate_user($username, $collect); #&authenticate_user($gsdl_cgi, $username, $collect);
|
|---|
| 2744 | }
|
|---|
| 2745 |
|
|---|
| 2746 | # Obtain the collect dir
|
|---|
| 2747 | my $site = $self->{'site'};
|
|---|
| 2748 | my $collect_dir = $gsdl_cgi->get_collection_dir($site);
|
|---|
| 2749 |
|
|---|
| 2750 | # Make sure the collection isn't locked by someone else
|
|---|
| 2751 | $self->lock_collection($username, $collect);
|
|---|
| 2752 |
|
|---|
| 2753 | $self->_erase_index_metadata(@_);
|
|---|
| 2754 |
|
|---|
| 2755 | # Release the lock once it is done
|
|---|
| 2756 | $self->unlock_collection($username, $collect);
|
|---|
| 2757 | }
|
|---|
| 2758 |
|
|---|
| 2759 |
|
|---|
| 2760 | # Given a metaname, removes ALL metadata with that metaname at the provided docid
|
|---|
| 2761 | sub _erase_archives_metadata
|
|---|
| 2762 | {
|
|---|
| 2763 | my $self = shift @_;
|
|---|
| 2764 |
|
|---|
| 2765 | my $collect = $self->{'collect'};
|
|---|
| 2766 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
|---|
| 2767 | my $infodbtype = $self->{'infodbtype'};
|
|---|
| 2768 |
|
|---|
| 2769 | # Obtain the collect and archive dir
|
|---|
| 2770 | my $site = $self->{'site'};
|
|---|
| 2771 | my $collect_dir = $gsdl_cgi->get_collection_dir($site);
|
|---|
| 2772 |
|
|---|
| 2773 | my $archive_dir = &util::filename_cat($collect_dir,$collect,"archives");
|
|---|
| 2774 |
|
|---|
| 2775 | # look up additional args
|
|---|
| 2776 | my ($docid, $docid_secnum) = ($self->{'d'} =~ m/^(.*?)(\..*)?$/);
|
|---|
| 2777 |
|
|---|
| 2778 | my $metaname = $self->{'metaname'};
|
|---|
| 2779 |
|
|---|
| 2780 | # metapos, metavalue and metamode are never provided and therefore all undefined for erase_archives_metadata
|
|---|
| 2781 | my $metapos = undef;
|
|---|
| 2782 | my $metavalue = undef;
|
|---|
| 2783 | my $metamode = "override"; # when override and no metapos or metaval, all values for metaname will be affected
|
|---|
| 2784 |
|
|---|
| 2785 | my $arcinfo_doc_filename = &dbutil::get_infodb_file_path($infodbtype, "archiveinf-doc", $archive_dir);
|
|---|
| 2786 | my $doc_rec = &dbutil::read_infodb_entry($infodbtype, $arcinfo_doc_filename, $docid);
|
|---|
| 2787 |
|
|---|
| 2788 | # This now stores the full pathname
|
|---|
| 2789 | my $doc_file = $doc_rec->{'doc-file'}->[0];
|
|---|
| 2790 |
|
|---|
| 2791 | # check if request if for file-level doc-version history 'nminus-<n>' version
|
|---|
| 2792 | my $dv = $self->{'dv'};
|
|---|
| 2793 | if (defined $dv && ($dv ne "")) {
|
|---|
| 2794 | # Need to insert '_fldv_history/nminus-<n>' into doc_filename
|
|---|
| 2795 |
|
|---|
| 2796 | my ($doc_tailname, $doc_dirname) = File::Basename::fileparse($doc_file);
|
|---|
| 2797 | $doc_file = &util::filename_cat($doc_dirname,$FLDV_HISTORY_DIR,$dv,$doc_tailname);
|
|---|
| 2798 | }
|
|---|
| 2799 |
|
|---|
| 2800 | my $doc_filename = &util::filename_cat($archive_dir, $doc_file);
|
|---|
| 2801 |
|
|---|
| 2802 | my $status = $self->remove_from_doc_xml($gsdl_cgi, $doc_filename, $metaname, $metapos, $metavalue, $docid_secnum, $metamode);
|
|---|
| 2803 |
|
|---|
| 2804 | if ($status == 0)
|
|---|
| 2805 | {
|
|---|
| 2806 | my $mess = "\nerase-archives-metadata successful: \nKey[$docid]\n";
|
|---|
| 2807 | $mess .= " $metaname";
|
|---|
| 2808 | $gsdl_cgi->generate_ok_message($mess);
|
|---|
| 2809 | }
|
|---|
| 2810 | else
|
|---|
| 2811 | {
|
|---|
| 2812 | my $mess .= "Failed to erase archives metadata key: $docid\n";
|
|---|
| 2813 | $mess .= "Exit status: $status\n";
|
|---|
| 2814 | $mess .= "System Error Message: $!\n";
|
|---|
| 2815 | $mess .= "-" x 20 . "\n";
|
|---|
| 2816 |
|
|---|
| 2817 | $gsdl_cgi->generate_error($mess);
|
|---|
| 2818 | }
|
|---|
| 2819 |
|
|---|
| 2820 | #return $status; # in case calling functions have a use for this
|
|---|
| 2821 | }
|
|---|
| 2822 |
|
|---|
| 2823 | # practically identical to remove_archives_metadata
|
|---|
| 2824 | sub erase_archives_metadata
|
|---|
| 2825 | {
|
|---|
| 2826 | my $self = shift @_;
|
|---|
| 2827 |
|
|---|
| 2828 | my $username = $self->{'username'};
|
|---|
| 2829 | my $collect = $self->{'collect'};
|
|---|
| 2830 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
|---|
| 2831 |
|
|---|
| 2832 | if ($baseaction::authentication_enabled)
|
|---|
| 2833 | {
|
|---|
| 2834 | # Ensure the user is allowed to edit this collection
|
|---|
| 2835 | $self->authenticate_user($username, $collect); #&authenticate_user($gsdl_cgi, $username, $collect);
|
|---|
| 2836 | }
|
|---|
| 2837 |
|
|---|
| 2838 | # Make sure the collection isn't locked by someone else
|
|---|
| 2839 | $self->lock_collection($username, $collect);
|
|---|
| 2840 |
|
|---|
| 2841 | $self->_erase_archives_metadata(@_);
|
|---|
| 2842 |
|
|---|
| 2843 | # Release the lock once it is done
|
|---|
| 2844 | $self->unlock_collection($username, $collect);
|
|---|
| 2845 | }
|
|---|
| 2846 |
|
|---|
| 2847 | sub _erase_import_metadata
|
|---|
| 2848 | {
|
|---|
| 2849 | my $self = shift @_;
|
|---|
| 2850 |
|
|---|
| 2851 | my $collect = $self->{'collect'};
|
|---|
| 2852 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
|---|
| 2853 | my $infodbtype = $self->{'infodbtype'};
|
|---|
| 2854 |
|
|---|
| 2855 | # Obtain the collect dir
|
|---|
| 2856 | my $site = $self->{'site'};
|
|---|
| 2857 | my $collect_dir = $gsdl_cgi->get_collection_dir($site);
|
|---|
| 2858 |
|
|---|
| 2859 | my $archive_dir = &util::filename_cat($collect_dir,$collect,"archives");
|
|---|
| 2860 |
|
|---|
| 2861 | # look up additional args
|
|---|
| 2862 | my $docid = $self->{'d'};
|
|---|
| 2863 | if ((!defined $docid) || ($docid =~ m/^\s*$/))
|
|---|
| 2864 | {
|
|---|
| 2865 | $gsdl_cgi->generate_error("No docid (d=...) specified.\n");
|
|---|
| 2866 | }
|
|---|
| 2867 |
|
|---|
| 2868 | my $metaname = $self->{'metaname'};
|
|---|
| 2869 | my $metapos = undef;
|
|---|
| 2870 | my $metavalue = undef;
|
|---|
| 2871 | my $metamode = "override";
|
|---|
| 2872 |
|
|---|
| 2873 | # Obtain where the metadata.xml is from the archiveinfo-doc.gdb file
|
|---|
| 2874 | # If the doc oid is not specified, we assume the metadata.xml is next to the specified "f"
|
|---|
| 2875 | my $metadata_xml_file;
|
|---|
| 2876 | my $import_filename = undef;
|
|---|
| 2877 | if (defined $docid)
|
|---|
| 2878 | {
|
|---|
| 2879 | my $arcinfo_doc_filename = &dbutil::get_infodb_file_path($infodbtype, "archiveinf-doc", $archive_dir);
|
|---|
| 2880 | my $doc_rec = &dbutil::read_infodb_entry($infodbtype, $arcinfo_doc_filename, $docid);
|
|---|
| 2881 |
|
|---|
| 2882 | # This now stores the full pathname
|
|---|
| 2883 | $import_filename = $doc_rec->{'src-file'}->[0];
|
|---|
| 2884 | $import_filename = &util::placeholders_to_abspath($import_filename);
|
|---|
| 2885 | }
|
|---|
| 2886 |
|
|---|
| 2887 | if((!defined $import_filename) || ($import_filename =~ m/^\s*$/))
|
|---|
| 2888 | {
|
|---|
| 2889 | $gsdl_cgi->generate_error("There is no metadata\n");
|
|---|
| 2890 | }
|
|---|
| 2891 |
|
|---|
| 2892 | # figure out correct metadata.xml file [?]
|
|---|
| 2893 | # Assuming the metadata.xml file is next to the source file
|
|---|
| 2894 | # Note: This will not work if it is using the inherited metadata from the parent folder
|
|---|
| 2895 | my ($import_tailname, $import_dirname) = File::Basename::fileparse($import_filename);
|
|---|
| 2896 | my $metadata_xml_filename = &util::filename_cat($import_dirname,"metadata.xml");
|
|---|
| 2897 |
|
|---|
| 2898 | $self->remove_from_metadata_xml($gsdl_cgi, $metadata_xml_filename, $metaname, $metapos, $metavalue, $import_tailname, $metamode); # metamode=override, means all values of metaname are affected
|
|---|
| 2899 |
|
|---|
| 2900 | my $mess = "erase-import-metadata successful: Key[$docid] -> $metadata_xml_filename\n";
|
|---|
| 2901 | $mess .= " $metaname";
|
|---|
| 2902 | $mess .= " = $metavalue\n";
|
|---|
| 2903 |
|
|---|
| 2904 | $gsdl_cgi->generate_ok_message($mess);
|
|---|
| 2905 |
|
|---|
| 2906 | #return $status; # in case calling functions have a use for this
|
|---|
| 2907 | }
|
|---|
| 2908 |
|
|---|
| 2909 | sub erase_import_metadata
|
|---|
| 2910 | {
|
|---|
| 2911 | my $self = shift @_;
|
|---|
| 2912 |
|
|---|
| 2913 | my $username = $self->{'username'};
|
|---|
| 2914 | my $collect = $self->{'collect'};
|
|---|
| 2915 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
|---|
| 2916 |
|
|---|
| 2917 | if ($baseaction::authentication_enabled) {
|
|---|
| 2918 | # Ensure the user is allowed to edit this collection
|
|---|
| 2919 | $self->authenticate_user($username, $collect); #&authenticate_user($gsdl_cgi, $username, $collect);
|
|---|
| 2920 | }
|
|---|
| 2921 |
|
|---|
| 2922 | # Make sure the collection isn't locked by someone else
|
|---|
| 2923 | $self->lock_collection($username, $collect);
|
|---|
| 2924 |
|
|---|
| 2925 | $self->_erase_import_metadata(@_);
|
|---|
| 2926 |
|
|---|
| 2927 | # Release the lock once it is done
|
|---|
| 2928 | $self->unlock_collection($username, $collect);
|
|---|
| 2929 |
|
|---|
| 2930 | }
|
|---|
| 2931 |
|
|---|
| 2932 | # removes all meta with matching metaname for the specified docid from the live metadata index
|
|---|
| 2933 | sub _erase_live_metadata
|
|---|
| 2934 | {
|
|---|
| 2935 | my $self = shift @_;
|
|---|
| 2936 | # At present, _remove_live_metadata doesn't take metaval or metapos either and therefore has the desired effect
|
|---|
| 2937 | $self->_remove_live_metadata(@_);
|
|---|
| 2938 |
|
|---|
| 2939 | }
|
|---|
| 2940 |
|
|---|
| 2941 | sub erase_live_metadata
|
|---|
| 2942 | {
|
|---|
| 2943 | my $self = shift @_;
|
|---|
| 2944 |
|
|---|
| 2945 | my $username = $self->{'username'};
|
|---|
| 2946 | my $collect = $self->{'collect'};
|
|---|
| 2947 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
|---|
| 2948 |
|
|---|
| 2949 | if ($baseaction::authentication_enabled) {
|
|---|
| 2950 | # Ensure the user is allowed to edit this collection
|
|---|
| 2951 | $self->authenticate_user($username, $collect); #&authenticate_user($gsdl_cgi, $username, $collect);
|
|---|
| 2952 | }
|
|---|
| 2953 |
|
|---|
| 2954 | # Make sure the collection isn't locked by someone else
|
|---|
| 2955 | $self->lock_collection($username, $collect);
|
|---|
| 2956 |
|
|---|
| 2957 | $self->_erase_live_metadata(@_);
|
|---|
| 2958 |
|
|---|
| 2959 | $self->unlock_collection($username, $collect);
|
|---|
| 2960 | }
|
|---|
| 2961 |
|
|---|
| 2962 |
|
|---|
| 2963 | # Was trying to reused the codes, but the functions need to be broken
|
|---|
| 2964 | # down more before they can be reused, otherwise there will be too
|
|---|
| 2965 | # much overhead and duplicate process...
|
|---|
| 2966 | sub insert_metadata
|
|---|
| 2967 | {
|
|---|
| 2968 | my $self = shift @_;
|
|---|
| 2969 |
|
|---|
| 2970 | my $username = $self->{'username'};
|
|---|
| 2971 | my $collect = $self->{'collect'};
|
|---|
| 2972 | my $gsdl_cgi = $self->{'gsdl_cgi'};
|
|---|
| 2973 | my $infodbtype = $self->{'infodbtype'};
|
|---|
| 2974 |
|
|---|
| 2975 | # If the import metadata and gdbm database have been updated, we
|
|---|
| 2976 | # need to insert some notification to warn user that the the text
|
|---|
| 2977 | # they see at the moment is not indexed and require a rebuild.
|
|---|
| 2978 | my $rebuild_pending_macro = "_rebuildpendingmessage_";
|
|---|
| 2979 |
|
|---|
| 2980 | if ($baseaction::authentication_enabled) {
|
|---|
| 2981 | # Ensure the user is allowed to edit this collection
|
|---|
| 2982 | $self->authenticate_user($username, $collect);
|
|---|
| 2983 | }
|
|---|
| 2984 |
|
|---|
| 2985 | # Obtain the collect and archive dir
|
|---|
| 2986 | my $site = $self->{'site'};
|
|---|
| 2987 | my $collect_dir = $gsdl_cgi->get_collection_dir($site);
|
|---|
| 2988 | my $archive_dir = &util::filename_cat($collect_dir,$collect,"archives");
|
|---|
| 2989 |
|
|---|
| 2990 | # Make sure the collection isn't locked by someone else
|
|---|
| 2991 | $self->lock_collection($username, $collect);
|
|---|
| 2992 |
|
|---|
| 2993 | # Check additional args
|
|---|
| 2994 | my $docid = $self->{'d'};
|
|---|
| 2995 | if (!defined($docid) || ($docid =~ m/^\s*$/)) {
|
|---|
| 2996 | $gsdl_cgi->generate_error("No document id is specified: d=...");
|
|---|
| 2997 | }
|
|---|
| 2998 | my $metaname = $self->{'metaname'};
|
|---|
| 2999 | if (!defined($metaname) || ($metaname =~ m/^\s*$/)) {
|
|---|
| 3000 | $gsdl_cgi->generate_error("No metaname is specified: metadataname=...");
|
|---|
| 3001 | }
|
|---|
| 3002 | my $metavalue = $self->{'metavalue'};
|
|---|
| 3003 | if (!defined($metavalue) || ($metavalue =~ m/^\s*$/)) {
|
|---|
| 3004 | $gsdl_cgi->generate_error("No metavalue or empty metavalue is specified: metadataname=...");
|
|---|
| 3005 | }
|
|---|
| 3006 | # make "accumulate" the default (less destructive, as it won't actually
|
|---|
| 3007 | # delete any existing values)
|
|---|
| 3008 | my $metamode = "accumulate";
|
|---|
| 3009 |
|
|---|
| 3010 | # metapos/prevmetavalue were never before used in this subroutine, so set them to undefined
|
|---|
| 3011 | my $metapos = undef;
|
|---|
| 3012 | my $prevmetavalue = undef;
|
|---|
| 3013 |
|
|---|
| 3014 | #=======================================================================#
|
|---|
| 3015 | # set_import_metadata [START]
|
|---|
| 3016 | #=======================================================================#
|
|---|
| 3017 | # Obtain where the metadata.xml is from the archiveinfo-doc.gdb file
|
|---|
| 3018 | # If the doc oid is not specified, we assume the metadata.xml is next to the specified "f"
|
|---|
| 3019 | my $metadata_xml_file;
|
|---|
| 3020 | my $arcinfo_doc_filename = &dbutil::get_infodb_file_path($infodbtype, "archiveinf-doc", $archive_dir);
|
|---|
| 3021 | my $archive_doc_rec = &dbutil::read_infodb_entry($infodbtype, $arcinfo_doc_filename, $docid);
|
|---|
| 3022 |
|
|---|
| 3023 | # This now stores the full pathname
|
|---|
| 3024 | my $import_filename = $archive_doc_rec->{'src-file'}->[0];
|
|---|
| 3025 | $import_filename = &util::placeholders_to_abspath($import_filename);
|
|---|
| 3026 |
|
|---|
| 3027 | # figure out correct metadata.xml file [?]
|
|---|
| 3028 | # Assuming the metadata.xml file is next to the source file
|
|---|
| 3029 | # Note: This will not work if it is using the inherited metadata from the parent folder
|
|---|
| 3030 | my ($import_tailname, $import_dirname)
|
|---|
| 3031 | = File::Basename::fileparse($import_filename);
|
|---|
| 3032 | my $metadata_xml_filename = &util::filename_cat($import_dirname,"metadata.xml");
|
|---|
| 3033 |
|
|---|
| 3034 | # Shane's escape characters
|
|---|
| 3035 | $metavalue = pack "U0C*", unpack "C*", $metavalue;
|
|---|
| 3036 | $metavalue =~ s/\,/,/g;
|
|---|
| 3037 | $metavalue =~ s/\:/:/g;
|
|---|
| 3038 | $metavalue =~ s/\|/|/g;
|
|---|
| 3039 | $metavalue =~ s/\(/(/g;
|
|---|
| 3040 | $metavalue =~ s/\)/)/g;
|
|---|
| 3041 | $metavalue =~ s/\[/[/g;
|
|---|
| 3042 | $metavalue =~ s/\\/\/g;
|
|---|
| 3043 | $metavalue =~ s/\]/]/g;
|
|---|
| 3044 | $metavalue =~ s/\{/{/g;
|
|---|
| 3045 | $metavalue =~ s/\}/}/g;
|
|---|
| 3046 | $metavalue =~ s/\"/"/g;
|
|---|
| 3047 | $metavalue =~ s/\`/`/g;
|
|---|
| 3048 | $metavalue =~ s/\n/_newline_/g;
|
|---|
| 3049 |
|
|---|
| 3050 | # Edit the metadata.xml
|
|---|
| 3051 | # Modified by Jeffrey from DL Consulting
|
|---|
| 3052 | # Handle the case where there is one metadata.xml file for multiple FileSets
|
|---|
| 3053 | # The XML filter needs to know whether it is in the right FileSet
|
|---|
| 3054 | # TODO: This doesn't fix the problem where the metadata.xml is not next to the src file.
|
|---|
| 3055 | # TODO: This doesn't handle the common metadata (where FileName doesn't point to a single file)
|
|---|
| 3056 | $self->edit_metadata_xml($gsdl_cgi, $metadata_xml_filename, $metaname,
|
|---|
| 3057 | $metapos, $metavalue, $metamode, $import_tailname, $prevmetavalue);
|
|---|
| 3058 | #=======================================================================#
|
|---|
| 3059 | # set_import_metadata [END]
|
|---|
| 3060 | #=======================================================================#
|
|---|
| 3061 |
|
|---|
| 3062 |
|
|---|
| 3063 | #=======================================================================#
|
|---|
| 3064 | # set_metadata (accumulate version) [START]
|
|---|
| 3065 | #=======================================================================#
|
|---|
| 3066 | # To people who know $collect_tail please add some comments
|
|---|
| 3067 | # Obtain path to the database
|
|---|
| 3068 | my $collect_tail = $collect;
|
|---|
| 3069 | $collect_tail =~ s/^.*[\/|\\]//;
|
|---|
| 3070 | my $index_text_directory = &util::filename_cat($collect_dir,$collect,"index","text");
|
|---|
| 3071 | my $infodb_file_path = &dbutil::get_infodb_file_path($infodbtype, $collect_tail, $index_text_directory);
|
|---|
| 3072 |
|
|---|
| 3073 | # Read the docid entry
|
|---|
| 3074 | my $doc_rec = &dbutil::read_infodb_entry($infodbtype, $infodb_file_path, $docid);
|
|---|
| 3075 |
|
|---|
| 3076 | # Protect the quotes
|
|---|
| 3077 | $metavalue =~ s/\"/\\\"/g;
|
|---|
| 3078 |
|
|---|
| 3079 | # Adds the pending macro
|
|---|
| 3080 | my $macro_metavalue = $rebuild_pending_macro . $metavalue;
|
|---|
| 3081 |
|
|---|
| 3082 | # If the metadata doesn't exist, create a new one
|
|---|
| 3083 | if (!defined($doc_rec->{$metaname})){
|
|---|
| 3084 | $doc_rec->{$metaname} = [ $macro_metavalue ];
|
|---|
| 3085 | }
|
|---|
| 3086 | # Else, let's acculumate the values
|
|---|
| 3087 | else {
|
|---|
| 3088 | push(@{$doc_rec->{$metaname}},$macro_metavalue);
|
|---|
| 3089 | }
|
|---|
| 3090 |
|
|---|
| 3091 | ## Use the dbutil set_entry method instead of assuming the database is gdbm
|
|---|
| 3092 | my $status = &dbutil::set_infodb_entry($infodbtype, $infodb_file_path, $docid, $doc_rec);
|
|---|
| 3093 |
|
|---|
| 3094 | if ($status != 0) {
|
|---|
| 3095 | # Catch error if gdbmget failed
|
|---|
| 3096 | my $mess = "Failed to set metadata key: $docid\n";
|
|---|
| 3097 |
|
|---|
| 3098 | $mess .= "PATH: $ENV{'PATH'}\n";
|
|---|
| 3099 | $mess .= "Exit status: $status\n";
|
|---|
| 3100 | $mess .= "System Error Message: $!\n";
|
|---|
| 3101 |
|
|---|
| 3102 | $gsdl_cgi->generate_error($mess);
|
|---|
| 3103 | }
|
|---|
| 3104 | else {
|
|---|
| 3105 | my $mess = "insert-metadata successful: Key[$docid]\n";
|
|---|
| 3106 | $mess .= " [In metadata.xml] $metaname";
|
|---|
| 3107 | $mess .= " = $metavalue\n";
|
|---|
| 3108 | $mess .= " [In database] $metaname";
|
|---|
| 3109 | $mess .= " = $macro_metavalue\n";
|
|---|
| 3110 | $mess .= " The new text has not been indexed, rebuilding collection is required\n";
|
|---|
| 3111 | $gsdl_cgi->generate_ok_message($mess);
|
|---|
| 3112 | }
|
|---|
| 3113 | #=======================================================================#
|
|---|
| 3114 | # set_metadata (accumulate version) [END]
|
|---|
| 3115 | #=======================================================================#
|
|---|
| 3116 |
|
|---|
| 3117 | # Release the lock once it is done
|
|---|
| 3118 | $self->unlock_collection($username, $collect);
|
|---|
| 3119 | }
|
|---|
| 3120 |
|
|---|
| 3121 | # not returning 1; here since this file is conditionally included by metadataction.pm
|
|---|
| 3122 | # and not otherwise meant to be used on its own
|
|---|
| 3123 |
|
|---|
