[4726] | 1 | ###########################################################################
|
---|
| 2 | #
|
---|
| 3 | # OAIPlug.pm -- basic Open Archives Initiative (OAI) plugin
|
---|
| 4 | #
|
---|
| 5 | # A component of the Greenstone digital library software
|
---|
| 6 | # from the New Zealand Digital Library Project at the
|
---|
| 7 | # University of Waikato, New Zealand.
|
---|
| 8 | #
|
---|
| 9 | # Copyright (C) 1999 New Zealand Digital Library Project
|
---|
| 10 | #
|
---|
| 11 | # This program is free software; you can redistribute it and/or modify
|
---|
| 12 | # it under the terms of the GNU General Public License as published by
|
---|
| 13 | # the Free Software Foundation; either version 2 of the License, or
|
---|
| 14 | # (at your option) any later version.
|
---|
| 15 | #
|
---|
| 16 | # This program is distributed in the hope that it will be useful,
|
---|
| 17 | # but WITHOUT ANY WARRANTY; without even the implied warranty of
|
---|
| 18 | # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
---|
| 19 | # GNU General Public License for more details.
|
---|
| 20 | #
|
---|
| 21 | # You should have received a copy of the GNU General Public License
|
---|
| 22 | # along with this program; if not, write to the Free Software
|
---|
| 23 | # Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
|
---|
| 24 | #
|
---|
| 25 | ###########################################################################
|
---|
| 26 |
|
---|
[15872] | 27 | package OAIPlugin;
|
---|
[4726] | 28 |
|
---|
| 29 | use unicode;
|
---|
| 30 | use util;
|
---|
| 31 |
|
---|
[10254] | 32 | use strict;
|
---|
| 33 | no strict 'refs'; # allow filehandles to be variables and viceversa
|
---|
| 34 |
|
---|
[15872] | 35 | use ReadXMLFile;
|
---|
[17066] | 36 | use ReadTextFile; # needed for subroutine textcat_get_language_encoding
|
---|
[17216] | 37 | use metadatautil;
|
---|
[9958] | 38 |
|
---|
[4726] | 39 | sub BEGIN {
|
---|
[17066] | 40 | @OAIPlugin::ISA = ('ReadXMLFile', 'ReadTextFile');
|
---|
[4726] | 41 | }
|
---|
| 42 |
|
---|
[19213] | 43 | my $set_list =
|
---|
| 44 | [ { 'name' => "auto",
|
---|
| 45 | 'desc' => "{OAIPlugin.metadata_set.auto}" },
|
---|
| 46 | { 'name' => "dc",
|
---|
| 47 | 'desc' => "{OAIPlugin.metadata_set.dc}" }
|
---|
| 48 | ];
|
---|
[9958] | 49 |
|
---|
[6408] | 50 | my $arguments =
|
---|
| 51 | [ { 'name' => "process_exp",
|
---|
[16013] | 52 | 'desc' => "{BasePlugin.process_exp}",
|
---|
[6408] | 53 | 'type' => "regexp",
|
---|
| 54 | 'reqd' => "no",
|
---|
[17290] | 55 | 'deft' => &get_default_process_exp() },
|
---|
[19213] | 56 | { 'name' => "metadata_set",
|
---|
| 57 | 'desc' => "{OAIPlugin.metadata_set}",
|
---|
| 58 | 'type' => "enumstring",
|
---|
| 59 | 'reqd' => "no",
|
---|
| 60 | 'list' => $set_list,
|
---|
| 61 | 'deft' => "dc" },
|
---|
[17319] | 62 | { 'name' => "document_field",
|
---|
| 63 | 'desc' => "{OAIPlugin.document_field}",
|
---|
[17290] | 64 | 'type' => "metadata",
|
---|
| 65 | 'reqd' => "no",
|
---|
| 66 | 'deft' => "gi.Sourcedoc" }
|
---|
[6408] | 67 | ];
|
---|
| 68 |
|
---|
[15872] | 69 | my $options = { 'name' => "OAIPlugin",
|
---|
| 70 | 'desc' => "{OAIPlugin.desc}",
|
---|
[6408] | 71 | 'abstract' => "no",
|
---|
| 72 | 'inherits' => "yes",
|
---|
[17103] | 73 | 'explodes' => "yes",
|
---|
[6408] | 74 | 'args' => $arguments };
|
---|
[4747] | 75 |
|
---|
[10254] | 76 |
|
---|
[4726] | 77 | sub new {
|
---|
[10218] | 78 | my ($class) = shift (@_);
|
---|
| 79 | my ($pluginlist,$inputargs,$hashArgOptLists) = @_;
|
---|
| 80 | push(@$pluginlist, $class);
|
---|
[4873] | 81 |
|
---|
[15872] | 82 | push(@{$hashArgOptLists->{"ArgList"}},@{$arguments});
|
---|
| 83 | push(@{$hashArgOptLists->{"OptList"}},$options);
|
---|
[4726] | 84 |
|
---|
[17126] | 85 | new ReadTextFile($pluginlist, $inputargs, $hashArgOptLists,1);
|
---|
[15872] | 86 | my $self = new ReadXMLFile($pluginlist, $inputargs, $hashArgOptLists);
|
---|
[4726] | 87 |
|
---|
| 88 | return bless $self, $class;
|
---|
| 89 | }
|
---|
| 90 |
|
---|
| 91 | sub get_default_process_exp {
|
---|
| 92 | my $self = shift (@_);
|
---|
| 93 |
|
---|
| 94 | return q^(?i)(\.oai)$^;
|
---|
| 95 | }
|
---|
| 96 |
|
---|
[13222] | 97 | sub get_doctype {
|
---|
| 98 | my $self = shift(@_);
|
---|
| 99 |
|
---|
| 100 | return "OAI-PMH";
|
---|
| 101 | }
|
---|
| 102 |
|
---|
[9958] | 103 | sub xml_start_document {
|
---|
[10254] | 104 | my $self = shift (@_);
|
---|
[9958] | 105 | $self->{'in_metadata_node'} = 0;
|
---|
| 106 | $self->{'rawxml'} = "";
|
---|
[17290] | 107 | $self->{'saved_metadata'} = {};
|
---|
[9958] | 108 | }
|
---|
[4726] | 109 |
|
---|
[9958] | 110 | sub xml_end_document {
|
---|
| 111 | }
|
---|
[4726] | 112 |
|
---|
[9958] | 113 | sub xml_doctype {
|
---|
| 114 | my $self = shift(@_);
|
---|
| 115 |
|
---|
| 116 | my ($expat, $name, $sysid, $pubid, $internal) = @_;
|
---|
| 117 |
|
---|
[13886] | 118 | ##die "" if ($name !~ /^OAI-PMH$/);
|
---|
[9958] | 119 |
|
---|
[4726] | 120 | my $outhandle = $self->{'outhandle'};
|
---|
[15872] | 121 | print $outhandle "OAIPlugin: processing $self->{'file'}\n" if $self->{'verbosity'} > 1;
|
---|
| 122 | print STDERR "<Processing n='$self->{'file'}' p='OAIPlugin'>\n" if $self->{'gli'};
|
---|
[4726] | 123 |
|
---|
[9958] | 124 | }
|
---|
[4726] | 125 |
|
---|
[9958] | 126 |
|
---|
| 127 | sub xml_start_tag {
|
---|
| 128 | my $self = shift(@_);
|
---|
| 129 | my ($expat,$element) = @_;
|
---|
| 130 |
|
---|
| 131 | my %attr_hash = %_;
|
---|
| 132 |
|
---|
| 133 | my $attr = "";
|
---|
| 134 | map { $attr .= " $_=$attr_hash{$_}"; } keys %attr_hash;
|
---|
| 135 |
|
---|
| 136 | $self->{'rawxml'} .= "<$element$attr>";
|
---|
| 137 |
|
---|
| 138 | if ($element eq "metadata") {
|
---|
| 139 | $self->{'in_metadata_node'} = 1;
|
---|
| 140 | $self->{'metadata_xml'} = "";
|
---|
[4726] | 141 | }
|
---|
[9958] | 142 |
|
---|
| 143 | if ($self->{'in_metadata_node'}) {
|
---|
| 144 | $self->{'metadata_xml'} .= "<$element$attr>";
|
---|
[4726] | 145 | }
|
---|
[9958] | 146 | }
|
---|
[4726] | 147 |
|
---|
[9958] | 148 | sub xml_end_tag {
|
---|
| 149 | my $self = shift(@_);
|
---|
| 150 | my ($expat, $element) = @_;
|
---|
[4726] | 151 |
|
---|
[9958] | 152 | $self->{'rawxml'} .= "</$element>";
|
---|
[4726] | 153 |
|
---|
[9958] | 154 | if ($self->{'in_metadata_node'}) {
|
---|
| 155 | $self->{'metadata_xml'} .= "</$element>";
|
---|
[4726] | 156 | }
|
---|
| 157 |
|
---|
[9958] | 158 | if ($element eq "metadata") {
|
---|
| 159 | my $textref = \$self->{'metadata_xml'};
|
---|
[17290] | 160 | #my $metadata = $self->{'metadata'};
|
---|
| 161 | my $metadata = $self->{'saved_metadata'};
|
---|
[9958] | 162 | $self->extract_oai_metadata($textref,$metadata);
|
---|
[4726] | 163 |
|
---|
[9958] | 164 | $self->{'in_metadata_node'} = 0;
|
---|
| 165 | }
|
---|
[4726] | 166 |
|
---|
| 167 |
|
---|
[9958] | 168 | }
|
---|
[4726] | 169 |
|
---|
[9958] | 170 | sub xml_text {
|
---|
| 171 | my $self = shift(@_);
|
---|
| 172 | my ($expat) = @_;
|
---|
[8684] | 173 |
|
---|
[9958] | 174 | $self->{'rawxml'} .= $_;
|
---|
[4726] | 175 |
|
---|
[9958] | 176 | if ($self->{'in_metadata_node'}) {
|
---|
| 177 | $self->{'metadata_xml'} .= $_;
|
---|
[4726] | 178 | }
|
---|
[9958] | 179 | }
|
---|
[4726] | 180 |
|
---|
[8121] | 181 |
|
---|
[17216] | 182 | sub metadata_read {
|
---|
| 183 | my $self = shift (@_);
|
---|
[4726] | 184 |
|
---|
[19493] | 185 | my ($pluginfo, $base_dir, $file, $block_hash,
|
---|
| 186 | $extrametakeys, $extrametadata, $extrametafile,
|
---|
| 187 | $processor, $maxdocs, $gli) = @_;
|
---|
[5919] | 188 |
|
---|
[17216] | 189 | # can we process this file??
|
---|
| 190 | my ($filename_full_path, $filename_no_path) = &util::get_full_filenames($base_dir, $file);
|
---|
| 191 | return undef unless $self->can_process_this_file($filename_full_path);
|
---|
| 192 |
|
---|
[17290] | 193 | if (!$self->parse_file($filename_full_path, $file, $gli)) {
|
---|
| 194 | $self->{'saved_metadata'} = undef;
|
---|
| 195 | return undef;
|
---|
| 196 | }
|
---|
| 197 |
|
---|
| 198 | my $new_metadata = $self->{'saved_metadata'};
|
---|
| 199 | $self->{'saved_metadata'} = undef;
|
---|
[17319] | 200 |
|
---|
[17290] | 201 | # add the pretty metadata table as metadata
|
---|
| 202 | my $ppmd_table = $self->{'ppmd_table'};
|
---|
| 203 | $new_metadata->{'prettymd'} = $ppmd_table;
|
---|
| 204 | $self->{'ppmd_table'} = undef;
|
---|
[17319] | 205 |
|
---|
| 206 | my $document_metadata_field = $self->{'document_field'};
|
---|
| 207 | my $url_array = $new_metadata->{$document_metadata_field};
|
---|
[17290] | 208 | my $num_urls = (defined $url_array) ? scalar(@$url_array) : 0;
|
---|
[17319] | 209 | ##print STDERR "$num_urls urls for $file\n";
|
---|
[17290] | 210 | my $srcdoc_exists = 0;
|
---|
| 211 | my $srcdoc_pos = 0;
|
---|
| 212 | my $filename_dir = &util::filename_head($filename_full_path);
|
---|
[17590] | 213 |
|
---|
[17591] | 214 | # filenames in extrametadata must be relative to current dir, as
|
---|
| 215 | # DirectoryPlugin adds path info on itself
|
---|
[17590] | 216 | my ($filename_for_metadata) = $file =~ /([^\\\/]+)$/; # this assumes there will only be one record per oai file - is this always the case??
|
---|
[17290] | 217 | for (my $i=0; $i<$num_urls; $i++) {
|
---|
[17216] | 218 |
|
---|
[17290] | 219 | if ($url_array->[$i] !~ m/^(https?|ftp):/) {
|
---|
[17216] | 220 |
|
---|
[17290] | 221 | my $src_filename = &util::filename_cat($filename_dir, $url_array->[$i]);
|
---|
[17216] | 222 |
|
---|
[17290] | 223 | if (-e $src_filename) {
|
---|
| 224 | $srcdoc_pos = $i;
|
---|
| 225 | $srcdoc_exists = 1;
|
---|
| 226 | $filename_for_metadata = $url_array->[$i];
|
---|
| 227 | last;
|
---|
[17216] | 228 | }
|
---|
| 229 | }
|
---|
| 230 | }
|
---|
[17290] | 231 |
|
---|
[17319] | 232 | if ($srcdoc_exists) {
|
---|
[17290] | 233 | $self->{'oai-files'}->{$file}->{'srcdoc_exists'} = 1;
|
---|
| 234 | }
|
---|
[17216] | 235 | else {
|
---|
[17290] | 236 | # save the rawxml for the source document
|
---|
| 237 | $self->{'oai-files'}->{$file}->{'srcdoc_exists'} = 0;
|
---|
| 238 | $self->{'oai-files'}->{$file}->{'rawxml'} = $self->{'rawxml'};
|
---|
| 239 | $self->{'rawxml'} = "";
|
---|
[17216] | 240 | }
|
---|
[17290] | 241 |
|
---|
| 242 | # return all the metadata we have extracted to the caller.
|
---|
| 243 | # Directory plug will pass it back in at read time, so we don't need to extract it again.
|
---|
[17513] | 244 | # extrametadata keys should be regular expressions
|
---|
| 245 | $filename_for_metadata = &util::filename_to_regex($filename_for_metadata);
|
---|
[17290] | 246 | $extrametadata->{$filename_for_metadata} = $new_metadata;
|
---|
| 247 | push(@$extrametakeys, $filename_for_metadata);
|
---|
| 248 |
|
---|
| 249 | return 1;
|
---|
| 250 |
|
---|
[17216] | 251 | }
|
---|
| 252 |
|
---|
| 253 |
|
---|
[9958] | 254 | sub read {
|
---|
| 255 | my $self = shift (@_);
|
---|
[17290] | 256 |
|
---|
[16392] | 257 | my ($pluginfo, $base_dir, $file, $block_hash, $metadata, $processor, $maxdocs, $total_count, $gli) = @_;
|
---|
[4726] | 258 |
|
---|
[17290] | 259 | if (!defined $self->{'oai-files'}->{$file}) {
|
---|
| 260 | return undef;
|
---|
| 261 | }
|
---|
[17319] | 262 |
|
---|
[17290] | 263 | my $srcdoc_exists = $self->{'oai-files'}->{$file}->{'srcdoc_exists'};
|
---|
| 264 | if ($srcdoc_exists) {
|
---|
[17319] | 265 | # do nothing more - all the metadata has been extracted and associated with the srcdoc
|
---|
[17216] | 266 | # no more need to access details of this $file => tidy up as you go
|
---|
| 267 | delete $self->{'oai-files'}->{$file};
|
---|
[17290] | 268 | return 0; # not processed here, but don't pass on to rest of plugins
|
---|
| 269 | }
|
---|
[17216] | 270 |
|
---|
[17290] | 271 | my $filename = $file;
|
---|
| 272 | $filename = &util::filename_cat ($base_dir, $file) if $base_dir =~ /\w/;
|
---|
[17216] | 273 |
|
---|
[17290] | 274 | # Do encoding stuff on metadata
|
---|
| 275 | my ($language, $encoding) = $self->textcat_get_language_encoding ($filename);
|
---|
[17216] | 276 |
|
---|
[17290] | 277 | # create a new document
|
---|
[18327] | 278 | my $doc_obj = new doc ($filename, "indexed_doc", $self->{'file_rename_method'});
|
---|
[17290] | 279 | my $top_section = $doc_obj->get_top_section;
|
---|
| 280 | my $plugin_type = $self->{'plugin_type'};
|
---|
| 281 |
|
---|
[17588] | 282 | my ($filemeta) = $file =~ /([^\\\/]+)$/;
|
---|
| 283 | $self->set_Source_metadata($doc_obj, $filemeta, $encoding);
|
---|
[17290] | 284 | $doc_obj->add_utf8_metadata($top_section, "Language", $language);
|
---|
| 285 | $doc_obj->add_utf8_metadata($top_section, "Encoding", $encoding);
|
---|
| 286 | $doc_obj->add_utf8_metadata($top_section, "Plugin", $plugin_type);
|
---|
| 287 | $doc_obj->add_metadata($top_section, "FileFormat", "OAI");
|
---|
| 288 | $doc_obj->add_metadata($top_section, "FileSize", (-s $filename));
|
---|
| 289 |
|
---|
| 290 | # include any metadata passed in from previous plugins
|
---|
| 291 | # note that this metadata is associated with the top level section
|
---|
[17319] | 292 | # this will include all the metadata from the oai file that we extracted
|
---|
| 293 | # during metadata_read
|
---|
[17290] | 294 | $self->extra_metadata ($doc_obj, $doc_obj->get_top_section(), $metadata);
|
---|
| 295 |
|
---|
| 296 | # do plugin specific processing of doc_obj
|
---|
| 297 | my $text = $self->{'oai-files'}->{$file}->{'rawxml'};
|
---|
| 298 | delete $self->{'oai-files'}->{$file};
|
---|
[17216] | 299 |
|
---|
[17290] | 300 | unless (defined ($self->process(\$text, $pluginfo, $base_dir, $file, $metadata, $doc_obj))) {
|
---|
| 301 | print STDERR "<ProcessingError n='$file'>\n" if ($gli);
|
---|
| 302 | return -1;
|
---|
[17216] | 303 | }
|
---|
[17290] | 304 |
|
---|
| 305 | # do any automatic metadata extraction
|
---|
| 306 | $self->auto_extract_metadata ($doc_obj);
|
---|
| 307 |
|
---|
| 308 | # add an OID
|
---|
| 309 | $self->add_OID($doc_obj);
|
---|
| 310 |
|
---|
| 311 | # process the document
|
---|
| 312 | $processor->process($doc_obj);
|
---|
| 313 |
|
---|
| 314 | $self->{'num_processed'} ++;
|
---|
| 315 |
|
---|
| 316 | return 1; # processed the file
|
---|
[17216] | 317 | }
|
---|
| 318 |
|
---|
| 319 |
|
---|
[4726] | 320 | # do plugin specific processing of doc_obj
|
---|
| 321 | sub process {
|
---|
| 322 | my $self = shift (@_);
|
---|
[6332] | 323 | my ($textref, $pluginfo, $base_dir, $file, $metadata, $doc_obj, $gli) = @_;
|
---|
[4726] | 324 | my $outhandle = $self->{'outhandle'};
|
---|
| 325 |
|
---|
[15872] | 326 | print STDERR "<Processing n='$file' p='OAIPlugin'>\n" if ($gli);
|
---|
| 327 | print $outhandle "OAIPlugin: processing $file\n"
|
---|
[4726] | 328 | if $self->{'verbosity'} > 1;
|
---|
| 329 |
|
---|
| 330 | my $cursection = $doc_obj->get_top_section();
|
---|
| 331 |
|
---|
| 332 | ## $self->extract_metadata ($textref, $metadata, $doc_obj, $cursection);
|
---|
| 333 |
|
---|
| 334 | # add text to document object
|
---|
| 335 |
|
---|
| 336 | # $$textref =~ s/<(.*?)>/$1 /g;
|
---|
| 337 | $$textref =~ s/</</g;
|
---|
| 338 | $$textref =~ s/>/>/g;
|
---|
[14963] | 339 | $$textref =~ s/\[/[/g;
|
---|
| 340 | $$textref =~ s/\]/]/g;
|
---|
[4726] | 341 |
|
---|
| 342 | $doc_obj->add_utf8_text($cursection, $$textref);
|
---|
| 343 |
|
---|
| 344 | return 1;
|
---|
| 345 | }
|
---|
| 346 |
|
---|
| 347 |
|
---|
[9958] | 348 | # Improvement is to merge this with newer version in MetadataPass
|
---|
[4726] | 349 |
|
---|
[9958] | 350 | sub open_prettyprint_metadata_table
|
---|
| 351 | {
|
---|
| 352 | my $self = shift(@_);
|
---|
| 353 |
|
---|
| 354 | my $att = "width=100% cellspacing=2";
|
---|
| 355 | my $style = "style=\'border-bottom: 4px solid #000080\'";
|
---|
| 356 |
|
---|
[17290] | 357 | $self->{'ppmd_table'} = "\n<table $att $style>";
|
---|
[9958] | 358 | }
|
---|
| 359 |
|
---|
| 360 | sub add_prettyprint_metadata_line
|
---|
| 361 | {
|
---|
| 362 | my $self = shift(@_);
|
---|
| 363 | my ($metaname, $metavalue_utf8) = @_;
|
---|
| 364 |
|
---|
| 365 | $metavalue_utf8 = &util::hyperlink_text($metavalue_utf8);
|
---|
| 366 |
|
---|
| 367 | $self->{'ppmd_table'} .= " <tr bgcolor=#b5d3cd>\n";
|
---|
| 368 | $self->{'ppmd_table'} .= " <td width=30%>\n";
|
---|
| 369 | $self->{'ppmd_table'} .= " $metaname\n";
|
---|
| 370 | $self->{'ppmd_table'} .= " </td>\n";
|
---|
| 371 | $self->{'ppmd_table'} .= " <td>\n";
|
---|
| 372 | $self->{'ppmd_table'} .= " $metavalue_utf8\n";
|
---|
| 373 | $self->{'ppmd_table'} .= " </td>\n";
|
---|
| 374 | $self->{'ppmd_table'} .= " </tr>\n";
|
---|
| 375 |
|
---|
| 376 | }
|
---|
| 377 |
|
---|
| 378 | sub close_prettyprint_metadata_table
|
---|
| 379 | {
|
---|
| 380 | my $self = shift(@_);
|
---|
| 381 |
|
---|
| 382 | $self->{'ppmd_table'} .= "</table>\n";
|
---|
| 383 | }
|
---|
| 384 |
|
---|
[19213] | 385 | my $qualified_dc_mapping = {
|
---|
| 386 | "alternative" => "dc.title",
|
---|
| 387 | "tableOfContents" => "dc.description",
|
---|
| 388 | "abstract" => "dc.description",
|
---|
| 389 | "created" => "dc.date",
|
---|
| 390 | "valid" => "dc.date",
|
---|
| 391 | "available" => "dc.date",
|
---|
| 392 | "issued" => "dc.date",
|
---|
| 393 | "modified" => "dc.date",
|
---|
| 394 | "dateAccepted" => "dc.date",
|
---|
| 395 | "dateCopyrighted" => "dc.date",
|
---|
| 396 | "dateSubmitted" => "dc.date",
|
---|
| 397 | "extent" => "dc.format",
|
---|
| 398 | "medium" => "dc.format",
|
---|
| 399 | "isVersionOf" => "dc.relation",
|
---|
| 400 | "hasVersion" => "dc.relation",
|
---|
| 401 | "isReplacedBy" => "dc.relation",
|
---|
| 402 | "replaces" => "dc.relation",
|
---|
| 403 | "isRequiredBy" => "dc.relation",
|
---|
| 404 | "requires" => "dc.relation",
|
---|
| 405 | "isPartOf" => "dc.relation",
|
---|
| 406 | "hasPart" => "dc.relation",
|
---|
| 407 | "isReferencedBy" => "dc.relation",
|
---|
| 408 | "references" => "dc.relation",
|
---|
| 409 | "isFormatOf" => "dc.relation",
|
---|
| 410 | "hasFormat" => "dc.relation",
|
---|
| 411 | "conformsTo" => "dc.relation",
|
---|
| 412 | "spatial" => "dc.coverage",
|
---|
| 413 | "temporal" => "dc.coverage",
|
---|
| 414 | # these are now top level elements in our qualified dc metadata set
|
---|
[18901] | 415 | # "audience" => "dc.any",
|
---|
| 416 | # "accrualMethod" => "dc.any",
|
---|
| 417 | # "accrualPeriodicity" => "dc.any",
|
---|
| 418 | # "accrualPolicy" => "dc.any",
|
---|
| 419 | # "instructionalMethod" => "dc.any",
|
---|
| 420 | # "provenance" => "dc.any",
|
---|
| 421 | # "rightsHolder" => "dc.any",
|
---|
[19213] | 422 | "mediator" => "dc.audience",
|
---|
| 423 | "educationLevel" => "dc.audience",
|
---|
| 424 | "accessRights" => "dc.rights",
|
---|
| 425 | "license" => "dc.rights",
|
---|
| 426 | "bibliographicCitation" => "dc.identifier"
|
---|
| 427 | };
|
---|
[14940] | 428 |
|
---|
[19213] | 429 | sub remap_dc_metadata
|
---|
| 430 | {
|
---|
| 431 | my $self = shift(@_);
|
---|
| 432 |
|
---|
| 433 | my ($metaname) = @_;
|
---|
| 434 |
|
---|
[14940] | 435 | my ($prefix,$name) = ($metaname =~ m/^(.*?)\.(.*?)$/);
|
---|
| 436 |
|
---|
[19213] | 437 | if (defined $qualified_dc_mapping->{$name}) {
|
---|
| 438 |
|
---|
| 439 | return $qualified_dc_mapping->{$name}."^".$name;
|
---|
[14940] | 440 | }
|
---|
[19213] | 441 |
|
---|
| 442 |
|
---|
[14940] | 443 | return $metaname; # didn't get a match, return param passed in unchanged
|
---|
| 444 | }
|
---|
| 445 |
|
---|
| 446 |
|
---|
[4726] | 447 | sub extract_oai_metadata {
|
---|
| 448 | my $self = shift (@_);
|
---|
| 449 | my ($textref, $metadata) = @_;
|
---|
| 450 | my $outhandle = $self->{'outhandle'};
|
---|
| 451 |
|
---|
[9958] | 452 | $self->open_prettyprint_metadata_table();
|
---|
| 453 |
|
---|
| 454 | if ($$textref =~ m/<metadata\s*>(.*?)<\/metadata\s*>/s)
|
---|
[4726] | 455 | {
|
---|
[10254] | 456 | my $metadata_text = $1;
|
---|
[4726] | 457 |
|
---|
[14940] | 458 | # locate and remove outermost tag (ignoring any attribute information in top-level tag)
|
---|
[19213] | 459 | my ($outer_tagname,$inner_metadata_text) = ($metadata_text =~ m/<([^ >]+).*?>(.*?)<\/\1>/s);
|
---|
[14940] | 460 | # split tag into namespace and tag name
|
---|
[19213] | 461 | my($namespace,$top_level_prefix) = ($outer_tagname =~ m/^(.*?):(.*?)$/);
|
---|
[17066] | 462 | # sometimes, the dc namespace is not specified as the prefix in each element (like <dc:title>)
|
---|
| 463 | # but is rather defined in the wrapper element containing the various dc meta elements,
|
---|
| 464 | # like <dc><title></title><creator></creator></dc>.
|
---|
| 465 | # In such a case, we use this wrapper element as the top_level_prefix
|
---|
[19213] | 466 |
|
---|
| 467 | # if there was no prefix, then the tag itself becomes the top_level_prefix
|
---|
| 468 | if(!defined $top_level_prefix && defined $outer_tagname) {
|
---|
| 469 | $top_level_prefix = $outer_tagname;
|
---|
[17066] | 470 | }
|
---|
| 471 |
|
---|
[19213] | 472 | #process each element one by one
|
---|
[14949] | 473 | while ($inner_metadata_text =~ m/<([^ >]+).*?>(.*?)<\/\1>(.*)/s)
|
---|
[4726] | 474 | {
|
---|
[9958] | 475 |
|
---|
[4726] | 476 | my $metaname = $1;
|
---|
| 477 | my $metavalue = $2;
|
---|
[14949] | 478 | $inner_metadata_text = $3;
|
---|
| 479 |
|
---|
[19213] | 480 | # greenstone uses . for namespace, while oai uses :
|
---|
[14940] | 481 | $metaname =~ s/:/\./;
|
---|
[19213] | 482 | # if there is no namespace, then we use the outer tag name or
|
---|
| 483 | # namespace for this element
|
---|
[14940] | 484 | if ($metaname !~ m/\./)
|
---|
[4726] | 485 | {
|
---|
[14940] | 486 | $metaname = "$top_level_prefix.$metaname";
|
---|
[4726] | 487 | }
|
---|
[19213] | 488 |
|
---|
| 489 | # if metadata set is auto, leave as is, otherwise convert to
|
---|
| 490 | # specified namespace
|
---|
| 491 | if ($self->{'metadata_set'} ne "auto") {
|
---|
| 492 | $metaname =~ s/^([^\.]*)\./$self->{'metadata_set'}\./;
|
---|
| 493 | if ($self->{'metadata_set'} eq "dc") {
|
---|
| 494 | # convert qualified dc terms to gs version, e.g.
|
---|
| 495 | # spatial becomes coverage^spatial
|
---|
| 496 | $metaname = $self->remap_dc_metadata($metaname);
|
---|
| 497 | }
|
---|
| 498 | }
|
---|
[4726] | 499 |
|
---|
[18901] | 500 | # uppercase the first char of the name
|
---|
| 501 | $metaname =~ s/\.(.)/\.\u$1/;
|
---|
[14963] | 502 | $metavalue =~ s/\[/[/g;
|
---|
| 503 | $metavalue =~ s/\]/]/g;
|
---|
| 504 |
|
---|
[4726] | 505 | if (defined $metadata->{$metaname})
|
---|
| 506 | {
|
---|
| 507 | push(@{$metadata->{$metaname}},$metavalue);
|
---|
[8121] | 508 |
|
---|
[4726] | 509 | }
|
---|
| 510 | else
|
---|
| 511 | {
|
---|
| 512 | $metadata->{$metaname} = [ $metavalue ];
|
---|
| 513 | }
|
---|
| 514 |
|
---|
[9958] | 515 | $self->add_prettyprint_metadata_line($metaname, $metavalue);
|
---|
| 516 |
|
---|
[4726] | 517 | }
|
---|
| 518 | }
|
---|
[9958] | 519 |
|
---|
| 520 | $self->close_prettyprint_metadata_table();
|
---|
[4726] | 521 | }
|
---|
| 522 |
|
---|
[13886] | 523 | ## we know from the file extension, so doesn't need to check the doctype
|
---|
| 524 | sub check_doctype {
|
---|
| 525 |
|
---|
| 526 | return 1;
|
---|
| 527 | }
|
---|
| 528 |
|
---|
[4726] | 529 | 1;
|
---|