| 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 |
|
|---|
| 27 | package OAIPlugin;
|
|---|
| 28 |
|
|---|
| 29 | # Devel::Peek's Dump($var) function is useful for debugging encoding issues.
|
|---|
| 30 | #use Devel::Peek;
|
|---|
| 31 | use Encode;
|
|---|
| 32 | use extrametautil;
|
|---|
| 33 | use unicode;
|
|---|
| 34 | use util;
|
|---|
| 35 |
|
|---|
| 36 | use strict;
|
|---|
| 37 | no strict 'refs'; # allow filehandles to be variables and viceversa
|
|---|
| 38 |
|
|---|
| 39 | use ReadXMLFile;
|
|---|
| 40 | use ReadTextFile; # needed for subroutine textcat_get_language_encoding
|
|---|
| 41 | use metadatautil;
|
|---|
| 42 | use MetadataRead;
|
|---|
| 43 | use util;
|
|---|
| 44 |
|
|---|
| 45 | # methods with identical signatures take precedence in the order given in the ISA list.
|
|---|
| 46 | sub BEGIN {
|
|---|
| 47 | @OAIPlugin::ISA = ('MetadataRead', 'ReadXMLFile', 'ReadTextFile');
|
|---|
| 48 | }
|
|---|
| 49 |
|
|---|
| 50 | my $set_list =
|
|---|
| 51 | [ { 'name' => "auto",
|
|---|
| 52 | 'desc' => "{OAIPlugin.metadata_set.auto}" },
|
|---|
| 53 | { 'name' => "dc",
|
|---|
| 54 | 'desc' => "{OAIPlugin.metadata_set.dc}" }
|
|---|
| 55 | ];
|
|---|
| 56 |
|
|---|
| 57 | my $arguments =
|
|---|
| 58 | [ { 'name' => "process_exp",
|
|---|
| 59 | 'desc' => "{BaseImporter.process_exp}",
|
|---|
| 60 | 'type' => "regexp",
|
|---|
| 61 | 'reqd' => "no",
|
|---|
| 62 | 'deft' => &get_default_process_exp() },
|
|---|
| 63 | { 'name' => "metadata_set",
|
|---|
| 64 | 'desc' => "{OAIPlugin.metadata_set}",
|
|---|
| 65 | 'type' => "enumstring",
|
|---|
| 66 | 'reqd' => "no",
|
|---|
| 67 | 'list' => $set_list,
|
|---|
| 68 | 'deft' => "dc" },
|
|---|
| 69 | { 'name' => "document_field",
|
|---|
| 70 | 'desc' => "{OAIPlugin.document_field}",
|
|---|
| 71 | 'type' => "metadata",
|
|---|
| 72 | 'reqd' => "no",
|
|---|
| 73 | 'deft' => "gi.Sourcedoc" }
|
|---|
| 74 | ];
|
|---|
| 75 |
|
|---|
| 76 | my $options = { 'name' => "OAIPlugin",
|
|---|
| 77 | 'desc' => "{OAIPlugin.desc}",
|
|---|
| 78 | 'abstract' => "no",
|
|---|
| 79 | 'inherits' => "yes",
|
|---|
| 80 | 'explodes' => "yes",
|
|---|
| 81 | 'args' => $arguments };
|
|---|
| 82 |
|
|---|
| 83 |
|
|---|
| 84 | sub new {
|
|---|
| 85 | my ($class) = shift (@_);
|
|---|
| 86 | my ($pluginlist,$inputargs,$hashArgOptLists) = @_;
|
|---|
| 87 | push(@$pluginlist, $class);
|
|---|
| 88 |
|
|---|
| 89 | push(@{$hashArgOptLists->{"ArgList"}},@{$arguments});
|
|---|
| 90 | push(@{$hashArgOptLists->{"OptList"}},$options);
|
|---|
| 91 |
|
|---|
| 92 | new ReadTextFile($pluginlist, $inputargs, $hashArgOptLists,1);
|
|---|
| 93 | my $self = new ReadXMLFile($pluginlist, $inputargs, $hashArgOptLists);
|
|---|
| 94 |
|
|---|
| 95 | if ($self->{'info_only'}) {
|
|---|
| 96 | # don't worry about modifying options
|
|---|
| 97 | return bless $self, $class;
|
|---|
| 98 | }
|
|---|
| 99 | # trim any ex. from document field iff it's the only metadata namespace prefix
|
|---|
| 100 | $self->{'document_field'} =~ s/^ex\.([^.]+)$/$1/;
|
|---|
| 101 | return bless $self, $class;
|
|---|
| 102 | }
|
|---|
| 103 |
|
|---|
| 104 | sub get_default_process_exp {
|
|---|
| 105 | my $self = shift (@_);
|
|---|
| 106 |
|
|---|
| 107 | return q^(?i)(\.oai)$^;
|
|---|
| 108 | }
|
|---|
| 109 |
|
|---|
| 110 | sub get_doctype {
|
|---|
| 111 | my $self = shift(@_);
|
|---|
| 112 |
|
|---|
| 113 | return "OAI-PMH";
|
|---|
| 114 | }
|
|---|
| 115 |
|
|---|
| 116 | sub xml_start_document {
|
|---|
| 117 | my $self = shift (@_);
|
|---|
| 118 | $self->{'in_metadata_node'} = 0;
|
|---|
| 119 | $self->{'rawxml'} = "";
|
|---|
| 120 | $self->{'saved_metadata'} = {};
|
|---|
| 121 | }
|
|---|
| 122 |
|
|---|
| 123 | sub xml_end_document {
|
|---|
| 124 | }
|
|---|
| 125 |
|
|---|
| 126 | sub xml_doctype {
|
|---|
| 127 | my $self = shift(@_);
|
|---|
| 128 |
|
|---|
| 129 | my ($expat, $name, $sysid, $pubid, $internal) = @_;
|
|---|
| 130 |
|
|---|
| 131 | ##die "" if ($name !~ /^OAI-PMH$/);
|
|---|
| 132 |
|
|---|
| 133 | my $outhandle = $self->{'outhandle'};
|
|---|
| 134 | print $outhandle "OAIPlugin: processing $self->{'file'}\n" if $self->{'verbosity'} > 1;
|
|---|
| 135 | print STDERR "<Processing n='$self->{'file'}' p='OAIPlugin'>\n" if $self->{'gli'};
|
|---|
| 136 |
|
|---|
| 137 | }
|
|---|
| 138 |
|
|---|
| 139 |
|
|---|
| 140 | sub xml_start_tag {
|
|---|
| 141 | my $self = shift(@_);
|
|---|
| 142 | my ($expat,$element) = @_;
|
|---|
| 143 |
|
|---|
| 144 | my %attr_hash = %_;
|
|---|
| 145 |
|
|---|
| 146 | my $attr = "";
|
|---|
| 147 | map { $attr .= " $_=$attr_hash{$_}"; } keys %attr_hash;
|
|---|
| 148 |
|
|---|
| 149 | $self->{'rawxml'} .= "<$element$attr>";
|
|---|
| 150 |
|
|---|
| 151 | if ($element eq "metadata") {
|
|---|
| 152 | $self->{'in_metadata_node'} = 1;
|
|---|
| 153 | $self->{'metadata_xml'} = "";
|
|---|
| 154 | }
|
|---|
| 155 |
|
|---|
| 156 | if ($self->{'in_metadata_node'}) {
|
|---|
| 157 | $self->{'metadata_xml'} .= "<$element$attr>";
|
|---|
| 158 | }
|
|---|
| 159 | }
|
|---|
| 160 |
|
|---|
| 161 | sub xml_end_tag {
|
|---|
| 162 | my $self = shift(@_);
|
|---|
| 163 | my ($expat, $element) = @_;
|
|---|
| 164 |
|
|---|
| 165 | $self->{'rawxml'} .= "</$element>";
|
|---|
| 166 |
|
|---|
| 167 | if ($self->{'in_metadata_node'}) {
|
|---|
| 168 | $self->{'metadata_xml'} .= "</$element>";
|
|---|
| 169 | }
|
|---|
| 170 |
|
|---|
| 171 | if ($element eq "metadata") {
|
|---|
| 172 | my $textref = \$self->{'metadata_xml'};
|
|---|
| 173 | #my $metadata = $self->{'metadata'};
|
|---|
| 174 | my $metadata = $self->{'saved_metadata'};
|
|---|
| 175 | $self->extract_oai_metadata($textref,$metadata);
|
|---|
| 176 |
|
|---|
| 177 | $self->{'in_metadata_node'} = 0;
|
|---|
| 178 | }
|
|---|
| 179 |
|
|---|
| 180 |
|
|---|
| 181 | }
|
|---|
| 182 |
|
|---|
| 183 | sub xml_text {
|
|---|
| 184 | my $self = shift(@_);
|
|---|
| 185 | my ($expat) = @_;
|
|---|
| 186 |
|
|---|
| 187 | $self->{'rawxml'} .= $_;
|
|---|
| 188 |
|
|---|
| 189 | if ($self->{'in_metadata_node'}) {
|
|---|
| 190 | $self->{'metadata_xml'} .= $_;
|
|---|
| 191 | }
|
|---|
| 192 | }
|
|---|
| 193 |
|
|---|
| 194 |
|
|---|
| 195 | sub metadata_read {
|
|---|
| 196 | my $self = shift (@_);
|
|---|
| 197 |
|
|---|
| 198 | my ($pluginfo, $base_dir, $file, $block_hash,
|
|---|
| 199 | $extrametakeys, $extrametadata, $extrametafile,
|
|---|
| 200 | $processor, $gli, $aux) = @_;
|
|---|
| 201 |
|
|---|
| 202 | # can we process this file??
|
|---|
| 203 | my ($filename_full_path, $filename_no_path) = &util::get_full_filenames($base_dir, $file);
|
|---|
| 204 | return undef unless $self->can_process_this_file_for_metadata($filename_full_path);
|
|---|
| 205 |
|
|---|
| 206 | print STDERR "\n<Processing n='$file' p='OAIPlugin'>\n" if ($gli);
|
|---|
| 207 | print STDERR "OAIPlugin: processing $file\n" if ($self->{'verbosity'}) > 1;
|
|---|
| 208 |
|
|---|
| 209 | if (!$self->parse_file($filename_full_path, $file, $gli)) {
|
|---|
| 210 | $self->{'saved_metadata'} = undef;
|
|---|
| 211 | return undef;
|
|---|
| 212 | }
|
|---|
| 213 |
|
|---|
| 214 | my $verbosity = $self->{'verbosity'};
|
|---|
| 215 | my $new_metadata = $self->{'saved_metadata'};
|
|---|
| 216 | $self->{'saved_metadata'} = undef;
|
|---|
| 217 |
|
|---|
| 218 | # add the pretty metadata table as metadata
|
|---|
| 219 | my $ppmd_table = $self->{'ppmd_table'};
|
|---|
| 220 | $new_metadata->{'prettymd'} = $ppmd_table;
|
|---|
| 221 | $self->{'ppmd_table'} = undef;
|
|---|
| 222 |
|
|---|
| 223 | my $document_metadata_field = $self->{'document_field'};
|
|---|
| 224 | my $url_array = $new_metadata->{$document_metadata_field};
|
|---|
| 225 | if (!defined $url_array) {
|
|---|
| 226 | # try ex.
|
|---|
| 227 | $url_array = $new_metadata->{"ex.$document_metadata_field"};
|
|---|
| 228 | }
|
|---|
| 229 | my $num_urls = (defined $url_array) ? scalar(@$url_array) : 0;
|
|---|
| 230 | ##print STDERR "$num_urls urls for $file\n";
|
|---|
| 231 | my $srcdoc_exists = 0;
|
|---|
| 232 | my $srcdoc_pos = 0;
|
|---|
| 233 | my $filename_dir = &util::filename_head($filename_full_path);
|
|---|
| 234 |
|
|---|
| 235 | # filenames in extrametadata must be relative to current dir, as
|
|---|
| 236 | # DirectoryPlugin adds path info on itself
|
|---|
| 237 | my ($filename_for_metadata) = $file =~ /([^\\\/]+)$/; # this assumes there will only be one record per oai file - is this always the case??
|
|---|
| 238 | for (my $i=0; $i<$num_urls; $i++) {
|
|---|
| 239 |
|
|---|
| 240 | if ($url_array->[$i] !~ m/^(https?|ftp):/) {
|
|---|
| 241 |
|
|---|
| 242 | my $src_filename = &util::filename_cat($filename_dir, $url_array->[$i]);
|
|---|
| 243 | if (-e $src_filename) {
|
|---|
| 244 | $srcdoc_pos = $i;
|
|---|
| 245 | $srcdoc_exists = 1;
|
|---|
| 246 | # get the slashes the right way, use filename_cat
|
|---|
| 247 | $filename_for_metadata = &util::filename_cat($url_array->[$i]);
|
|---|
| 248 | last;
|
|---|
| 249 | }
|
|---|
| 250 | }
|
|---|
| 251 | }
|
|---|
| 252 |
|
|---|
| 253 | if ($srcdoc_exists) {
|
|---|
| 254 | $self->{'oai-files'}->{$file}->{'srcdoc_exists'} = 1;
|
|---|
| 255 | }
|
|---|
| 256 | else {
|
|---|
| 257 | # save the rawxml for the source document
|
|---|
| 258 | $self->{'oai-files'}->{$file}->{'srcdoc_exists'} = 0;
|
|---|
| 259 | $self->{'oai-files'}->{$file}->{'rawxml'} = $self->{'rawxml'};
|
|---|
| 260 | $self->{'rawxml'} = "";
|
|---|
| 261 | }
|
|---|
| 262 |
|
|---|
| 263 | # return all the metadata we have extracted to the caller.
|
|---|
| 264 | # Directory plug will pass it back in at read time, so we don't need to extract it again.
|
|---|
| 265 |
|
|---|
| 266 | # Store the metadata for later in extrameta. if we have a srcdoc, then treat this file as a metadata file, and pass it in to the store_meta method.
|
|---|
| 267 | # If there is no srcdoc, then this is the actual doc, so we don't want it treated as a metadata file. - pass in undef.
|
|---|
| 268 | if ($srcdoc_exists) {
|
|---|
| 269 | $self->store_meta_in_extrametadata($filename_for_metadata, $new_metadata, $file, $filename_full_path, $extrametakeys, $extrametadata, $extrametafile);
|
|---|
| 270 | } else {
|
|---|
| 271 | $self->store_meta_in_extrametadata($filename_for_metadata, $new_metadata, undef, undef, $extrametakeys, $extrametadata, $extrametafile);
|
|---|
| 272 | }
|
|---|
| 273 | # Extrametadata keys should be regular expressions
|
|---|
| 274 | # Indexing into the extrameta data structures requires the filename's style of slashes to be in URL format
|
|---|
| 275 | # Then need to convert the filename to a regex, no longer to protect windows directory chars \, but for
|
|---|
| 276 | # protecting special characters like brackets in the filepath such as "C:\Program Files (x86)\Greenstone".
|
|---|
| 277 |
|
|---|
| 278 | if (0) {
|
|---|
| 279 | $filename_for_metadata = &util::filepath_to_url_format($filename_for_metadata);
|
|---|
| 280 | $filename_for_metadata = &util::filename_to_regex($filename_for_metadata);
|
|---|
| 281 |
|
|---|
| 282 | # Check that we haven't already got some metadata
|
|---|
| 283 | if (defined &extrametautil::getmetadata($extrametadata, $filename_for_metadata)) {
|
|---|
| 284 | print STDERR "\n**** OAIPlugin: Need to merge new metadata with existing stored metadata: file = $filename_for_metadata\n" if $verbosity > 3;
|
|---|
| 285 |
|
|---|
| 286 | my $file_metadata_table = &extrametautil::getmetadata($extrametadata, $filename_for_metadata);
|
|---|
| 287 |
|
|---|
| 288 | foreach my $metaname (keys %{$new_metadata}) {
|
|---|
| 289 | # will create new entry if one does not already exist
|
|---|
| 290 | push(@{$file_metadata_table->{$metaname}}, @{$new_metadata->{$metaname}});
|
|---|
| 291 | }
|
|---|
| 292 |
|
|---|
| 293 | } else {
|
|---|
| 294 | &extrametautil::setmetadata($extrametadata, $filename_for_metadata, $new_metadata);
|
|---|
| 295 | &extrametautil::addmetakey($extrametakeys, $filename_for_metadata);
|
|---|
| 296 | }
|
|---|
| 297 |
|
|---|
| 298 | if ($srcdoc_exists) {
|
|---|
| 299 | # if (!defined &extrametautil::getmetafile($extrametafile, $filename_for_metadata)) {
|
|---|
| 300 | # &extrametautil::setmetafile($extrametafile, $filename_for_metadata, {});
|
|---|
| 301 | # }
|
|---|
| 302 | #maps the file to full path
|
|---|
| 303 | # &extrametautil::setmetafile_for_named_file($extrametafile, $filename_for_metadata, $file, $filename_full_path);
|
|---|
| 304 | &extrametautil::addmetafile($extrametafile, $filename_for_metadata, $file, $filename_full_path);
|
|---|
| 305 |
|
|---|
| 306 | }
|
|---|
| 307 | } # if 0
|
|---|
| 308 |
|
|---|
| 309 | return 1;
|
|---|
| 310 |
|
|---|
| 311 | }
|
|---|
| 312 |
|
|---|
| 313 |
|
|---|
| 314 | sub read {
|
|---|
| 315 | my $self = shift (@_);
|
|---|
| 316 |
|
|---|
| 317 | my ($pluginfo, $base_dir, $file, $block_hash, $metadata, $processor, $maxdocs, $total_count, $gli) = @_;
|
|---|
| 318 |
|
|---|
| 319 | if (!defined $self->{'oai-files'}->{$file}) {
|
|---|
| 320 | return undef;
|
|---|
| 321 | }
|
|---|
| 322 |
|
|---|
| 323 | my $srcdoc_exists = $self->{'oai-files'}->{$file}->{'srcdoc_exists'};
|
|---|
| 324 | if ($srcdoc_exists) {
|
|---|
| 325 | # do nothing more - all the metadata has been extracted and associated with the srcdoc
|
|---|
| 326 | # no more need to access details of this $file => tidy up as you go
|
|---|
| 327 | delete $self->{'oai-files'}->{$file};
|
|---|
| 328 | return 0; # not processed here, but don't pass on to rest of plugins
|
|---|
| 329 | }
|
|---|
| 330 |
|
|---|
| 331 | my $filename = $file;
|
|---|
| 332 | $filename = &util::filename_cat ($base_dir, $file) if $base_dir =~ /\w/;
|
|---|
| 333 |
|
|---|
| 334 | # Do encoding stuff on metadata
|
|---|
| 335 | my ($language, $encoding) = $self->textcat_get_language_encoding ($filename);
|
|---|
| 336 |
|
|---|
| 337 | # create a new document
|
|---|
| 338 | my $doc_obj = new doc ($filename, "indexed_doc", $self->{'file_rename_method'});
|
|---|
| 339 | my $top_section = $doc_obj->get_top_section;
|
|---|
| 340 | my $plugin_type = $self->{'plugin_type'};
|
|---|
| 341 |
|
|---|
| 342 | my ($filemeta) = $file =~ /([^\\\/]+)$/;
|
|---|
| 343 | my $plugin_filename_encoding = $self->{'filename_encoding'};
|
|---|
| 344 | my $filename_encoding = $self->deduce_filename_encoding($file,$metadata,$plugin_filename_encoding);
|
|---|
| 345 | $self->set_Source_metadata($doc_obj, $filename, $filename_encoding);
|
|---|
| 346 |
|
|---|
| 347 | $doc_obj->add_utf8_metadata($top_section, "Language", $language);
|
|---|
| 348 | $doc_obj->add_utf8_metadata($top_section, "Encoding", $encoding);
|
|---|
| 349 | $doc_obj->add_utf8_metadata($top_section, "Plugin", $plugin_type);
|
|---|
| 350 | $doc_obj->add_metadata($top_section, "FileFormat", "OAI");
|
|---|
| 351 | $doc_obj->add_metadata($top_section, "FileSize", (-s $filename));
|
|---|
| 352 |
|
|---|
| 353 | # include any metadata passed in from previous plugins
|
|---|
| 354 | # note that this metadata is associated with the top level section
|
|---|
| 355 | # this will include all the metadata from the oai file that we extracted
|
|---|
| 356 | # during metadata_read
|
|---|
| 357 | $self->extra_metadata ($doc_obj, $doc_obj->get_top_section(), $metadata);
|
|---|
| 358 |
|
|---|
| 359 | # do plugin specific processing of doc_obj
|
|---|
| 360 | my $text = $self->{'oai-files'}->{$file}->{'rawxml'};
|
|---|
| 361 | delete $self->{'oai-files'}->{$file};
|
|---|
| 362 |
|
|---|
| 363 | unless (defined ($self->process(\$text, $pluginfo, $base_dir, $file, $metadata, $doc_obj))) {
|
|---|
| 364 | print STDERR "<ProcessingError n='$file'>\n" if ($gli);
|
|---|
| 365 | return -1;
|
|---|
| 366 | }
|
|---|
| 367 |
|
|---|
| 368 | # do any automatic metadata extraction
|
|---|
| 369 | $self->auto_extract_metadata ($doc_obj);
|
|---|
| 370 |
|
|---|
| 371 | # add an OID
|
|---|
| 372 | $self->add_OID($doc_obj);
|
|---|
| 373 |
|
|---|
| 374 | # process the document
|
|---|
| 375 | $processor->process($doc_obj);
|
|---|
| 376 |
|
|---|
| 377 | $self->{'num_processed'} ++;
|
|---|
| 378 |
|
|---|
| 379 | return 1; # processed the file
|
|---|
| 380 | }
|
|---|
| 381 |
|
|---|
| 382 |
|
|---|
| 383 | # do plugin specific processing of doc_obj
|
|---|
| 384 | sub process {
|
|---|
| 385 | my $self = shift (@_);
|
|---|
| 386 | my ($textref, $pluginfo, $base_dir, $file, $metadata, $doc_obj, $gli) = @_;
|
|---|
| 387 | my $outhandle = $self->{'outhandle'};
|
|---|
| 388 |
|
|---|
| 389 | print STDERR "<Processing n='$file' p='OAIPlugin'>\n" if ($gli);
|
|---|
| 390 | print $outhandle "OAIPlugin: processing $file\n"
|
|---|
| 391 | if $self->{'verbosity'} > 1;
|
|---|
| 392 |
|
|---|
| 393 | my $cursection = $doc_obj->get_top_section();
|
|---|
| 394 |
|
|---|
| 395 | ## $self->extract_metadata ($textref, $metadata, $doc_obj, $cursection);
|
|---|
| 396 |
|
|---|
| 397 | # add text to document object
|
|---|
| 398 |
|
|---|
| 399 | # $$textref =~ s/<(.*?)>/$1 /g;
|
|---|
| 400 | $$textref =~ s/</</g;
|
|---|
| 401 | $$textref =~ s/>/>/g;
|
|---|
| 402 | $$textref =~ s/\[/[/g;
|
|---|
| 403 | $$textref =~ s/\]/]/g;
|
|---|
| 404 |
|
|---|
| 405 | $doc_obj->add_utf8_text($cursection, $$textref);
|
|---|
| 406 |
|
|---|
| 407 | return 1;
|
|---|
| 408 | }
|
|---|
| 409 |
|
|---|
| 410 |
|
|---|
| 411 | # Improvement is to merge this with newer version in MetadataPass
|
|---|
| 412 |
|
|---|
| 413 | sub open_prettyprint_metadata_table
|
|---|
| 414 | {
|
|---|
| 415 | my $self = shift(@_);
|
|---|
| 416 |
|
|---|
| 417 | my $att = "width=100% cellspacing=2";
|
|---|
| 418 | my $style = "style=\'border-bottom: 4px solid #000080\'";
|
|---|
| 419 |
|
|---|
| 420 | $self->{'ppmd_table'} = "\n<table $att $style>";
|
|---|
| 421 | }
|
|---|
| 422 |
|
|---|
| 423 | sub add_prettyprint_metadata_line
|
|---|
| 424 | {
|
|---|
| 425 | my $self = shift(@_);
|
|---|
| 426 | my ($metaname, $metavalue_utf8) = @_;
|
|---|
| 427 |
|
|---|
| 428 | $metavalue_utf8 = &util::hyperlink_text($metavalue_utf8);
|
|---|
| 429 |
|
|---|
| 430 | $self->{'ppmd_table'} .= " <tr bgcolor=#b5d3cd>\n";
|
|---|
| 431 | $self->{'ppmd_table'} .= " <td width=30%>\n";
|
|---|
| 432 | $self->{'ppmd_table'} .= " $metaname\n";
|
|---|
| 433 | $self->{'ppmd_table'} .= " </td>\n";
|
|---|
| 434 | $self->{'ppmd_table'} .= " <td>\n";
|
|---|
| 435 | $self->{'ppmd_table'} .= " $metavalue_utf8\n";
|
|---|
| 436 | $self->{'ppmd_table'} .= " </td>\n";
|
|---|
| 437 | $self->{'ppmd_table'} .= " </tr>\n";
|
|---|
| 438 |
|
|---|
| 439 | }
|
|---|
| 440 |
|
|---|
| 441 | sub close_prettyprint_metadata_table
|
|---|
| 442 | {
|
|---|
| 443 | my $self = shift(@_);
|
|---|
| 444 |
|
|---|
| 445 | $self->{'ppmd_table'} .= "</table>\n";
|
|---|
| 446 | }
|
|---|
| 447 |
|
|---|
| 448 | my $qualified_dc_mapping = {
|
|---|
| 449 | "alternative" => "dc.title",
|
|---|
| 450 | "tableOfContents" => "dc.description",
|
|---|
| 451 | "abstract" => "dc.description",
|
|---|
| 452 | "created" => "dc.date",
|
|---|
| 453 | "valid" => "dc.date",
|
|---|
| 454 | "available" => "dc.date",
|
|---|
| 455 | "issued" => "dc.date",
|
|---|
| 456 | "modified" => "dc.date",
|
|---|
| 457 | "dateAccepted" => "dc.date",
|
|---|
| 458 | "dateCopyrighted" => "dc.date",
|
|---|
| 459 | "dateSubmitted" => "dc.date",
|
|---|
| 460 | "extent" => "dc.format",
|
|---|
| 461 | "medium" => "dc.format",
|
|---|
| 462 | "isVersionOf" => "dc.relation",
|
|---|
| 463 | "hasVersion" => "dc.relation",
|
|---|
| 464 | "isReplacedBy" => "dc.relation",
|
|---|
| 465 | "replaces" => "dc.relation",
|
|---|
| 466 | "isRequiredBy" => "dc.relation",
|
|---|
| 467 | "requires" => "dc.relation",
|
|---|
| 468 | "isPartOf" => "dc.relation",
|
|---|
| 469 | "hasPart" => "dc.relation",
|
|---|
| 470 | "isReferencedBy" => "dc.relation",
|
|---|
| 471 | "references" => "dc.relation",
|
|---|
| 472 | "isFormatOf" => "dc.relation",
|
|---|
| 473 | "hasFormat" => "dc.relation",
|
|---|
| 474 | "conformsTo" => "dc.relation",
|
|---|
| 475 | "spatial" => "dc.coverage",
|
|---|
| 476 | "temporal" => "dc.coverage",
|
|---|
| 477 | # these are now top level elements in our qualified dc metadata set
|
|---|
| 478 | # "audience" => "dc.any",
|
|---|
| 479 | # "accrualMethod" => "dc.any",
|
|---|
| 480 | # "accrualPeriodicity" => "dc.any",
|
|---|
| 481 | # "accrualPolicy" => "dc.any",
|
|---|
| 482 | # "instructionalMethod" => "dc.any",
|
|---|
| 483 | # "provenance" => "dc.any",
|
|---|
| 484 | # "rightsHolder" => "dc.any",
|
|---|
| 485 | "mediator" => "dc.audience",
|
|---|
| 486 | "educationLevel" => "dc.audience",
|
|---|
| 487 | "accessRights" => "dc.rights",
|
|---|
| 488 | "license" => "dc.rights",
|
|---|
| 489 | "bibliographicCitation" => "dc.identifier"
|
|---|
| 490 | };
|
|---|
| 491 |
|
|---|
| 492 | sub remap_dc_metadata
|
|---|
| 493 | {
|
|---|
| 494 | my $self = shift(@_);
|
|---|
| 495 |
|
|---|
| 496 | my ($metaname) = @_;
|
|---|
| 497 |
|
|---|
| 498 | my ($prefix,$name) = ($metaname =~ m/^(.*?)\.(.*?)$/);
|
|---|
| 499 |
|
|---|
| 500 | if (defined $qualified_dc_mapping->{$name}) {
|
|---|
| 501 |
|
|---|
| 502 | return $qualified_dc_mapping->{$name}."^".$name;
|
|---|
| 503 | }
|
|---|
| 504 |
|
|---|
| 505 |
|
|---|
| 506 | return $metaname; # didn't get a match, return param passed in unchanged
|
|---|
| 507 | }
|
|---|
| 508 |
|
|---|
| 509 |
|
|---|
| 510 | sub extract_oai_metadata {
|
|---|
| 511 | my $self = shift (@_);
|
|---|
| 512 | my ($textref, $metadata) = @_;
|
|---|
| 513 | my $outhandle = $self->{'outhandle'};
|
|---|
| 514 |
|
|---|
| 515 | $self->open_prettyprint_metadata_table();
|
|---|
| 516 |
|
|---|
| 517 | # need to decode the string, else it will be double-encoded at this point
|
|---|
| 518 | $$textref = decode("utf-8",$$textref);
|
|---|
| 519 |
|
|---|
| 520 | # Debugging encoding issues with Devel::Peek's Dump() which prints octal and hexcode
|
|---|
| 521 | # print STDERR "#### text ref: $$textref\n";
|
|---|
| 522 | # print STDERR "\n@@@\n";
|
|---|
| 523 | # Dump($$textref);
|
|---|
| 524 | # print STDERR "\n";
|
|---|
| 525 |
|
|---|
| 526 | if ($$textref =~ m/<metadata\s*>(.*?)<\/metadata\s*>/s)
|
|---|
| 527 | {
|
|---|
| 528 | my $metadata_text = $1;
|
|---|
| 529 |
|
|---|
| 530 | # locate and remove outermost tag (ignoring any attribute information in top-level tag)
|
|---|
| 531 | my ($outer_tagname,$inner_metadata_text) = ($metadata_text =~ m/<([^ >]+).*?>(.*?)<\/\1>/s);
|
|---|
| 532 | # split tag into namespace and tag name
|
|---|
| 533 | my($namespace,$top_level_prefix) = ($outer_tagname =~ m/^(.*?):(.*?)$/);
|
|---|
| 534 | # sometimes, the dc namespace is not specified as the prefix in each element (like <dc:title>)
|
|---|
| 535 | # but is rather defined in the wrapper element containing the various dc meta elements,
|
|---|
| 536 | # like <dc><title></title><creator></creator></dc>.
|
|---|
| 537 | # In such a case, we use this wrapper element as the top_level_prefix
|
|---|
| 538 |
|
|---|
| 539 | # if there was no prefix, then the tag itself becomes the top_level_prefix
|
|---|
| 540 | if(!defined $top_level_prefix && defined $outer_tagname) {
|
|---|
| 541 | $top_level_prefix = $outer_tagname;
|
|---|
| 542 | }
|
|---|
| 543 |
|
|---|
| 544 | #process each element one by one
|
|---|
| 545 | while ($inner_metadata_text =~ m/<([^ >]+).*?>(.*?)<\/\1>(.*)/s)
|
|---|
| 546 | {
|
|---|
| 547 |
|
|---|
| 548 | my $metaname = $1;
|
|---|
| 549 | my $metavalue = $2;
|
|---|
| 550 | $inner_metadata_text = $3;
|
|---|
| 551 |
|
|---|
| 552 | # greenstone uses . for namespace, while oai uses :
|
|---|
| 553 | $metaname =~ s/:/\./;
|
|---|
| 554 | # if there is no namespace, then we use the outer tag name or
|
|---|
| 555 | # namespace for this element
|
|---|
| 556 | if ($metaname !~ m/\./)
|
|---|
| 557 | {
|
|---|
| 558 | $metaname = "$top_level_prefix.$metaname";
|
|---|
| 559 | }
|
|---|
| 560 |
|
|---|
| 561 | # if metadata set is auto, leave as is, otherwise convert to
|
|---|
| 562 | # specified namespace
|
|---|
| 563 | if ($self->{'metadata_set'} ne "auto") {
|
|---|
| 564 | if ($metaname !~ /^gi\./) { # hack to not overwrite gi metadata
|
|---|
| 565 | $metaname =~ s/^([^\.]*)\./$self->{'metadata_set'}\./;
|
|---|
| 566 | if ($self->{'metadata_set'} eq "dc") {
|
|---|
| 567 | # convert qualified dc terms to gs version, e.g.
|
|---|
| 568 | # spatial becomes coverage^spatial
|
|---|
| 569 | $metaname = $self->remap_dc_metadata($metaname);
|
|---|
| 570 | }
|
|---|
| 571 | }
|
|---|
| 572 | }
|
|---|
| 573 |
|
|---|
| 574 | # uppercase the first char of the name
|
|---|
| 575 | $metaname =~ s/\.(.)/\.\u$1/;
|
|---|
| 576 | $metavalue =~ s/\[/[/g;
|
|---|
| 577 | $metavalue =~ s/\]/]/g;
|
|---|
| 578 |
|
|---|
| 579 | # so that GLI can see this metadata, store here as ex.dc.Title etc
|
|---|
| 580 | my $ex_metaname = $metaname;
|
|---|
| 581 | $ex_metaname =~ s/^ex\.//; # remove any pre-existing ex. prefix
|
|---|
| 582 | $ex_metaname = "ex.$ex_metaname"; # at last can prefix ex.
|
|---|
| 583 |
|
|---|
| 584 | if (defined $metadata->{$ex_metaname})
|
|---|
| 585 | {
|
|---|
| 586 | push(@{$metadata->{$ex_metaname}},$metavalue);
|
|---|
| 587 |
|
|---|
| 588 | }
|
|---|
| 589 | else
|
|---|
| 590 | {
|
|---|
| 591 | $metadata->{$ex_metaname} = [ $metavalue ];
|
|---|
| 592 | }
|
|---|
| 593 |
|
|---|
| 594 | # but don't add ex to the pretty print line
|
|---|
| 595 | $self->add_prettyprint_metadata_line($metaname, $metavalue);
|
|---|
| 596 |
|
|---|
| 597 | }
|
|---|
| 598 | }
|
|---|
| 599 |
|
|---|
| 600 | $self->close_prettyprint_metadata_table();
|
|---|
| 601 | }
|
|---|
| 602 |
|
|---|
| 603 | ## we know from the file extension, so doesn't need to check the doctype
|
|---|
| 604 | sub check_doctype {
|
|---|
| 605 |
|
|---|
| 606 | return 1;
|
|---|
| 607 | }
|
|---|
| 608 |
|
|---|
| 609 | 1;
|
|---|
