| 1 | ###########################################################################
|
|---|
| 2 | #
|
|---|
| 3 | # MetadataXMLPlugin.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) 2006 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 | # MetadataXMLPlugin process metadata.xml files in a collection
|
|---|
| 27 |
|
|---|
| 28 | # Here's an example of a metadata file that uses three FileSet structures
|
|---|
| 29 | # (ignore the # characters):
|
|---|
| 30 |
|
|---|
| 31 | #<?xml version="1.0" encoding="UTF-8" standalone="no"?>
|
|---|
| 32 | #<!DOCTYPE DirectoryMetadata SYSTEM "http://greenstone.org/dtd/DirectoryMetadata/1.0/DirectoryMetadata.dtd">
|
|---|
| 33 | #<DirectoryMetadata>
|
|---|
| 34 | # <FileSet>
|
|---|
| 35 | # <FileName>nugget.*</FileName>
|
|---|
| 36 | # <Description>
|
|---|
| 37 | # <Metadata name="Title">Nugget Point, The Catlins</Metadata>
|
|---|
| 38 | # <Metadata name="Place" mode="accumulate">Nugget Point</Metadata>
|
|---|
| 39 | # </Description>
|
|---|
| 40 | # </FileSet>
|
|---|
| 41 | # <FileSet>
|
|---|
| 42 | # <FileName>nugget-point-1.jpg</FileName>
|
|---|
| 43 | # <Description>
|
|---|
| 44 | # <Metadata name="Title">Nugget Point Lighthouse, The Catlins</Metadata>
|
|---|
| 45 | # <Metadata name="Subject">Lighthouse</Metadata>
|
|---|
| 46 | # </Description>
|
|---|
| 47 | # </FileSet>
|
|---|
| 48 | # <FileSet>
|
|---|
| 49 | # <FileName>kaka-point-dir</FileName>
|
|---|
| 50 | # <Description>
|
|---|
| 51 | # <Metadata name="Title">Kaka Point, The Catlins</Metadata>
|
|---|
| 52 | # </Description>
|
|---|
| 53 | # </FileSet>
|
|---|
| 54 | #</DirectoryMetadata>
|
|---|
| 55 |
|
|---|
| 56 | # Metadata elements are read and applied to files in the order they appear
|
|---|
| 57 | # in the file.
|
|---|
| 58 | #
|
|---|
| 59 | # The FileName element describes the subfiles in the directory that the
|
|---|
| 60 | # metadata applies to as a perl regular expression (a FileSet group may
|
|---|
| 61 | # contain multiple FileName elements). So, <FileName>nugget.*</FileName>
|
|---|
| 62 | # indicates that the metadata records in the following Description block
|
|---|
| 63 | # apply to every subfile that starts with "nugget". For these files, a
|
|---|
| 64 | # Title metadata element is set, overriding any old value that the Title
|
|---|
| 65 | # might have had.
|
|---|
| 66 | #
|
|---|
| 67 | # Occasionally, we want to have multiple metadata values applied to a
|
|---|
| 68 | # document; in this case we use the "mode=accumulate" attribute of the
|
|---|
| 69 | # particular Metadata element. In the second metadata element of the first
|
|---|
| 70 | # FileSet above, the "Place" metadata is accumulating, and may therefore be
|
|---|
| 71 | # given several values. If we wanted to override these values and use a
|
|---|
| 72 | # single metadata element again, we could set the mode attribute to
|
|---|
| 73 | # "override" instead. Remember: every element is assumed to be in override
|
|---|
| 74 | # mode unless you specify otherwise, so if you want to accumulate metadata
|
|---|
| 75 | # for some field, every occurance must have "mode=accumulate" specified.
|
|---|
| 76 | #
|
|---|
| 77 | # The second FileSet element above applies to a specific file, called
|
|---|
| 78 | # nugget-point-1.jpg. This element overrides the Title metadata set in the
|
|---|
| 79 | # first FileSet, and adds a "Subject" metadata field.
|
|---|
| 80 | #
|
|---|
| 81 | # The third and final FileSet sets metadata for a subdirectory rather than
|
|---|
| 82 | # a file. The metadata specified (a Title) will be passed into the
|
|---|
| 83 | # subdirectory and applied to every file that occurs in the subdirectory
|
|---|
| 84 | # (and to every subsubdirectory and its contents, and so on) unless the
|
|---|
| 85 | # metadata is explictly overridden later in the import.
|
|---|
| 86 |
|
|---|
| 87 | package MetadataXMLPlugin;
|
|---|
| 88 |
|
|---|
| 89 | use strict;
|
|---|
| 90 | no strict 'refs';
|
|---|
| 91 |
|
|---|
| 92 | use Encode;
|
|---|
| 93 |
|
|---|
| 94 | use BaseImporter;
|
|---|
| 95 | use extrametautil;
|
|---|
| 96 | use util;
|
|---|
| 97 | use FileUtils;
|
|---|
| 98 | use metadatautil;
|
|---|
| 99 |
|
|---|
| 100 | sub BEGIN {
|
|---|
| 101 | @MetadataXMLPlugin::ISA = ('BaseImporter');
|
|---|
| 102 | unshift (@INC, "$ENV{'GSDLHOME'}/perllib/cpan");
|
|---|
| 103 | }
|
|---|
| 104 |
|
|---|
| 105 | use XMLParser;
|
|---|
| 106 |
|
|---|
| 107 | my $arguments = [
|
|---|
| 108 | { 'name' => "process_exp",
|
|---|
| 109 | 'desc' => "{BaseImporter.process_exp}",
|
|---|
| 110 | 'type' => "regexp",
|
|---|
| 111 | 'reqd' => "no",
|
|---|
| 112 | 'deft' => &get_default_process_exp() }
|
|---|
| 113 |
|
|---|
| 114 | ];
|
|---|
| 115 |
|
|---|
| 116 | my $options = { 'name' => "MetadataXMLPlugin",
|
|---|
| 117 | 'desc' => "{MetadataXMLPlugin.desc}",
|
|---|
| 118 | 'abstract' => "no",
|
|---|
| 119 | 'inherits' => "yes",
|
|---|
| 120 | 'args' => $arguments };
|
|---|
| 121 |
|
|---|
| 122 | sub new {
|
|---|
| 123 | my ($class) = shift (@_);
|
|---|
| 124 | my ($pluginlist,$inputargs,$hashArgOptLists) = @_;
|
|---|
| 125 | push(@$pluginlist, $class);
|
|---|
| 126 |
|
|---|
| 127 | push(@{$hashArgOptLists->{"ArgList"}},@{$arguments});
|
|---|
| 128 | push(@{$hashArgOptLists->{"OptList"}},$options);
|
|---|
| 129 |
|
|---|
| 130 | my $self = new BaseImporter($pluginlist, $inputargs, $hashArgOptLists);
|
|---|
| 131 |
|
|---|
| 132 | if ($self->{'info_only'}) {
|
|---|
| 133 | # don't worry about any options or initialisations etc
|
|---|
| 134 | return bless $self, $class;
|
|---|
| 135 | }
|
|---|
| 136 |
|
|---|
| 137 | # The following used to be passed in as a parameter to XML::Parser,
|
|---|
| 138 | # if the version of perl was greater than or equal to 5.8.
|
|---|
| 139 | # The svn commit comment explaining the reason for adding this was
|
|---|
| 140 | # not very clear and also said that it was quick fix and hadn't
|
|---|
| 141 | # been tested under windows.
|
|---|
| 142 | # More recent work has been to make strings in Perl "Unicode-aware"
|
|---|
| 143 | # and so this line might actually be potentially harmful, however
|
|---|
| 144 | # it is not the case that we encountered an actual error leading to
|
|---|
| 145 | # its removal, rather it has been eliminated in an attempt to tighten
|
|---|
| 146 | # up the code. For example, this protocol encoding is not used in
|
|---|
| 147 | # ReadXMLFile.
|
|---|
| 148 | # 'ProtocolEncoding' => 'ISO-8859-1',
|
|---|
| 149 |
|
|---|
| 150 | # create XML::Parser object for parsing metadata.xml files
|
|---|
| 151 | my $parser = new XML::Parser('Style' => 'Stream',
|
|---|
| 152 | 'Pkg' => 'MetadataXMLPlugin',
|
|---|
| 153 | 'PluginObj' => $self,
|
|---|
| 154 | 'Handlers' => {'Char' => \&Char,
|
|---|
| 155 | 'Doctype' => \&Doctype
|
|---|
| 156 | });
|
|---|
| 157 |
|
|---|
| 158 | $self->{'parser'} = $parser;
|
|---|
| 159 | $self->{'in_filename'} = 0;
|
|---|
| 160 |
|
|---|
| 161 | return bless $self, $class;
|
|---|
| 162 | }
|
|---|
| 163 |
|
|---|
| 164 |
|
|---|
| 165 | sub get_default_process_exp
|
|---|
| 166 | {
|
|---|
| 167 | return q^metadata\.xml$^;
|
|---|
| 168 | }
|
|---|
| 169 |
|
|---|
| 170 | sub get_doctype {
|
|---|
| 171 | my $self = shift(@_);
|
|---|
| 172 |
|
|---|
| 173 | return "(Greenstone)?DirectoryMetadata"
|
|---|
| 174 | }
|
|---|
| 175 |
|
|---|
| 176 | sub can_process_this_file {
|
|---|
| 177 | my $self = shift(@_);
|
|---|
| 178 | my ($filename) = @_;
|
|---|
| 179 |
|
|---|
| 180 | if (-f $filename && $self->SUPER::can_process_this_file($filename) && $self->check_doctype($filename)) {
|
|---|
| 181 | return 1; # its a file for us
|
|---|
| 182 | }
|
|---|
| 183 | return 0;
|
|---|
| 184 | }
|
|---|
| 185 |
|
|---|
| 186 | sub check_doctype {
|
|---|
| 187 | my $self = shift (@_);
|
|---|
| 188 |
|
|---|
| 189 | my ($filename) = @_;
|
|---|
| 190 |
|
|---|
| 191 | if (open(XMLIN,"<$filename")) {
|
|---|
| 192 | my $doctype = $self->get_doctype();
|
|---|
| 193 | ## check whether the doctype has the same name as the root element tag
|
|---|
| 194 | while (defined (my $line = <XMLIN>)) {
|
|---|
| 195 | ## find the root element
|
|---|
| 196 | if ($line =~ /<([\w\d:]+)[\s\/>]/){
|
|---|
| 197 | my $root = $1;
|
|---|
| 198 | if ($root !~ $doctype){
|
|---|
| 199 | close(XMLIN);
|
|---|
| 200 | return 0;
|
|---|
| 201 | }
|
|---|
| 202 | else {
|
|---|
| 203 | close(XMLIN);
|
|---|
| 204 | return 1;
|
|---|
| 205 | }
|
|---|
| 206 | }
|
|---|
| 207 | }
|
|---|
| 208 | close(XMLIN);
|
|---|
| 209 | }
|
|---|
| 210 |
|
|---|
| 211 | return undef; # haven't found a valid line
|
|---|
| 212 |
|
|---|
| 213 | }
|
|---|
| 214 |
|
|---|
| 215 | sub file_block_read {
|
|---|
| 216 | my $self = shift (@_);
|
|---|
| 217 | my ($pluginfo, $base_dir, $file, $block_hash, $metadata, $gli) = @_;
|
|---|
| 218 |
|
|---|
| 219 | my $filename_full_path = &FileUtils::filenameConcatenate($base_dir, $file);
|
|---|
| 220 | return undef unless $self->can_process_this_file($filename_full_path);
|
|---|
| 221 |
|
|---|
| 222 | if (($ENV{'GSDLOS'} =~ m/^windows$/) && ($^O ne "cygwin")) {
|
|---|
| 223 | # convert to full name - paths stored in block hash are long filenames
|
|---|
| 224 | $filename_full_path = &util::upgrade_if_dos_filename($filename_full_path);
|
|---|
| 225 | my $lower_drive = $filename_full_path;
|
|---|
| 226 | $lower_drive =~ s/^([A-Z]):/\l$1:/i;
|
|---|
| 227 |
|
|---|
| 228 | my $upper_drive = $filename_full_path;
|
|---|
| 229 | $upper_drive =~ s/^([A-Z]):/\u$1:/i;
|
|---|
| 230 |
|
|---|
| 231 | $block_hash->{'metadata_files'}->{$lower_drive} = 1;
|
|---|
| 232 | $block_hash->{'metadata_files'}->{$upper_drive} = 1;
|
|---|
| 233 |
|
|---|
| 234 | }
|
|---|
| 235 | else {
|
|---|
| 236 | $block_hash->{'metadata_files'}->{$filename_full_path} = 1;
|
|---|
| 237 | }
|
|---|
| 238 |
|
|---|
| 239 | return 1;
|
|---|
| 240 | }
|
|---|
| 241 |
|
|---|
| 242 | sub metadata_read
|
|---|
| 243 | {
|
|---|
| 244 | my $self = shift (@_);
|
|---|
| 245 | my ($pluginfo, $base_dir, $file, $block_hash,
|
|---|
| 246 | $extrametakeys, $extrametadata,$extrametafile,
|
|---|
| 247 | $processor, $gli, $aux) = @_;
|
|---|
| 248 |
|
|---|
| 249 | my $filename = &FileUtils::filenameConcatenate($base_dir, $file);
|
|---|
| 250 | return undef unless $self->can_process_this_file($filename);
|
|---|
| 251 |
|
|---|
| 252 | $self->{'metadata-file'} = $file;
|
|---|
| 253 | $self->{'metadata-filename'} = $filename;
|
|---|
| 254 |
|
|---|
| 255 | my $outhandle = $self->{'outhandle'};
|
|---|
| 256 |
|
|---|
| 257 | print STDERR "\n<Processing n='$file' p='MetadataXMLPlugin'>\n" if ($gli);
|
|---|
| 258 | print $outhandle "MetadataXMLPlugin: processing $file\n" if ($self->{'verbosity'})> 1;
|
|---|
| 259 | # add the file to the block list so that it won't be processed in read, as we will do all we can with it here
|
|---|
| 260 | $self->block_raw_filename($block_hash,$filename);
|
|---|
| 261 |
|
|---|
| 262 | $self->{'metadataref'} = $extrametadata;
|
|---|
| 263 | $self->{'metafileref'} = $extrametafile;
|
|---|
| 264 | $self->{'metakeysref'} = $extrametakeys;
|
|---|
| 265 |
|
|---|
| 266 | eval {
|
|---|
| 267 | $self->{'parser'}->parsefile($filename);
|
|---|
| 268 | };
|
|---|
| 269 |
|
|---|
| 270 | if ($@) {
|
|---|
| 271 | print STDERR "**** Error is: $@\n";
|
|---|
| 272 | my $plugin_name = ref ($self);
|
|---|
| 273 | my $failhandle = $self->{'failhandle'};
|
|---|
| 274 | print $outhandle "$plugin_name failed to process $file ($@)\n";
|
|---|
| 275 | print $failhandle "$plugin_name failed to process $file ($@)\n";
|
|---|
| 276 | print STDERR "<ProcessingError n='$file'>\n" if ($gli);
|
|---|
| 277 | return -1; #error
|
|---|
| 278 | }
|
|---|
| 279 |
|
|---|
| 280 | return 1;
|
|---|
| 281 |
|
|---|
| 282 | }
|
|---|
| 283 |
|
|---|
| 284 |
|
|---|
| 285 | # Updated by Jeffrey 2010/04/16 @ DL Consulting Ltd.
|
|---|
| 286 | # Get rid off the global $self as it cause problems when there are 2+ MetadataXMLPlugin in your collect.cfg...
|
|---|
| 287 | # For example when you have an OAIMetadataXMLPlugin that is a child of MetadataXMLPlugin
|
|---|
| 288 | sub Doctype {$_[0]->{'PluginObj'}->xml_doctype(@_);}
|
|---|
| 289 | sub StartTag {$_[0]->{'PluginObj'}->xml_start_tag(@_);}
|
|---|
| 290 | sub EndTag {$_[0]->{'PluginObj'}->xml_end_tag(@_);}
|
|---|
| 291 | sub Text {$_[0]->{'PluginObj'}->xml_text(@_);}
|
|---|
| 292 |
|
|---|
| 293 |
|
|---|
| 294 | sub xml_doctype {
|
|---|
| 295 | my $self = shift(@_);
|
|---|
| 296 | my ($expat, $name, $sysid, $pubid, $internal) = @_;
|
|---|
| 297 |
|
|---|
| 298 | # allow the short-lived and badly named "GreenstoneDirectoryMetadata" files
|
|---|
| 299 | # to be processed as well as the "DirectoryMetadata" files which should now
|
|---|
| 300 | # be created by import.pl
|
|---|
| 301 | die if ($name !~ /^(Greenstone)?DirectoryMetadata$/);
|
|---|
| 302 | }
|
|---|
| 303 |
|
|---|
| 304 | sub xml_start_tag {
|
|---|
| 305 | my $self = shift(@_);
|
|---|
| 306 | my ($expat, $element) = @_;
|
|---|
| 307 |
|
|---|
| 308 | if ($element eq "FileSet") {
|
|---|
| 309 | $self->{'saved_targets'} = [];
|
|---|
| 310 | $self->{'saved_metadata'} = {};
|
|---|
| 311 | }
|
|---|
| 312 | elsif ($element eq "FileName") {
|
|---|
| 313 | $self->{'in_filename'} = 1;
|
|---|
| 314 | }
|
|---|
| 315 | elsif ($element eq "Metadata") {
|
|---|
| 316 | $self->{'metadata_name'} = $_{'name'};
|
|---|
| 317 | $self->{'metadata_value'} = "";
|
|---|
| 318 | if ((defined $_{'mode'}) && ($_{'mode'} eq "accumulate")) {
|
|---|
| 319 | $self->{'metadata_accumulate'} = 1;
|
|---|
| 320 | } else {
|
|---|
| 321 | $self->{'metadata_accumulate'} = 0;
|
|---|
| 322 | }
|
|---|
| 323 | }
|
|---|
| 324 | }
|
|---|
| 325 |
|
|---|
| 326 | sub xml_end_tag {
|
|---|
| 327 | my $self = shift(@_);
|
|---|
| 328 | my ($expat, $element) = @_;
|
|---|
| 329 |
|
|---|
| 330 | if ($element eq "FileSet") {
|
|---|
| 331 | foreach my $target (@{$self->{'saved_targets'}}) {
|
|---|
| 332 |
|
|---|
| 333 | # FileNames must be regex, but we allow \\ for path separator on windows. convert to /
|
|---|
| 334 | $target = &util::filepath_regex_to_url_format($target);
|
|---|
| 335 |
|
|---|
| 336 | # we want proper unicode for the regex, so convert url-encoded chars
|
|---|
| 337 | if (&unicode::is_url_encoded($target)) {
|
|---|
| 338 | $target = &unicode::url_decode($target);
|
|---|
| 339 | }
|
|---|
| 340 |
|
|---|
| 341 | my $file_metadata = &extrametautil::getmetadata($self->{'metadataref'}, $target);
|
|---|
| 342 | my $saved_metadata = $self->{'saved_metadata'};
|
|---|
| 343 |
|
|---|
| 344 | if (!defined $file_metadata) {
|
|---|
| 345 | &extrametautil::setmetadata($self->{'metadataref'}, $target, $saved_metadata);
|
|---|
| 346 |
|
|---|
| 347 | # not had target before
|
|---|
| 348 | &extrametautil::addmetakey($self->{'metakeysref'}, $target);
|
|---|
| 349 | }
|
|---|
| 350 | else {
|
|---|
| 351 | &metadatautil::combine_metadata_structures($file_metadata,$saved_metadata);
|
|---|
| 352 | }
|
|---|
| 353 |
|
|---|
| 354 |
|
|---|
| 355 | # now record which metadata.xml file it came from
|
|---|
| 356 |
|
|---|
| 357 | my $file = $self->{'metadata-file'};
|
|---|
| 358 | my $filename = $self->{'metadata-filename'};
|
|---|
| 359 |
|
|---|
| 360 | if (!defined &extrametautil::getmetafile($self->{'metafileref'}, $target)) {
|
|---|
| 361 | &extrametautil::setmetafile($self->{'metafileref'}, $target, {});
|
|---|
| 362 | }
|
|---|
| 363 |
|
|---|
| 364 | &extrametautil::setmetafile_for_named_file($self->{'metafileref'}, $target, $file, $filename);
|
|---|
| 365 | }
|
|---|
| 366 | }
|
|---|
| 367 | elsif ($element eq "FileName") {
|
|---|
| 368 | $self->{'in_filename'} = 0;
|
|---|
| 369 | }
|
|---|
| 370 | elsif ($element eq "Metadata") {
|
|---|
| 371 | # text read in by XML::Parser is in Perl's binary byte value
|
|---|
| 372 | # form ... need to explicitly make it UTF-8
|
|---|
| 373 |
|
|---|
| 374 | my $metadata_name = $self->{'metadata_name'};
|
|---|
| 375 | my $metadata_value = $self->{'metadata_value'};
|
|---|
| 376 | #my $metadata_name = decode("utf-8",$self->{'metadata_name'});
|
|---|
| 377 | #my $metadata_value = decode("utf-8",$self->{'metadata_value'});
|
|---|
| 378 |
|
|---|
| 379 | &metadatautil::store_saved_metadata($self,
|
|---|
| 380 | $metadata_name, $metadata_value,
|
|---|
| 381 | $self->{'metadata_accumulate'});
|
|---|
| 382 | $self->{'metadata_name'} = "";
|
|---|
| 383 | }
|
|---|
| 384 |
|
|---|
| 385 | }
|
|---|
| 386 |
|
|---|
| 387 | sub xml_text {
|
|---|
| 388 | my $self = shift(@_);
|
|---|
| 389 |
|
|---|
| 390 | if ($self->{'in_filename'}) {
|
|---|
| 391 | # $_ == FileName content
|
|---|
| 392 | push (@{$self->{'saved_targets'}}, $_);
|
|---|
| 393 | }
|
|---|
| 394 | elsif (defined ($self->{'metadata_name'}) && $self->{'metadata_name'} ne "") {
|
|---|
| 395 | # $_ == Metadata content
|
|---|
| 396 | $self->{'metadata_value'} = $_;
|
|---|
| 397 | }
|
|---|
| 398 | }
|
|---|
| 399 |
|
|---|
| 400 | # This Char function overrides the one in XML::Parser::Stream to overcome a
|
|---|
| 401 | # problem where $expat->{Text} is treated as the return value, slowing
|
|---|
| 402 | # things down significantly in some cases.
|
|---|
| 403 | sub Char {
|
|---|
| 404 | # use bytes; # Necessary to prevent encoding issues with XML::Parser 2.31+
|
|---|
| 405 |
|
|---|
| 406 | # if ($]<5.008) {
|
|---|
| 407 | # use bytes; # Necessary to prevent encoding issues with XML::Parser 2.31+ and Perl 5.6
|
|---|
| 408 | # }
|
|---|
| 409 | $_[0]->{'Text'} .= $_[1];
|
|---|
| 410 | return undef;
|
|---|
| 411 | }
|
|---|
| 412 |
|
|---|
| 413 |
|
|---|
| 414 |
|
|---|
| 415 | 1;
|
|---|
