[10549] | 1 |
|
---|
[8511] | 2 | ###########################################################################
|
---|
| 3 | #
|
---|
[15872] | 4 | # DSpacePlugin.pm -- plugin for importing a collection from DSpace
|
---|
[8511] | 5 | #
|
---|
| 6 | # A component of the Greenstone digital library software
|
---|
| 7 | # from the New Zealand Digital Library Project at the
|
---|
| 8 | # University of Waikato, New Zealand.
|
---|
| 9 | #
|
---|
[15872] | 10 | # Copyright (C) 2004 New Zealand Digital Library Project
|
---|
[8511] | 11 | #
|
---|
| 12 | # This program is free software; you can redistribute it and/or modify
|
---|
| 13 | # it under the terms of the GNU General Public License as published by
|
---|
| 14 | # the Free Software Foundation; either version 2 of the License, or
|
---|
| 15 | # (at your option) any later version.
|
---|
| 16 | #
|
---|
| 17 | # This program is distributed in the hope that it will be useful,
|
---|
| 18 | # but WITHOUT ANY WARRANTY; without even the implied warranty of
|
---|
| 19 | # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
---|
| 20 | # GNU General Public License for more details.
|
---|
| 21 | #
|
---|
| 22 | # You should have received a copy of the GNU General Public License
|
---|
| 23 | # along with this program; if not, write to the Free Software
|
---|
| 24 | # Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
|
---|
| 25 | #
|
---|
| 26 | ###########################################################################
|
---|
| 27 |
|
---|
[15872] | 28 |
|
---|
[8511] | 29 | # This plugin takes "contents" and dublin_core.xml file, which contain
|
---|
| 30 | # Metadata and lists of associated files for a particular document
|
---|
| 31 | # and produces a document containing sections, one for each page.
|
---|
| 32 | # The files should be named "contents" and "dublin_core.xml". For each of
|
---|
| 33 | # document in DSpace, it is stored in one directory
|
---|
| 34 | #
|
---|
| 35 | # The format of the "contents" file is as follows:
|
---|
| 36 | #
|
---|
| 37 | # File.type bundle:ORIGINAL
|
---|
| 38 | # license.txt bundle:LICENSE
|
---|
| 39 | # The format of the "dublin_core.xml" file is as follows:
|
---|
| 40 | # The first line contains any metadata for the whole document
|
---|
| 41 | # <dublin_core>
|
---|
| 42 | # eg.
|
---|
| 43 | # <dcvalue element="Title" qualifier="">Snail farming</dcvalue>
|
---|
| 44 | # <dcvalue element="date" qualifier="">2004-10-15</dcvalue>
|
---|
| 45 | #
|
---|
| 46 |
|
---|
[15872] | 47 | package DSpacePlugin;
|
---|
[8511] | 48 |
|
---|
[19746] | 49 | use ReadTextFile;
|
---|
[8511] | 50 | use plugin;
|
---|
| 51 | use XMLParser;
|
---|
[10254] | 52 | use strict;
|
---|
| 53 | no strict 'refs'; # allow filehandles to be variables and viceversa
|
---|
[8511] | 54 |
|
---|
| 55 | sub BEGIN {
|
---|
[19746] | 56 | @DSpacePlugin::ISA = ('ReadTextFile');
|
---|
[8511] | 57 | }
|
---|
| 58 |
|
---|
| 59 | my $arguments =
|
---|
| 60 | [ { 'name' => "process_exp",
|
---|
[15872] | 61 | 'desc' => "{BasePlugin.process_exp}",
|
---|
[8511] | 62 | 'type' => "string",
|
---|
| 63 | 'deft' => &get_default_process_exp(),
|
---|
| 64 | 'reqd' => "no" },
|
---|
| 65 | { 'name' => "only_first_doc",
|
---|
[15872] | 66 | 'desc' => "{DSpacePlugin.only_first_doc}",
|
---|
[8511] | 67 | 'type' => "flag",
|
---|
| 68 | 'reqd' => "no" },
|
---|
[9168] | 69 | { 'name' => "first_inorder_ext",
|
---|
[15872] | 70 | 'desc' => "{DSpacePlugin.first_inorder_ext}",
|
---|
[8891] | 71 | 'type' => "string",
|
---|
[8511] | 72 | 'reqd' => "no" },
|
---|
| 73 | { 'name' => "first_inorder_mime",
|
---|
[15872] | 74 | 'desc' => "{DSpacePlugin.first_inorder_mime}",
|
---|
[8511] | 75 | 'type' => "flag",
|
---|
| 76 | 'reqd' => "no" },
|
---|
| 77 | { 'name' => "block_exp",
|
---|
[15872] | 78 | 'desc' => "{BasePlugin.block_exp}",
|
---|
[8891] | 79 | 'type' => "regexp",
|
---|
[8511] | 80 | 'deft' => &get_default_block_exp(),
|
---|
| 81 | 'reqd' => "no" }];
|
---|
| 82 |
|
---|
| 83 |
|
---|
[15872] | 84 | my $options = { 'name' => "DSpacePlugin",
|
---|
| 85 | 'desc' => "{DSpacePlugin.desc}",
|
---|
[8511] | 86 | 'inherits' => "yes",
|
---|
[11676] | 87 | 'abstract' => "no",
|
---|
[8511] | 88 | 'args' => $arguments };
|
---|
| 89 |
|
---|
| 90 |
|
---|
| 91 | my $primary_doc_lookup = { 'text/html' => '(?i)\.(gif|jpe?g|jpe|jpg|png|css)$' };
|
---|
| 92 |
|
---|
[8913] | 93 | # Important variation to regular plugin structure. Need to declare
|
---|
| 94 | # $self as global variable to file so XMLParser callback routines
|
---|
| 95 | # can access the content of the object.
|
---|
| 96 | my ($self);
|
---|
| 97 |
|
---|
[8511] | 98 | sub new {
|
---|
[10218] | 99 | my ($class) = shift (@_);
|
---|
| 100 | my ($pluginlist,$inputargs,$hashArgOptLists) = @_;
|
---|
| 101 | push(@$pluginlist, $class);
|
---|
[8511] | 102 |
|
---|
[15872] | 103 | push(@{$hashArgOptLists->{"ArgList"}},@{$arguments});
|
---|
| 104 | push(@{$hashArgOptLists->{"OptList"}},$options);
|
---|
[8511] | 105 |
|
---|
[19746] | 106 | $self = new ReadTextFile($pluginlist, $inputargs, $hashArgOptLists);
|
---|
[8511] | 107 |
|
---|
[15872] | 108 | if ($self->{'info_only'}) {
|
---|
| 109 | # don't worry about creating the XML parser as all we want is the
|
---|
| 110 | # list of plugin options
|
---|
| 111 | return bless $self, $class;
|
---|
| 112 | }
|
---|
| 113 |
|
---|
[8511] | 114 | #create XML::Parser object for parsing dublin_core.xml files
|
---|
| 115 | my $parser = new XML::Parser('Style' => 'Stream',
|
---|
| 116 | 'Handlers' => {'Char' => \&Char,
|
---|
| 117 | 'Doctype' => \&Doctype
|
---|
| 118 | });
|
---|
| 119 | $self->{'parser'} = $parser;
|
---|
| 120 | $self->{'extra_blocks'} = {};
|
---|
| 121 |
|
---|
| 122 | return bless $self, $class;
|
---|
| 123 | }
|
---|
| 124 |
|
---|
| 125 | sub get_default_process_exp {
|
---|
| 126 | my $self = shift (@_);
|
---|
| 127 |
|
---|
[19746] | 128 | return q^(?i)(contents)$^;
|
---|
[8511] | 129 | }
|
---|
| 130 |
|
---|
| 131 | # want to block all files except the "contents"
|
---|
| 132 | sub get_default_block_exp {
|
---|
| 133 | my $self = shift (@_);
|
---|
| 134 |
|
---|
[19746] | 135 | # Block handle and txt files if present. Specifically don't block dublin_core xml
|
---|
| 136 | return q^(?i)(handle|\.tx?t)$^;
|
---|
[8511] | 137 | }
|
---|
| 138 |
|
---|
[19746] | 139 | sub store_block_files_BACKUP
|
---|
| 140 | {
|
---|
| 141 | # Option of making blocking sensitive to files that are in directory
|
---|
| 142 | # This subroutine is not currently used! (relies on default block expression stopping all handle and .txt files)
|
---|
| 143 |
|
---|
| 144 | my $self =shift (@_);
|
---|
| 145 | my ($filename_full_path, $block_hash) = @_;
|
---|
| 146 |
|
---|
| 147 | my ($tailname, $contents_basedir, $suffix) = &File::Basename::fileparse($filename_full_path, "\\.[^\\.]+\$");
|
---|
| 148 | my $handle_filename = &util::filename_cat($contents_basedir,"handle");
|
---|
| 149 |
|
---|
| 150 | $block_hash->{'file_blocks'}->{$handle_filename} = 1 if (-e $handle_filename);
|
---|
| 151 | }
|
---|
| 152 |
|
---|
[8511] | 153 | sub read_content
|
---|
| 154 | {
|
---|
| 155 | my $self = shift (@_);
|
---|
| 156 | my ($dir, $only_first_doc, $first_inorder_ext, $first_inorder_mime, $mimetype_list) = @_;
|
---|
| 157 | my $outhandle = $self->{'outhandle'};
|
---|
| 158 |
|
---|
| 159 | my @fnamemime_list = ();
|
---|
| 160 | my @assocmime_list = ();
|
---|
| 161 |
|
---|
| 162 | my $content_fname = &util::filename_cat($dir,"contents");
|
---|
| 163 |
|
---|
| 164 | open(CIN,"<$content_fname")
|
---|
| 165 | || die "Unable to open $content_fname: $!\n";
|
---|
| 166 |
|
---|
| 167 | my $line;
|
---|
| 168 | my $pos = 0;
|
---|
| 169 |
|
---|
| 170 | while (defined ($line = <CIN>)) {
|
---|
| 171 | if ($line =~ m/^(.*)\s+bundle:ORIGINAL\s*$/) {
|
---|
| 172 | my $fname = $1;
|
---|
| 173 | my $mtype = $mimetype_list->[$pos];
|
---|
| 174 | my $fm_rec = { 'file' => $fname, 'mimetype' => $mtype};
|
---|
| 175 | push(@fnamemime_list,$fm_rec);
|
---|
| 176 | $pos++;
|
---|
| 177 | }
|
---|
| 178 | }
|
---|
| 179 | close CIN;
|
---|
| 180 |
|
---|
| 181 | if ($only_first_doc){
|
---|
| 182 | my ($first_fname, @rest_fnames) = @fnamemime_list;
|
---|
| 183 | @fnamemime_list = ($first_fname);
|
---|
| 184 | @assocmime_list = @rest_fnames;
|
---|
| 185 | }
|
---|
| 186 |
|
---|
| 187 | # allow user to specify the types of files (inorder)they would like to assign as
|
---|
| 188 | # a primary bitstream
|
---|
| 189 | if ($first_inorder_ext) {
|
---|
| 190 | # parse user-define file extension names
|
---|
| 191 | my @extfiles_list = split /,/, $first_inorder_ext;
|
---|
| 192 | my (@rest_fnames) = @fnamemime_list;
|
---|
| 193 | my @matched_list = ();
|
---|
| 194 | foreach my $file_ext (@extfiles_list) {
|
---|
| 195 | $pos = 0;
|
---|
| 196 | foreach my $allfiles (@fnamemime_list){
|
---|
| 197 | $allfiles->{'file'} =~ /^(.*)\.(.*?)$/;
|
---|
| 198 | my $allfiles_ext = $2;
|
---|
| 199 |
|
---|
| 200 | if ($allfiles_ext =~ /$file_ext/) {
|
---|
| 201 | print $outhandle "Existing file:$allfiles->{'file'} match the user-define File Extension:$file_ext\n";
|
---|
| 202 | push (@matched_list, $allfiles);
|
---|
| 203 |
|
---|
| 204 | # delete the matched extension file from the array
|
---|
| 205 | splice(@rest_fnames,$pos,1);
|
---|
| 206 |
|
---|
| 207 | return (\@matched_list, \@rest_fnames);
|
---|
| 208 |
|
---|
| 209 | }
|
---|
| 210 | $pos++;
|
---|
| 211 | }
|
---|
| 212 | }
|
---|
| 213 | }
|
---|
| 214 |
|
---|
| 215 | if ($first_inorder_mime) {
|
---|
| 216 | # parse user-define file mimetype
|
---|
| 217 | my @file_mime_list = split /,/, $first_inorder_mime;
|
---|
| 218 | my (@rest_fnames) = @fnamemime_list;
|
---|
| 219 | my @matched_list = ();
|
---|
| 220 | foreach my $file_mime (@file_mime_list) {
|
---|
| 221 | $pos = 0;
|
---|
| 222 | foreach my $allfiles (@fnamemime_list){
|
---|
| 223 | my $allfiles_mime = $allfiles->{'mimetype'};
|
---|
| 224 |
|
---|
| 225 | if ($allfiles_mime =~ /$file_mime/) {
|
---|
| 226 | print $outhandle "Existing file:$allfiles->{'file'} match the user-defined File MimeType:$file_mime\n";
|
---|
| 227 | push (@matched_list, $allfiles);
|
---|
| 228 |
|
---|
| 229 | # delete the matched MIMEType file from the array
|
---|
| 230 | splice(@rest_fnames,$pos,1);
|
---|
| 231 | return (\@matched_list, \@rest_fnames);
|
---|
| 232 | }
|
---|
| 233 | $pos++;
|
---|
| 234 | }
|
---|
| 235 | }
|
---|
| 236 | }
|
---|
| 237 | return (\@fnamemime_list, \@assocmime_list);
|
---|
| 238 | }
|
---|
| 239 |
|
---|
| 240 |
|
---|
| 241 | sub filemime_list_to_re
|
---|
| 242 | {
|
---|
| 243 | my $self = shift (@_);
|
---|
| 244 | my ($fnamemime_list) = @_;
|
---|
| 245 |
|
---|
| 246 | my @fname_list = map { "(".$_->{'file'}.")" } @$fnamemime_list;
|
---|
| 247 | my $fname_re = join("|",@fname_list);
|
---|
| 248 | $fname_re =~ s/\./\\\./g;
|
---|
| 249 |
|
---|
| 250 | return $fname_re;
|
---|
| 251 | }
|
---|
| 252 |
|
---|
| 253 | # Read dublin_core metadata from DSpace collection
|
---|
| 254 | sub metadata_read {
|
---|
| 255 | my $self = shift (@_);
|
---|
[19493] | 256 | my ($pluginfo, $base_dir, $file, $block_hash,
|
---|
| 257 | $extrametakeys, $extrametadata, $extrametafile,
|
---|
| 258 | $processor, $maxdocs, $gli) = @_;
|
---|
[8511] | 259 |
|
---|
| 260 | my $only_first_doc = $self->{'only_first_doc'};
|
---|
| 261 | my $first_inorder_ext = $self->{'first_inorder_ext'};
|
---|
| 262 | my $first_inorder_mime = $self->{'first_inorder_mime'};
|
---|
| 263 |
|
---|
| 264 | my $outhandle = $self->{'outhandle'};
|
---|
| 265 |
|
---|
| 266 | my $filename = &util::filename_cat($base_dir, $file);
|
---|
| 267 | # return 0 if $self->{'block_exp'} ne "" && $filename =~ /$self->{'block_exp'}/;
|
---|
| 268 |
|
---|
| 269 | if ($filename !~ /dublin_core\.xml$/ || !-f $filename) {
|
---|
| 270 | return undef;
|
---|
| 271 | }
|
---|
| 272 |
|
---|
[15872] | 273 | print $outhandle "DSpacePlugin: extracting metadata from $file\n"
|
---|
[8511] | 274 | if $self->{'verbosity'} > 1;
|
---|
| 275 |
|
---|
| 276 | my ($dir) = $filename =~ /^(.*?)[^\/\\]*$/;
|
---|
| 277 |
|
---|
| 278 | eval {
|
---|
| 279 | $self->{'parser'}->parsefile($filename);
|
---|
| 280 | };
|
---|
| 281 |
|
---|
| 282 | if ($@) {
|
---|
[15872] | 283 | die "DSpacePlugin: ERROR $filename is not a well formed dublin_core.xml file ($@)\n";
|
---|
[8511] | 284 | }
|
---|
| 285 |
|
---|
| 286 | my $mimetype_list = $self->{'saved_metadata'}->{'dc.Format^mimetype'};
|
---|
| 287 | my ($doc_file_mimes, $assoc_file_mimes) = $self->read_content($dir, $only_first_doc, $first_inorder_ext,
|
---|
| 288 | $first_inorder_mime, $mimetype_list);
|
---|
[8913] | 289 |
|
---|
[8511] | 290 | my $file_re = $self->filemime_list_to_re($doc_file_mimes);
|
---|
| 291 |
|
---|
| 292 | if ($only_first_doc || $first_inorder_ext || $first_inorder_mime) {
|
---|
| 293 | foreach my $afm ( @$assoc_file_mimes ) {
|
---|
| 294 | my $full_af = &util::filename_cat($dir,$afm->{'file'});
|
---|
| 295 | $self->{'extra_blocks'}->{$full_af} = 1;
|
---|
| 296 | }
|
---|
| 297 | }
|
---|
| 298 | push(@$extrametakeys,$file_re);
|
---|
| 299 |
|
---|
| 300 | if (defined $self->{'saved_metadata'}->{'dc.Format^extent'}) {
|
---|
| 301 | delete $self->{'saved_metadata'}->{'dc.Format^extent'};
|
---|
| 302 | }
|
---|
| 303 |
|
---|
| 304 | if (defined $mimetype_list) {
|
---|
| 305 | delete $self->{'saved_metadata'}->{'dc.Format^mimetype'};
|
---|
| 306 |
|
---|
| 307 | # Temporarily store associate file info in metadata table
|
---|
[15872] | 308 | # This will be removed in 'extra_metadata' in BasePlugin and used
|
---|
[8511] | 309 | # to perform the actual file association (once the doc obj has
|
---|
| 310 | # been formed
|
---|
| 311 |
|
---|
| 312 | my $main_doc = $doc_file_mimes->[0];
|
---|
| 313 | my $md_mimetype = $main_doc->{'mimetype'};
|
---|
| 314 |
|
---|
| 315 | my $pd_lookup = $primary_doc_lookup->{$md_mimetype};
|
---|
| 316 |
|
---|
| 317 | if (defined $pd_lookup) {
|
---|
| 318 | my $filter_re = $pd_lookup;
|
---|
| 319 | @$assoc_file_mimes = grep { $_->{'file'} !~ m/$filter_re/ } @$assoc_file_mimes;
|
---|
| 320 | }
|
---|
| 321 |
|
---|
| 322 | my @gsdlassocfile_tobe
|
---|
| 323 | = map { &util::filename_cat($dir,$_->{'file'}).":".$_->{'mimetype'}.":" } @$assoc_file_mimes;
|
---|
| 324 | $self->{'saved_metadata'}->{'gsdlassocfile_tobe'} = \@gsdlassocfile_tobe;
|
---|
| 325 |
|
---|
| 326 | }
|
---|
[8913] | 327 |
|
---|
[8511] | 328 | $extrametadata->{$file_re} = $self->{'saved_metadata'};
|
---|
| 329 |
|
---|
| 330 | return 1;
|
---|
| 331 | }
|
---|
| 332 |
|
---|
| 333 |
|
---|
[15872] | 334 | # The DSpacePlugin read() function. We are not actually reading any documents
|
---|
| 335 | # here, just blocking ones that have been processed by metadata read.
|
---|
[8511] | 336 | #
|
---|
[15872] | 337 | # Returns 0 for a file its blocking, undef for any other
|
---|
[8511] | 338 | sub read {
|
---|
| 339 | my $self = shift (@_);
|
---|
[16392] | 340 | my ($pluginfo, $base_dir, $file, $block_hash, $metadata, $processor, $maxdocs, $total_count, $gli) = @_;
|
---|
[8511] | 341 | my $outhandle = $self->{'outhandle'};
|
---|
| 342 |
|
---|
| 343 | # Block all files except contents
|
---|
| 344 | my $filename = &util::filename_cat($base_dir, $file);
|
---|
| 345 | return 0 if $self->{'block_exp'} ne "" && $filename =~ /$self->{'block_exp'}/;
|
---|
| 346 |
|
---|
| 347 | my $assocfile = $metadata->{'assocfile'};
|
---|
| 348 |
|
---|
| 349 | return 0 if (($filename =~ /dublin_core\.xml$/) || ($filename =~ /contents$/));
|
---|
| 350 | return 0 if (defined $self->{'extra_blocks'}->{$filename});
|
---|
| 351 | return undef;
|
---|
| 352 | }
|
---|
| 353 |
|
---|
| 354 | sub Doctype {
|
---|
| 355 | my ($expat, $name, $sysid, $pubid, $internal) = @_;
|
---|
| 356 |
|
---|
| 357 | die if ($name !~ /^dublin_core$/);
|
---|
| 358 | }
|
---|
| 359 |
|
---|
| 360 | sub StartTag {
|
---|
| 361 | my ($expat, $element) = @_;
|
---|
| 362 | if ($element eq "dublin_core") {
|
---|
| 363 | $self->{'saved_metadata'} = {};
|
---|
[8913] | 364 | } elsif ($element eq "dcvalue") {
|
---|
[8511] | 365 | my $metaname = $_{'element'};
|
---|
[10549] | 366 | my $qualifier = $_{'qualifier'}||"";
|
---|
[21367] | 367 | if ($metaname ne "description" || $qualifier ne "provenance") {
|
---|
[9694] | 368 | $metaname .= "^$qualifier" if ($qualifier ne "none" && $qualifier ne "");
|
---|
[8511] | 369 | $self->{'metaname'} = "dc.\u$metaname";
|
---|
| 370 | }
|
---|
| 371 | }
|
---|
| 372 | }
|
---|
| 373 |
|
---|
| 374 | sub EndTag {
|
---|
| 375 | my ($expat, $element) = @_;
|
---|
| 376 |
|
---|
| 377 | if ($element eq "dcvalue") {
|
---|
| 378 | $self->{'metaname'} = "";
|
---|
| 379 | }
|
---|
| 380 | }
|
---|
| 381 |
|
---|
| 382 | sub Text {
|
---|
| 383 | if (defined ($self->{'metaname'}) && $self->{'metaname'} ne "") {
|
---|
| 384 | # $_ == Metadata content
|
---|
| 385 | my $mname = $self->{'metaname'};
|
---|
[21367] | 386 | my $mvalue = prepareMetadataValue($_);
|
---|
[8511] | 387 | if (defined $self->{'saved_metadata'}->{$mname}) {
|
---|
| 388 | # accumulate - add value to existing value(s)
|
---|
| 389 | if (ref ($self->{'saved_metadata'}->{$mname}) eq "ARRAY") {
|
---|
[21367] | 390 | push (@{$self->{'saved_metadata'}->{$mname}}, $mvalue);
|
---|
[8511] | 391 | } else {
|
---|
| 392 | $self->{'saved_metadata'}->{$mname} =
|
---|
[21367] | 393 | [$self->{'saved_metadata'}->{$mname}, $mvalue];
|
---|
[8511] | 394 | }
|
---|
| 395 | } else {
|
---|
| 396 | # accumulate - add value into (currently empty) array
|
---|
[21367] | 397 | $self->{'saved_metadata'}->{$mname} = [$mvalue];
|
---|
[8511] | 398 | }
|
---|
| 399 |
|
---|
| 400 | }
|
---|
| 401 | }
|
---|
| 402 |
|
---|
[21367] | 403 | # Prepare DSpace metadata for using with Greenstone.
|
---|
| 404 | # Some value must be escaped.
|
---|
| 405 | sub prepareMetadataValue {
|
---|
| 406 | my ($value) = @_;
|
---|
| 407 |
|
---|
| 408 | $value =~ s/\[/[/g;
|
---|
| 409 | $value =~ s/\]/]/g;
|
---|
| 410 |
|
---|
| 411 | return $value;
|
---|
| 412 | }
|
---|
[8511] | 413 | # This Char function overrides the one in XML::Parser::Stream to overcome a
|
---|
| 414 | # problem where $expat->{Text} is treated as the return value, slowing
|
---|
| 415 | # things down significantly in some cases.
|
---|
| 416 | sub Char {
|
---|
[9462] | 417 | use bytes; # Necessary to prevent encoding issues with XML::Parser 2.31+
|
---|
| 418 | $_[0]->{'Text'} .= $_[1];
|
---|
| 419 | return undef;
|
---|
[8511] | 420 | }
|
---|
| 421 |
|
---|
| 422 | 1;
|
---|