[12330] | 1 | ###########################################################################
|
---|
| 2 | #
|
---|
| 3 | # BasPlugout.pm -- base class for all the plugout modules
|
---|
| 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 | package BasPlugout;
|
---|
| 27 |
|
---|
| 28 | eval {require bytes};
|
---|
| 29 |
|
---|
| 30 | use strict;
|
---|
| 31 | no strict 'subs';
|
---|
[12459] | 32 | no strict 'refs';
|
---|
[12330] | 33 |
|
---|
| 34 | use gsprintf 'gsprintf';
|
---|
| 35 | use printusage;
|
---|
[12546] | 36 | use parse2;
|
---|
[12330] | 37 |
|
---|
| 38 | # suppress the annoying "subroutine redefined" warning that various
|
---|
| 39 | # gets cause under perl 5.6
|
---|
| 40 | $SIG{__WARN__} = sub {warn($_[0]) unless ($_[0] =~ /Subroutine\s+\S+\sredefined/)};
|
---|
| 41 |
|
---|
| 42 | my $arguments = [
|
---|
[13172] | 43 | { 'name' => "group_size",
|
---|
[12330] | 44 | 'desc' => "{BasPlugout.group_size}",
|
---|
| 45 | 'type' => "int",
|
---|
| 46 | 'deft' => "1",
|
---|
| 47 | 'reqd' => "no",
|
---|
| 48 | 'hiddengli' => "no"},
|
---|
| 49 | { 'name' => "output_info",
|
---|
| 50 | 'desc' => "{BasPlugout.output_info}",
|
---|
| 51 | 'type' => "string",
|
---|
| 52 | 'reqd' => "yes",
|
---|
| 53 | 'hiddengli' => "yes"},
|
---|
| 54 | { 'name' => "xslt_file",
|
---|
| 55 | 'desc' => "{BasPlugout.xslt_file}",
|
---|
| 56 | 'type' => "string",
|
---|
| 57 | 'reqd' => "no",
|
---|
| 58 | 'hiddengli' => "no"},
|
---|
| 59 | { 'name' => "output_handle",
|
---|
| 60 | 'desc' => "{BasPlugout.output_handle}",
|
---|
| 61 | 'type' => "string",
|
---|
| 62 | 'deft' => 'STDERR',
|
---|
| 63 | 'reqd' => "no",
|
---|
| 64 | 'hiddengli' => "yes"},
|
---|
| 65 | { 'name' => "verbosity",
|
---|
| 66 | 'desc' => "{BasPlugout.verbosity}",
|
---|
| 67 | 'type' => "int",
|
---|
| 68 | 'deft' => "0",
|
---|
| 69 | 'reqd' => "no",
|
---|
| 70 | 'hiddengli' => "no"},
|
---|
| 71 | { 'name' => "gzip_output",
|
---|
| 72 | 'desc' => "{BasPlugout.gzip_output}",
|
---|
| 73 | 'type' => "flag",
|
---|
| 74 | 'reqd' => "no",
|
---|
[13172] | 75 | 'hiddengli' => "no"},
|
---|
| 76 | { 'name' => "debug",
|
---|
| 77 | 'desc' => "{BasPlugout.debug}",
|
---|
| 78 | 'type' => "flag",
|
---|
| 79 | 'reqd' => "no",
|
---|
| 80 | 'hiddengli' => "yes"}
|
---|
[12330] | 81 | ];
|
---|
| 82 |
|
---|
| 83 | my $options = { 'name' => "BasPlugout",
|
---|
| 84 | 'desc' => "{BasPlugout.desc}",
|
---|
| 85 | 'abstract' => "yes",
|
---|
| 86 | 'inherits' => "no",
|
---|
| 87 | 'args' => $arguments};
|
---|
| 88 |
|
---|
| 89 | sub new
|
---|
| 90 | {
|
---|
| 91 | my $class = shift (@_);
|
---|
| 92 |
|
---|
| 93 | my ($plugoutlist,$args,$hashArgOptLists) = @_;
|
---|
| 94 | push(@$plugoutlist, $class);
|
---|
| 95 |
|
---|
| 96 | my $strPlugoutName = (defined $plugoutlist->[0]) ? $plugoutlist->[0] : $class;
|
---|
| 97 |
|
---|
| 98 | if(defined $arguments){ push(@{$hashArgOptLists->{"ArgList"}},@{$arguments});}
|
---|
| 99 | if(defined $options) { push(@{$hashArgOptLists->{"OptList"}},$options)};
|
---|
| 100 |
|
---|
| 101 | my $self = {};
|
---|
| 102 | $self->{'plugout_type'} = $class;
|
---|
| 103 | $self->{'option_list'} = $hashArgOptLists->{"OptList"};
|
---|
| 104 | $self->{"info_only"} = 0;
|
---|
| 105 |
|
---|
| 106 | # Check if gsdlinfo is in the argument list or not - if it is, don't parse
|
---|
| 107 | # the args, just return the object.
|
---|
| 108 | foreach my $strArg (@{$args})
|
---|
| 109 | {
|
---|
| 110 | if(defined $strArg && $strArg eq "-gsdlinfo")
|
---|
| 111 | {
|
---|
| 112 | $self->{"info_only"} = 1;
|
---|
| 113 | return bless $self, $class;
|
---|
| 114 | }
|
---|
| 115 | }
|
---|
| 116 |
|
---|
| 117 | delete $self->{"info_only"};
|
---|
| 118 |
|
---|
[12546] | 119 | if(parse2::parse($args,$hashArgOptLists->{"ArgList"},$self) == -1)
|
---|
[12330] | 120 | {
|
---|
| 121 | my $classTempClass = bless $self, $class;
|
---|
| 122 | print STDERR "<BadPlugout d=$self->{'plugout_name'}>\n";
|
---|
| 123 | &gsprintf(STDERR, "\n{BasPlugout.bad_general_option}\n", $self->{'plugout_name'});
|
---|
| 124 | $classTempClass->print_txt_usage(""); # Use default resource bundle
|
---|
| 125 | die "\n";
|
---|
| 126 | }
|
---|
| 127 |
|
---|
| 128 |
|
---|
[12603] | 129 | if(defined $self->{'xslt_file'} && $self->{'xslt_file'} ne "")
|
---|
[12330] | 130 | {
|
---|
| 131 | ##$self->{'xslt_file'} =~ s/\"//g;##working on Windows???
|
---|
| 132 | print STDERR "Can not find $self->{'xslt_file'}, please make sure you have supplied the correct file path\n" and die "\n" unless (-e $self->{'xslt_file'});
|
---|
| 133 | }
|
---|
| 134 |
|
---|
| 135 | $self->{'gs_count'} = 0;
|
---|
| 136 |
|
---|
[12459] | 137 | $self->{'keep_import_structure'} = 0;
|
---|
[12330] | 138 |
|
---|
| 139 | return bless $self, $class;
|
---|
| 140 |
|
---|
| 141 | }
|
---|
| 142 |
|
---|
| 143 | sub print_xml_usage
|
---|
| 144 | {
|
---|
| 145 | my $self = shift(@_);
|
---|
| 146 | my $header = shift(@_);
|
---|
[12628] | 147 | my $high_level_information_only = shift(@_);
|
---|
[12330] | 148 |
|
---|
| 149 | # XML output is always in UTF-8
|
---|
| 150 | gsprintf::output_strings_in_UTF8;
|
---|
| 151 |
|
---|
| 152 | if ($header) {
|
---|
| 153 | &PrintUsage::print_xml_header("plugout");
|
---|
| 154 | }
|
---|
[12628] | 155 | $self->print_xml($high_level_information_only);
|
---|
[12330] | 156 | }
|
---|
| 157 |
|
---|
| 158 |
|
---|
| 159 | sub print_xml
|
---|
| 160 | {
|
---|
| 161 | my $self = shift(@_);
|
---|
[12628] | 162 | my $high_level_information_only = shift(@_);
|
---|
| 163 |
|
---|
[12330] | 164 | my $optionlistref = $self->{'option_list'};
|
---|
| 165 | my @optionlist = @$optionlistref;
|
---|
| 166 | my $plugoutoptions = shift(@$optionlistref);
|
---|
| 167 | return if (!defined($plugoutoptions));
|
---|
| 168 |
|
---|
| 169 | gsprintf(STDERR, "<PlugoutInfo>\n");
|
---|
| 170 | gsprintf(STDERR, " <Name>$plugoutoptions->{'name'}</Name>\n");
|
---|
| 171 | my $desc = gsprintf::lookup_string($plugoutoptions->{'desc'});
|
---|
| 172 | $desc =~ s/</&lt;/g; # doubly escaped
|
---|
| 173 | $desc =~ s/>/&gt;/g;
|
---|
| 174 | gsprintf(STDERR, " <Desc>$desc</Desc>\n");
|
---|
| 175 | gsprintf(STDERR, " <Abstract>$plugoutoptions->{'abstract'}</Abstract>\n");
|
---|
| 176 | gsprintf(STDERR, " <Inherits>$plugoutoptions->{'inherits'}</Inherits>\n");
|
---|
[12628] | 177 | unless (defined($high_level_information_only)) {
|
---|
| 178 | gsprintf(STDERR, " <Arguments>\n");
|
---|
| 179 | if (defined($plugoutoptions->{'args'})) {
|
---|
| 180 | &PrintUsage::print_options_xml($plugoutoptions->{'args'});
|
---|
| 181 | }
|
---|
| 182 | gsprintf(STDERR, " </Arguments>\n");
|
---|
[12330] | 183 |
|
---|
[12628] | 184 | # Recurse up the plugout hierarchy
|
---|
| 185 | $self->print_xml();
|
---|
[12330] | 186 | }
|
---|
| 187 | gsprintf(STDERR, "</PlugoutInfo>\n");
|
---|
| 188 | }
|
---|
| 189 |
|
---|
| 190 |
|
---|
| 191 | sub print_txt_usage
|
---|
| 192 | {
|
---|
| 193 | my $self = shift(@_);
|
---|
| 194 |
|
---|
| 195 | # Print the usage message for a plugout (recursively)
|
---|
| 196 | my $descoffset = $self->determine_description_offset(0);
|
---|
| 197 | $self->print_plugout_usage($descoffset, 1);
|
---|
| 198 | }
|
---|
| 199 |
|
---|
| 200 | sub determine_description_offset
|
---|
| 201 | {
|
---|
| 202 | my $self = shift(@_);
|
---|
| 203 | my $maxoffset = shift(@_);
|
---|
| 204 |
|
---|
| 205 | my $optionlistref = $self->{'option_list'};
|
---|
| 206 | my @optionlist = @$optionlistref;
|
---|
| 207 | my $plugoutoptions = pop(@$optionlistref);
|
---|
| 208 | return $maxoffset if (!defined($plugoutoptions));
|
---|
| 209 |
|
---|
| 210 | # Find the length of the longest option string of this download
|
---|
| 211 | my $plugoutargs = $plugoutoptions->{'args'};
|
---|
| 212 | if (defined($plugoutargs)) {
|
---|
| 213 | my $longest = &PrintUsage::find_longest_option_string($plugoutargs);
|
---|
| 214 | if ($longest > $maxoffset) {
|
---|
| 215 | $maxoffset = $longest;
|
---|
| 216 | }
|
---|
| 217 | }
|
---|
| 218 |
|
---|
| 219 | # Recurse up the download hierarchy
|
---|
| 220 | $maxoffset = $self->determine_description_offset($maxoffset);
|
---|
| 221 | $self->{'option_list'} = \@optionlist;
|
---|
| 222 | return $maxoffset;
|
---|
| 223 | }
|
---|
| 224 |
|
---|
| 225 |
|
---|
| 226 | sub print_plugout_usage
|
---|
| 227 | {
|
---|
| 228 | my $self = shift(@_);
|
---|
| 229 | my $descoffset = shift(@_);
|
---|
| 230 | my $isleafclass = shift(@_);
|
---|
| 231 |
|
---|
| 232 | my $optionlistref = $self->{'option_list'};
|
---|
| 233 | my @optionlist = @$optionlistref;
|
---|
| 234 | my $plugoutoptions = shift(@$optionlistref);
|
---|
| 235 | return if (!defined($plugoutoptions));
|
---|
| 236 |
|
---|
| 237 | my $plugoutname = $plugoutoptions->{'name'};
|
---|
| 238 | my $plugoutargs = $plugoutoptions->{'args'};
|
---|
| 239 | my $plugoutdesc = $plugoutoptions->{'desc'};
|
---|
| 240 |
|
---|
| 241 | # Produce the usage information using the data structure above
|
---|
| 242 | if ($isleafclass) {
|
---|
| 243 | if (defined($plugoutdesc)) {
|
---|
| 244 | gsprintf(STDERR, "$plugoutdesc\n\n");
|
---|
| 245 | }
|
---|
| 246 | gsprintf(STDERR, " {common.usage}: plugout $plugoutname [{common.options}]\n\n");
|
---|
| 247 | }
|
---|
| 248 |
|
---|
| 249 | # Display the download options, if there are some
|
---|
| 250 | if (defined($plugoutargs)) {
|
---|
| 251 | # Calculate the column offset of the option descriptions
|
---|
| 252 | my $optiondescoffset = $descoffset + 2; # 2 spaces between options & descriptions
|
---|
| 253 |
|
---|
| 254 | if ($isleafclass) {
|
---|
| 255 | gsprintf(STDERR, " {common.specific_options}:\n");
|
---|
| 256 | }
|
---|
| 257 | else {
|
---|
| 258 | gsprintf(STDERR, " {common.general_options}:\n", $plugoutname);
|
---|
| 259 | }
|
---|
| 260 |
|
---|
| 261 | # Display the download options
|
---|
| 262 | &PrintUsage::print_options_txt($plugoutargs, $optiondescoffset);
|
---|
| 263 | }
|
---|
| 264 |
|
---|
| 265 | # Recurse up the download hierarchy
|
---|
| 266 | $self->print_plugout_usage($descoffset, 0);
|
---|
| 267 | $self->{'option_list'} = \@optionlist;
|
---|
| 268 | }
|
---|
| 269 |
|
---|
| 270 |
|
---|
| 271 | sub error
|
---|
| 272 | {
|
---|
| 273 | my ($strFunctionName,$strError) = @_;
|
---|
| 274 | {
|
---|
| 275 | print "Error occoured in BasPlugout.pm\n".
|
---|
| 276 | "In Function: ".$strFunctionName."\n".
|
---|
| 277 | "Error Message: ".$strError."\n";
|
---|
| 278 | exit(-1);
|
---|
| 279 | }
|
---|
| 280 | }
|
---|
| 281 |
|
---|
| 282 | # OIDtype may be "hash" or "incremental" or "dirname" or "assigned"
|
---|
| 283 | sub set_OIDtype {
|
---|
| 284 | my $self = shift (@_);
|
---|
[12618] | 285 | my ($type, $metadata) = @_;
|
---|
[12330] | 286 |
|
---|
| 287 | if ($type =~ /^(hash|incremental|dirname|assigned)$/) {
|
---|
| 288 | $self->{'OIDtype'} = $type;
|
---|
| 289 | } else {
|
---|
| 290 | $self->{'OIDtype'} = "hash";
|
---|
| 291 | }
|
---|
[12618] | 292 | if ($type =~ /^assigned$/) {
|
---|
| 293 | if (defined $metadata) {
|
---|
| 294 | $self->{'OIDmetadata'} = $metadata;
|
---|
| 295 | } else {
|
---|
| 296 | $self->{'OIDmetadata'} = "dc.Identifier";
|
---|
| 297 | }
|
---|
| 298 | }
|
---|
[12330] | 299 | }
|
---|
| 300 |
|
---|
| 301 | sub set_output_dir
|
---|
| 302 | {
|
---|
| 303 | my $self = shift @_;
|
---|
| 304 | my ($output_dir) = @_;
|
---|
| 305 |
|
---|
| 306 | $self->{'output_dir'} = $output_dir;
|
---|
| 307 | }
|
---|
| 308 |
|
---|
| 309 | sub setoutputdir
|
---|
| 310 | {
|
---|
| 311 | my $self = shift @_;
|
---|
| 312 | my ($output_dir) = @_;
|
---|
| 313 |
|
---|
| 314 | $self->{'output_dir'} = $output_dir;
|
---|
| 315 | }
|
---|
| 316 |
|
---|
| 317 | sub get_output_dir
|
---|
| 318 | {
|
---|
| 319 | my $self = shift (@_);
|
---|
| 320 |
|
---|
| 321 | return $self->{'output_dir'};
|
---|
| 322 | }
|
---|
| 323 |
|
---|
| 324 | sub getoutputdir
|
---|
| 325 | {
|
---|
| 326 | my $self = shift (@_);
|
---|
| 327 |
|
---|
| 328 | return $self->{'output_dir'};
|
---|
| 329 | }
|
---|
| 330 |
|
---|
| 331 | sub getoutputinfo
|
---|
| 332 | {
|
---|
| 333 | my $self = shift (@_);
|
---|
| 334 |
|
---|
| 335 | return $self->{'output_info'};
|
---|
| 336 | }
|
---|
| 337 |
|
---|
| 338 |
|
---|
| 339 | sub get_output_handler
|
---|
| 340 | {
|
---|
| 341 | my $self = shift (@_);
|
---|
| 342 |
|
---|
| 343 | my ($output_file_name) = @_;
|
---|
| 344 |
|
---|
| 345 | open(*OUTPUT, ">$output_file_name") or die "Can not open a file handler for $output_file_name\n";
|
---|
| 346 |
|
---|
| 347 | return *OUTPUT;
|
---|
| 348 | }
|
---|
| 349 |
|
---|
| 350 | sub release_output_handler
|
---|
| 351 | {
|
---|
| 352 | my $self = shift (@_);
|
---|
| 353 | my ($outhandler) = @_;
|
---|
| 354 |
|
---|
| 355 | close($outhandler);
|
---|
| 356 |
|
---|
| 357 | }
|
---|
| 358 |
|
---|
| 359 | sub output_xml_header {
|
---|
| 360 | my $self = shift (@_);
|
---|
| 361 | my ($handle,$docroot,$nondoctype) = @_;
|
---|
| 362 |
|
---|
| 363 | print $handle '<?xml version="1.0" encoding="UTF-8" standalone="no"?>' . "\n";
|
---|
| 364 |
|
---|
| 365 | if (!defined $nondoctype){
|
---|
| 366 | print $handle '<!DOCTYPE Archive SYSTEM "http://greenstone.org/dtd/Archive/1.0/Archive.dtd">' . "\n";
|
---|
| 367 | }
|
---|
| 368 |
|
---|
| 369 | print $handle "<$docroot>\n" if defined $docroot;
|
---|
| 370 | }
|
---|
| 371 |
|
---|
| 372 | sub output_xml_footer {
|
---|
| 373 | my $self = shift (@_);
|
---|
| 374 | my ($handle,$docroot) = @_;
|
---|
| 375 | print $handle "</$docroot>\n" if defined $docroot;
|
---|
| 376 | }
|
---|
| 377 |
|
---|
| 378 | sub process {
|
---|
| 379 | my $self = shift (@_);
|
---|
| 380 | my ($doc_obj) = @_;
|
---|
[12459] | 381 |
|
---|
[12330] | 382 | $doc_obj->set_lastmodified();
|
---|
| 383 |
|
---|
| 384 | if ($self->{'group_size'} > 1) {
|
---|
[12459] | 385 | $self->group_process ($doc_obj);
|
---|
[12330] | 386 | return;
|
---|
| 387 | }
|
---|
| 388 |
|
---|
| 389 | my $OID = $doc_obj->get_OID();
|
---|
| 390 | $OID = "NULL" unless defined $OID;
|
---|
| 391 |
|
---|
| 392 | my $top_section = $doc_obj->get_top_section();
|
---|
| 393 |
|
---|
| 394 | #get document's directory
|
---|
| 395 | my $doc_dir = $self->get_doc_dir ($OID, $doc_obj->get_source_filename());
|
---|
| 396 |
|
---|
| 397 | my $output_info = $self->{'output_info'};
|
---|
| 398 | return if (!defined $output_info);
|
---|
| 399 |
|
---|
| 400 | ##############################
|
---|
| 401 | # call subclass' saveas method
|
---|
| 402 | ##############################
|
---|
| 403 | $self->saveas($doc_obj,$doc_dir);
|
---|
| 404 |
|
---|
| 405 | }
|
---|
| 406 |
|
---|
[12363] | 407 | sub store_output_info_reference {
|
---|
[12330] | 408 | my $self = shift (@_);
|
---|
| 409 | my ($doc_obj) = @_;
|
---|
| 410 |
|
---|
[12363] | 411 | my $output_info = $self->{'output_info'};
|
---|
| 412 | my $metaname = $self->{'sortmeta'};
|
---|
| 413 | if (!defined $metaname || $metaname !~ /\S/) {
|
---|
| 414 | $output_info->add_info($doc_obj->get_OID(),$self->{'short_doc_file'}, undef, "");
|
---|
| 415 | return;
|
---|
| 416 | }
|
---|
[12330] | 417 |
|
---|
[12363] | 418 | my $metadata = "";
|
---|
| 419 | my $top_section = $doc_obj->get_top_section();
|
---|
| 420 |
|
---|
| 421 | my @commameta_list = split(/,/, $metaname);
|
---|
| 422 | foreach my $cmn (@commameta_list) {
|
---|
| 423 | my $meta = $doc_obj->get_metadata_element($top_section, $cmn);
|
---|
| 424 | if ($meta) {
|
---|
| 425 | # do remove prefix/suffix - this will apply to all values
|
---|
| 426 | $meta =~ s/^$self->{'removeprefix'}// if defined $self->{'removeprefix'};
|
---|
| 427 | $meta =~ s/$self->{'removesuffix'}$// if defined $self->{'removesuffix'};
|
---|
| 428 | $meta = &sorttools::format_metadata_for_sorting($cmn, $meta, $doc_obj);
|
---|
| 429 | $metadata .= $meta if ($meta);
|
---|
| 430 | }
|
---|
[12330] | 431 | }
|
---|
| 432 |
|
---|
| 433 | # store reference in the output_info
|
---|
| 434 | $output_info->add_info($doc_obj->get_OID(),$self->{'short_doc_file'}, undef, $metadata);
|
---|
| 435 |
|
---|
| 436 | }
|
---|
| 437 |
|
---|
| 438 | sub group_process {
|
---|
| 439 |
|
---|
| 440 | my $self = shift (@_);
|
---|
| 441 | my ($doc_obj) = @_;
|
---|
| 442 |
|
---|
| 443 | my $OID = $doc_obj->get_OID();
|
---|
| 444 | $OID = "NULL" unless defined $OID;
|
---|
| 445 |
|
---|
| 446 | my $groupsize = $self->{'group_size'};
|
---|
| 447 | my $gs_count = $self->{'gs_count'};
|
---|
| 448 | my $open_new_file = (($gs_count % $groupsize)==0);
|
---|
| 449 | my $outhandle = $self->{'output_handle'};
|
---|
| 450 |
|
---|
| 451 | # opening a new file, or document has assoicated files => directory needed
|
---|
[12459] | 452 | if (($open_new_file) || (scalar(@{$doc_obj->get_assoc_files()})>0)) {
|
---|
| 453 |
|
---|
[12330] | 454 | # The directory the archive file (doc.xml) and all associated files
|
---|
| 455 | # should end up in
|
---|
| 456 | my $doc_dir;
|
---|
| 457 | # If we've determined its time for a new file, open it now
|
---|
| 458 | if ($open_new_file || !defined($self->{'gs_doc_dir'}))
|
---|
| 459 | {
|
---|
| 460 | $doc_dir = $self->get_doc_dir ($OID, $doc_obj->get_source_filename());
|
---|
| 461 | # only if opening new file
|
---|
| 462 | my $output_dir = $self->get_output_dir();
|
---|
[12459] | 463 | &util::mk_all_dir ($output_dir) unless -e $output_dir;
|
---|
[12330] | 464 | my $doc_file = &util::filename_cat ($output_dir, $doc_dir, "doc.xml");
|
---|
| 465 | my $short_doc_file = &util::filename_cat ($doc_dir, "doc.xml");
|
---|
[12459] | 466 |
|
---|
[12330] | 467 | if ($gs_count>0)
|
---|
| 468 | {
|
---|
| 469 | return if (!$self->close_file_output());
|
---|
| 470 | }
|
---|
| 471 |
|
---|
| 472 | open (GROUPPROCESS, ">$doc_file") or (print $outhandle "BasPlugout::group_process could not write to file $doc_file\n" and return);
|
---|
| 473 |
|
---|
| 474 |
|
---|
| 475 | $self->{'gs_filename'} = $doc_file;
|
---|
[12459] | 476 | $self->{'short_doc_file'} = $short_doc_file;
|
---|
[12330] | 477 | $self->{'gs_OID'} = $OID;
|
---|
| 478 | $self->{'gs_doc_dir'} = $doc_dir;
|
---|
| 479 |
|
---|
[12459] | 480 | $self->output_xml_header('BasPlugout::GROUPPROCESS','Archive');
|
---|
[12330] | 481 | }
|
---|
| 482 | # Otherwise load the same archive document directory used last time
|
---|
| 483 | else
|
---|
| 484 | {
|
---|
| 485 | $doc_dir = $self->{'gs_doc_dir'};
|
---|
| 486 | }
|
---|
| 487 |
|
---|
| 488 | # copy all the associated files, add this information as metadata
|
---|
| 489 | # to the document
|
---|
[13172] | 490 | print $outhandle "Writing associated files to $doc_dir\n";
|
---|
[12330] | 491 | $self->process_assoc_files ($doc_obj, $doc_dir);
|
---|
| 492 | }
|
---|
| 493 |
|
---|
| 494 | # save this document
|
---|
[13172] | 495 | my $section_text = &docprint::get_section_xml($doc_obj,$doc_obj->get_top_section());
|
---|
| 496 | print GROUPPROCESS $section_text;
|
---|
[12330] | 497 |
|
---|
| 498 | $self->{'gs_count'}++;
|
---|
| 499 | }
|
---|
| 500 |
|
---|
| 501 |
|
---|
| 502 | sub saveas {
|
---|
| 503 | my $self = shift (@_);
|
---|
| 504 |
|
---|
| 505 | die "Basplug::saveas function must be implemented in sub classes\n";
|
---|
| 506 | }
|
---|
| 507 |
|
---|
| 508 | sub get_doc_dir {
|
---|
| 509 | my $self = shift (@_);
|
---|
| 510 | my ($OID, $source_filename) = @_;
|
---|
| 511 |
|
---|
| 512 | my $working_dir = $self->get_output_dir();
|
---|
| 513 | my $working_info = $self->{output_info};
|
---|
| 514 | return if (!defined $working_info);
|
---|
| 515 |
|
---|
| 516 | my $doc_info = $working_info->get_info($OID);
|
---|
| 517 | my $doc_dir = '';
|
---|
| 518 |
|
---|
| 519 | if (defined $doc_info && scalar(@$doc_info) >= 1) {
|
---|
| 520 | # this OID already has an assigned directory, use the
|
---|
| 521 | # same one.
|
---|
| 522 | $doc_dir = $doc_info->[0];
|
---|
| 523 | $doc_dir =~ s/\/?((doc(mets)?)|(dublin_core))\.xml(\.gz)?$//;
|
---|
| 524 | } elsif ($self->{'keep_import_structure'}) {
|
---|
| 525 | $source_filename = &File::Basename::dirname($source_filename);
|
---|
| 526 | $source_filename =~ s/[\\\/]+/\//g;
|
---|
| 527 | $source_filename =~ s/\/$//;
|
---|
| 528 |
|
---|
| 529 | $doc_dir = substr($source_filename, length($ENV{'GSDLIMPORTDIR'}) + 1);
|
---|
| 530 |
|
---|
| 531 | }
|
---|
| 532 |
|
---|
| 533 | # have to get a new document directory
|
---|
| 534 | $doc_dir = $self->get_new_doc_dir($working_info,$working_dir,$OID) unless $doc_dir ne "";
|
---|
| 535 |
|
---|
| 536 | $doc_dir .= ".dir";
|
---|
[12603] | 537 | if (!defined $self->{'group'} || !$self->{'group'}){
|
---|
| 538 | &util::mk_all_dir (&util::filename_cat ($working_dir, $doc_dir));
|
---|
| 539 | }
|
---|
[12330] | 540 | return $doc_dir;
|
---|
| 541 | }
|
---|
| 542 |
|
---|
| 543 | sub get_new_doc_dir{
|
---|
| 544 | my $self = shift (@_);
|
---|
| 545 | my($working_info,$working_dir,$OID) = @_;
|
---|
| 546 |
|
---|
| 547 | my $doc_dir = "";
|
---|
| 548 | my $doc_dir_rest = $OID;
|
---|
| 549 | my $doc_dir_num = 0;
|
---|
| 550 |
|
---|
| 551 | do {
|
---|
| 552 | $doc_dir .= "/" if $doc_dir_num > 0;
|
---|
| 553 | if ($doc_dir_rest =~ s/^(.{1,8})//) {
|
---|
| 554 | $doc_dir .= $1;
|
---|
| 555 | $doc_dir_num++;
|
---|
| 556 | }
|
---|
| 557 | } while ($doc_dir_rest ne "" &&
|
---|
| 558 | ((-d &util::filename_cat ($working_dir, "$doc_dir.dir")) ||
|
---|
| 559 | ($working_info->size() >= 1024 && $doc_dir_num < 2)));
|
---|
| 560 |
|
---|
| 561 |
|
---|
| 562 | return $doc_dir;
|
---|
| 563 | }
|
---|
| 564 |
|
---|
| 565 | sub process_assoc_files {
|
---|
| 566 | my $self = shift (@_);
|
---|
| 567 | my ($doc_obj, $doc_dir, $handle) = @_;
|
---|
| 568 |
|
---|
| 569 | my $outhandle = $self->{'output_handle'};
|
---|
| 570 |
|
---|
| 571 | my $output_dir = $self->get_output_dir();
|
---|
| 572 | return if (!defined $output_dir);
|
---|
| 573 |
|
---|
| 574 | &util::mk_all_dir ($output_dir) unless -e $output_dir;
|
---|
| 575 |
|
---|
| 576 | my $working_dir = &util::filename_cat($output_dir, $doc_dir);
|
---|
| 577 | &util::mk_all_dir ($working_dir) unless -e $working_dir;
|
---|
| 578 |
|
---|
| 579 | my @assoc_files = ();
|
---|
| 580 | my $filename;;
|
---|
| 581 |
|
---|
| 582 | my $source_filename = $doc_obj->get_source_filename();
|
---|
| 583 |
|
---|
| 584 | my $collect_dir = $ENV{'GSDLCOLLECTDIR'};
|
---|
| 585 |
|
---|
| 586 | if (defined $collect_dir) {
|
---|
| 587 | my $dirsep_regexp = &util::get_os_dirsep();
|
---|
| 588 |
|
---|
| 589 | if ($collect_dir !~ /$dirsep_regexp$/) {
|
---|
| 590 | $collect_dir .= &util::get_dirsep(); # ensure there is a slash at the end
|
---|
| 591 | }
|
---|
| 592 |
|
---|
| 593 | # This test is never going to fail on Windows -- is this a problem?
|
---|
| 594 |
|
---|
| 595 | if ($source_filename !~ /^$dirsep_regexp/) {
|
---|
| 596 | $source_filename = &util::filename_cat($collect_dir, $source_filename);
|
---|
| 597 | }
|
---|
| 598 | }
|
---|
| 599 |
|
---|
| 600 |
|
---|
| 601 | # set the assocfile path (even if we have no assoc files - need this for lucene)
|
---|
| 602 | $doc_obj->set_utf8_metadata_element ($doc_obj->get_top_section(),
|
---|
| 603 | "assocfilepath",
|
---|
| 604 | "$doc_dir");
|
---|
| 605 | foreach my $assoc_file_rec (@{$doc_obj->get_assoc_files()}) {
|
---|
| 606 | my ($dir, $afile) = $assoc_file_rec->[1] =~ /^(.*?)([^\/\\]+)$/;
|
---|
| 607 | $dir = "" unless defined $dir;
|
---|
| 608 |
|
---|
| 609 |
|
---|
| 610 | my $real_filename = $assoc_file_rec->[0];
|
---|
| 611 | # for some reasons the image associate file has / before the full path
|
---|
| 612 | $real_filename =~ s/^\\(.*)/$1/i;
|
---|
| 613 | if (-e $real_filename) {
|
---|
| 614 |
|
---|
| 615 | $filename = &util::filename_cat($working_dir, $afile);
|
---|
| 616 |
|
---|
| 617 | &util::hard_link ($real_filename, $filename);
|
---|
| 618 |
|
---|
| 619 | $doc_obj->add_utf8_metadata ($doc_obj->get_top_section(),
|
---|
| 620 | "gsdlassocfile",
|
---|
| 621 | "$afile:$assoc_file_rec->[2]:$dir");
|
---|
| 622 | } elsif ($self->{'verbosity'} > 2) {
|
---|
| 623 | print $outhandle "BasPlugout::process couldn't copy the associated file " .
|
---|
| 624 | "$real_filename to $afile\n";
|
---|
| 625 | }
|
---|
| 626 | }
|
---|
| 627 | }
|
---|
| 628 |
|
---|
| 629 | sub set_sortmeta {
|
---|
| 630 | my $self = shift (@_);
|
---|
| 631 | my ($sortmeta, $removeprefix, $removesuffix) = @_;
|
---|
| 632 |
|
---|
| 633 | $self->{'sortmeta'} = $sortmeta;
|
---|
| 634 | if (defined ($removeprefix) && $removeprefix ) {
|
---|
| 635 | $removeprefix =~ s/^\^//; # don't need a leading ^
|
---|
| 636 | $self->{'removeprefix'} = $removeprefix;
|
---|
| 637 | }
|
---|
| 638 | if (defined ($removesuffix) && $removesuffix) {
|
---|
| 639 | $removesuffix =~ s/\$$//; # don't need a trailing $
|
---|
| 640 | $self->{'removesuffix'} = $removesuffix;
|
---|
| 641 | }
|
---|
| 642 | }
|
---|
| 643 |
|
---|
| 644 | sub open_xslt_pipe
|
---|
| 645 | {
|
---|
| 646 | my $self = shift @_;
|
---|
| 647 | my ($output_file_name, $xslt_file)=@_;
|
---|
| 648 |
|
---|
| 649 | return unless defined $xslt_file and $xslt_file ne "" and -e $xslt_file;
|
---|
| 650 |
|
---|
| 651 | my $java_class_path = &util::filename_cat ($ENV{'GSDLHOME'},"bin","java");
|
---|
[13013] | 652 |
|
---|
[13225] | 653 | if ($ENV{'GSDLOS'} eq "windows"){
|
---|
| 654 | $java_class_path .=";".&util::filename_cat ($ENV{'GSDLHOME'},"bin","java","xalan.jar");
|
---|
| 655 | }
|
---|
| 656 | else{
|
---|
| 657 | $java_class_path .=":".&util::filename_cat ($ENV{'GSDLHOME'},"bin","java","xalan.jar");
|
---|
| 658 | }
|
---|
| 659 |
|
---|
| 660 |
|
---|
[13013] | 661 | $java_class_path = "\"".$java_class_path."\"";
|
---|
[13024] | 662 |
|
---|
[13013] | 663 | $xslt_file = "\"".$xslt_file."\"";
|
---|
[13024] | 664 |
|
---|
[14969] | 665 | my $cmd = "| java -cp $java_class_path org.nzdl.gsdl.ApplyXSLT -t $xslt_file ";
|
---|
[12330] | 666 |
|
---|
[12603] | 667 | if (defined $self->{'mapping_file'} and $self->{'mapping_file'} ne ""){
|
---|
[13064] | 668 | my $mapping_file_path = "\"".$self->{'mapping_file'}."\"";
|
---|
[14969] | 669 | $cmd .= "-m $mapping_file_path";
|
---|
[12603] | 670 | }
|
---|
[14969] | 671 |
|
---|
[12330] | 672 | open(*XMLWRITER, $cmd)
|
---|
| 673 | or die "can't open pipe to xslt: $!";
|
---|
| 674 |
|
---|
| 675 |
|
---|
| 676 | $self->{'xslt_writer'} = *XMLWRITER;
|
---|
| 677 |
|
---|
| 678 | print XMLWRITER "<?DocStart?>\n";
|
---|
| 679 | print XMLWRITER "$output_file_name\n";
|
---|
[14969] | 680 |
|
---|
[12330] | 681 |
|
---|
| 682 | }
|
---|
| 683 |
|
---|
| 684 |
|
---|
| 685 | sub close_xslt_pipe
|
---|
| 686 | {
|
---|
| 687 | my $self = shift @_;
|
---|
| 688 |
|
---|
| 689 |
|
---|
| 690 | return unless defined $self->{'xslt_writer'} ;
|
---|
| 691 |
|
---|
| 692 | my $xsltwriter = $self->{'xslt_writer'};
|
---|
| 693 |
|
---|
| 694 | print $xsltwriter "<?DocEnd?>\n";
|
---|
| 695 | close($xsltwriter);
|
---|
[13024] | 696 |
|
---|
| 697 | undef $self->{'xslt_writer'};
|
---|
| 698 |
|
---|
[12330] | 699 | }
|
---|
| 700 |
|
---|
| 701 | sub close_file_output
|
---|
| 702 | {
|
---|
| 703 | my ($self) = @_;
|
---|
| 704 |
|
---|
| 705 | # make sure that the handle has been opened - it won't be if we failed
|
---|
| 706 | # to import any documents...
|
---|
| 707 | if (defined(fileno(GROUPPROCESS))) {
|
---|
| 708 | $self->output_xml_footer('GROUPPROCESS','Archive');
|
---|
| 709 | close GROUPPROCESS;
|
---|
| 710 | }
|
---|
| 711 |
|
---|
| 712 | my $OID = $self->{'gs_OID'};
|
---|
| 713 | my $short_doc_file = $self->{'short_doc_file'};
|
---|
| 714 |
|
---|
| 715 | if ($self->{'gzip'}) {
|
---|
| 716 | my $doc_file = $self->{'gs_filename'};
|
---|
| 717 | `gzip $doc_file`;
|
---|
| 718 | $doc_file .= ".gz";
|
---|
| 719 | $short_doc_file .= ".gz";
|
---|
| 720 | if (!-e $doc_file) {
|
---|
| 721 | my $outhandle = $self->{'output_handle'};
|
---|
| 722 | print $outhandle "error while gzipping: $doc_file doesn't exist\n";
|
---|
| 723 | return 0;
|
---|
| 724 | }
|
---|
| 725 | }
|
---|
| 726 |
|
---|
| 727 | # store reference in output_info
|
---|
| 728 | my $output_info = $self->{'output_info'};
|
---|
| 729 | return 0 if (!defined $output_info);
|
---|
| 730 | $output_info->add_info($OID, $short_doc_file, undef, undef);
|
---|
| 731 | return 1;
|
---|
| 732 | }
|
---|
| 733 |
|
---|
[13172] | 734 |
|
---|
[12330] | 735 | #the subclass should implement this method if is_group method could return 1.
|
---|
| 736 | sub close_group_output{
|
---|
| 737 | my $self = shift (@_);
|
---|
| 738 | }
|
---|
| 739 |
|
---|
| 740 | sub is_group {
|
---|
| 741 | my $self = shift (@_);
|
---|
| 742 | return 0;
|
---|
| 743 | }
|
---|
| 744 |
|
---|
[13172] | 745 | my $dc_set = { Title => 1,
|
---|
| 746 | Creator => 1,
|
---|
| 747 | Subject => 1,
|
---|
| 748 | Description => 1,
|
---|
| 749 | Publisher => 1,
|
---|
| 750 | Contributor => 1,
|
---|
| 751 | Date => 1,
|
---|
| 752 | Type => 1,
|
---|
| 753 | Format => 1,
|
---|
| 754 | Identifier => 1,
|
---|
| 755 | Source => 1,
|
---|
| 756 | Language => 1,
|
---|
| 757 | Relation => 1,
|
---|
| 758 | Coverage => 1,
|
---|
| 759 | Rights => 1};
|
---|
| 760 |
|
---|
| 761 |
|
---|
| 762 | # returns an XML representation of the dublin core metadata
|
---|
| 763 | # if dc meta is not found, try ex mete
|
---|
| 764 | sub get_dc_metadata {
|
---|
| 765 | my $self = shift(@_);
|
---|
| 766 | my ($doc_obj, $section, $version) = @_;
|
---|
| 767 |
|
---|
| 768 | # build up string of dublin core metadata
|
---|
| 769 | $section="" unless defined $section;
|
---|
| 770 |
|
---|
| 771 | my $section_ptr = $doc_obj->_lookup_section($section);
|
---|
| 772 | return "" unless defined $section_ptr;
|
---|
| 773 |
|
---|
| 774 |
|
---|
| 775 | my $explicit_dc = {};
|
---|
| 776 | my $explicit_ex = {};
|
---|
| 777 |
|
---|
| 778 | my $all_text="";
|
---|
| 779 | foreach my $data (@{$section_ptr->{'metadata'}}){
|
---|
| 780 | my $escaped_value = &docprint::escape_text($data->[1]);
|
---|
| 781 | if ($data->[0]=~ m/^dc\./) {
|
---|
| 782 | $data->[0] =~ tr/[A-Z]/[a-z]/;
|
---|
| 783 |
|
---|
| 784 | $data->[0] =~ m/^dc\.(.*)/;
|
---|
| 785 | my $dc_element = $1;
|
---|
| 786 |
|
---|
| 787 | if (!defined $explicit_dc->{$dc_element}) {
|
---|
| 788 | $explicit_dc->{$dc_element} = [];
|
---|
| 789 | }
|
---|
| 790 | push(@{$explicit_dc->{$dc_element}},$escaped_value);
|
---|
| 791 |
|
---|
| 792 | if (defined $version && ($version eq "oai_dc")) {
|
---|
| 793 | $all_text .= " <dc:$dc_element>$escaped_value</dc:$dc_element>\n";
|
---|
| 794 | }
|
---|
| 795 | else {
|
---|
| 796 | # qualifier???
|
---|
| 797 | $all_text .= ' <dcvalue element="'. $dc_element.'">'. $escaped_value. "</dcvalue>\n";
|
---|
| 798 | }
|
---|
| 799 |
|
---|
| 800 | }
|
---|
| 801 | elsif (($data->[0] =~ m/^ex\./) || ($data->[0] !~ m/\./)) {
|
---|
| 802 | $data->[0] =~ m/^(ex\.)?(.*)/;
|
---|
| 803 | my $ex_element = $2;
|
---|
| 804 | my $lc_ex_element = lc($ex_element);
|
---|
| 805 |
|
---|
| 806 | if (defined $dc_set->{$ex_element}) {
|
---|
| 807 | if (!defined $explicit_ex->{$lc_ex_element}) {
|
---|
| 808 | $explicit_ex->{$lc_ex_element} = [];
|
---|
| 809 | }
|
---|
| 810 | push(@{$explicit_ex->{$lc_ex_element}},$escaped_value);
|
---|
| 811 | }
|
---|
| 812 | }
|
---|
| 813 | }
|
---|
| 814 |
|
---|
| 815 | # go through dc_set and for any element *not* defined in explicit_dc
|
---|
[14932] | 816 | # that does exist in explicit_ex, add it in as metadata
|
---|
[13172] | 817 | foreach my $k ( keys %$dc_set ) {
|
---|
| 818 | my $lc_k = lc($k);
|
---|
| 819 |
|
---|
| 820 | if (!defined $explicit_dc->{$lc_k}) {
|
---|
| 821 | if (defined $explicit_ex->{$lc_k}) {
|
---|
| 822 |
|
---|
| 823 | foreach my $v (@{$explicit_ex->{$lc_k}}) {
|
---|
| 824 | my $dc_element = $lc_k;
|
---|
| 825 | my $escaped_value = $v;
|
---|
| 826 |
|
---|
| 827 | if (defined $version && ($version eq "oai_dc")) {
|
---|
| 828 | $all_text .= " <dc:$dc_element>$escaped_value</dc:$dc_element>\n";
|
---|
| 829 | }
|
---|
| 830 | else {
|
---|
| 831 | $all_text .= ' <dcvalue element="'. $dc_element.'">'. $escaped_value. "</dcvalue>\n";
|
---|
| 832 | }
|
---|
| 833 |
|
---|
| 834 | }
|
---|
| 835 | }
|
---|
| 836 | }
|
---|
| 837 | }
|
---|
| 838 |
|
---|
| 839 | if ($all_text eq "") {
|
---|
| 840 | $all_text .= " There is no Dublin Core metatdata in this document\n";
|
---|
| 841 | }
|
---|
| 842 | $all_text =~ s/[\x00-\x09\x0B\x0C\x0E-\x1F]//g;
|
---|
| 843 |
|
---|
| 844 | return $all_text;
|
---|
| 845 | }
|
---|
| 846 |
|
---|
| 847 | # Build up dublin_core metadata. Priority given to dc.* over ex.*
|
---|
| 848 | # This method was apparently added by Jeffrey and committed by Shaoqun.
|
---|
| 849 | # But we don't know why it was added, so not using it anymore.
|
---|
| 850 | sub new_get_dc_metadata {
|
---|
| 851 |
|
---|
| 852 | my $self = shift(@_);
|
---|
| 853 | my ($doc_obj, $section, $version) = @_;
|
---|
| 854 |
|
---|
| 855 | # build up string of dublin core metadata
|
---|
| 856 | $section="" unless defined $section;
|
---|
| 857 |
|
---|
| 858 | my $section_ptr=$doc_obj->_lookup_section($section);
|
---|
| 859 | return "" unless defined $section_ptr;
|
---|
| 860 |
|
---|
| 861 | my $all_text = "";
|
---|
| 862 | foreach my $data (@{$section_ptr->{'metadata'}}){
|
---|
| 863 | my $escaped_value = &docprint::escape_text($data->[1]);
|
---|
| 864 | my $dc_element = $data->[0];
|
---|
| 865 |
|
---|
| 866 | my @array = split('\.',$dc_element);
|
---|
| 867 | my ($type,$name);
|
---|
| 868 |
|
---|
| 869 | if(defined $array[1])
|
---|
| 870 | {
|
---|
| 871 | $type = $array[0];
|
---|
| 872 | $name = $array[1];
|
---|
| 873 | }
|
---|
| 874 | else
|
---|
| 875 | {
|
---|
| 876 | $type = "ex";
|
---|
| 877 | $name = $array[0];
|
---|
| 878 | }
|
---|
| 879 |
|
---|
| 880 | $all_text .= ' <Metadata Type="'. $type.'" Name="'.$name.'">'. $escaped_value. "</Metadata>\n";
|
---|
| 881 | }
|
---|
| 882 | return $all_text;
|
---|
| 883 | }
|
---|
| 884 |
|
---|
| 885 |
|
---|
[12330] | 886 | 1;
|
---|