| 1 | ###########################################################################
|
|---|
| 2 | #
|
|---|
| 3 | # LOMPlugin.pm -- plugin for import the collection from LOM
|
|---|
| 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) 2005 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 | ### Note this plugin currently can't download source documents from outside if you are behind a firewall.
|
|---|
| 28 | # Unless, you set the http_proxy environment variable to be your proxy server,
|
|---|
| 29 | # and set proxy_user and proxy_password in .wgetrc file in home directory.
|
|---|
| 30 | # (does that work on windows??)
|
|---|
| 31 |
|
|---|
| 32 | package LOMPlugin;
|
|---|
| 33 |
|
|---|
| 34 | use ReadTextFile;
|
|---|
| 35 | use MetadataPass;
|
|---|
| 36 | use MetadataRead;
|
|---|
| 37 | use XMLParser;
|
|---|
| 38 | use Cwd;
|
|---|
| 39 |
|
|---|
| 40 | # methods with identical signatures take precedence in the order given in the ISA list.
|
|---|
| 41 | sub BEGIN {
|
|---|
| 42 | @ISA = ('MetadataRead', 'ReadTextFile', 'MetadataPass');
|
|---|
| 43 | }
|
|---|
| 44 |
|
|---|
| 45 | use strict; # every perl program should have this!
|
|---|
| 46 | no strict 'refs'; # make an exception so we can use variables as filehandles
|
|---|
| 47 |
|
|---|
| 48 |
|
|---|
| 49 | my $arguments =
|
|---|
| 50 | [ { 'name' => "process_exp",
|
|---|
| 51 | 'desc' => "{BasePlugin.process_exp}",
|
|---|
| 52 | 'type' => "string",
|
|---|
| 53 | 'deft' => &get_default_process_exp(),
|
|---|
| 54 | 'reqd' => "no" },
|
|---|
| 55 | { 'name' => "root_tag",
|
|---|
| 56 | 'desc' => "{LOMPlugin.root_tag}",
|
|---|
| 57 | 'type' => "regexp",
|
|---|
| 58 | 'deft' => q/^(?i)lom$/,
|
|---|
| 59 | 'reqd' => "no" },
|
|---|
| 60 | { 'name' => "check_timestamp",
|
|---|
| 61 | 'desc' => "{LOMPlugin.check_timestamp}",
|
|---|
| 62 | 'type' => "flag" },
|
|---|
| 63 | { 'name' => "download_srcdocs",
|
|---|
| 64 | 'desc' => "{LOMPlugin.download_srcdocs}",
|
|---|
| 65 | 'type' => "regexp",
|
|---|
| 66 | 'deft' => "",
|
|---|
| 67 | 'reqd' => "no" }];
|
|---|
| 68 |
|
|---|
| 69 | my $options = { 'name' => "LOMPlugin",
|
|---|
| 70 | 'desc' => "{LOMPlugin.desc}",
|
|---|
| 71 | 'abstract' => "no",
|
|---|
| 72 | 'inherits' => "yes",
|
|---|
| 73 | 'args' => $arguments };
|
|---|
| 74 |
|
|---|
| 75 |
|
|---|
| 76 |
|
|---|
| 77 | my ($self);
|
|---|
| 78 | sub new {
|
|---|
| 79 | my $class = shift (@_);
|
|---|
| 80 | my ($pluginlist,$inputargs,$hashArgOptLists) = @_;
|
|---|
| 81 | push(@$pluginlist, $class);
|
|---|
| 82 |
|
|---|
| 83 | push(@{$hashArgOptLists->{"ArgList"}},@{$arguments});
|
|---|
| 84 | push(@{$hashArgOptLists->{"OptList"}},$options);
|
|---|
| 85 |
|
|---|
| 86 | $self = new ReadTextFile($pluginlist, $inputargs, $hashArgOptLists);
|
|---|
| 87 |
|
|---|
| 88 | if ($self->{'info_only'}) {
|
|---|
| 89 | # don't worry about creating the XML parser as all we want is the
|
|---|
| 90 | # list of plugin options
|
|---|
| 91 | return bless $self, $class;
|
|---|
| 92 | }
|
|---|
| 93 |
|
|---|
| 94 | #create XML::Parser object for parsing dublin_core.xml files
|
|---|
| 95 | my $parser = new XML::Parser('Style' => 'Stream',
|
|---|
| 96 | 'Handlers' => {'Char' => \&Char,
|
|---|
| 97 | 'Doctype' => \&Doctype
|
|---|
| 98 | });
|
|---|
| 99 | $self->{'parser'} = $parser;
|
|---|
| 100 |
|
|---|
| 101 | $self->{'extra_blocks'} = {};
|
|---|
| 102 |
|
|---|
| 103 | return bless $self, $class;
|
|---|
| 104 | }
|
|---|
| 105 |
|
|---|
| 106 | sub get_default_process_exp {
|
|---|
| 107 | my $self = shift (@_);
|
|---|
| 108 |
|
|---|
| 109 | return q^(?i)\.xml$^;
|
|---|
| 110 | }
|
|---|
| 111 |
|
|---|
| 112 |
|
|---|
| 113 | sub can_process_this_file {
|
|---|
| 114 | my $self = shift(@_);
|
|---|
| 115 | my ($filename) = @_;
|
|---|
| 116 |
|
|---|
| 117 | if ($self->SUPER::can_process_this_file($filename) && $self->check_doctype($filename)) {
|
|---|
| 118 | return 1; # its a file for us
|
|---|
| 119 | }
|
|---|
| 120 | return 0;
|
|---|
| 121 | }
|
|---|
| 122 |
|
|---|
| 123 | sub metadata_read {
|
|---|
| 124 | my $self = shift (@_);
|
|---|
| 125 | my ($pluginfo, $base_dir, $file, $block_hash,
|
|---|
| 126 | $extrametakeys, $extrametadata, $extrametafile,
|
|---|
| 127 | $processor, $gli, $aux) = @_;
|
|---|
| 128 |
|
|---|
| 129 | my $outhandle = $self->{'outhandle'};
|
|---|
| 130 |
|
|---|
| 131 | # can we process this file??
|
|---|
| 132 | my ($filename_full_path, $filename_no_path) = &util::get_full_filenames($base_dir, $file);
|
|---|
| 133 | return undef unless $self->can_process_this_file_for_metadata($filename_full_path);
|
|---|
| 134 |
|
|---|
| 135 | $file =~ s/^[\/\\]+//; # $file often begins with / so we'll tidy it up
|
|---|
| 136 |
|
|---|
| 137 | print $outhandle "LOMPlugin: extracting metadata from $file\n"
|
|---|
| 138 | if $self->{'verbosity'} > 1;
|
|---|
| 139 |
|
|---|
| 140 | my ($dir,$tail) = $filename_full_path =~ /^(.*?)([^\/\\]*)$/;
|
|---|
| 141 | $self->{'output_dir'} = $dir;
|
|---|
| 142 |
|
|---|
| 143 | eval {
|
|---|
| 144 | $self->{'parser'}->parsefile($filename_full_path);
|
|---|
| 145 | };
|
|---|
| 146 |
|
|---|
| 147 | if ($@) {
|
|---|
| 148 | print $outhandle "LOMPlugin: skipping $filename_full_path as not conformant to LOM syntax\n" if ($self->{'verbosity'} > 1);
|
|---|
| 149 | print $outhandle "\n Perl Error:\n $@\n" if ($self->{'verbosity'}>2);
|
|---|
| 150 | return 0;
|
|---|
| 151 | }
|
|---|
| 152 |
|
|---|
| 153 | $self->{'output_dir'} = undef;
|
|---|
| 154 |
|
|---|
| 155 | my $file_re;
|
|---|
| 156 | my $lom_srcdoc = $self->{'lom_srcdoc'};
|
|---|
| 157 |
|
|---|
| 158 | if (defined $lom_srcdoc) {
|
|---|
| 159 | my $dirsep = &util::get_re_dirsep();
|
|---|
| 160 | $lom_srcdoc =~ s/^$base_dir($dirsep)//;
|
|---|
| 161 | $self->{'extra_blocks'}->{$file}++;
|
|---|
| 162 | $file_re = $lom_srcdoc;
|
|---|
| 163 | }
|
|---|
| 164 | else {
|
|---|
| 165 | $file_re = $tail;
|
|---|
| 166 | }
|
|---|
| 167 | $file_re = &util::filename_to_regex($file_re);
|
|---|
| 168 | $self->{'lom_srcdoc'} = undef; # reset for next file to be processed
|
|---|
| 169 |
|
|---|
| 170 | push(@$extrametakeys,$file_re);
|
|---|
| 171 | $extrametadata->{$file_re} = $self->{'saved_metadata'};
|
|---|
| 172 | if (defined $lom_srcdoc) {
|
|---|
| 173 | # copied from oaiplugin
|
|---|
| 174 | if (!defined $extrametafile->{$file_re}) {
|
|---|
| 175 | $extrametafile->{$file_re} = {};
|
|---|
| 176 | }
|
|---|
| 177 | #maps the file to full path
|
|---|
| 178 | $extrametafile->{$file_re}->{$file} = $filename_full_path;
|
|---|
| 179 | }
|
|---|
| 180 |
|
|---|
| 181 | return 1;
|
|---|
| 182 | }
|
|---|
| 183 |
|
|---|
| 184 | sub check_doctype {
|
|---|
| 185 | $self = shift (@_);
|
|---|
| 186 |
|
|---|
| 187 | my ($filename) = @_;
|
|---|
| 188 |
|
|---|
| 189 | if (open(XMLIN,"<$filename")) {
|
|---|
| 190 | my $doctype = $self->{'root_tag'};
|
|---|
| 191 | ## check whether the doctype has the same name as the root element tag
|
|---|
| 192 | while (defined (my $line = <XMLIN>)) {
|
|---|
| 193 | ## find the root element
|
|---|
| 194 | if ($line =~ /<([\w\d:]+)[\s>]/){
|
|---|
| 195 | my $root = $1;
|
|---|
| 196 | if ($root !~ $doctype){
|
|---|
| 197 | close(XMLIN);
|
|---|
| 198 | return 0;
|
|---|
| 199 | }
|
|---|
| 200 | else {
|
|---|
| 201 | close(XMLIN);
|
|---|
| 202 | return 1;
|
|---|
| 203 | }
|
|---|
| 204 | }
|
|---|
| 205 | }
|
|---|
| 206 | close(XMLIN);
|
|---|
| 207 | }
|
|---|
| 208 |
|
|---|
| 209 | return undef; # haven't found a valid line
|
|---|
| 210 |
|
|---|
| 211 | }
|
|---|
| 212 |
|
|---|
| 213 | sub read_file {
|
|---|
| 214 | my $self = shift (@_);
|
|---|
| 215 | my ($filename, $encoding, $language, $textref) = @_;
|
|---|
| 216 |
|
|---|
| 217 | my $metadata_table = $self->{'metadata_table'};
|
|---|
| 218 |
|
|---|
| 219 | my $rawtext = $metadata_table->{'rawtext'};
|
|---|
| 220 |
|
|---|
| 221 | delete $metadata_table->{'rawtext'};
|
|---|
| 222 |
|
|---|
| 223 | $$textref = $rawtext;
|
|---|
| 224 | }
|
|---|
| 225 |
|
|---|
| 226 | sub read {
|
|---|
| 227 | my $self = shift (@_);
|
|---|
| 228 | my ($pluginfo, $base_dir, $file, $block_hash, $metadata, $processor, $maxdocs, $total_count, $gli) = @_;
|
|---|
| 229 |
|
|---|
| 230 | my $outhandle = $self->{'outhandle'};
|
|---|
| 231 |
|
|---|
| 232 | return 0 if (defined $self->{'extra_blocks'}->{$file});
|
|---|
| 233 |
|
|---|
| 234 | # can we process this file??
|
|---|
| 235 | my ($filename_full_path, $filename_no_path) = &util::get_full_filenames($base_dir, $file);
|
|---|
| 236 | return undef unless $self->can_process_this_file($filename_full_path);
|
|---|
| 237 |
|
|---|
| 238 | $self->{'metadata_table'} = $metadata;
|
|---|
| 239 |
|
|---|
| 240 | my $lom_language = $metadata->{'lom_language'};
|
|---|
| 241 |
|
|---|
| 242 | my $store_input_encoding;
|
|---|
| 243 | my $store_extract_language;
|
|---|
| 244 | my $store_default_language;
|
|---|
| 245 | my $store_default_encoding;
|
|---|
| 246 |
|
|---|
| 247 | if (defined $lom_language) {
|
|---|
| 248 | delete $metadata->{'lom_language'};
|
|---|
| 249 |
|
|---|
| 250 | $store_input_encoding = $self->{'input_encoding'};
|
|---|
| 251 | $store_extract_language = $self->{'extract_language'};
|
|---|
| 252 | $store_default_language = $self->{'default_language'};
|
|---|
| 253 | $store_default_encoding = $self->{'default_encoding'};
|
|---|
| 254 |
|
|---|
| 255 | $self->{'input_encoding'} = "utf8";
|
|---|
| 256 | $self->{'extract_language'} = 0;
|
|---|
| 257 | $self->{'default_language'} = $lom_language;
|
|---|
| 258 | $self->{'default_encoding'} = "utf8";
|
|---|
| 259 | }
|
|---|
| 260 |
|
|---|
| 261 | my $rv = $self->SUPER::read(@_);
|
|---|
| 262 |
|
|---|
| 263 | if (defined $lom_language) {
|
|---|
| 264 | $self->{'input_encoding'} = $store_input_encoding;
|
|---|
| 265 | $self->{'extract_language'} = $store_extract_language;
|
|---|
| 266 | $self->{'default_language'} = $store_default_language;
|
|---|
| 267 | $self->{'default_encoding'} = $store_default_encoding;
|
|---|
| 268 | }
|
|---|
| 269 |
|
|---|
| 270 | $self->{'metadata_table'} = undef;
|
|---|
| 271 |
|
|---|
| 272 | return $rv;
|
|---|
| 273 | }
|
|---|
| 274 |
|
|---|
| 275 | # do plugin specific processing of doc_obj
|
|---|
| 276 | sub process {
|
|---|
| 277 | my $self = shift (@_);
|
|---|
| 278 | my ($textref, $pluginfo, $base_dir, $file, $metadata, $doc_obj, $gli) = @_;
|
|---|
| 279 | my $outhandle = $self->{'outhandle'};
|
|---|
| 280 |
|
|---|
| 281 | my $cursection = $doc_obj->get_top_section();
|
|---|
| 282 | $doc_obj->add_utf8_text($cursection, $$textref);
|
|---|
| 283 |
|
|---|
| 284 | return 1;
|
|---|
| 285 | }
|
|---|
| 286 |
|
|---|
| 287 | sub Doctype {
|
|---|
| 288 | my ($expat, $name, $sysid, $pubid, $internal) = @_;
|
|---|
| 289 |
|
|---|
| 290 | my $root_tag = $self->{'root_tag'};
|
|---|
| 291 |
|
|---|
| 292 | if ($name !~ /$root_tag/) {
|
|---|
| 293 | die "Root tag $name does not match regular expression $root_tag";
|
|---|
| 294 | }
|
|---|
| 295 | }
|
|---|
| 296 |
|
|---|
| 297 | sub StartTag {
|
|---|
| 298 | my ($expat, $element) = @_;
|
|---|
| 299 |
|
|---|
| 300 | my %attr = %_;
|
|---|
| 301 |
|
|---|
| 302 | my $raw_tag = "<$element";
|
|---|
| 303 | map { $raw_tag .= " $_=\"$attr{$_}\""; } keys %attr;
|
|---|
| 304 | $raw_tag .= ">";
|
|---|
| 305 |
|
|---|
| 306 | if ($element =~ m/$self->{'root_tag'}/) {
|
|---|
| 307 | $self->{'raw_text'} = $raw_tag;
|
|---|
| 308 |
|
|---|
| 309 | $self->{'saved_metadata'} = {};
|
|---|
| 310 | $self->{'metaname_stack'} = [];
|
|---|
| 311 | $self->{'lom_datatype'} = "";
|
|---|
| 312 | $self->{'lom_language'} = undef;
|
|---|
| 313 | $self->{'metadatatext'} = "<table class=\"metadata\" width=\"_pagewidth_\" >\n";
|
|---|
| 314 | }
|
|---|
| 315 | else {
|
|---|
| 316 | my $xml_depth = scalar(@{$self->{'metaname_stack'}});
|
|---|
| 317 | $self->{'raw_text'} .= "\n";
|
|---|
| 318 | $self->{'raw_text'} .= " " x $xml_depth;
|
|---|
| 319 | $self->{'raw_text'} .= $raw_tag;
|
|---|
| 320 |
|
|---|
| 321 | my $metaname_stack = $self->{'metaname_stack'};
|
|---|
| 322 | push(@$metaname_stack,$element);
|
|---|
| 323 | if (scalar(@$metaname_stack)==1) {
|
|---|
| 324 | # top level LOM category
|
|---|
| 325 | my $style = "class=\"metadata\"";
|
|---|
| 326 | my $open_close
|
|---|
| 327 | = "<a id=\"${element}opencloselink\" href=\"javascript:hideTBodyArea('$element')\">\n";
|
|---|
| 328 | $open_close
|
|---|
| 329 | .= "<img id=\"${element}openclose\" border=\"0\" src=\"_httpopenmdicon_\"></a>\n";
|
|---|
| 330 |
|
|---|
| 331 | my $header_line = " <tr $style ><th $style colspan=\"3\">$open_close \u$element</th></tr>\n";
|
|---|
| 332 | my $md_tbody = "<tbody id=\"$element\">\n";
|
|---|
| 333 |
|
|---|
| 334 | $self->{'mdheader'} = $header_line;
|
|---|
| 335 | $self->{'mdtbody'} = $md_tbody;
|
|---|
| 336 | $self->{'mdtbody_text'} = "";
|
|---|
| 337 | }
|
|---|
| 338 | }
|
|---|
| 339 | }
|
|---|
| 340 |
|
|---|
| 341 | sub EndTag {
|
|---|
| 342 | my ($expat, $element) = @_;
|
|---|
| 343 |
|
|---|
| 344 | my $raw_tag = "</$element>";
|
|---|
| 345 |
|
|---|
| 346 | if ($element =~ m/$self->{'root_tag'}/) {
|
|---|
| 347 | $self->{'raw_text'} .= $raw_tag;
|
|---|
| 348 |
|
|---|
| 349 | my $metadatatext = $self->{'metadatatext'};
|
|---|
| 350 | $metadatatext .= "</table>";
|
|---|
| 351 |
|
|---|
| 352 | my $raw_text = $self->{'raw_text'};
|
|---|
| 353 |
|
|---|
| 354 | $self->{'saved_metadata'}->{'MetadataTable'} = $metadatatext;
|
|---|
| 355 | $self->{'metadatatext'} = "";
|
|---|
| 356 |
|
|---|
| 357 | $self->{'saved_metadata'}->{'rawtext'} = $raw_text;
|
|---|
| 358 | $self->{'raw_text'} = "";
|
|---|
| 359 |
|
|---|
| 360 | if (defined $self->{'lom_language'}) {
|
|---|
| 361 | $self->{'saved_metadata'}->{'lom_language'} = $self->{'lom_language'};
|
|---|
| 362 | $self->{'lom_language'} = undef;
|
|---|
| 363 | }
|
|---|
| 364 | }
|
|---|
| 365 | else {
|
|---|
| 366 | my $metaname_stack = $self->{'metaname_stack'};
|
|---|
| 367 |
|
|---|
| 368 | if (scalar(@$metaname_stack)==1) {
|
|---|
| 369 | my $header_line = $self->{'mdheader'};
|
|---|
| 370 | my $tbody_start = $self->{'mdtbody'};
|
|---|
| 371 | my $tbody_text = $self->{'mdtbody_text'};
|
|---|
| 372 | if ($tbody_text !~ m/^\s*$/s) {
|
|---|
| 373 | my $tbody_end = "</tbody>\n";
|
|---|
| 374 | my $table_chunk
|
|---|
| 375 | = $header_line.$tbody_start.$tbody_text.$tbody_end;
|
|---|
| 376 |
|
|---|
| 377 | $self->{'metadatatext'} .= $table_chunk;
|
|---|
| 378 | }
|
|---|
| 379 | $self->{'mdtheader'} = "";
|
|---|
| 380 | $self->{'mdtbody'} = "";
|
|---|
| 381 | $self->{'mdtbody_text'} = "";
|
|---|
| 382 | }
|
|---|
| 383 |
|
|---|
| 384 | pop(@$metaname_stack);
|
|---|
| 385 |
|
|---|
| 386 | my $xml_depth = scalar(@{$self->{'metaname_stack'}});
|
|---|
| 387 | $self->{'raw_text'} .= "\n";
|
|---|
| 388 | $self->{'raw_text'} .= " " x $xml_depth;
|
|---|
| 389 | $self->{'raw_text'} .= $raw_tag;
|
|---|
| 390 | }
|
|---|
| 391 | }
|
|---|
| 392 |
|
|---|
| 393 | sub process_datatype_info
|
|---|
| 394 | {
|
|---|
| 395 | my $self = shift(@_);
|
|---|
| 396 | my ($metaname_stack,$md_content) = @_;
|
|---|
| 397 |
|
|---|
| 398 | my @without_dt_stack = @$metaname_stack; # without datatype stack
|
|---|
| 399 |
|
|---|
| 400 | my $innermost_element = $without_dt_stack[$#without_dt_stack];
|
|---|
| 401 |
|
|---|
| 402 | # Loose last item if encoding datatype information
|
|---|
| 403 | if ($innermost_element =~ m/^(lang)?string$/) {
|
|---|
| 404 | $self->{'lom_datatype'} = $innermost_element;
|
|---|
| 405 |
|
|---|
| 406 | pop @without_dt_stack;
|
|---|
| 407 | $innermost_element = $without_dt_stack[$#without_dt_stack];
|
|---|
| 408 | }
|
|---|
| 409 | elsif ($innermost_element =~ m/^date(Time)?$/i) {
|
|---|
| 410 | if ($innermost_element =~ m/^date$/i) {
|
|---|
| 411 | $self->{'lom_datatype'} = "dateTime";
|
|---|
| 412 | }
|
|---|
| 413 | else {
|
|---|
| 414 | $self->{'lom_datatype'} = $innermost_element;
|
|---|
| 415 |
|
|---|
| 416 | pop @without_dt_stack;
|
|---|
| 417 | $innermost_element = $without_dt_stack[$#without_dt_stack];
|
|---|
| 418 | }
|
|---|
| 419 |
|
|---|
| 420 | if ($md_content =~ m/^(\d{1,2})\s*(Jan|Feb|Mar|Apr|May|Jun|Jul|Aug|Sep|Oct|Nov|Dec)\w*\s*(\d{4})/i) {
|
|---|
| 421 | my ($day,$mon,$year) = ($1,$2,$3);
|
|---|
| 422 |
|
|---|
| 423 | my %month_lookup = ( 'jan' => 1, 'feb' => 2, 'mar' => 3,
|
|---|
| 424 | 'apr' => 4, 'may' => 5, 'jun' => 6,
|
|---|
| 425 | 'jul' => 7, 'aug' => 8, 'sep' => 9,
|
|---|
| 426 | 'oct' => 10, 'nov' => 11, 'dec' => 12 );
|
|---|
| 427 |
|
|---|
| 428 | my $mon_num = $month_lookup{lc($mon)};
|
|---|
| 429 |
|
|---|
| 430 | $md_content = sprintf("%d%02d%02d",$year,$mon_num,$day);
|
|---|
| 431 | }
|
|---|
| 432 |
|
|---|
| 433 | $md_content =~ s/\-//g;
|
|---|
| 434 | }
|
|---|
| 435 |
|
|---|
| 436 | if ($innermost_element eq "source") {
|
|---|
| 437 | $self->{'lom_source'} = $md_content;
|
|---|
| 438 | }
|
|---|
| 439 | elsif ($innermost_element eq "value") {
|
|---|
| 440 | $self->{'lom_value'} = $md_content;
|
|---|
| 441 | }
|
|---|
| 442 |
|
|---|
| 443 | return (\@without_dt_stack,$innermost_element,$md_content);
|
|---|
| 444 | }
|
|---|
| 445 |
|
|---|
| 446 | sub reset_datatype_info
|
|---|
| 447 | {
|
|---|
| 448 | my $self = shift(@_);
|
|---|
| 449 |
|
|---|
| 450 | $self->{'lom_datatype'} = "";
|
|---|
| 451 | }
|
|---|
| 452 |
|
|---|
| 453 |
|
|---|
| 454 | sub pretty_print_text
|
|---|
| 455 | {
|
|---|
| 456 | my $self = shift(@_);
|
|---|
| 457 |
|
|---|
| 458 | my ($pretty_print_text) = @_;
|
|---|
| 459 |
|
|---|
| 460 | ## $metavalue_utf8 = &util::hyperlink_text($metavalue_utf8);
|
|---|
| 461 | $pretty_print_text = &util::hyperlink_text($pretty_print_text);
|
|---|
| 462 |
|
|---|
| 463 | #### $pretty_print_text =~ s/(BEGIN:vCard.*END:vCard)/<pre>$1<\/pre>/sg;
|
|---|
| 464 |
|
|---|
| 465 | if ($self->{'lom_datatype'} eq "dateTime") {
|
|---|
| 466 | if ($pretty_print_text =~ m/^(\d{4})(\d{2})(\d{2})$/) {
|
|---|
| 467 | $pretty_print_text = "$1-$2-$3";
|
|---|
| 468 | }
|
|---|
| 469 | }
|
|---|
| 470 |
|
|---|
| 471 | return $pretty_print_text;
|
|---|
| 472 | }
|
|---|
| 473 |
|
|---|
| 474 | sub pretty_print_table_tr
|
|---|
| 475 | {
|
|---|
| 476 | my $self = shift (@_);
|
|---|
| 477 | my ($without_dt_stack) = @_;
|
|---|
| 478 |
|
|---|
| 479 | my $style = "class=\"metadata\"";
|
|---|
| 480 |
|
|---|
| 481 | my $innermost_element = $without_dt_stack->[scalar(@$without_dt_stack)-1];
|
|---|
| 482 | my $outermost_element = $without_dt_stack->[0];
|
|---|
| 483 |
|
|---|
| 484 | # Loose top level stack item (already named in pretty print table)
|
|---|
| 485 | my @pretty_print_stack = @$without_dt_stack;
|
|---|
| 486 | shift @pretty_print_stack;
|
|---|
| 487 |
|
|---|
| 488 | if ($innermost_element eq "source") {
|
|---|
| 489 | return if (!defined $self->{'lom_value'});
|
|---|
| 490 | }
|
|---|
| 491 |
|
|---|
| 492 | if ($innermost_element eq "value") {
|
|---|
| 493 | return if (!defined $self->{'lom_source'});
|
|---|
| 494 | }
|
|---|
| 495 |
|
|---|
| 496 | my $pretty_print_text = "";
|
|---|
| 497 |
|
|---|
| 498 | if (($innermost_element eq "value") || ($innermost_element eq "source")) {
|
|---|
| 499 | my $source = $self->{'lom_source'};
|
|---|
| 500 | my $value = $self->pretty_print_text($self->{'lom_value'});
|
|---|
| 501 |
|
|---|
| 502 | $self->{'lom_source'} = undef;
|
|---|
| 503 | $self->{'lom_value'} = undef;
|
|---|
| 504 |
|
|---|
| 505 | pop @pretty_print_stack;
|
|---|
| 506 |
|
|---|
| 507 | $pretty_print_text = "<td $style>$source</td><td $style>$value</td>";
|
|---|
| 508 | }
|
|---|
| 509 | else {
|
|---|
| 510 | $pretty_print_text = $self->pretty_print_text($_);
|
|---|
| 511 | $pretty_print_text = "<td $style colspan=2>$pretty_print_text</td>";
|
|---|
| 512 | }
|
|---|
| 513 | my $pretty_print_fmn = join(' : ',map { "\u$_"; } @pretty_print_stack);
|
|---|
| 514 |
|
|---|
| 515 |
|
|---|
| 516 | # my $tr_attr = "id=\"$outermost_element\" style=\"display:block;\"";
|
|---|
| 517 | my $tr_attr = "$style id=\"$outermost_element\"";
|
|---|
| 518 |
|
|---|
| 519 | my $mdtext_line = " <tr $tr_attr><td $style><nobr>$pretty_print_fmn</nobr></td>$pretty_print_text</tr>\n";
|
|---|
| 520 | $self->{'mdtbody_text'} .= $mdtext_line;
|
|---|
| 521 | }
|
|---|
| 522 |
|
|---|
| 523 |
|
|---|
| 524 | sub check_for_language
|
|---|
| 525 | {
|
|---|
| 526 | my $self = shift(@_);
|
|---|
| 527 | my ($innermost_element,$md_content) = @_;
|
|---|
| 528 |
|
|---|
| 529 | # Look for 'language' tag
|
|---|
| 530 | if ($innermost_element eq "language") {
|
|---|
| 531 | my $lom_lang = $self->{'lom_language'};
|
|---|
| 532 |
|
|---|
| 533 | if (defined $lom_lang) {
|
|---|
| 534 | my $new_lom_lang = $md_content;
|
|---|
| 535 | $new_lom_lang =~ s/-.*//; # remove endings like -US or -GB
|
|---|
| 536 |
|
|---|
| 537 | if ($lom_lang ne $new_lom_lang) {
|
|---|
| 538 | my $outhandle = $self->{'outhandle'};
|
|---|
| 539 |
|
|---|
| 540 | print $outhandle "Warning: Conflicting general language in record\n";
|
|---|
| 541 | print $outhandle " $new_lom_lang (previous value for language = $lom_lang)\n";
|
|---|
| 542 | }
|
|---|
| 543 | # otherwise, existing value OK => do nothing
|
|---|
| 544 | }
|
|---|
| 545 | else {
|
|---|
| 546 | $lom_lang = $md_content;
|
|---|
| 547 | $lom_lang =~ s/-.*//; # remove endings like -US or -GB
|
|---|
| 548 |
|
|---|
| 549 | $self->{'lom_language'} = $lom_lang;
|
|---|
| 550 | }
|
|---|
| 551 | }
|
|---|
| 552 | }
|
|---|
| 553 |
|
|---|
| 554 | sub found_specific_identifier
|
|---|
| 555 | {
|
|---|
| 556 | my $self = shift(@_);
|
|---|
| 557 | my ($specific_id,$full_mname,$md_content) = @_;
|
|---|
| 558 |
|
|---|
| 559 | my $found_id = 0;
|
|---|
| 560 | if ($full_mname eq $specific_id) {
|
|---|
| 561 | if ($md_content =~ m/^(http|ftp):/) {
|
|---|
| 562 | $found_id = 1;
|
|---|
| 563 | }
|
|---|
| 564 | }
|
|---|
| 565 |
|
|---|
| 566 | return $found_id;
|
|---|
| 567 | }
|
|---|
| 568 |
|
|---|
| 569 | sub download_srcdoc
|
|---|
| 570 | {
|
|---|
| 571 | my $self = shift(@_);
|
|---|
| 572 | my ($doc_url) = @_;
|
|---|
| 573 |
|
|---|
| 574 | my $outhandle = $self->{'outhandle'};
|
|---|
| 575 | my $output_dir = $self->{'output_dir'};
|
|---|
| 576 |
|
|---|
| 577 | $output_dir = &util::filename_cat($output_dir,"_gsdldown.all");
|
|---|
| 578 |
|
|---|
| 579 | if (! -d $output_dir) {
|
|---|
| 580 | mkdir $output_dir;
|
|---|
| 581 | }
|
|---|
| 582 |
|
|---|
| 583 | my $re_dirsep = &util::get_re_dirsep();
|
|---|
| 584 | my $os_dirsep = &util::get_dirsep();
|
|---|
| 585 |
|
|---|
| 586 | my $file_url = $doc_url;
|
|---|
| 587 | $file_url =~ s/$re_dirsep/$os_dirsep/g;
|
|---|
| 588 | $file_url =~ s/^(http|ftp):\/\///;
|
|---|
| 589 | $file_url .= "index.html" if ($file_url =~ m/\/$/);
|
|---|
| 590 |
|
|---|
| 591 | my $full_file_url = &util::filename_cat($output_dir,$file_url);
|
|---|
| 592 | # the path to srcdoc will be used later in extrametadata to associate
|
|---|
| 593 | # the lom metadata with the document. Needs to be relative to current
|
|---|
| 594 | # directory.
|
|---|
| 595 | my $srcdoc_path = &util::filename_cat("_gsdldown.all", $file_url);
|
|---|
| 596 | my $check_timestamp = $self->{'check_timestamp'};
|
|---|
| 597 | my $status;
|
|---|
| 598 |
|
|---|
| 599 | if (($check_timestamp) || (!$check_timestamp && !-e $full_file_url)) {
|
|---|
| 600 | if (!-e $full_file_url) {
|
|---|
| 601 | print $outhandle "Mirroring $doc_url\n";
|
|---|
| 602 | }
|
|---|
| 603 | else {
|
|---|
| 604 | print $outhandle "Checking to see if update needed for $doc_url\n";
|
|---|
| 605 | }
|
|---|
| 606 |
|
|---|
| 607 | # on linux, if we pass an absolute path as -P arg to wget, then it
|
|---|
| 608 | # stuffs up the
|
|---|
| 609 | # URL rewriting in the file. Need a relative path or none, so now
|
|---|
| 610 | # we change working directory first.
|
|---|
| 611 | my $changed_dir = 0;
|
|---|
| 612 | my $current_dir = cwd();
|
|---|
| 613 | my $wget_cmd = "";
|
|---|
| 614 | if ($ENV{'GSDLOS'} ne "windows") {
|
|---|
| 615 | $changed_dir = 1;
|
|---|
| 616 |
|
|---|
| 617 | chdir "$output_dir";
|
|---|
| 618 | $wget_cmd = "wget -nv --timestamping -k -p \"$doc_url\"";
|
|---|
| 619 | } else {
|
|---|
| 620 | $wget_cmd = "wget -nv -P \"$output_dir\" --timestamping -k -p \"$doc_url\"";
|
|---|
| 621 | }
|
|---|
| 622 | ##print STDERR "**** wget = $wget_cmd\n";
|
|---|
| 623 |
|
|---|
| 624 |
|
|---|
| 625 | $status = system($wget_cmd);
|
|---|
| 626 | if ($changed_dir) {
|
|---|
| 627 | chdir $current_dir;
|
|---|
| 628 | }
|
|---|
| 629 | if ($status==0) {
|
|---|
| 630 | $self->{'lom_srcdoc'} = $srcdoc_path;
|
|---|
| 631 | }
|
|---|
| 632 | else {
|
|---|
| 633 | $self->{'lom_srcdoc'} = undef;
|
|---|
| 634 | print $outhandle "Error: failed to execute $wget_cmd\n";
|
|---|
| 635 | }
|
|---|
| 636 | }
|
|---|
| 637 | else {
|
|---|
| 638 | # not time-stamping and file already exists
|
|---|
| 639 | $status=0;
|
|---|
| 640 | $self->{'lom_srcdoc'} = $srcdoc_path;
|
|---|
| 641 | }
|
|---|
| 642 |
|
|---|
| 643 | return $status==0;
|
|---|
| 644 |
|
|---|
| 645 | }
|
|---|
| 646 |
|
|---|
| 647 |
|
|---|
| 648 | sub check_for_identifier
|
|---|
| 649 | {
|
|---|
| 650 | my $self = shift(@_);
|
|---|
| 651 | my ($full_mname,$md_content) = @_;
|
|---|
| 652 |
|
|---|
| 653 | my $success = 0;
|
|---|
| 654 |
|
|---|
| 655 | my $download_re = $self->{'download_srcdocs'};
|
|---|
| 656 | if (($download_re ne "") && $md_content =~ m/$download_re/) {
|
|---|
| 657 |
|
|---|
| 658 | if ($self->found_specific_identifier("general^identifier^entry",$full_mname,$md_content)) {
|
|---|
| 659 | $success = $self->download_srcdoc($md_content);
|
|---|
| 660 | }
|
|---|
| 661 |
|
|---|
| 662 | if (!$success) {
|
|---|
| 663 | if ($self->found_specific_identifier("technical^location",$full_mname,$md_content)) {
|
|---|
| 664 | $success = $self->download_srcdoc($md_content);
|
|---|
| 665 | }
|
|---|
| 666 | }
|
|---|
| 667 | }
|
|---|
| 668 |
|
|---|
| 669 | return $success;
|
|---|
| 670 | }
|
|---|
| 671 |
|
|---|
| 672 |
|
|---|
| 673 | sub Text {
|
|---|
| 674 | if ($_ !~ m/^\s*$/) {
|
|---|
| 675 | #
|
|---|
| 676 | # Work out indentations and line wraps for raw XML
|
|---|
| 677 | #
|
|---|
| 678 | my $xml_depth = scalar(@{$self->{'metaname_stack'}})+1;
|
|---|
| 679 | my $indent = " " x $xml_depth;
|
|---|
| 680 |
|
|---|
| 681 | my $formatted_text = "\n".$_;
|
|---|
| 682 |
|
|---|
| 683 | # break into lines < 80 chars on space
|
|---|
| 684 | $formatted_text =~ s/(.{50,80})\s+/$1\n/mg;
|
|---|
| 685 | $formatted_text =~ s/^/$indent/mg;
|
|---|
| 686 | ## $formatted_text =~ s/\s+$//s;
|
|---|
| 687 |
|
|---|
| 688 | $self->{'raw_text'} .= $formatted_text;
|
|---|
| 689 | }
|
|---|
| 690 |
|
|---|
| 691 | my $metaname_stack = $self->{'metaname_stack'};
|
|---|
| 692 | if (($_ !~ /^\s*$/) && (scalar(@$metaname_stack)>0)) {
|
|---|
| 693 |
|
|---|
| 694 | my ($without_dt_stack,$innermost_element,$md_content)
|
|---|
| 695 | = $self->process_datatype_info($metaname_stack,$_);
|
|---|
| 696 |
|
|---|
| 697 | $self->pretty_print_table_tr($without_dt_stack);
|
|---|
| 698 |
|
|---|
| 699 | my $full_mname = join('^',@{$without_dt_stack});
|
|---|
| 700 | $self->set_filere_metadata(lc($full_mname),$md_content);
|
|---|
| 701 |
|
|---|
| 702 | $self->check_for_language($innermost_element,$md_content);
|
|---|
| 703 | $self->check_for_identifier($full_mname,$md_content); # source doc
|
|---|
| 704 |
|
|---|
| 705 | $self->reset_datatype_info();
|
|---|
| 706 | }
|
|---|
| 707 | }
|
|---|
| 708 |
|
|---|
| 709 | # This Char function overrides the one in XML::Parser::Stream to overcome a
|
|---|
| 710 | # problem where $expat->{Text} is treated as the return value, slowing
|
|---|
| 711 | # things down significantly in some cases.
|
|---|
| 712 | sub Char {
|
|---|
| 713 | $_[0]->{'Text'} .= $_[1];
|
|---|
| 714 | return undef;
|
|---|
| 715 | }
|
|---|
| 716 |
|
|---|
| 717 | 1;
|
|---|
