[13324] | 1 | ###########################################################################
|
---|
| 2 | #
|
---|
[15872] | 3 | # LOMPlugin.pm -- plugin for import the collection from LOM
|
---|
[13324] | 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 |
|
---|
[13334] | 27 | ### Note this plugin currently can't download source documents from outside if you are behind a firewall.
|
---|
[22900] | 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??)
|
---|
[13334] | 31 |
|
---|
[15872] | 32 | package LOMPlugin;
|
---|
[13324] | 33 |
|
---|
[24951] | 34 | use extrametautil;
|
---|
[15872] | 35 | use ReadTextFile;
|
---|
[13324] | 36 | use MetadataPass;
|
---|
[24548] | 37 | use MetadataRead;
|
---|
[24971] | 38 | use util;
|
---|
[13324] | 39 | use XMLParser;
|
---|
| 40 | use Cwd;
|
---|
| 41 |
|
---|
[24548] | 42 | # methods with identical signatures take precedence in the order given in the ISA list.
|
---|
[13324] | 43 | sub BEGIN {
|
---|
[24548] | 44 | @ISA = ('MetadataRead', 'ReadTextFile', 'MetadataPass');
|
---|
[13324] | 45 | }
|
---|
| 46 |
|
---|
| 47 | use strict; # every perl program should have this!
|
---|
| 48 | no strict 'refs'; # make an exception so we can use variables as filehandles
|
---|
| 49 |
|
---|
| 50 |
|
---|
| 51 | my $arguments =
|
---|
| 52 | [ { 'name' => "process_exp",
|
---|
[16019] | 53 | 'desc' => "{BasePlugin.process_exp}",
|
---|
[13324] | 54 | 'type' => "string",
|
---|
| 55 | 'deft' => &get_default_process_exp(),
|
---|
| 56 | 'reqd' => "no" },
|
---|
| 57 | { 'name' => "root_tag",
|
---|
[15872] | 58 | 'desc' => "{LOMPlugin.root_tag}",
|
---|
[13324] | 59 | 'type' => "regexp",
|
---|
| 60 | 'deft' => q/^(?i)lom$/,
|
---|
| 61 | 'reqd' => "no" },
|
---|
| 62 | { 'name' => "check_timestamp",
|
---|
[15872] | 63 | 'desc' => "{LOMPlugin.check_timestamp}",
|
---|
[13324] | 64 | 'type' => "flag" },
|
---|
| 65 | { 'name' => "download_srcdocs",
|
---|
[15872] | 66 | 'desc' => "{LOMPlugin.download_srcdocs}",
|
---|
[13324] | 67 | 'type' => "regexp",
|
---|
| 68 | 'deft' => "",
|
---|
| 69 | 'reqd' => "no" }];
|
---|
| 70 |
|
---|
[15872] | 71 | my $options = { 'name' => "LOMPlugin",
|
---|
| 72 | 'desc' => "{LOMPlugin.desc}",
|
---|
[15961] | 73 | 'abstract' => "no",
|
---|
[13324] | 74 | 'inherits' => "yes",
|
---|
| 75 | 'args' => $arguments };
|
---|
| 76 |
|
---|
| 77 |
|
---|
| 78 |
|
---|
| 79 | my ($self);
|
---|
| 80 | sub new {
|
---|
| 81 | my $class = shift (@_);
|
---|
| 82 | my ($pluginlist,$inputargs,$hashArgOptLists) = @_;
|
---|
| 83 | push(@$pluginlist, $class);
|
---|
| 84 |
|
---|
[15872] | 85 | push(@{$hashArgOptLists->{"ArgList"}},@{$arguments});
|
---|
| 86 | push(@{$hashArgOptLists->{"OptList"}},$options);
|
---|
[13324] | 87 |
|
---|
[15872] | 88 | $self = new ReadTextFile($pluginlist, $inputargs, $hashArgOptLists);
|
---|
[13324] | 89 |
|
---|
[15872] | 90 | if ($self->{'info_only'}) {
|
---|
| 91 | # don't worry about creating the XML parser as all we want is the
|
---|
| 92 | # list of plugin options
|
---|
| 93 | return bless $self, $class;
|
---|
| 94 | }
|
---|
| 95 |
|
---|
[13324] | 96 | #create XML::Parser object for parsing dublin_core.xml files
|
---|
| 97 | my $parser = new XML::Parser('Style' => 'Stream',
|
---|
| 98 | 'Handlers' => {'Char' => \&Char,
|
---|
| 99 | 'Doctype' => \&Doctype
|
---|
| 100 | });
|
---|
| 101 | $self->{'parser'} = $parser;
|
---|
| 102 |
|
---|
| 103 | $self->{'extra_blocks'} = {};
|
---|
| 104 |
|
---|
| 105 | return bless $self, $class;
|
---|
| 106 | }
|
---|
| 107 |
|
---|
| 108 | sub get_default_process_exp {
|
---|
| 109 | my $self = shift (@_);
|
---|
| 110 |
|
---|
| 111 | return q^(?i)\.xml$^;
|
---|
| 112 | }
|
---|
| 113 |
|
---|
| 114 |
|
---|
[16392] | 115 | sub can_process_this_file {
|
---|
| 116 | my $self = shift(@_);
|
---|
| 117 | my ($filename) = @_;
|
---|
[13324] | 118 |
|
---|
[16392] | 119 | if ($self->SUPER::can_process_this_file($filename) && $self->check_doctype($filename)) {
|
---|
| 120 | return 1; # its a file for us
|
---|
| 121 | }
|
---|
| 122 | return 0;
|
---|
| 123 | }
|
---|
[13324] | 124 |
|
---|
| 125 | sub metadata_read {
|
---|
| 126 | my $self = shift (@_);
|
---|
[19493] | 127 | my ($pluginfo, $base_dir, $file, $block_hash,
|
---|
| 128 | $extrametakeys, $extrametadata, $extrametafile,
|
---|
[23212] | 129 | $processor, $gli, $aux) = @_;
|
---|
[13324] | 130 |
|
---|
| 131 | my $outhandle = $self->{'outhandle'};
|
---|
| 132 |
|
---|
[16392] | 133 | # can we process this file??
|
---|
| 134 | my ($filename_full_path, $filename_no_path) = &util::get_full_filenames($base_dir, $file);
|
---|
[24548] | 135 | return undef unless $self->can_process_this_file_for_metadata($filename_full_path);
|
---|
[13324] | 136 |
|
---|
| 137 | $file =~ s/^[\/\\]+//; # $file often begins with / so we'll tidy it up
|
---|
| 138 |
|
---|
[15872] | 139 | print $outhandle "LOMPlugin: extracting metadata from $file\n"
|
---|
[13324] | 140 | if $self->{'verbosity'} > 1;
|
---|
| 141 |
|
---|
[16392] | 142 | my ($dir,$tail) = $filename_full_path =~ /^(.*?)([^\/\\]*)$/;
|
---|
[13324] | 143 | $self->{'output_dir'} = $dir;
|
---|
| 144 |
|
---|
| 145 | eval {
|
---|
[16392] | 146 | $self->{'parser'}->parsefile($filename_full_path);
|
---|
[13324] | 147 | };
|
---|
| 148 |
|
---|
| 149 | if ($@) {
|
---|
[16392] | 150 | print $outhandle "LOMPlugin: skipping $filename_full_path as not conformant to LOM syntax\n" if ($self->{'verbosity'} > 1);
|
---|
[13324] | 151 | print $outhandle "\n Perl Error:\n $@\n" if ($self->{'verbosity'}>2);
|
---|
| 152 | return 0;
|
---|
| 153 | }
|
---|
| 154 |
|
---|
| 155 | $self->{'output_dir'} = undef;
|
---|
| 156 |
|
---|
| 157 | my $file_re;
|
---|
| 158 | my $lom_srcdoc = $self->{'lom_srcdoc'};
|
---|
| 159 |
|
---|
| 160 | if (defined $lom_srcdoc) {
|
---|
| 161 | my $dirsep = &util::get_re_dirsep();
|
---|
| 162 | $lom_srcdoc =~ s/^$base_dir($dirsep)//;
|
---|
| 163 | $self->{'extra_blocks'}->{$file}++;
|
---|
| 164 | $file_re = $lom_srcdoc;
|
---|
| 165 | }
|
---|
| 166 | else {
|
---|
| 167 | $file_re = $tail;
|
---|
| 168 | }
|
---|
[24971] | 169 |
|
---|
| 170 | # Indexing into the extrameta data structures requires the filename's style of slashes to be in URL format
|
---|
| 171 | # Then need to convert the filename to a regex, no longer to protect windows directory chars \, but for
|
---|
| 172 | # protecting special characters like brackets in the filepath such as "C:\Program Files (x86)\Greenstone".
|
---|
| 173 | $file_re = &util::filepath_to_url_format($file_re);
|
---|
[22900] | 174 | $file_re = &util::filename_to_regex($file_re);
|
---|
[13324] | 175 | $self->{'lom_srcdoc'} = undef; # reset for next file to be processed
|
---|
| 176 |
|
---|
[24951] | 177 | &extrametautil::addmetakey($extrametakeys, $file_re);
|
---|
| 178 | &extrametautil::setmetadata($extrametadata, $file_re, $self->{'saved_metadata'});
|
---|
[22900] | 179 | if (defined $lom_srcdoc) {
|
---|
| 180 | # copied from oaiplugin
|
---|
[24951] | 181 | if (!defined &extrametautil::getmetafile($extrametafile, $file_re)) {
|
---|
| 182 | &extrametautil::setmetafile($extrametafile, $file_re, {});
|
---|
[22900] | 183 | }
|
---|
| 184 | #maps the file to full path
|
---|
[24951] | 185 | &extrametautil::setmetafile_for_named_file($extrametafile, $file_re, $file, $filename_full_path);
|
---|
[22900] | 186 | }
|
---|
| 187 |
|
---|
[13324] | 188 | return 1;
|
---|
| 189 | }
|
---|
| 190 |
|
---|
| 191 | sub check_doctype {
|
---|
| 192 | $self = shift (@_);
|
---|
| 193 |
|
---|
| 194 | my ($filename) = @_;
|
---|
| 195 |
|
---|
| 196 | if (open(XMLIN,"<$filename")) {
|
---|
| 197 | my $doctype = $self->{'root_tag'};
|
---|
| 198 | ## check whether the doctype has the same name as the root element tag
|
---|
| 199 | while (defined (my $line = <XMLIN>)) {
|
---|
| 200 | ## find the root element
|
---|
| 201 | if ($line =~ /<([\w\d:]+)[\s>]/){
|
---|
| 202 | my $root = $1;
|
---|
| 203 | if ($root !~ $doctype){
|
---|
| 204 | close(XMLIN);
|
---|
| 205 | return 0;
|
---|
| 206 | }
|
---|
| 207 | else {
|
---|
| 208 | close(XMLIN);
|
---|
| 209 | return 1;
|
---|
| 210 | }
|
---|
| 211 | }
|
---|
| 212 | }
|
---|
| 213 | close(XMLIN);
|
---|
| 214 | }
|
---|
| 215 |
|
---|
| 216 | return undef; # haven't found a valid line
|
---|
| 217 |
|
---|
| 218 | }
|
---|
| 219 |
|
---|
| 220 | sub read_file {
|
---|
| 221 | my $self = shift (@_);
|
---|
| 222 | my ($filename, $encoding, $language, $textref) = @_;
|
---|
| 223 |
|
---|
| 224 | my $metadata_table = $self->{'metadata_table'};
|
---|
| 225 |
|
---|
| 226 | my $rawtext = $metadata_table->{'rawtext'};
|
---|
| 227 |
|
---|
| 228 | delete $metadata_table->{'rawtext'};
|
---|
| 229 |
|
---|
| 230 | $$textref = $rawtext;
|
---|
| 231 | }
|
---|
| 232 |
|
---|
| 233 | sub read {
|
---|
| 234 | my $self = shift (@_);
|
---|
[16392] | 235 | my ($pluginfo, $base_dir, $file, $block_hash, $metadata, $processor, $maxdocs, $total_count, $gli) = @_;
|
---|
[13324] | 236 |
|
---|
| 237 | my $outhandle = $self->{'outhandle'};
|
---|
| 238 |
|
---|
| 239 | return 0 if (defined $self->{'extra_blocks'}->{$file});
|
---|
| 240 |
|
---|
[16392] | 241 | # can we process this file??
|
---|
| 242 | my ($filename_full_path, $filename_no_path) = &util::get_full_filenames($base_dir, $file);
|
---|
| 243 | return undef unless $self->can_process_this_file($filename_full_path);
|
---|
[13324] | 244 |
|
---|
| 245 | $self->{'metadata_table'} = $metadata;
|
---|
| 246 |
|
---|
| 247 | my $lom_language = $metadata->{'lom_language'};
|
---|
| 248 |
|
---|
| 249 | my $store_input_encoding;
|
---|
| 250 | my $store_extract_language;
|
---|
| 251 | my $store_default_language;
|
---|
| 252 | my $store_default_encoding;
|
---|
| 253 |
|
---|
| 254 | if (defined $lom_language) {
|
---|
| 255 | delete $metadata->{'lom_language'};
|
---|
| 256 |
|
---|
| 257 | $store_input_encoding = $self->{'input_encoding'};
|
---|
| 258 | $store_extract_language = $self->{'extract_language'};
|
---|
| 259 | $store_default_language = $self->{'default_language'};
|
---|
| 260 | $store_default_encoding = $self->{'default_encoding'};
|
---|
| 261 |
|
---|
| 262 | $self->{'input_encoding'} = "utf8";
|
---|
| 263 | $self->{'extract_language'} = 0;
|
---|
| 264 | $self->{'default_language'} = $lom_language;
|
---|
| 265 | $self->{'default_encoding'} = "utf8";
|
---|
| 266 | }
|
---|
| 267 |
|
---|
| 268 | my $rv = $self->SUPER::read(@_);
|
---|
| 269 |
|
---|
| 270 | if (defined $lom_language) {
|
---|
| 271 | $self->{'input_encoding'} = $store_input_encoding;
|
---|
| 272 | $self->{'extract_language'} = $store_extract_language;
|
---|
| 273 | $self->{'default_language'} = $store_default_language;
|
---|
| 274 | $self->{'default_encoding'} = $store_default_encoding;
|
---|
| 275 | }
|
---|
| 276 |
|
---|
| 277 | $self->{'metadata_table'} = undef;
|
---|
| 278 |
|
---|
| 279 | return $rv;
|
---|
| 280 | }
|
---|
| 281 |
|
---|
| 282 | # do plugin specific processing of doc_obj
|
---|
| 283 | sub process {
|
---|
| 284 | my $self = shift (@_);
|
---|
| 285 | my ($textref, $pluginfo, $base_dir, $file, $metadata, $doc_obj, $gli) = @_;
|
---|
| 286 | my $outhandle = $self->{'outhandle'};
|
---|
| 287 |
|
---|
| 288 | my $cursection = $doc_obj->get_top_section();
|
---|
| 289 | $doc_obj->add_utf8_text($cursection, $$textref);
|
---|
| 290 |
|
---|
| 291 | return 1;
|
---|
| 292 | }
|
---|
| 293 |
|
---|
| 294 | sub Doctype {
|
---|
| 295 | my ($expat, $name, $sysid, $pubid, $internal) = @_;
|
---|
| 296 |
|
---|
| 297 | my $root_tag = $self->{'root_tag'};
|
---|
| 298 |
|
---|
| 299 | if ($name !~ /$root_tag/) {
|
---|
| 300 | die "Root tag $name does not match regular expression $root_tag";
|
---|
| 301 | }
|
---|
| 302 | }
|
---|
| 303 |
|
---|
| 304 | sub StartTag {
|
---|
| 305 | my ($expat, $element) = @_;
|
---|
| 306 |
|
---|
| 307 | my %attr = %_;
|
---|
| 308 |
|
---|
| 309 | my $raw_tag = "<$element";
|
---|
| 310 | map { $raw_tag .= " $_=\"$attr{$_}\""; } keys %attr;
|
---|
| 311 | $raw_tag .= ">";
|
---|
| 312 |
|
---|
| 313 | if ($element =~ m/$self->{'root_tag'}/) {
|
---|
| 314 | $self->{'raw_text'} = $raw_tag;
|
---|
| 315 |
|
---|
| 316 | $self->{'saved_metadata'} = {};
|
---|
| 317 | $self->{'metaname_stack'} = [];
|
---|
| 318 | $self->{'lom_datatype'} = "";
|
---|
| 319 | $self->{'lom_language'} = undef;
|
---|
| 320 | $self->{'metadatatext'} = "<table class=\"metadata\" width=\"_pagewidth_\" >\n";
|
---|
| 321 | }
|
---|
| 322 | else {
|
---|
| 323 | my $xml_depth = scalar(@{$self->{'metaname_stack'}});
|
---|
| 324 | $self->{'raw_text'} .= "\n";
|
---|
| 325 | $self->{'raw_text'} .= " " x $xml_depth;
|
---|
| 326 | $self->{'raw_text'} .= $raw_tag;
|
---|
| 327 |
|
---|
| 328 | my $metaname_stack = $self->{'metaname_stack'};
|
---|
| 329 | push(@$metaname_stack,$element);
|
---|
| 330 | if (scalar(@$metaname_stack)==1) {
|
---|
| 331 | # top level LOM category
|
---|
| 332 | my $style = "class=\"metadata\"";
|
---|
| 333 | my $open_close
|
---|
| 334 | = "<a id=\"${element}opencloselink\" href=\"javascript:hideTBodyArea('$element')\">\n";
|
---|
| 335 | $open_close
|
---|
| 336 | .= "<img id=\"${element}openclose\" border=\"0\" src=\"_httpopenmdicon_\"></a>\n";
|
---|
| 337 |
|
---|
| 338 | my $header_line = " <tr $style ><th $style colspan=\"3\">$open_close \u$element</th></tr>\n";
|
---|
| 339 | my $md_tbody = "<tbody id=\"$element\">\n";
|
---|
| 340 |
|
---|
| 341 | $self->{'mdheader'} = $header_line;
|
---|
| 342 | $self->{'mdtbody'} = $md_tbody;
|
---|
| 343 | $self->{'mdtbody_text'} = "";
|
---|
| 344 | }
|
---|
| 345 | }
|
---|
| 346 | }
|
---|
| 347 |
|
---|
| 348 | sub EndTag {
|
---|
| 349 | my ($expat, $element) = @_;
|
---|
| 350 |
|
---|
| 351 | my $raw_tag = "</$element>";
|
---|
| 352 |
|
---|
| 353 | if ($element =~ m/$self->{'root_tag'}/) {
|
---|
| 354 | $self->{'raw_text'} .= $raw_tag;
|
---|
| 355 |
|
---|
| 356 | my $metadatatext = $self->{'metadatatext'};
|
---|
| 357 | $metadatatext .= "</table>";
|
---|
| 358 |
|
---|
| 359 | my $raw_text = $self->{'raw_text'};
|
---|
| 360 |
|
---|
| 361 | $self->{'saved_metadata'}->{'MetadataTable'} = $metadatatext;
|
---|
| 362 | $self->{'metadatatext'} = "";
|
---|
| 363 |
|
---|
| 364 | $self->{'saved_metadata'}->{'rawtext'} = $raw_text;
|
---|
| 365 | $self->{'raw_text'} = "";
|
---|
| 366 |
|
---|
| 367 | if (defined $self->{'lom_language'}) {
|
---|
| 368 | $self->{'saved_metadata'}->{'lom_language'} = $self->{'lom_language'};
|
---|
| 369 | $self->{'lom_language'} = undef;
|
---|
| 370 | }
|
---|
| 371 | }
|
---|
| 372 | else {
|
---|
| 373 | my $metaname_stack = $self->{'metaname_stack'};
|
---|
| 374 |
|
---|
| 375 | if (scalar(@$metaname_stack)==1) {
|
---|
| 376 | my $header_line = $self->{'mdheader'};
|
---|
| 377 | my $tbody_start = $self->{'mdtbody'};
|
---|
| 378 | my $tbody_text = $self->{'mdtbody_text'};
|
---|
| 379 | if ($tbody_text !~ m/^\s*$/s) {
|
---|
| 380 | my $tbody_end = "</tbody>\n";
|
---|
| 381 | my $table_chunk
|
---|
| 382 | = $header_line.$tbody_start.$tbody_text.$tbody_end;
|
---|
| 383 |
|
---|
| 384 | $self->{'metadatatext'} .= $table_chunk;
|
---|
| 385 | }
|
---|
| 386 | $self->{'mdtheader'} = "";
|
---|
| 387 | $self->{'mdtbody'} = "";
|
---|
| 388 | $self->{'mdtbody_text'} = "";
|
---|
| 389 | }
|
---|
| 390 |
|
---|
| 391 | pop(@$metaname_stack);
|
---|
| 392 |
|
---|
| 393 | my $xml_depth = scalar(@{$self->{'metaname_stack'}});
|
---|
| 394 | $self->{'raw_text'} .= "\n";
|
---|
| 395 | $self->{'raw_text'} .= " " x $xml_depth;
|
---|
| 396 | $self->{'raw_text'} .= $raw_tag;
|
---|
| 397 | }
|
---|
| 398 | }
|
---|
| 399 |
|
---|
| 400 | sub process_datatype_info
|
---|
| 401 | {
|
---|
| 402 | my $self = shift(@_);
|
---|
| 403 | my ($metaname_stack,$md_content) = @_;
|
---|
| 404 |
|
---|
| 405 | my @without_dt_stack = @$metaname_stack; # without datatype stack
|
---|
| 406 |
|
---|
| 407 | my $innermost_element = $without_dt_stack[$#without_dt_stack];
|
---|
| 408 |
|
---|
| 409 | # Loose last item if encoding datatype information
|
---|
| 410 | if ($innermost_element =~ m/^(lang)?string$/) {
|
---|
| 411 | $self->{'lom_datatype'} = $innermost_element;
|
---|
| 412 |
|
---|
| 413 | pop @without_dt_stack;
|
---|
| 414 | $innermost_element = $without_dt_stack[$#without_dt_stack];
|
---|
| 415 | }
|
---|
| 416 | elsif ($innermost_element =~ m/^date(Time)?$/i) {
|
---|
| 417 | if ($innermost_element =~ m/^date$/i) {
|
---|
| 418 | $self->{'lom_datatype'} = "dateTime";
|
---|
| 419 | }
|
---|
| 420 | else {
|
---|
| 421 | $self->{'lom_datatype'} = $innermost_element;
|
---|
| 422 |
|
---|
| 423 | pop @without_dt_stack;
|
---|
| 424 | $innermost_element = $without_dt_stack[$#without_dt_stack];
|
---|
| 425 | }
|
---|
| 426 |
|
---|
| 427 | if ($md_content =~ m/^(\d{1,2})\s*(Jan|Feb|Mar|Apr|May|Jun|Jul|Aug|Sep|Oct|Nov|Dec)\w*\s*(\d{4})/i) {
|
---|
| 428 | my ($day,$mon,$year) = ($1,$2,$3);
|
---|
| 429 |
|
---|
| 430 | my %month_lookup = ( 'jan' => 1, 'feb' => 2, 'mar' => 3,
|
---|
| 431 | 'apr' => 4, 'may' => 5, 'jun' => 6,
|
---|
| 432 | 'jul' => 7, 'aug' => 8, 'sep' => 9,
|
---|
| 433 | 'oct' => 10, 'nov' => 11, 'dec' => 12 );
|
---|
| 434 |
|
---|
| 435 | my $mon_num = $month_lookup{lc($mon)};
|
---|
| 436 |
|
---|
| 437 | $md_content = sprintf("%d%02d%02d",$year,$mon_num,$day);
|
---|
| 438 | }
|
---|
| 439 |
|
---|
| 440 | $md_content =~ s/\-//g;
|
---|
| 441 | }
|
---|
| 442 |
|
---|
| 443 | if ($innermost_element eq "source") {
|
---|
| 444 | $self->{'lom_source'} = $md_content;
|
---|
| 445 | }
|
---|
| 446 | elsif ($innermost_element eq "value") {
|
---|
| 447 | $self->{'lom_value'} = $md_content;
|
---|
| 448 | }
|
---|
| 449 |
|
---|
| 450 | return (\@without_dt_stack,$innermost_element,$md_content);
|
---|
| 451 | }
|
---|
| 452 |
|
---|
| 453 | sub reset_datatype_info
|
---|
| 454 | {
|
---|
| 455 | my $self = shift(@_);
|
---|
| 456 |
|
---|
| 457 | $self->{'lom_datatype'} = "";
|
---|
| 458 | }
|
---|
| 459 |
|
---|
| 460 |
|
---|
| 461 | sub pretty_print_text
|
---|
| 462 | {
|
---|
| 463 | my $self = shift(@_);
|
---|
| 464 |
|
---|
| 465 | my ($pretty_print_text) = @_;
|
---|
| 466 |
|
---|
| 467 | ## $metavalue_utf8 = &util::hyperlink_text($metavalue_utf8);
|
---|
| 468 | $pretty_print_text = &util::hyperlink_text($pretty_print_text);
|
---|
| 469 |
|
---|
| 470 | #### $pretty_print_text =~ s/(BEGIN:vCard.*END:vCard)/<pre>$1<\/pre>/sg;
|
---|
| 471 |
|
---|
| 472 | if ($self->{'lom_datatype'} eq "dateTime") {
|
---|
| 473 | if ($pretty_print_text =~ m/^(\d{4})(\d{2})(\d{2})$/) {
|
---|
| 474 | $pretty_print_text = "$1-$2-$3";
|
---|
| 475 | }
|
---|
| 476 | }
|
---|
| 477 |
|
---|
| 478 | return $pretty_print_text;
|
---|
| 479 | }
|
---|
| 480 |
|
---|
| 481 | sub pretty_print_table_tr
|
---|
| 482 | {
|
---|
| 483 | my $self = shift (@_);
|
---|
| 484 | my ($without_dt_stack) = @_;
|
---|
| 485 |
|
---|
| 486 | my $style = "class=\"metadata\"";
|
---|
| 487 |
|
---|
| 488 | my $innermost_element = $without_dt_stack->[scalar(@$without_dt_stack)-1];
|
---|
| 489 | my $outermost_element = $without_dt_stack->[0];
|
---|
| 490 |
|
---|
| 491 | # Loose top level stack item (already named in pretty print table)
|
---|
| 492 | my @pretty_print_stack = @$without_dt_stack;
|
---|
| 493 | shift @pretty_print_stack;
|
---|
| 494 |
|
---|
| 495 | if ($innermost_element eq "source") {
|
---|
| 496 | return if (!defined $self->{'lom_value'});
|
---|
| 497 | }
|
---|
| 498 |
|
---|
| 499 | if ($innermost_element eq "value") {
|
---|
| 500 | return if (!defined $self->{'lom_source'});
|
---|
| 501 | }
|
---|
| 502 |
|
---|
| 503 | my $pretty_print_text = "";
|
---|
| 504 |
|
---|
| 505 | if (($innermost_element eq "value") || ($innermost_element eq "source")) {
|
---|
| 506 | my $source = $self->{'lom_source'};
|
---|
| 507 | my $value = $self->pretty_print_text($self->{'lom_value'});
|
---|
| 508 |
|
---|
| 509 | $self->{'lom_source'} = undef;
|
---|
| 510 | $self->{'lom_value'} = undef;
|
---|
| 511 |
|
---|
| 512 | pop @pretty_print_stack;
|
---|
| 513 |
|
---|
| 514 | $pretty_print_text = "<td $style>$source</td><td $style>$value</td>";
|
---|
| 515 | }
|
---|
| 516 | else {
|
---|
| 517 | $pretty_print_text = $self->pretty_print_text($_);
|
---|
| 518 | $pretty_print_text = "<td $style colspan=2>$pretty_print_text</td>";
|
---|
| 519 | }
|
---|
| 520 | my $pretty_print_fmn = join(' : ',map { "\u$_"; } @pretty_print_stack);
|
---|
| 521 |
|
---|
| 522 |
|
---|
| 523 | # my $tr_attr = "id=\"$outermost_element\" style=\"display:block;\"";
|
---|
| 524 | my $tr_attr = "$style id=\"$outermost_element\"";
|
---|
| 525 |
|
---|
| 526 | my $mdtext_line = " <tr $tr_attr><td $style><nobr>$pretty_print_fmn</nobr></td>$pretty_print_text</tr>\n";
|
---|
| 527 | $self->{'mdtbody_text'} .= $mdtext_line;
|
---|
| 528 | }
|
---|
| 529 |
|
---|
| 530 |
|
---|
| 531 | sub check_for_language
|
---|
| 532 | {
|
---|
| 533 | my $self = shift(@_);
|
---|
| 534 | my ($innermost_element,$md_content) = @_;
|
---|
| 535 |
|
---|
| 536 | # Look for 'language' tag
|
---|
| 537 | if ($innermost_element eq "language") {
|
---|
| 538 | my $lom_lang = $self->{'lom_language'};
|
---|
| 539 |
|
---|
| 540 | if (defined $lom_lang) {
|
---|
| 541 | my $new_lom_lang = $md_content;
|
---|
| 542 | $new_lom_lang =~ s/-.*//; # remove endings like -US or -GB
|
---|
| 543 |
|
---|
| 544 | if ($lom_lang ne $new_lom_lang) {
|
---|
| 545 | my $outhandle = $self->{'outhandle'};
|
---|
| 546 |
|
---|
| 547 | print $outhandle "Warning: Conflicting general language in record\n";
|
---|
| 548 | print $outhandle " $new_lom_lang (previous value for language = $lom_lang)\n";
|
---|
| 549 | }
|
---|
| 550 | # otherwise, existing value OK => do nothing
|
---|
| 551 | }
|
---|
| 552 | else {
|
---|
| 553 | $lom_lang = $md_content;
|
---|
| 554 | $lom_lang =~ s/-.*//; # remove endings like -US or -GB
|
---|
| 555 |
|
---|
| 556 | $self->{'lom_language'} = $lom_lang;
|
---|
| 557 | }
|
---|
| 558 | }
|
---|
| 559 | }
|
---|
| 560 |
|
---|
| 561 | sub found_specific_identifier
|
---|
| 562 | {
|
---|
| 563 | my $self = shift(@_);
|
---|
| 564 | my ($specific_id,$full_mname,$md_content) = @_;
|
---|
| 565 |
|
---|
| 566 | my $found_id = 0;
|
---|
| 567 | if ($full_mname eq $specific_id) {
|
---|
| 568 | if ($md_content =~ m/^(http|ftp):/) {
|
---|
| 569 | $found_id = 1;
|
---|
| 570 | }
|
---|
| 571 | }
|
---|
| 572 |
|
---|
| 573 | return $found_id;
|
---|
| 574 | }
|
---|
| 575 |
|
---|
| 576 | sub download_srcdoc
|
---|
| 577 | {
|
---|
| 578 | my $self = shift(@_);
|
---|
| 579 | my ($doc_url) = @_;
|
---|
| 580 |
|
---|
| 581 | my $outhandle = $self->{'outhandle'};
|
---|
| 582 | my $output_dir = $self->{'output_dir'};
|
---|
| 583 |
|
---|
| 584 | $output_dir = &util::filename_cat($output_dir,"_gsdldown.all");
|
---|
| 585 |
|
---|
| 586 | if (! -d $output_dir) {
|
---|
| 587 | mkdir $output_dir;
|
---|
| 588 | }
|
---|
| 589 |
|
---|
| 590 | my $re_dirsep = &util::get_re_dirsep();
|
---|
| 591 | my $os_dirsep = &util::get_dirsep();
|
---|
| 592 |
|
---|
| 593 | my $file_url = $doc_url;
|
---|
| 594 | $file_url =~ s/$re_dirsep/$os_dirsep/g;
|
---|
| 595 | $file_url =~ s/^(http|ftp):\/\///;
|
---|
| 596 | $file_url .= "index.html" if ($file_url =~ m/\/$/);
|
---|
| 597 |
|
---|
| 598 | my $full_file_url = &util::filename_cat($output_dir,$file_url);
|
---|
[22900] | 599 | # the path to srcdoc will be used later in extrametadata to associate
|
---|
| 600 | # the lom metadata with the document. Needs to be relative to current
|
---|
| 601 | # directory.
|
---|
| 602 | my $srcdoc_path = &util::filename_cat("_gsdldown.all", $file_url);
|
---|
[13324] | 603 | my $check_timestamp = $self->{'check_timestamp'};
|
---|
| 604 | my $status;
|
---|
| 605 |
|
---|
| 606 | if (($check_timestamp) || (!$check_timestamp && !-e $full_file_url)) {
|
---|
| 607 | if (!-e $full_file_url) {
|
---|
| 608 | print $outhandle "Mirroring $doc_url\n";
|
---|
| 609 | }
|
---|
| 610 | else {
|
---|
| 611 | print $outhandle "Checking to see if update needed for $doc_url\n";
|
---|
| 612 | }
|
---|
| 613 |
|
---|
[13332] | 614 | # on linux, if we pass an absolute path as -P arg to wget, then it
|
---|
| 615 | # stuffs up the
|
---|
[13324] | 616 | # URL rewriting in the file. Need a relative path or none, so now
|
---|
| 617 | # we change working directory first.
|
---|
[13332] | 618 | my $changed_dir = 0;
|
---|
[13324] | 619 | my $current_dir = cwd();
|
---|
[13332] | 620 | my $wget_cmd = "";
|
---|
| 621 | if ($ENV{'GSDLOS'} ne "windows") {
|
---|
| 622 | $changed_dir = 1;
|
---|
| 623 |
|
---|
| 624 | chdir "$output_dir";
|
---|
| 625 | $wget_cmd = "wget -nv --timestamping -k -p \"$doc_url\"";
|
---|
| 626 | } else {
|
---|
[13333] | 627 | $wget_cmd = "wget -nv -P \"$output_dir\" --timestamping -k -p \"$doc_url\"";
|
---|
[13332] | 628 | }
|
---|
[22900] | 629 | ##print STDERR "**** wget = $wget_cmd\n";
|
---|
[13324] | 630 |
|
---|
| 631 |
|
---|
| 632 | $status = system($wget_cmd);
|
---|
[13332] | 633 | if ($changed_dir) {
|
---|
| 634 | chdir $current_dir;
|
---|
| 635 | }
|
---|
[13324] | 636 | if ($status==0) {
|
---|
[22900] | 637 | $self->{'lom_srcdoc'} = $srcdoc_path;
|
---|
[13324] | 638 | }
|
---|
| 639 | else {
|
---|
| 640 | $self->{'lom_srcdoc'} = undef;
|
---|
| 641 | print $outhandle "Error: failed to execute $wget_cmd\n";
|
---|
| 642 | }
|
---|
| 643 | }
|
---|
| 644 | else {
|
---|
| 645 | # not time-stamping and file already exists
|
---|
| 646 | $status=0;
|
---|
[22900] | 647 | $self->{'lom_srcdoc'} = $srcdoc_path;
|
---|
[13324] | 648 | }
|
---|
| 649 |
|
---|
| 650 | return $status==0;
|
---|
| 651 |
|
---|
| 652 | }
|
---|
| 653 |
|
---|
| 654 |
|
---|
| 655 | sub check_for_identifier
|
---|
| 656 | {
|
---|
| 657 | my $self = shift(@_);
|
---|
| 658 | my ($full_mname,$md_content) = @_;
|
---|
| 659 |
|
---|
| 660 | my $success = 0;
|
---|
| 661 |
|
---|
| 662 | my $download_re = $self->{'download_srcdocs'};
|
---|
| 663 | if (($download_re ne "") && $md_content =~ m/$download_re/) {
|
---|
| 664 |
|
---|
| 665 | if ($self->found_specific_identifier("general^identifier^entry",$full_mname,$md_content)) {
|
---|
| 666 | $success = $self->download_srcdoc($md_content);
|
---|
| 667 | }
|
---|
| 668 |
|
---|
| 669 | if (!$success) {
|
---|
| 670 | if ($self->found_specific_identifier("technical^location",$full_mname,$md_content)) {
|
---|
| 671 | $success = $self->download_srcdoc($md_content);
|
---|
| 672 | }
|
---|
| 673 | }
|
---|
| 674 | }
|
---|
| 675 |
|
---|
| 676 | return $success;
|
---|
| 677 | }
|
---|
| 678 |
|
---|
| 679 |
|
---|
| 680 | sub Text {
|
---|
| 681 | if ($_ !~ m/^\s*$/) {
|
---|
| 682 | #
|
---|
| 683 | # Work out indentations and line wraps for raw XML
|
---|
| 684 | #
|
---|
| 685 | my $xml_depth = scalar(@{$self->{'metaname_stack'}})+1;
|
---|
| 686 | my $indent = " " x $xml_depth;
|
---|
| 687 |
|
---|
| 688 | my $formatted_text = "\n".$_;
|
---|
| 689 |
|
---|
| 690 | # break into lines < 80 chars on space
|
---|
| 691 | $formatted_text =~ s/(.{50,80})\s+/$1\n/mg;
|
---|
| 692 | $formatted_text =~ s/^/$indent/mg;
|
---|
| 693 | ## $formatted_text =~ s/\s+$//s;
|
---|
| 694 |
|
---|
| 695 | $self->{'raw_text'} .= $formatted_text;
|
---|
| 696 | }
|
---|
| 697 |
|
---|
| 698 | my $metaname_stack = $self->{'metaname_stack'};
|
---|
| 699 | if (($_ !~ /^\s*$/) && (scalar(@$metaname_stack)>0)) {
|
---|
| 700 |
|
---|
| 701 | my ($without_dt_stack,$innermost_element,$md_content)
|
---|
| 702 | = $self->process_datatype_info($metaname_stack,$_);
|
---|
| 703 |
|
---|
| 704 | $self->pretty_print_table_tr($without_dt_stack);
|
---|
| 705 |
|
---|
| 706 | my $full_mname = join('^',@{$without_dt_stack});
|
---|
| 707 | $self->set_filere_metadata(lc($full_mname),$md_content);
|
---|
| 708 |
|
---|
| 709 | $self->check_for_language($innermost_element,$md_content);
|
---|
| 710 | $self->check_for_identifier($full_mname,$md_content); # source doc
|
---|
| 711 |
|
---|
| 712 | $self->reset_datatype_info();
|
---|
| 713 | }
|
---|
| 714 | }
|
---|
| 715 |
|
---|
| 716 | # This Char function overrides the one in XML::Parser::Stream to overcome a
|
---|
| 717 | # problem where $expat->{Text} is treated as the return value, slowing
|
---|
| 718 | # things down significantly in some cases.
|
---|
| 719 | sub Char {
|
---|
| 720 | $_[0]->{'Text'} .= $_[1];
|
---|
| 721 | return undef;
|
---|
| 722 | }
|
---|
| 723 |
|
---|
| 724 | 1;
|
---|