| 1 | ###########################################################################
|
|---|
| 2 | #
|
|---|
| 3 | # ReadXMLFile.pm -- base class for XML plugins
|
|---|
| 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) 2001 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 ReadXMLFile;
|
|---|
| 27 |
|
|---|
| 28 | use BasePlugin;
|
|---|
| 29 | use doc;
|
|---|
| 30 | use strict;
|
|---|
| 31 | no strict 'refs'; # allow filehandles to be variables and viceversa
|
|---|
| 32 |
|
|---|
| 33 | sub BEGIN {
|
|---|
| 34 | @ReadXMLFile::ISA = ('BasePlugin');
|
|---|
| 35 | unshift (@INC, "$ENV{'GSDLHOME'}/perllib/cpan");
|
|---|
| 36 | }
|
|---|
| 37 |
|
|---|
| 38 | use XMLParser;
|
|---|
| 39 |
|
|---|
| 40 | my $arguments =
|
|---|
| 41 | [ { 'name' => "process_exp",
|
|---|
| 42 | 'desc' => "{BasePlugin.process_exp}",
|
|---|
| 43 | 'type' => "regexp",
|
|---|
| 44 | 'deft' => &get_default_process_exp(),
|
|---|
| 45 | 'reqd' => "no" },
|
|---|
| 46 | { 'name' => "xslt",
|
|---|
| 47 | 'desc' => "{ReadXMLFile.xslt}",
|
|---|
| 48 | 'type' => "string",
|
|---|
| 49 | 'deft' => "",
|
|---|
| 50 | 'reqd' => "no" } ];
|
|---|
| 51 |
|
|---|
| 52 | my $options = { 'name' => "ReadXMLFile",
|
|---|
| 53 | 'desc' => "{ReadXMLFile.desc}",
|
|---|
| 54 | 'abstract' => "yes",
|
|---|
| 55 | 'inherits' => "yes",
|
|---|
| 56 | 'args' => $arguments };
|
|---|
| 57 |
|
|---|
| 58 | sub new {
|
|---|
| 59 | my ($class) = shift (@_);
|
|---|
| 60 | my ($pluginlist,$inputargs,$hashArgOptLists) = @_;
|
|---|
| 61 | push(@$pluginlist, $class);
|
|---|
| 62 |
|
|---|
| 63 | push(@{$hashArgOptLists->{"ArgList"}},@{$arguments});
|
|---|
| 64 | push(@{$hashArgOptLists->{"OptList"}},$options);
|
|---|
| 65 |
|
|---|
| 66 | my $self = new BasePlugin($pluginlist, $inputargs, $hashArgOptLists);
|
|---|
| 67 |
|
|---|
| 68 | if ($self->{'info_only'}) {
|
|---|
| 69 | # don't worry about creating the XML parser as all we want is the
|
|---|
| 70 | # list of plugin options
|
|---|
| 71 | return bless $self, $class;
|
|---|
| 72 | }
|
|---|
| 73 |
|
|---|
| 74 | my $parser = new XML::Parser('Style' => 'Stream',
|
|---|
| 75 | 'Pkg' => 'ReadXMLFile',
|
|---|
| 76 | 'PluginObj' => $self,
|
|---|
| 77 | 'Handlers' => {'Char' => \&Char,
|
|---|
| 78 | 'XMLDecl' => \&XMLDecl,
|
|---|
| 79 | 'Entity' => \&Entity,
|
|---|
| 80 | 'Doctype' => \&Doctype,
|
|---|
| 81 | 'Default' => \&Default
|
|---|
| 82 | });
|
|---|
| 83 |
|
|---|
| 84 | $self->{'parser'} = $parser;
|
|---|
| 85 |
|
|---|
| 86 | return bless $self, $class;
|
|---|
| 87 | }
|
|---|
| 88 |
|
|---|
| 89 | # the inheriting class must implement this method to tell whether to parse this doc type
|
|---|
| 90 | sub get_doctype {
|
|---|
| 91 | my $self = shift(@_);
|
|---|
| 92 | die "$self The inheriting class must implement get_doctype method";
|
|---|
| 93 | }
|
|---|
| 94 |
|
|---|
| 95 |
|
|---|
| 96 | sub apply_xslt
|
|---|
| 97 | {
|
|---|
| 98 | my $self = shift @_;
|
|---|
| 99 | my ($xslt,$filename) = @_;
|
|---|
| 100 |
|
|---|
| 101 | my $outhandle = $self->{'outhandle'};
|
|---|
| 102 |
|
|---|
| 103 | my $xslt_filename = $xslt;
|
|---|
| 104 |
|
|---|
| 105 | if (! -e $xslt_filename) {
|
|---|
| 106 | # Look in main site directory
|
|---|
| 107 | my $gsdlhome = $ENV{'GSDLHOME'};
|
|---|
| 108 | $xslt_filename = &util::filename_cat($gsdlhome,$xslt);
|
|---|
| 109 | }
|
|---|
| 110 |
|
|---|
| 111 | if (! -e $xslt_filename) {
|
|---|
| 112 | # Look in collection directory
|
|---|
| 113 | my $coldir = $ENV{'GSDLCOLLECTDIR'};
|
|---|
| 114 | $xslt_filename = &util::filename_cat($coldir,$xslt);
|
|---|
| 115 | }
|
|---|
| 116 |
|
|---|
| 117 | if (! -e $xslt_filename) {
|
|---|
| 118 | print $outhandle "Warning: Unable to find XSLT $xslt\n";
|
|---|
| 119 | if (open(XMLIN,"<$filename")) {
|
|---|
| 120 |
|
|---|
| 121 | my $untransformed_xml = "";
|
|---|
| 122 | while (defined (my $line = <XMLIN>)) {
|
|---|
| 123 |
|
|---|
| 124 | $untransformed_xml .= $line;
|
|---|
| 125 | }
|
|---|
| 126 | close(XMLIN);
|
|---|
| 127 |
|
|---|
| 128 | return $untransformed_xml;
|
|---|
| 129 | }
|
|---|
| 130 | else {
|
|---|
| 131 | print $outhandle "Error: Unable to open file $filename\n";
|
|---|
| 132 | print $outhandle " $!\n";
|
|---|
| 133 | return "";
|
|---|
| 134 | }
|
|---|
| 135 |
|
|---|
| 136 | }
|
|---|
| 137 |
|
|---|
| 138 | my $bin_java = &util::filename_cat($ENV{'GSDLHOME'},"bin","java");
|
|---|
| 139 | my $jar_filename = &util::filename_cat($bin_java,"xalan.jar");
|
|---|
| 140 | my $xslt_base_cmd = "java -jar $jar_filename";
|
|---|
| 141 | my $xslt_cmd = "$xslt_base_cmd -IN \"$filename\" -XSL \"$xslt_filename\"";
|
|---|
| 142 |
|
|---|
| 143 | my $transformed_xml = "";
|
|---|
| 144 |
|
|---|
| 145 | if (open(XSLT_IN,"$xslt_cmd |")) {
|
|---|
| 146 | while (defined (my $line = <XSLT_IN>)) {
|
|---|
| 147 |
|
|---|
| 148 | $transformed_xml .= $line;
|
|---|
| 149 | }
|
|---|
| 150 | close(XSLT_IN);
|
|---|
| 151 | }
|
|---|
| 152 | else {
|
|---|
| 153 | print $outhandle "Error: Unable to run command $xslt_cmd\n";
|
|---|
| 154 | print $outhandle " $!\n";
|
|---|
| 155 | }
|
|---|
| 156 |
|
|---|
| 157 | return $transformed_xml;
|
|---|
| 158 |
|
|---|
| 159 | }
|
|---|
| 160 |
|
|---|
| 161 | sub can_process_this_file {
|
|---|
| 162 | my $self = shift(@_);
|
|---|
| 163 | my ($filename) = @_;
|
|---|
| 164 |
|
|---|
| 165 | if (-f $filename && $self->SUPER::can_process_this_file($filename) && $self->check_doctype($filename)) {
|
|---|
| 166 | return 1; # its a file for us
|
|---|
| 167 | }
|
|---|
| 168 | return 0;
|
|---|
| 169 | }
|
|---|
| 170 |
|
|---|
| 171 | sub check_doctype {
|
|---|
| 172 | my $self = shift (@_);
|
|---|
| 173 |
|
|---|
| 174 | my ($filename) = @_;
|
|---|
| 175 |
|
|---|
| 176 | if (open(XMLIN,"<$filename")) {
|
|---|
| 177 | my $doctype = $self->get_doctype();
|
|---|
| 178 | ## check whether the doctype has the same name as the root element tag
|
|---|
| 179 | while (defined (my $line = <XMLIN>)) {
|
|---|
| 180 | ## find the root element
|
|---|
| 181 | if ($line =~ /<([\w\d:]+)[\s>]/){
|
|---|
| 182 | my $root = $1;
|
|---|
| 183 | if ($root !~ $doctype){
|
|---|
| 184 | close(XMLIN);
|
|---|
| 185 | return 0;
|
|---|
| 186 | }
|
|---|
| 187 | else {
|
|---|
| 188 | close(XMLIN);
|
|---|
| 189 | return 1;
|
|---|
| 190 | }
|
|---|
| 191 | }
|
|---|
| 192 | }
|
|---|
| 193 | close(XMLIN);
|
|---|
| 194 | }
|
|---|
| 195 |
|
|---|
| 196 | return undef; # haven't found a valid line
|
|---|
| 197 |
|
|---|
| 198 | }
|
|---|
| 199 |
|
|---|
| 200 | sub read {
|
|---|
| 201 | my $self = shift (@_);
|
|---|
| 202 |
|
|---|
| 203 | my ($pluginfo, $base_dir, $file, $block_hash, $metadata, $processor, $maxdocs, $total_count, $gli) = @_;
|
|---|
| 204 |
|
|---|
| 205 | # can we process this file??
|
|---|
| 206 | my ($filename_full_path, $filename_no_path) = &util::get_full_filenames($base_dir, $file);
|
|---|
| 207 | return undef unless $self->can_process_this_file($filename_full_path);
|
|---|
| 208 |
|
|---|
| 209 | $file =~ s/^[\/\\]+//; # $file often begins with / so we'll tidy it up
|
|---|
| 210 | $self->{'base_dir'} = $base_dir;
|
|---|
| 211 | $self->{'file'} = $file;
|
|---|
| 212 | $self->{'filename'} = $filename_full_path;
|
|---|
| 213 | $self->{'filename_no_path'} = $filename_no_path;
|
|---|
| 214 | $self->{'processor'} = $processor;
|
|---|
| 215 |
|
|---|
| 216 | # this contains metadata passed in from running metadata_read with other plugins (eg from MetadataXMLPlugin)
|
|---|
| 217 | # we are also using it to store up any metadata found during parsing the XML, so that it can be added to the doc obj.
|
|---|
| 218 | $self->{'metadata'} = $metadata;
|
|---|
| 219 |
|
|---|
| 220 | if ($self->parse_file($filename_full_path)) {
|
|---|
| 221 | return 1; # processed the file
|
|---|
| 222 | }
|
|---|
| 223 | return -1;
|
|---|
| 224 | }
|
|---|
| 225 |
|
|---|
| 226 |
|
|---|
| 227 | sub parse_file {
|
|---|
| 228 | my $self = shift (@_);
|
|---|
| 229 | my ($filename_full_path, $file, $gli) = @_;
|
|---|
| 230 | eval {
|
|---|
| 231 | my $xslt = $self->{'xslt'};
|
|---|
| 232 | if (defined $xslt && ($xslt ne "")) {
|
|---|
| 233 | # perform xslt
|
|---|
| 234 | my $transformed_xml = $self->apply_xslt($xslt,$filename_full_path);
|
|---|
| 235 |
|
|---|
| 236 | # feed transformed file (now in memory as string) into XML parser
|
|---|
| 237 | $self->{'parser'}->parse($transformed_xml);
|
|---|
| 238 | }
|
|---|
| 239 | else {
|
|---|
| 240 | $self->{'parser'}->parsefile($filename_full_path);
|
|---|
| 241 | }
|
|---|
| 242 | };
|
|---|
| 243 |
|
|---|
| 244 | if ($@) {
|
|---|
| 245 |
|
|---|
| 246 | # parsefile may either croak somewhere in XML::Parser (e.g. because
|
|---|
| 247 | # the document is not well formed) or die somewhere in ReadXMLFile or a
|
|---|
| 248 | # derived plugin (e.g. because we're attempting to process a
|
|---|
| 249 | # document whose DOCTYPE is not meant for this plugin). For the
|
|---|
| 250 | # first case we'll print a warning and continue, for the second
|
|---|
| 251 | # we'll just continue quietly
|
|---|
| 252 |
|
|---|
| 253 | print STDERR "**** Error is: $@\n";
|
|---|
| 254 |
|
|---|
| 255 | my ($msg) = $@ =~ /Carp::croak\(\'(.*?)\'\)/;
|
|---|
| 256 | if (defined $msg) {
|
|---|
| 257 | my $outhandle = $self->{'outhandle'};
|
|---|
| 258 | my $plugin_name = ref ($self);
|
|---|
| 259 | print $outhandle "$plugin_name failed to process $file ($msg)\n";
|
|---|
| 260 | }
|
|---|
| 261 |
|
|---|
| 262 | # reset ourself for the next document
|
|---|
| 263 | $self->{'section_level'}=0;
|
|---|
| 264 | print STDERR "<ProcessingError n='$file'>\n" if ($gli);
|
|---|
| 265 | return -1; # error during processing
|
|---|
| 266 | }
|
|---|
| 267 | return 1; # parsing was successful
|
|---|
| 268 | }
|
|---|
| 269 |
|
|---|
| 270 | sub get_default_process_exp {
|
|---|
| 271 | my $self = shift (@_);
|
|---|
| 272 |
|
|---|
| 273 | return q^(?i)\.xml$^;
|
|---|
| 274 | }
|
|---|
| 275 |
|
|---|
| 276 | sub StartDocument {$_[0]->{'PluginObj'}->xml_start_document(@_);}
|
|---|
| 277 | sub XMLDecl {$_[0]->{'PluginObj'}->xml_xmldecl(@_);}
|
|---|
| 278 | sub Entity {$_[0]->{'PluginObj'}->xml_entity(@_);}
|
|---|
| 279 | sub Doctype {$_[0]->{'PluginObj'}->xml_doctype(@_);}
|
|---|
| 280 | sub StartTag {$_[0]->{'PluginObj'}->xml_start_tag(@_);}
|
|---|
| 281 | sub EndTag {$_[0]->{'PluginObj'}->xml_end_tag(@_);}
|
|---|
| 282 | sub Text {$_[0]->{'PluginObj'}->xml_text(@_);}
|
|---|
| 283 | sub PI {$_[0]->{'PluginObj'}->xml_pi(@_);}
|
|---|
| 284 | sub EndDocument {$_[0]->{'PluginObj'}->xml_end_document(@_);}
|
|---|
| 285 | sub Default {$_[0]->{'PluginObj'}->xml_default(@_);}
|
|---|
| 286 |
|
|---|
| 287 | # This Char function overrides the one in XML::Parser::Stream to overcome a
|
|---|
| 288 | # problem where $expat->{Text} is treated as the return value, slowing
|
|---|
| 289 | # things down significantly in some cases.
|
|---|
| 290 | sub Char {
|
|---|
| 291 | use bytes; # Necessary to prevent encoding issues with XML::Parser 2.31+
|
|---|
| 292 | $_[0]->{'Text'} .= $_[1];
|
|---|
| 293 | return undef;
|
|---|
| 294 | }
|
|---|
| 295 |
|
|---|
| 296 |
|
|---|
| 297 | # Called at the beginning of the XML document.
|
|---|
| 298 | sub xml_start_document {
|
|---|
| 299 | my $self = shift(@_);
|
|---|
| 300 | my ($expat) = @_;
|
|---|
| 301 |
|
|---|
| 302 | $self->open_document();
|
|---|
| 303 | }
|
|---|
| 304 |
|
|---|
| 305 | # Called for XML declarations
|
|---|
| 306 | sub xml_xmldecl {
|
|---|
| 307 | my $self = shift(@_);
|
|---|
| 308 | my ($expat, $version, $encoding, $standalone) = @_;
|
|---|
| 309 | }
|
|---|
| 310 |
|
|---|
| 311 | # Called for XML entities
|
|---|
| 312 | sub xml_entity {
|
|---|
| 313 | my $self = shift(@_);
|
|---|
| 314 | my ($expat, $name, $val, $sysid, $pubid, $ndata) = @_;
|
|---|
| 315 | }
|
|---|
| 316 |
|
|---|
| 317 | # Called for DOCTYPE declarations - use die to bail out if this doctype
|
|---|
| 318 | # is not meant for this plugin
|
|---|
| 319 | sub xml_doctype {
|
|---|
| 320 | my $self = shift(@_);
|
|---|
| 321 |
|
|---|
| 322 | my ($expat, $name, $sysid, $pubid, $internal) = @_;
|
|---|
| 323 | die "ReadXMLFile Cannot process XML document with DOCTYPE of $name";
|
|---|
| 324 | }
|
|---|
| 325 |
|
|---|
| 326 |
|
|---|
| 327 | # Called for every start tag. The $_ variable will contain a copy of the
|
|---|
| 328 | # tag and the %_ variable will contain the element's attributes.
|
|---|
| 329 | sub xml_start_tag {
|
|---|
| 330 | my $self = shift(@_);
|
|---|
| 331 | my ($expat, $element) = @_;
|
|---|
| 332 | }
|
|---|
| 333 |
|
|---|
| 334 | # Called for every end tag. The $_ variable will contain a copy of the tag.
|
|---|
| 335 | sub xml_end_tag {
|
|---|
| 336 | my $self = shift(@_);
|
|---|
| 337 | my ($expat, $element) = @_;
|
|---|
| 338 | }
|
|---|
| 339 |
|
|---|
| 340 | # Called just before start or end tags with accumulated non-markup text in
|
|---|
| 341 | # the $_ variable.
|
|---|
| 342 | sub xml_text {
|
|---|
| 343 | my $self = shift(@_);
|
|---|
| 344 | my ($expat) = @_;
|
|---|
| 345 | }
|
|---|
| 346 |
|
|---|
| 347 | # Called for processing instructions. The $_ variable will contain a copy
|
|---|
| 348 | # of the pi.
|
|---|
| 349 | sub xml_pi {
|
|---|
| 350 | my $self = shift(@_);
|
|---|
| 351 | my ($expat, $target, $data) = @_;
|
|---|
| 352 | }
|
|---|
| 353 |
|
|---|
| 354 | # Called at the end of the XML document.
|
|---|
| 355 | sub xml_end_document {
|
|---|
| 356 | my $self = shift(@_);
|
|---|
| 357 | my ($expat) = @_;
|
|---|
| 358 |
|
|---|
| 359 | $self->close_document();
|
|---|
| 360 | }
|
|---|
| 361 |
|
|---|
| 362 | # Called for any characters not handled by the above functions.
|
|---|
| 363 | sub xml_default {
|
|---|
| 364 | my $self = shift(@_);
|
|---|
| 365 | my ($expat, $text) = @_;
|
|---|
| 366 | }
|
|---|
| 367 |
|
|---|
| 368 | sub open_document {
|
|---|
| 369 | my $self = shift(@_);
|
|---|
| 370 |
|
|---|
| 371 | my $metadata = $self->{'metadata'};
|
|---|
| 372 | my $filename_full_path = $self->{'filename'};
|
|---|
| 373 |
|
|---|
| 374 | # create a new document
|
|---|
| 375 | my $doc_obj = $self->{'doc_obj'} = new doc ($filename_full_path, "indexed_doc", $self->{'file_rename_method'});
|
|---|
| 376 |
|
|---|
| 377 | $doc_obj->add_utf8_metadata($doc_obj->get_top_section(), "Plugin", "$self->{'plugin_type'}");
|
|---|
| 378 |
|
|---|
| 379 | my $filename_no_path = $self->{'filename_no_path'};
|
|---|
| 380 | my $plugin_filename_encoding = $self->{'filename_encoding'};
|
|---|
| 381 | my $filename_encoding = $self->deduce_filename_encoding($filename_no_path,$metadata,$plugin_filename_encoding);
|
|---|
| 382 |
|
|---|
| 383 | $self->set_Source_metadata($doc_obj, $filename_full_path, $filename_encoding);
|
|---|
| 384 |
|
|---|
| 385 | # do we want other auto metadata here (see BasePlugin.read_into_doc_obj)
|
|---|
| 386 | }
|
|---|
| 387 |
|
|---|
| 388 | sub close_document {
|
|---|
| 389 | my $self = shift(@_);
|
|---|
| 390 | my $doc_obj = $self->{'doc_obj'};
|
|---|
| 391 |
|
|---|
| 392 | # do we want other auto stuff here, see BasePlugin.read_into_doc_obj
|
|---|
| 393 |
|
|---|
| 394 | # include any metadata passed in from previous plugins
|
|---|
| 395 | # note that this metadata is associated with the top level section
|
|---|
| 396 | $self->extra_metadata ($doc_obj,
|
|---|
| 397 | $doc_obj->get_top_section(),
|
|---|
| 398 | $self->{'metadata'});
|
|---|
| 399 |
|
|---|
| 400 | # do any automatic metadata extraction
|
|---|
| 401 | $self->auto_extract_metadata ($doc_obj);
|
|---|
| 402 |
|
|---|
| 403 | # add an OID
|
|---|
| 404 | $self->add_OID($doc_obj);
|
|---|
| 405 |
|
|---|
| 406 | $doc_obj->add_utf8_metadata($doc_obj->get_top_section(), "Plugin", "$self->{'plugin_type'}");
|
|---|
| 407 | $doc_obj->add_metadata($doc_obj->get_top_section(), "FileFormat", "XML");
|
|---|
| 408 |
|
|---|
| 409 | # process the document
|
|---|
| 410 | $self->{'processor'}->process($doc_obj);
|
|---|
| 411 |
|
|---|
| 412 | $self->{'num_processed'} ++;
|
|---|
| 413 | undef $self->{'doc_obj'};
|
|---|
| 414 | undef $doc_obj; # is this the same as above??
|
|---|
| 415 | }
|
|---|
| 416 |
|
|---|
| 417 | 1;
|
|---|
| 418 |
|
|---|
| 419 |
|
|---|
| 420 |
|
|---|
| 421 |
|
|---|
