| 1 | ###########################################################################
|
|---|
| 2 | #
|
|---|
| 3 | # OpenDocumentPlug.pm -- The Open Document plugin
|
|---|
| 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 | # Processes OASIS Open Document format.
|
|---|
| 27 | # Word processing document: .odt, template: .ott
|
|---|
| 28 | # Spreadsheet document: .ods, template: .ots
|
|---|
| 29 | # Presentation document: .odp, template: .otp
|
|---|
| 30 | # Graphics document: .odg, template: .otg
|
|---|
| 31 | # Formulas document: .odf, template: .otf (not supported)
|
|---|
| 32 |
|
|---|
| 33 | #This basically extracts any text out of the document, but not much else.
|
|---|
| 34 |
|
|---|
| 35 | package OpenDocumentPlug;
|
|---|
| 36 |
|
|---|
| 37 | use strict;
|
|---|
| 38 | no strict 'refs'; # allow filehandles to be variables and viceversa
|
|---|
| 39 |
|
|---|
| 40 | use XMLPlug;
|
|---|
| 41 | use XML::XPath;
|
|---|
| 42 | use XML::XPath::XMLParser;
|
|---|
| 43 | use Cwd;
|
|---|
| 44 | use util;
|
|---|
| 45 | use ghtml;
|
|---|
| 46 |
|
|---|
| 47 | sub BEGIN {
|
|---|
| 48 | @OpenDocumentPlug::ISA = ('XMLPlug');
|
|---|
| 49 | }
|
|---|
| 50 |
|
|---|
| 51 |
|
|---|
| 52 | #our @filesAssoc = ();
|
|---|
| 53 | our @filesProcess = ( "content.xml" , "meta.xml" );
|
|---|
| 54 | #XML plug has this so we need it too
|
|---|
| 55 | our ($self);
|
|---|
| 56 |
|
|---|
| 57 | my $arguments = [
|
|---|
| 58 | { 'name' => "process_exp",
|
|---|
| 59 | 'desc' => "{BasPlug.process_exp}",
|
|---|
| 60 | 'type' => "regexp",
|
|---|
| 61 | 'deft' => &get_default_process_exp() }
|
|---|
| 62 | ];
|
|---|
| 63 |
|
|---|
| 64 | my $options = { 'name' => "OpenDocumentPlug",
|
|---|
| 65 | 'desc' => "{OpenDocumentPlug.desc}",
|
|---|
| 66 | 'abstract' => "no",
|
|---|
| 67 | 'inherits' => "yes",
|
|---|
| 68 | 'args' => $arguments};
|
|---|
| 69 |
|
|---|
| 70 | sub get_default_process_exp { return q^(?i)\.o(?:d|t)(?:t|s|p|g)$^; }
|
|---|
| 71 |
|
|---|
| 72 | sub new {
|
|---|
| 73 | my ($class) = shift (@_);
|
|---|
| 74 | my ($pluginlist,$inputargs,$hashArgOptLists) = @_;
|
|---|
| 75 | push(@$pluginlist, $class);
|
|---|
| 76 |
|
|---|
| 77 | if(defined $arguments){ push(@{$hashArgOptLists->{"ArgList"}},@{$arguments});}
|
|---|
| 78 | if(defined $options) { push(@{$hashArgOptLists->{"OptList"}},$options)};
|
|---|
| 79 |
|
|---|
| 80 | my $self = new XMLPlug($pluginlist, $inputargs, $hashArgOptLists);
|
|---|
| 81 |
|
|---|
| 82 | $self->{'section'} = "";
|
|---|
| 83 | $self->{'office:meta'} = "";
|
|---|
| 84 |
|
|---|
| 85 | return bless $self, $class;
|
|---|
| 86 | }
|
|---|
| 87 |
|
|---|
| 88 | sub get_doctype {
|
|---|
| 89 | my $self = shift(@_);
|
|---|
| 90 |
|
|---|
| 91 | return "manifest:manifest";
|
|---|
| 92 | }
|
|---|
| 93 |
|
|---|
| 94 | sub xml_doctype {
|
|---|
| 95 | my $self = shift(@_);
|
|---|
| 96 | my ($expat, $name, $sysid, $pubid, $internal) = @_;
|
|---|
| 97 | die "The only valid doctype is manifest, $name is not valid" if ($name ne "manifest:manifest");
|
|---|
| 98 | }
|
|---|
| 99 |
|
|---|
| 100 | # Called for every start tag. The $_ variable will contain a copy of the
|
|---|
| 101 | # tag and the %_ variable will contain the element's attributes.
|
|---|
| 102 | sub xml_start_tag {
|
|---|
| 103 | my $self = shift(@_);
|
|---|
| 104 | my ($expat, $element) = @_;
|
|---|
| 105 | my %atts = %_;
|
|---|
| 106 | $self->{'office:meta'} = $element if $self->{'office:meta'} eq "Start";
|
|---|
| 107 | if($element eq 'office:text') {
|
|---|
| 108 | $self->{'collectedText'} = "";
|
|---|
| 109 | }elsif($element eq 'office:meta') {
|
|---|
| 110 | $self->{'collectedText'} = "";
|
|---|
| 111 | $self->{'office:meta'} = "Start";
|
|---|
| 112 | }elsif($element eq 'meta:document-statistic'){
|
|---|
| 113 | foreach my $att (keys %atts) {
|
|---|
| 114 | $self->{'doc_obj'}->add_utf8_metadata("",$att,$atts{$att});
|
|---|
| 115 | }
|
|---|
| 116 |
|
|---|
| 117 | }
|
|---|
| 118 | }
|
|---|
| 119 |
|
|---|
| 120 | sub xml_end_tag {
|
|---|
| 121 | my $self = shift(@_);
|
|---|
| 122 | my ($expat, $element) = @_;
|
|---|
| 123 |
|
|---|
| 124 | if($element eq 'office:text') {
|
|---|
| 125 | $self->{'doc_obj'}->add_utf8_text("",$self->{'collectedText'});
|
|---|
| 126 | $self->{'collectedText'} = "";
|
|---|
| 127 | }elsif($element eq $self->{'office:meta'}) {
|
|---|
| 128 | if( $self->{'collectedText'} ne "") {
|
|---|
| 129 | $self->{'doc_obj'}->add_utf8_metadata("",$self->{'office:meta'},$self->{'collectedText'});
|
|---|
| 130 | $self->{'doc_obj'}->add_utf8_metadata("","Title",$self->{'collectedText'}) if $self->{'office:meta'} =~ m/:title$/;
|
|---|
| 131 | $self->{'doc_obj'}->add_utf8_metadata("","Language",$self->{'collectedText'}) if $self->{'office:meta'} =~ m/:language$/;
|
|---|
| 132 | $self->{'doc_obj'}->add_utf8_metadata("","GENERATOR",$self->{'collectedText'}) if $self->{'office:meta'} =~ m/:generator$/;
|
|---|
| 133 |
|
|---|
| 134 | }
|
|---|
| 135 | $self->{'collectedText'} = "";
|
|---|
| 136 | $self->{'office:meta'} = "Start";
|
|---|
| 137 | }elsif($element eq 'office:meta'){
|
|---|
| 138 | $self->{'office:meta'} = "";
|
|---|
| 139 | }elsif($element eq 'office:body'){
|
|---|
| 140 | #some documents have text in other places that should probably be indexed if we can't find any doc text
|
|---|
| 141 |
|
|---|
| 142 | if( defined $self->{'collectedText'} && $self->{'collectedText'} ne "" && $self->{'doc_obj'}->get_text("") eq "") {
|
|---|
| 143 | $self->{'doc_obj'}->add_utf8_text("",$self->{'collectedText'});
|
|---|
| 144 | }
|
|---|
| 145 | }
|
|---|
| 146 | }
|
|---|
| 147 |
|
|---|
| 148 | sub xml_text {
|
|---|
| 149 | my $self = shift(@_);
|
|---|
| 150 | my ($expat) = @_;
|
|---|
| 151 | if($_ =~ m/\w/i) {
|
|---|
| 152 | $self->{'collectedText'} .= "<br/>" if $self->{'collectedText'} ne "";
|
|---|
| 153 | $self->{'collectedText'} .= "$_";
|
|---|
| 154 | }
|
|---|
| 155 | }
|
|---|
| 156 |
|
|---|
| 157 | #trap start and end document so we do not get our doc_obj closed too soon
|
|---|
| 158 | sub xml_start_document {}
|
|---|
| 159 | sub xml_end_document {}
|
|---|
| 160 |
|
|---|
| 161 | sub read {
|
|---|
| 162 | # $self must be global to work with XML callback routines.
|
|---|
| 163 | $self = shift (@_);
|
|---|
| 164 |
|
|---|
| 165 | my ($pluginfo, $base_dir, $file, $metadata, $processor, $maxdocs, $total_count, $gli) = @_;
|
|---|
| 166 |
|
|---|
| 167 | # check process and block exps, smart block, associate_ext etc
|
|---|
| 168 | my ($block_status,$filename) = $self->read_block(@_);
|
|---|
| 169 | return $block_status if ((!defined $block_status) || ($block_status==0));
|
|---|
| 170 |
|
|---|
| 171 | $file =~ s/^[\/\\]+//; # $file often begins with / so we'll tidy it up
|
|---|
| 172 | $self->{'file'} = $file;
|
|---|
| 173 | $self->{'filename'} = $filename;
|
|---|
| 174 | $self->{'processor'} = $processor;
|
|---|
| 175 | $self->{'metadata'} = $metadata;
|
|---|
| 176 |
|
|---|
| 177 | eval{
|
|---|
| 178 | my ($file_only) = $file =~ /([^\\\/]*)$/;
|
|---|
| 179 | my $tmpdir = &util::get_tmp_filename ();
|
|---|
| 180 | &util::mk_all_dir ($tmpdir);
|
|---|
| 181 |
|
|---|
| 182 | $self->open_document();
|
|---|
| 183 |
|
|---|
| 184 | # save current working directory
|
|---|
| 185 | my $cwd = getcwd();
|
|---|
| 186 | chdir ($tmpdir) || die "Unable to change to $tmpdir";
|
|---|
| 187 | &util::cp ($filename, $tmpdir);
|
|---|
| 188 |
|
|---|
| 189 | $self->unzip ("\"$file_only\"");
|
|---|
| 190 | foreach my $xmlFile (@OpenDocumentPlug::filesProcess) {
|
|---|
| 191 | if (-e $xmlFile) {
|
|---|
| 192 | $self->parse_file($xmlFile);
|
|---|
| 193 | }
|
|---|
| 194 | }
|
|---|
| 195 | $self->close_document($filename,$file_only);
|
|---|
| 196 |
|
|---|
| 197 | chdir ($cwd) || die "Unable to change back to $cwd";
|
|---|
| 198 | &util::rm_r ($tmpdir);
|
|---|
| 199 |
|
|---|
| 200 | };
|
|---|
| 201 |
|
|---|
| 202 | if ($@) {
|
|---|
| 203 |
|
|---|
| 204 | # parsefile may either croak somewhere in XML::Parser (e.g. because
|
|---|
| 205 | # the document is not well formed) or die somewhere in XMLPlug or a
|
|---|
| 206 | # derived plugin (e.g. because we're attempting to process a
|
|---|
| 207 | # document whose DOCTYPE is not meant for this plugin). For the
|
|---|
| 208 | # first case we'll print a warning and continue, for the second
|
|---|
| 209 | # we'll just continue quietly
|
|---|
| 210 |
|
|---|
| 211 | print STDERR "**** Error is: $@\n";
|
|---|
| 212 |
|
|---|
| 213 | my ($msg) = $@ =~ /Carp::croak\(\'(.*?)\'\)/;
|
|---|
| 214 | if (defined $msg) {
|
|---|
| 215 | my $outhandle = $self->{'outhandle'};
|
|---|
| 216 | my $plugin_name = ref ($self);
|
|---|
| 217 | print $outhandle "$plugin_name failed to process $file ($msg)\n";
|
|---|
| 218 | }
|
|---|
| 219 |
|
|---|
| 220 | # reset ourself for the next document
|
|---|
| 221 | print STDERR "<ProcessingError n='$file'>\n" if ($gli);
|
|---|
| 222 | return -1; # error during processing
|
|---|
| 223 | }
|
|---|
| 224 |
|
|---|
| 225 | return 1;
|
|---|
| 226 | }
|
|---|
| 227 |
|
|---|
| 228 | sub unzip {
|
|---|
| 229 | my $self = shift (@_);
|
|---|
| 230 | my ($file) = @_;
|
|---|
| 231 |
|
|---|
| 232 | system ("unzip $file");
|
|---|
| 233 | &util::rm ($file) if -e $file;
|
|---|
| 234 | }
|
|---|
| 235 |
|
|---|
| 236 | sub close_document() {
|
|---|
| 237 | my $self = shift(@_);
|
|---|
| 238 | my ($filename,$file_only) = @_;
|
|---|
| 239 |
|
|---|
| 240 | my $doc_obj = $self->{'doc_obj'};
|
|---|
| 241 |
|
|---|
| 242 | my $mimetype = $self->get_mimetype();
|
|---|
| 243 |
|
|---|
| 244 | $doc_obj->associate_file($filename, $file_only, $mimetype, "");
|
|---|
| 245 | $doc_obj->associate_file("Thumbnails/thumbnail.png", "thumbnail.png", "image/png", "");
|
|---|
| 246 | my $doc_ext = $filename;
|
|---|
| 247 | $doc_ext =~ s/.*\.od(.)/od$1/;
|
|---|
| 248 |
|
|---|
| 249 | # We use set instead of add here because we only want one value
|
|---|
| 250 | $doc_obj->set_utf8_metadata_element("", "FileFormat", "Open Document");
|
|---|
| 251 |
|
|---|
| 252 | #setup to doclink thingi
|
|---|
| 253 | my $doclink = "<a href=\"_httpprefix_/collect/[collection]/index/assoc/[archivedir]/$file_only\">";
|
|---|
| 254 | $doc_obj->add_utf8_metadata ("", "srclink", $doclink);
|
|---|
| 255 | $doc_obj->add_utf8_metadata ("", "srcicon", "<img border=\"0\" align=\"absmiddle\" src=\"_httpprefix_/collect/[collection]/index/assoc/[archivedir]/thumbnail.png\" alt=\"View the Open document\" title=\"View the Open document\">");
|
|---|
| 256 | $doc_obj->add_utf8_metadata ("", "/srclink", "</a>");
|
|---|
| 257 | $doc_obj->add_utf8_metadata ("", "Source", &ghtml::dmsafe($file_only));
|
|---|
| 258 | $doc_obj->set_utf8_metadata_element("", "FileSize", (-s $filename));
|
|---|
| 259 |
|
|---|
| 260 | # include any metadata passed in from previous plugins
|
|---|
| 261 | # note that this metadata is associated with the top level section
|
|---|
| 262 | $self->extra_metadata ($doc_obj,
|
|---|
| 263 | $doc_obj->get_top_section(),
|
|---|
| 264 | $self->{'metadata'});
|
|---|
| 265 |
|
|---|
| 266 | # add a Title if none has been found yet
|
|---|
| 267 | $self->title_fallback($doc_obj,"",$file_only);
|
|---|
| 268 |
|
|---|
| 269 | # add an OID
|
|---|
| 270 | $doc_obj->set_OID();
|
|---|
| 271 |
|
|---|
| 272 | $doc_obj->add_utf8_metadata("", "Plugin", "$self->{'plugin_type'}");
|
|---|
| 273 |
|
|---|
| 274 | # process the document
|
|---|
| 275 | $self->{'processor'}->process($doc_obj);
|
|---|
| 276 |
|
|---|
| 277 | $self->{'num_processed'} ++;
|
|---|
| 278 | return 1;
|
|---|
| 279 | }
|
|---|
| 280 |
|
|---|
| 281 | sub get_mimetype(){
|
|---|
| 282 | my $filename = "mimetype";
|
|---|
| 283 | if (!open (FILEIN,"<$filename")){
|
|---|
| 284 | print STDERR "Warning: unable to open the $filename\n";
|
|---|
| 285 | return "Unknown OpenDocument Format";
|
|---|
| 286 | }
|
|---|
| 287 | else {
|
|---|
| 288 | my $text = "";
|
|---|
| 289 | while (defined (my $line = <FILEIN>)) {
|
|---|
| 290 | $text .= $line;
|
|---|
| 291 | }
|
|---|
| 292 | return $text;
|
|---|
| 293 | }
|
|---|
| 294 | }
|
|---|
| 295 | 1;
|
|---|
| 296 |
|
|---|
| 297 |
|
|---|
| 298 |
|
|---|
| 299 |
|
|---|
