1 package OpenILS::Utils::ModsParser;
2 use strict; use warnings;
4 use OpenSRF::EX qw/:try/;
7 use Time::HiRes qw(time);
8 use OpenILS::Utils::Fieldmapper;
9 use OpenSRF::Utils::SettingsClient;
12 my $parser = XML::LibXML->new();
13 my $xslt = XML::LibXSLT->new();
16 # ----------------------------------------------------------------------------------------
17 # XPATH for extracting info from a MODS doc
18 my $isbn_xpath = "//mods:mods/mods:identifier[\@type='isbn']";
19 my $resource_xpath = "//mods:mods/mods:typeOfResource";
20 my $pub_xpath = "//mods:mods/mods:originInfo//mods:dateIssued[\@encoding='marc']|" .
21 "//mods:mods/mods:originInfo//mods:dateIssued[1]";
22 my $tcn_xpath = "//mods:mods/mods:recordInfo/mods:recordIdentifier";
23 my $publisher_xpath = "//mods:mods/mods:originInfo//mods:publisher[1]";
24 my $edition_xpath = "//mods:mods/mods:originInfo//mods:edition[1]";
25 my $abstract_xpath = "//mods:mods/mods:abstract";
27 my $related_xpath = "";
28 my $online_loc_xpath = "(//mods:location/mods:url|//mods:location/mods:url/\@displayLabel)";
34 "//mods:mods/mods:titleInfo[mods:title and (\@type='abreviated')]",
36 "//mods:mods/mods:titleInfo[mods:title and (\@type='translated')]",
38 "//mods:mods/mods:titleInfo[mods:title and (\@type='uniform')]",
40 "//mods:mods/mods:titleInfo[mods:title and not (\@type)]",
45 "//mods:mods/mods:name[\@type='corporate']/*[local-name()='namePart']".
46 "[../mods:role/mods:text[text()='creator']][1]",
48 "//mods:mods/mods:name[\@type='personal']/*[local-name()='namePart']".
49 "[../mods:role/mods:text[text()='creator']][1]",
51 "//mods:mods/mods:name[\@type='conference']/*[local-name()='namePart']".
52 "[../mods:role/mods:text[text()='creator']][1]",
54 "//mods:mods/mods:name[\@type='personal']/*[local-name()='namePart']",
60 "//mods:mods/mods:subject/*[local-name()!='geographicCode']/parent::mods:subject",
63 # "//mods:mods/*[local-name()='subject']/*[local-name()='geographic']",
65 # "//mods:mods/*[local-name()='subject']/*[local-name()='name']",
67 # "//mods:mods/*[local-name()='subject']/*[local-name()='temporal']",
69 # "//mods:mods/*[local-name()='subject']/*[local-name()='topic']",
71 #keyword => { keyword => "//mods:mods/*[not(local-name()='originInfo')]", },
74 series => "//mods:mods/mods:relatedItem[\@type='series']/mods:titleInfo"
77 # ----------------------------------------------------------------------------------------
81 sub new { return bless( {}, shift() ); }
85 my( $self, $mods, $xpath ) = @_;
88 my $root = $mods->documentElement;
89 $root->setNamespace( "http://www.loc.gov/mods/v3", "mods", 1 );
91 # grab the set of matching nodes
92 my @nodes = $root->findnodes( $xpath );
93 for my $value (@nodes) {
95 # grab all children of the node
96 my @children = $value->childNodes();
98 for my $child (@children) {
99 next unless( $child->nodeType != 3 );
101 if($child->childNodes) {
103 for my $c (@{$child->childNodes}){
104 push @a, $c->textContent;
106 push(@child_text, join(' ', @a));
109 push(@child_text, $child->textContent);
114 push(@string, \@child_text);
118 push(@string, $value->textContent );
125 sub _modsdoc_to_values {
126 my( $self, $mods ) = @_;
128 for my $class (keys %$xpathset) {
129 $data->{$class} = {};
130 for my $type(keys %{$xpathset->{$class}}) {
131 my @value = $self->get_field_value( $mods, $xpathset->{$class}->{$type} );
132 if( $class eq "subject" ) {
133 push( @{$data->{$class}->{$type}}, @value );
135 $data->{$class}->{$type} = $value[0];
143 sub modsdoc_to_values {
144 my( $self, $mods ) = @_;
148 my $class = "subject";
149 $data->{$class} = {};
150 for my $type(keys %{$xpathset->{$class}}) {
151 my @value = $self->get_field_value( $mods, $xpathset->{$class}->{$type} );
152 for my $arr (@value) {
153 push( @{$data->{$class}->{$type}}, $arr);
160 $data->{$class} = {};
161 for my $type(keys %{$xpathset->{$class}}) {
162 my @value = $self->get_field_value( $mods, $xpathset->{$class}->{$type} );
163 for my $arr (@value) {
165 $data->{$class}->{$type} = join(" ", @$arr);
167 $data->{$class}->{$type} = $arr;
174 my $class = "author";
175 $data->{$class} = {};
176 for my $type(keys %{$xpathset->{$class}}) {
177 my @value = $self->get_field_value( $mods, $xpathset->{$class}->{$type} );
178 $data->{$class}->{$type} = $value[0];
183 my $class = "series";
184 $data->{$class} = {};
185 for my $type(keys %{$xpathset->{$class}}) {
186 my @value = $self->get_field_value( $mods, $xpathset->{$class}->{$type} );
187 for my $arr (@value) {
189 push(@{$data->{$class}->{$type}}, join(" ", @$arr));
191 push( @{$data->{$class}->{$type}}, $arr );
204 # ---------------------------------------------------------------------------
205 # Grabs the data 'we want' from the MODS doc and returns it in hash form
206 # ---------------------------------------------------------------------------
207 sub mods_values_to_mods_slim {
208 my( $self, $modsperl ) = @_;
215 my $tmp = $modsperl->{title};
218 if(!$tmp) { $title = ""; }
220 ($title = $tmp->{proper}) ||
221 ($title = $tmp->{translated}) ||
222 ($title = $tmp->{abbreviated}) ||
223 ($title = $tmp->{uniform});
226 $tmp = $modsperl->{author};
227 if(!$tmp) { $author = ""; }
229 ($author = $tmp->{personal}) ||
230 ($author = $tmp->{other}) ||
231 ($author = $tmp->{corporate}) ||
232 ($author = $tmp->{conference});
235 $tmp = $modsperl->{subject};
236 if(!$tmp) { $subject = []; }
238 for my $key( keys %{$tmp}) {
239 push(@$subject, @{$tmp->{$key}}) if ($tmp->{$key});
243 $tmp = $modsperl->{'series'};
244 if(!$tmp) { $series = []; }
245 else { $series = $tmp->{'series'}; }
248 return { series => $series, title => $title,
249 author => $author, subject => $subject };
254 # ---------------------------------------------------------------------------
255 # Initializes a MARC -> Unified MODS batch process
256 # ---------------------------------------------------------------------------
258 sub start_mods_batch {
260 my( $self, $master_doc ) = @_;
264 my $xslt_doc = $parser->parse_file(
265 OpenSRF::Utils::SettingsClient->new->config_value(dirs => 'xsl') . "/MARC21slim2MODS3.xsl");
266 $mods_sheet = $xslt->parse_stylesheet( $xslt_doc );
270 my $xmldoc = $parser->parse_string($master_doc);
271 my $mods = $mods_sheet->transform($xmldoc);
273 # warn "-" x 100 . "\n";
274 # warn "MODS " . $mods->toString(1) . "\n";
275 # warn "-" x 100 . "\n";
277 $self->{master_doc} = $self->modsdoc_to_values( $mods );
278 $self->{master_doc} = $self->mods_values_to_mods_slim( $self->{master_doc} );
280 ($self->{master_doc}->{isbn}) =
281 $self->get_field_value( $mods, $isbn_xpath );
283 $self->{master_doc}->{type_of_resource} =
284 [ $self->get_field_value( $mods, $resource_xpath ) ];
286 ($self->{master_doc}->{tcn}) =
287 $self->get_field_value( $mods, $tcn_xpath );
289 ($self->{master_doc}->{pubdate}) =
290 $self->get_field_value( $mods, $pub_xpath );
292 ($self->{master_doc}->{publisher}) =
293 $self->get_field_value( $mods, $publisher_xpath );
295 ($self->{master_doc}->{edition}) =
296 $self->get_field_value( $mods, $edition_xpath );
300 # ------------------------------
301 # holds an array of [ link, title, link, title, ... ]
302 $self->{master_doc}->{online_loc} = [];
303 push(@{$self->{master_doc}->{online_loc}},
304 $self->get_field_value( $mods, $online_loc_xpath ));
306 ($self->{master_doc}->{synopsis}) =
307 $self->get_field_value( $mods, $abstract_xpath );
313 # ---------------------------------------------------------------------------
314 # Takes a MARCXML string and adds it to the growing MODS doc
315 # ---------------------------------------------------------------------------
316 sub push_mods_batch {
317 my( $self, $marcxml ) = @_;
319 my $xmldoc = $parser->parse_string($marcxml);
320 my $mods = $mods_sheet->transform($xmldoc);
322 my $xmlperl = $self->modsdoc_to_values( $mods );
323 $xmlperl = $self->mods_values_to_mods_slim( $xmlperl );
325 for my $subject( @{$xmlperl->{subject}} ) {
326 push @{$self->{master_doc}->{subject}}, $subject;
329 push( @{$self->{master_doc}->{type_of_resource}},
330 $self->get_field_value( $mods, $resource_xpath ));
332 if(!($self->{master_doc}->{isbn}) ) {
333 ($self->{master_doc}->{isbn}) =
334 $self->get_field_value( $mods, $isbn_xpath );
339 # ---------------------------------------------------------------------------
340 # Completes a MARC -> Unified MODS batch process and returns the perl hash
341 # ---------------------------------------------------------------------------
342 sub init_virtual_record {
343 my $record = new Fieldmapper::metabib::virtual_record;
344 $record->subject([]);
345 $record->types_of_resource([]);
346 $record->call_numbers([]);
350 sub finish_mods_batch {
352 my $perl = $self->{master_doc};
353 my $record = init_virtual_record();
355 # turn the hash into a fieldmapper object
356 (my $title = $perl->{title}) =~ s/\[.*?\]//og;
357 (my $author = $perl->{author}) =~ s/\(.*?\)//og;
360 for my $s (@{$perl->{series}}) {
361 push @series, (split( /\s*;/, $s ))[0];
364 # uniquify the types of resource
365 my $rtypes = $perl->{type_of_resource};
366 my %hash = map { ($_ => 1) } @$rtypes;
367 $rtypes = [ keys %hash ];
369 $record->title($title);
370 $record->author($author);
372 $record->doc_id($perl->{doc_id});
373 $record->isbn($perl->{isbn});
374 $record->pubdate($perl->{pubdate});
375 $record->publisher($perl->{publisher});
376 $record->tcn($perl->{tcn});
378 $record->edition($perl->{edition});
380 $record->subject($perl->{subject});
381 $record->types_of_resource($rtypes);
382 $record->series(\@series);
384 $record->online_loc($perl->{online_loc});
385 $record->synopsis($perl->{synopsis});
387 $self->{master_doc} = undef;