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}->{$type} = shift @$arr;
166 $data->{$class}->{$type} = shift @$arr if (lc($data->{$class}->{$type}) =~ /^the|an?/o);
168 $data->{$class}->{$type} .= ' : ' if ($data->{$class}->{$type} =~ /\w\s*$/o);
169 $data->{$class}->{$type} .= " $t";
172 $data->{$class}->{$type} = $arr;
179 my $class = "author";
180 $data->{$class} = {};
181 for my $type(keys %{$xpathset->{$class}}) {
182 my @value = $self->get_field_value( $mods, $xpathset->{$class}->{$type} );
183 $data->{$class}->{$type} = $value[0];
188 my $class = "series";
189 $data->{$class} = {};
190 for my $type(keys %{$xpathset->{$class}}) {
191 my @value = $self->get_field_value( $mods, $xpathset->{$class}->{$type} );
192 for my $arr (@value) {
194 push(@{$data->{$class}->{$type}}, join(" ", @$arr));
196 push( @{$data->{$class}->{$type}}, $arr );
209 # ---------------------------------------------------------------------------
210 # Grabs the data 'we want' from the MODS doc and returns it in hash form
211 # ---------------------------------------------------------------------------
212 sub mods_values_to_mods_slim {
213 my( $self, $modsperl ) = @_;
220 my $tmp = $modsperl->{title};
223 if(!$tmp) { $title = ""; }
225 ($title = $tmp->{proper}) ||
226 ($title = $tmp->{translated}) ||
227 ($title = $tmp->{abbreviated}) ||
228 ($title = $tmp->{uniform});
231 $tmp = $modsperl->{author};
232 if(!$tmp) { $author = ""; }
234 ($author = $tmp->{personal}) ||
235 ($author = $tmp->{other}) ||
236 ($author = $tmp->{corporate}) ||
237 ($author = $tmp->{conference});
240 $tmp = $modsperl->{subject};
241 if(!$tmp) { $subject = {}; }
243 for my $key( keys %{$tmp}) {
244 push(@$subject, @{$tmp->{$key}}) if ($tmp->{$key});
247 for my $s (@$subject) {
248 if(defined($subh->{$s})) { $subh->{$s->[0]}++ } else { $subh->{$s->[0]} = 1;}
253 $tmp = $modsperl->{'series'};
254 if(!$tmp) { $series = []; }
255 else { $series = $tmp->{'series'}; }
258 return { series => $series, title => $title,
259 author => $author, subject => $subject };
264 # ---------------------------------------------------------------------------
265 # Initializes a MARC -> Unified MODS batch process
266 # ---------------------------------------------------------------------------
268 sub start_mods_batch {
270 my( $self, $master_doc ) = @_;
274 my $xslt_doc = $parser->parse_file(
275 OpenSRF::Utils::SettingsClient->new->config_value(dirs => 'xsl') . "/MARC21slim2MODS3.xsl");
276 $mods_sheet = $xslt->parse_stylesheet( $xslt_doc );
280 my $xmldoc = $parser->parse_string($master_doc);
281 my $mods = $mods_sheet->transform($xmldoc);
283 # warn "-" x 100 . "\n";
284 # warn "MODS " . $mods->toString(1) . "\n";
285 # warn "-" x 100 . "\n";
287 $self->{master_doc} = $self->modsdoc_to_values( $mods );
288 $self->{master_doc} = $self->mods_values_to_mods_slim( $self->{master_doc} );
290 ($self->{master_doc}->{isbn}) =
291 $self->get_field_value( $mods, $isbn_xpath );
293 $self->{master_doc}->{type_of_resource} =
294 [ $self->get_field_value( $mods, $resource_xpath ) ];
296 ($self->{master_doc}->{tcn}) =
297 $self->get_field_value( $mods, $tcn_xpath );
299 ($self->{master_doc}->{pubdate}) =
300 $self->get_field_value( $mods, $pub_xpath );
302 ($self->{master_doc}->{publisher}) =
303 $self->get_field_value( $mods, $publisher_xpath );
305 ($self->{master_doc}->{edition}) =
306 $self->get_field_value( $mods, $edition_xpath );
310 # ------------------------------
311 # holds an array of [ link, title, link, title, ... ]
312 $self->{master_doc}->{online_loc} = [];
313 push(@{$self->{master_doc}->{online_loc}},
314 $self->get_field_value( $mods, $online_loc_xpath ));
316 ($self->{master_doc}->{synopsis}) =
317 $self->get_field_value( $mods, $abstract_xpath );
323 # ---------------------------------------------------------------------------
324 # Takes a MARCXML string and adds it to the growing MODS doc
325 # ---------------------------------------------------------------------------
326 sub push_mods_batch {
327 my( $self, $marcxml ) = @_;
329 my $xmldoc = $parser->parse_string($marcxml);
330 my $mods = $mods_sheet->transform($xmldoc);
332 my $xmlperl = $self->modsdoc_to_values( $mods );
333 $xmlperl = $self->mods_values_to_mods_slim( $xmlperl );
335 # for backwards compatibility, remove the array part when all is decided
336 if(ref($xmlperl->{subject}) eq 'ARRAY' ) {
337 for my $subject( @{$xmlperl->{subject}} ) {
338 push @{$self->{master_doc}->{subject}}, $subject;
341 for my $subject ( keys %{$xmlperl->{subject}} ) {
342 my $s = $self->{master_doc}->{subject};
343 if(defined($s->{$subject})) { $s->{$subject}++; } else { $s->{$subject} = 1; }
347 push( @{$self->{master_doc}->{type_of_resource}},
348 $self->get_field_value( $mods, $resource_xpath ));
350 if(!($self->{master_doc}->{isbn}) ) {
351 ($self->{master_doc}->{isbn}) =
352 $self->get_field_value( $mods, $isbn_xpath );
357 # ---------------------------------------------------------------------------
358 # Completes a MARC -> Unified MODS batch process and returns the perl hash
359 # ---------------------------------------------------------------------------
360 sub init_virtual_record {
361 my $record = new Fieldmapper::metabib::virtual_record;
362 $record->subject([]);
363 $record->types_of_resource([]);
364 $record->call_numbers([]);
368 sub finish_mods_batch {
370 my $perl = $self->{master_doc};
371 my $record = init_virtual_record();
373 # turn the hash into a fieldmapper object
374 (my $title = $perl->{title}) =~ s/\[.*?\]//og;
375 (my $author = $perl->{author}) =~ s/\(.*?\)//og;
378 for my $s (@{$perl->{series}}) {
379 push @series, (split( /\s*;/, $s ))[0];
382 # uniquify the types of resource
383 my $rtypes = $perl->{type_of_resource};
384 my %hash = map { ($_ => 1) } @$rtypes;
385 $rtypes = [ keys %hash ];
387 $record->title($title);
388 $record->author($author);
390 $record->doc_id($perl->{doc_id});
391 $record->isbn($perl->{isbn});
392 $record->pubdate($perl->{pubdate});
393 $record->publisher($perl->{publisher});
394 $record->tcn($perl->{tcn});
396 $record->edition($perl->{edition});
398 $record->subject($perl->{subject});
399 $record->types_of_resource($rtypes);
400 $record->series(\@series);
402 $record->online_loc($perl->{online_loc});
403 $record->synopsis($perl->{synopsis});
405 $self->{master_doc} = undef;