tagno == 100) || ($field->tagno == 110)){
$temp = ereg_replace(',$', '', $field->subfields['a'] .' '. $field->subfields['d']);
$atomic['author'][] = $temp;
$tags[] = 'scrib:author='. trim($temp);
}else if($field->tagno == 110){
$temp = $field->subfields['a'];
$atomic['author'][] = $temp;
$tags[] = 'scrib:author='. trim($temp);
}else if(($field->tagno > 699) && ($field->tagno < 721)){
$temp = ereg_replace(',$', '', $field->subfields['a'] .' '. $field->subfields['d']);
$atomic['author'][] = $temp;
$tags[] = 'scrib:author='. trim($temp);
//Standard Numbers
}else if($field->tagno == 10){
$temp = explode(' ', trim($field->subfields['a']));
$atomic['lccn'][] = ereg_replace('[^0-9]', '', $temp[0]);
$tags[] = 'scrib:lccn='. trim($temp[0]);
}else if($field->tagno == 20){
$temp = explode(' ', trim($field->subfields['a']));
$temp = ereg_replace('[^0-9|x|X]', '', $temp[0]);
//echo 'i';
$isbn_xml = simplexml_load_string(file_get_contents('http://api.scriblio.net/v01a/isbn1013/' . $temp));
foreach ($isbn_xml->xpath('/idlist/isbn') as $temp) {
$atomic['isbn'][] = strtolower($temp);
}
//Call Numbers
}else if($field->tagno == 852){ // callnums from InfoCenter
$temp = trim($field->subfields['h']);
$atomic['callnumber'][] = $temp;
$tags[] = 'scrib:callnumber='. $temp;
$temp = trim($field->subfields['b']);
$atomic['location'][] = $temp;
$tags[] = 'scrib:location='. $temp;
$atomic['acqdate'][] = $field->subfields[x]{14}.$field->subfields[x]{15}.$field->subfields[x]{16}.$field->subfields[x]{17} .'-'. $field->subfields[x]{18}.$field->subfields[x]{19} .'-'. $field->subfields[x]{20}.$field->subfields[x]{21};
//Titles
}else if($field->tagno == 245){
$temp = ucwords(trim(trim(trim(ereg_replace('/$', '', $field->subfields['a']) .' '. trim(ereg_replace('/$', '', $field->subfields['b']))), ',.;:/')));
$atomic['title'][] = $temp;
$tags[] = 'scrib:title='. trim($temp);
$atomic['attribution'][] = trim(trim(trim($field->subfields['c']), ',.;:/'));
}else if($field->tagno == 240){
$temp = ucwords(trim(trim(trim(ereg_replace('/$', '', $field->subfields['a']) .' '. trim(ereg_replace('/$', '', $field->subfields['b']))), ',.;:/')));
$atomic['alttitle'][] = $temp;
$tags[] = 'scrib:title='. trim($temp);
}else if(($field->tagno > 719) && ($field->tagno < 741)){
$temp = $field->subfields['a'];
$atomic['alttitle'][] = $field->subfields['a'];
$tags[] = 'scrib:title='. trim($temp);
//Dates
}else if($field->tagno == 260){
$temp = str_pad(substr(ereg_replace('[^0-9]', '', $field->subfields['c']), 0, 4), 4 , '5');
$atomic['pubyear'][] = $temp;
}else if($field->tagno == 005){
$atomic['catdate'][] = $field->data{0}.$field->data{1}.$field->data{2}.$field->data{3} .'-'. $field->data{4}.$field->data{5} .'-'. $field->data{6}.$field->data{7};
}else if($field->tagno == 008){
$atomic['pubyear'][] = substr($field->data, 14, 4);
//Subjects
}else if(($field->tagno > 599) && ($field->tagno < 700)){
$atomic['subject'][] = trim(trim(implode(' -- ', $field->subfields)), '.');
if($atomic['subjkey']){
$atomic['subjkey'] = array_unique(array_merge($atomic['subjkey'], array_map($trimmer, array_values($field->subfields))));
}else{
$atomic['subjkey'] = array_map($trimmer, array_values($field->subfields));
}
//URLs
}else if($field->tagno == 856){
unset($temp);
$temp['href'] = $temp['title'] = str_replace(' ', '', $field->subfields['u']);
if($field->subfields['3'])
$temp['title'] = $field->subfields['3'];
if($field->subfields['z'])
$temp['title'] = $field->subfields['z'];
$atomic['url'][] = ''. $temp['title'] .'';
//Notes
}else if(($field->tagno > 299) && ($field->tagno < 400)){
$atomic['physdesc'][] = implode(' ', array_values($field->subfields));
}else if(($field->tagno > 399) && ($field->tagno < 500)){
$atomic['title'][] = implode("\n", array_values($field->subfields));
}else if(($field->tagno > 799) && ($field->tagno < 841)){
$atomic['series'][] = implode("\n", array_values($field->subfields));
}else if(($field->tagno > 499) && ($field->tagno < 600)){
$line = implode("\n", array_values($field->subfields));
if($field->tagno == 504)
continue;
if($field->tagno == 505){
$atomic['contents'][] = str_replace(array('> ','> ','> '), '>', ''. str_replace('--', "\n", trim(str_replace(array(' ', ' ', ' '), ' ', $line))) .'');
continue;
}
$atomic['notes'][] = str_replace(' ', ' ', $line);
//echo "Notes:
$line
";
}
//Format
if((!$atomic['format']) && ($field->tagno > 239) && ($field->tagno < 246)){
$temp = ucwords(strtolower(str_replace('[', '', str_replace(']', '', $field->subfields['h']))));
if(eregi('^book', $temp)){
$format = 'Book';
$formats = 'Books';
}else if(eregi('^micr', $temp)){
$format = 'Microform';
}else if(eregi('^electr', $temp)){
$format = 'Website';
$formats = 'Websites';
}else if(eregi('^vid', $temp)){
$format = 'Video';
}else if(eregi('^motion', $temp)){
$format = 'Video';
}else if(eregi('^audi', $temp)){
$format = 'Audio';
}else if(eregi('^cass', $temp)){
$format = 'Audio';
}else if(eregi('^phono', $temp)){
$format = 'Audio';
}else if(eregi('^record', $temp)){
$format = 'Audio';
}else if(eregi('^sound', $temp)){
$format = 'Audio';
}else if(eregi('^carto', $temp)){
$format = 'Map';
$formats = 'Maps';
}else if(eregi('^map', $temp)){
$format = 'Map';
$formats = 'Maps';
}else if(eregi('^globe', $temp)){
$format = 'Map';
$formats = 'Maps';
}else if($temp){
$format = 'Classroom Material';
//$format = $temp;
}
if(!$formats)
$formats = $format;
if($format){
$atomic['format'][] = $format;
$tags[] = 'scrib:format='. trim($formats);
}
}
}
}
if(!$atomic['format'][0]){
$atomic['format'][0] = 'Book';
$tags[] = 'scrib:format=Books';
}
if(!$atomic['acqdate'])
$atomic['acqdate'] = $atomic['catdate'];
if(!$atomic['catdate'][0])
$atomic['catdate'][0] = '1984-01-01';
if($atomic['pubyear'][0] > (date(Y) + 5))
$atomic['pubyear'][0] = substr($atomic['catdate'][0],0,4);
if($atomic['pubyear'][0]){
$atomic['pubdate'][] = $atomic['pubyear'][0].substr($atomic['catdate'][0],4);
$tags[] = 'scrib:pubyear='. $atomic['pubyear'][0];
}
foreach($atomic['isbn'] as $temp){
$tags[] = 'scrib:isbn='. trim($temp);
}
if($atomic['alttitle'])
$atomic['title'] = array_unique(array_merge($atomic['title'], $atomic['alttitle']));
foreach($atomic['subjkey'] as $temp){
$tags[] = 'scrib:subject='. trim($temp);
}
$atomic['tags'] = $tags;
$atomic['sourceid'] = eval('return($marcrecord'. str_replace(array('(',')','$'), '', $_REQUEST['sourceidfield']) .');');
// print_r($atomic);
return($atomic);
}
public function scrapeit($marcrecord){
global $scrib, $wpdb, $scrib_importer, $scrib_templates;
$record = $this->parse_marc($marcrecord);
if($record['title'] && $record['sourceid']){
echo 'a';
if($enrichinfo = $scrib_importer->enrich($record['isbn'])){
//print_r($enrichinfo);
$record['img'] = $enrichinfo['img'];
if($enrichinfo['summary']){
echo 's';
$record['shortdescription'] = $enrichinfo['summary'];
}
}
$record['the_title'] = $record['title'][0];
$record['the_pubdate'] = $record['pubdate'][0];
$record['the_acqdate'] = $record['acqdate'][0];
$record['the_sourceid'] = substr(ereg_replace('[^a-z|0-9]', '', strtolower($_REQUEST['sourceid'])), 0, 2) . $record['sourceid'];
$record['tags'][] = 'scrib:sourceid='. $record['the_sourceid'];
$record['the_excerpt'] = $scrib_templates->summary($record);
$record['the_content'] = $scrib_templates->full($record);
//print_r($record);
$scrib_importer->insertharvest($record);
}else{
echo '.';
}
flush();
}
}
if($_REQUEST['source'] && $_REQUEST['sourceid']){
$scrib_marc_reader = new Scrib_marc_reader;
$scrib_importer = new Scrib_importer;
$scrib_templates = new Scrib_default_templates;
$prefs['lastsource'] = $_REQUEST['source'];
$prefs['lastsourceid'] = $_REQUEST['sourceid'];
$prefs['lastsourceidfield'] = $_REQUEST['sourceidfield'];
update_option('scrib_marcimporter', $prefs);
$count = 0;
if($_REQUEST['run']){
$file = new File($_REQUEST['source'], $_REQUEST['start'], $_REQUEST['run']);
}else{
$file = new File($_REQUEST['source']);
}
while($record = $file->next()){
$count++;
$scrib_marc_reader->scrapeit($record->fields());
echo ' '. $count .' ';
while (@ob_end_flush());
flush();
//if($count > 200)
//break;
}
if(is_numeric($_REQUEST['start']) && $file->num_records()){
$_REQUEST['start'] = $_REQUEST['start'] + $_REQUEST['run'];
echo '
';
}
}else{
$prefs = get_option('scrib_marcimporter');
?>