public function generate_record($id, $result)
{
// Well... really generate some record.
$record = new MsRecord($this);
$record->set_data('msdb_generator_id', $id);
// just for debugging
$record->set_data('pdb_id', $result->abstract_records[$id]);
// generate the neccessary data on the fly
$name = $this->soap_client->call('getPrimaryCitationTitle', array('in0' => $result->abstract_records[$id]));
$record->set_data('name', $name);
return $record;
}
/**
* The standarized way to print out redcords. This should
* RETURN the string to print out.
*/
public function print_record(MsRecord $record)
{
global $msConfiguration;
$tmpl = $this->record_template;
if ($record->get_type()) {
$tmpl .= '-' . $record->get_type();
}
return wfMsgData($tmpl, $record->data);
/*
$tmpl = wfMsg($this->record_template);
if($tmpl == '<'.$this->record_template.'>') { # the output when there's no such page
print $this->record_template." does not exist.<p>";
$tmpl = wgMsg($msConfiguration['default-record-message']);
}
$ret = strtr($tmpl, $record->data);
return $ret;
*/
}
function fetch($query)
{
$fetch_url = str_replace('$1', urlencode($query->keyword), self::$url);
$page = file_get_contents($fetch_url);
if (!$page) {
throw MWException('Performing viral zone search failed');
}
// typische Eintraege haben so einen Form:
// alles ohne Leerzeichen!
/*
<h4 class="column_name">VIRUSNAME <small>[SHORTFORM]</small></h4>
<div class="column_hr"><div class="context">
<a href="/viralzone/all_by_species/236.html">dsDNA (baltimore)</a><br />
<a href="/viralzone/all_by_species/176.html">Herpesviridae (family)</a><br />
<a href="/viralzone/all_by_species/526.html">Macavirus (genus)</a><br /></div></div>
*/
// Let's get the genus... don't know if thats right ;-)
if (!preg_match_all(self::$parser, strstr($page, 'h3'), $matches, PREG_SET_ORDER)) {
// no matches... so let's check wether we got anything
if (preg_match('/NO VIRUS FOUND/i', $page)) {
// return... an empty record array.
return array();
} else {
// the parsing seems to be broken
throw new MWException('Regex on viral zone output did not match!');
}
}
// Don't need to much memory -- delete page
unset($page);
// We'll have pretty much matchings. Parse them
$records = array();
// MsRecord array // wanted format: virusname => array(link label=>full link url)
$latest_record = false;
// will hold the latest record
#var_dump($matches); exit(0);
foreach ($matches as $entry) {
// since the regex always contains all parentheses, [1] is only
// filled when we have matched a h4, starting a new record.
if (strlen($entry[1])) {
#print("Adding new record; $entry[1].\n");
if ($latest_record) {
$records[] = $latest_record;
}
$latest_record = new MsRecord($this);
$latest_record->set_data('name', $entry[1]);
# fill fields stupidly with default values:
$latest_record->set_data('baltimore_link', '');
$latest_record->set_data('family_link', '');
$latest_record->set_data('genus_link', '');
$latest_record->set_data('baltimore', "''Baltimore nicht angegeben''");
$latest_record->set_data('family', "''Familie nicht angegeben''");
$latest_record->set_data('genus', "''Genus nicht angegeben''");
#$latest_virus = $entry[1];
#$records[$latest_virus] = array();
} else {
#print("Found nice other information concerning $entry[4]\n");
# something like 'family' => 'Herpesviridae'
$latest_record->set_data($entry[4], $entry[3]);
# and something like 'family_link' => '/viralzone/all_by_species/...'
$latest_record->set_data($entry[4] . '_link', self::$domain . $entry[2]);
}
#flush();
}
# and for the very last item:
if ($latest_record) {
$records[] = $latest_record;
}
# print "These are the records:<pre>"; var_dump($records);
# create "best links" for each record:
/*foreach($records as $record) {
if($record->get_data('genus'))
$record->set_data('link', $record->get_data('genus_link'));
else if($record->get_data('family'))
$record->set_data('link', $record->get_data('family_link'));
else if($record->get_data('baltimore'))
$record->set_data('link', $record->get_data('baltimore_link'));
else $record->set_data('link', 'http://www.google.dei);
}*/
return $records;
}
