function addMOProperty($unifiedName, $client)
{
$mysqli = new mysqli(HOSTNAME, USERNAME, PASSWD, DATABASE);
if ($mysqli->connect_errno) {
die("error: " . $mysqli->connect_error);
}
$localTable = "MOSpacePeople";
$parser = new HumanNameParser_Parser($unifiedName);
$last = $parser->getLast();
$first = $parser->getFirst();
//@Debug
//echo "Checking ".$last.", ".$first.".....";
$q = "SELECT * from " . $localTable . " where firstname='" . $mysqli->real_escape_string($first) . "' and lastname='" . $mysqli->real_escape_string($last) . "'";
$fromMU = 0;
$isProfessor = 0;
if ($result = $mysqli->query($q)) {
if ($result->num_rows == 1) {
//@Debug
//echo "found in local database.....";
$fromMU = 1;
if ($resAssocArray = $result->fetch_assoc()) {
if (isset($resAssocArray['Title'])) {
$isProfessor = findIfProfessor($resAssocArray['Title']) == 1 ? 1 : 0;
//@Debug
//if($isProfessor == 1) echo "is a prof\n"; else echo "NOT a prof\n";
} else {
$isProfessor = 0;
//@Debu
//echo "NOT a prof\n";
}
} else {
die("fetch result from MOSpacePeople failed");
}
} else {
$peopleFinderURL = "https://webservices.doit.missouri.edu/peoplefinderWS/peoplefinderws.asmx/PeopleFinderXml?firstName=" . urlencode($first) . "&lastname=" . urlencode($last) . "&department=&phoneno=&email=";
$url_parser = new URLParser($peopleFinderURL);
$retArr = $url_parser->XMLToArray();
if (intval($retArr['@attributes']['found']) == 1) {
//@Debug
//echo "found in Peoplefinder...";
$fromMU = 1;
$title = array_key_exists("Title", $retArr['Person']) && !empty($retArr['Person']['Title']) ? $retArr['Person']['Title'] : "";
$isProfessor = findIfProfessor($title) == 1 ? 1 : 0;
//@Debug
//if($isProfessor == 1) echo "is a prof\n"; else echo "NOT a prof\n";
} else {
$isProfessor = 0;
//@Debug
//echo "NOT a prof\n";
}
}
} else {
die("query: " . $q . "\nFailed");
}
$q_str = "match (u:Person {name: \"" . $unifiedName . "\"}) set u.fromMU = " . $fromMU . ", u.isProfessor = " . $isProfessor;
$query = new Query($client, $q_str);
$client->executeCypherQuery($query);
}
function IEEEHandler($query_str, $sphinxPort, $numOfResults = 10)
{
$table_name = "ieee";
$mysqli = sphinxDB_connect($sphinxPort);
$IEEE_url_prefix = "http://ieeexplore.ieee.org/gateway/ipsSearch.jsp?hc=" . $numOfResults . "&ti=";
$url_parser = new URLParser($IEEE_url_prefix . urlencode($query_str));
$IEEE_item_array = $url_parser->XMLToArray();
$ins_qry = "INSERT INTO {$table_name} (id,title, authors, pubtitle, pubtype, volume, issue, abstract, affiliation, issn, mdurl, pdf) VALUES";
if (array_key_exists('document', $IEEE_item_array) && !empty($IEEE_item_array['document'])) {
$IEEE_item_array['document'] = oneRecordArrayFormation($IEEE_item_array['document']);
$id = 1;
foreach ($IEEE_item_array['document'] as $ind => $attr_array) {
$title = array_key_exists('title', $attr_array) ? $attr_array['title'] : "";
$authors = array_key_exists('authors', $attr_array) ? $attr_array['authors'] : "";
if (!is_string($authors) || empty($authors)) {
continue;
}
$pubtitle = array_key_exists('pubtitle', $attr_array) ? $attr_array['pubtitle'] : "";
$pubtype = array_key_exists('pubtype', $attr_array) ? $attr_array['pubtype'] : "";
$volume = array_key_exists('volume', $attr_array) ? $attr_array['volume'] : "";
$issue = array_key_exists('issue', $attr_array) ? $attr_array['issue'] : "";
$abstract = array_key_exists('abstract', $attr_array) ? $attr_array['abstract'] : "";
if ($abstract != strip_tags($abstract)) {
$abstract = "";
}
$affiliation = array_key_exists('affiliation', $attr_array) ? $attr_array['affiliation'] : "";
$issn = array_key_exists('issn', $attr_array) ? $attr_array['issn'] : "";
$mdurl = array_key_exists('mdurl', $attr_array) ? $attr_array['mdurl'] : "";
$pdf = array_key_exists('pdf', $attr_array) ? $attr_array['pdf'] : "";
$title = $mysqli->real_escape_string($title);
$authors = $mysqli->real_escape_string($authors);
$pubtitle = $mysqli->real_escape_string($pubtitle);
$pubtype = $mysqli->real_escape_string($pubtype);
$volume = $mysqli->real_escape_string($volume);
$issue = $mysqli->real_escape_string($issue);
$abstract = $mysqli->real_escape_string($abstract);
$affiliation = $mysqli->real_escape_string($affiliation);
$issn = $mysqli->real_escape_string($issn);
$mdurl = $mysqli->real_escape_string($mdurl);
$pdf = $mysqli->real_escape_string($pdf);
$ins_qry .= "({$id}, '{$title}', '{$authors}','{$pubtitle}','{$pubtype}','{$volume}','{$issue}','{$abstract}', '{$affiliation}', '{$issn}','{$mdurl}','{$pdf}'),";
++$id;
}
$ins_qry = rtrim($ins_qry, ",");
if (!$mysqli->query("TRUNCATE RTINDEX " . $table_name)) {
exit("Error truncate: " . $mysqli->error);
}
if (!$mysqli->query($ins_qry)) {
exit("Error insert: " . $mysqli->error);
}
//Rerank the result
$rank_query = "SELECT *,weight() AS weight FROM {$table_name} where MATCH('{$query_str}') LIMIT 0,1000 OPTION ranker=MATCHANY;";
if (!($ranked_result = $mysqli->query($rank_query))) {
exit("Error rank: " . $mysqli->error);
}
//Fetch the re-ranked result
while ($row = $ranked_result->fetch_assoc()) {
$IEEEResultArray[] = array("title" => $row['title'], "authors" => $row['authors'], "pubtitle" => $row['pubtitle'], "pubtyle" => $row['pubtype'], "volume" => $row['volume'], "issue" => $row['issue'], "abstract" => $row['abstract'], "issn" => $row['issn'], "mdurl" => $row['mdurl'], "pdf" => $row['pdf']);
}
return $IEEEResultArray;
}
}
echo "is a prof\n";
} else {
echo "NOT a prof\n";
}
} else {
$isProfessor = 0;
//@Debu
echo "NOT a prof\n";
}
} else {
die("fetch result from MOSpacePeople failed");
}
} else {
$peopleFinderURL = "https://webservices.doit.missouri.edu/peoplefinderWS/peoplefinderws.asmx/PeopleFinderXml?firstName=" . urlencode($first) . "&lastname=" . urlencode($last) . "&department=&phoneno=&email=";
$url_parser = new URLParser($peopleFinderURL);
$retArr = $url_parser->XMLToArray();
if (intval($retArr['@attributes']['found']) == 1) {
//@Debug
echo "found in Peoplefinder...";
$fromMU = 1;
$title = array_key_exists("Title", $retArr['Person']) && !empty($retArr['Person']['Title']) ? $retArr['Person']['Title'] : "";
$isProfessor = findIfProfessor($title) == 1 ? 1 : 0;
//@Debug
if ($isProfessor == 1) {
echo "is a prof\n";
} else {
echo "NOT a prof\n";
}
} else {
$isProfessor = 0;
//@Debug
function pubmedHandler($query_str, $sphinxPort, $numOfResults = 10)
{
$table_name = "pubmed";
$mysqli = sphinxDB_connect($sphinxPort);
$pubmedSearchURLPrefix = "http://eutils.ncbi.nlm.nih.gov/entrez/eutils/esearch.fcgi?RetMax=" . $numOfResults . "&db=pubmed&term=";
$url_parser = new URLParser($pubmedSearchURLPrefix . urlencode($query_str));
$pubmedSearchResultArray = $url_parser->XMLToArray();
if (array_key_exists('Count', $pubmedSearchResultArray) && $pubmedSearchResultArray['Count'] > 0) {
//if pubmed search returns results
if ($pubmedSearchResultArray['Count'] == 1 || $pubmedSearchResultArray['RetMax'] == 1) {
$pubmedSearchResultArray['IdList']['Id'] = array($pubmedSearchResultArray['IdList']['Id']);
}
$pubmedFetchURL = "http://eutils.ncbi.nlm.nih.gov/entrez/eutils/efetch.fcgi?db=pubmed&retmode=xml&id=";
foreach ($pubmedSearchResultArray['IdList']['Id'] as $ind => $pubid) {
$pubmedFetchURL .= $pubid . ",";
}
$pubmedFetchURL = rtrim($pubmedFetchURL, ",");
$pubmedFetchXML = file_get_contents($pubmedFetchURL);
if ($pubmedFetchXML == FALSE) {
die("file_get_contents pubmed fetch failed");
}
$pubmedFetchResult = simplexml_load_string($pubmedFetchXML);
$pubmedFetchResultArray = json_decode(json_encode($pubmedFetchResult), TRUE);
if (!array_key_exists('PubmedArticle', $pubmedFetchResultArray)) {
trigger_error("PubmedArticle not set in pubmedFetchResultArray", E_USER_ERROR);
return FALSE;
}
$pubmedFetchResultArray['PubmedArticle'] = isset($pubmedFetchResultArray['PubmedArticle'][0]) ? $pubmedFetchResultArray['PubmedArticle'] : array($pubmedFetchResultArray['PubmedArticle']);
$id = 1;
$ins_qry = "INSERT INTO {$table_name} (id,pubid,title,authors,abstract,url,date,keywords,affiliations) VALUES";
foreach ($pubmedFetchResultArray['PubmedArticle'] as $record) {
if (!array_key_exists('MedlineCitation', $record)) {
trigger_error("MedlineCitation not set in one pubmed record", E_USER_ERROR);
return FALSE;
}
$pubid = $record['MedlineCitation']['PMID'];
$title = $record['MedlineCitation']['Article']['ArticleTitle'];
$abstract = "";
if (array_key_exists("Abstract", $record['MedlineCitation']['Article']) && array_key_exists("AbstractText", $record['MedlineCitation']['Article']['Abstract'])) {
$record['MedlineCitation']['Article']['Abstract']['AbstractText'] = is_array($record['MedlineCitation']['Article']['Abstract']['AbstractText']) ? $record['MedlineCitation']['Article']['Abstract']['AbstractText'] : array($record['MedlineCitation']['Article']['Abstract']['AbstractText']);
foreach ($record['MedlineCitation']['Article']['Abstract']['AbstractText'] as $ab) {
if (is_string($ab)) {
$abstract .= $ab . " ";
}
}
}
$authors = "";
$affiliations = "";
if (array_key_exists('AuthorList', $record['MedlineCitation']['Article']) && array_key_exists('Author', $record['MedlineCitation']['Article']['AuthorList']) && !empty($record['MedlineCitation']['Article']['AuthorList'])) {
$record['MedlineCitation']['Article']['AuthorList']['Author'] = isset($record['MedlineCitation']['Article']['AuthorList']['Author'][0]) ? $record['MedlineCitation']['Article']['AuthorList']['Author'] : array($record['MedlineCitation']['Article']['AuthorList']['Author']);
foreach ($record['MedlineCitation']['Article']['AuthorList']['Author'] as $a) {
if (array_key_exists('LastName', $a) && array_key_exists('ForeName', $a)) {
$authors .= $a['LastName'] . ", " . $a['ForeName'] . " | ";
if (array_key_exists('AffiliationInfo', $a) && !empty($a['AffiliationInfo'])) {
$a['AffiliationInfo'] = isset($a['AffiliationInfo'][0]) ? $a['AffiliationInfo'] : array($a['AffiliationInfo']);
foreach ($a['AffiliationInfo'] as $af) {
$affiliations .= $af['Affiliation'] . ";";
}
$affiliations = rtrim($affiliations, ";");
$affiliations .= " | ";
} else {
$affiliations .= "NULL" . " | ";
}
}
}
$authors = rtrim($authors, "|");
$affiliations = rtrim($affiliations, "|");
}
$keywords = "";
if (array_key_exists("KeywordList", $record['MedlineCitation']) && !empty($record['MedlineCitation']['KeywordList'])) {
$record['MedlineCitation']['KeywordList']['Keyword'] = isset($record['MedlineCitation']['KeywordList']['Keyword'][0]) ? $record['MedlineCitation']['KeywordList']['Keyword'] : array($record['MedlineCitation']['KeywordList']['Keyword']);
foreach ($record['MedlineCitation']['KeywordList']['Keyword'] as $k) {
$keywords .= $k . " | ";
}
$keywords = rtrim($keywords, "|");
}
$date = "";
if (array_key_exists("ArticleDate", $record['MedlineCitation']['Article']) && !empty($record['MedlineCitation']['Article']['ArticleDate'])) {
$date = $record['MedlineCitation']['Article']['ArticleDate']['Year'] . "-" . $record['MedlineCitation']['Article']['ArticleDate']['Month'] . "-" . $record['MedlineCitation']['Article']['ArticleDate']['Day'];
}
$url = 'http://www.ncbi.nlm.nih.gov/pubmed/' . $pubid;
$title = $mysqli->real_escape_string($title);
$abstract = $mysqli->real_escape_string($abstract);
$authors = $mysqli->real_escape_string($authors);
$affiliations = $mysqli->real_escape_string($affiliations);
$keywords = $mysqli->real_escape_string($keywords);
$ins_qry .= "({$id}, {$pubid}, '{$title}', '{$authors}', '{$abstract}', '{$url}', '{$date}', '{$keywords}', '{$affiliations}'),";
++$id;
}
$ins_qry = rtrim($ins_qry, ",");
if (!$mysqli->query("TRUNCATE RTINDEX " . $table_name)) {
exit("Error truncate: " . $mysqli->error);
}
if (!$mysqli->query($ins_qry)) {
exit("Error insert: " . $mysqli->error);
}
//Rerank the result
$rank_query = "SELECT *,weight() AS weight FROM {$table_name} where MATCH('{$query_str}') LIMIT 0,1000 OPTION ranker=MATCHANY;";
if (!($ranked_result = $mysqli->query($rank_query))) {
exit("Error rank: " . $mysqli->error);
}
//Fetch the re-ranked result
while ($row = $ranked_result->fetch_assoc()) {
$pubmedResultArray[] = array("pubid" => $row['pubid'], "title" => $row['title'], "abstract" => $row['abstract'], "authors" => $row['authors'], "keywords" => $row['keywords'], "date" => $row['date'], "url" => $row['url'], "affiliations" => $row['affiliations']);
}
return $pubmedResultArray;
}
return FALSE;
}
public function crawlPubmed()
{
$search_url = $this->pubmed_search_url . "&term=" . urlencode($this->firstName . " " . $this->lastName . "[Author]" . " AND " . $this->affiliation . "[Affiliation]");
$search_url_parser = new URLParser($search_url);
$pubmedSearch_item_array = $search_url_parser->XMLToArray();
$returnVal = array();
if (array_key_exists('Count', $pubmedSearch_item_array) && $pubmedSearch_item_array['Count'] > 0) {
if ($pubmedSearch_item_array['Count'] == 1 || $pubmedSearch_item_array['RetMax'] == 1) {
$pubmedSearch_item_array['IdList']['Id'] = array($pubmedSearch_item_array['IdList']['Id']);
}
$pid_url = "";
foreach ($pubmedSearch_item_array['IdList']['Id'] as $ind => $pubid) {
$pid_url .= $pubid . ",";
}
$pid_url = rtrim($pid_url, ",");
$fetch_url = $this->pubmed_fetch_url . $pid_url;
$fetch_url_parser = new URLParser($fetch_url);
$pubmedFetchResultArray = $fetch_url_parser->XMLToArray();
if (!array_key_exists('PubmedArticle', $pubmedFetchResultArray)) {
trigger_error("PubmedArticle not set in pubmedFetchResultArray", E_USER_ERROR);
return null;
}
$pubmedFetchResultArray['PubmedArticle'] = isset($pubmedFetchResultArray['PubmedArticle'][0]) ? $pubmedFetchResultArray['PubmedArticle'] : array($pubmedFetchResultArray['PubmedArticle']);
foreach ($pubmedFetchResultArray['PubmedArticle'] as $record) {
if (!array_key_exists('MedlineCitation', $record)) {
trigger_error("MedlineCitation not set in one pubmed record", E_USER_ERROR);
continue;
}
$coauthorList = array();
$title = $record['MedlineCitation']['Article']['ArticleTitle'];
$url = 'http://www.ncbi.nlm.nih.gov/pubmed/' . $record['MedlineCitation']['PMID'];
if (array_key_exists('AuthorList', $record['MedlineCitation']['Article']) && array_key_exists('Author', $record['MedlineCitation']['Article']['AuthorList']) && !empty($record['MedlineCitation']['Article']['AuthorList'])) {
$authorList = $record['MedlineCitation']['Article']['AuthorList']['Author'];
} else {
trigger_error("authorlist/author not set or author is empty", E_USER_ERROR);
continue;
}
if (!isset($authorList[0])) {
continue;
}
if (count($authorList) > 20) {
continue;
}
foreach ($authorList as $a) {
//converting pubmed name to neo4j compatible format
$parser = new HumanNameParser_Parser($a['LastName'] . ", " . $a['ForeName']);
$first = $parser->getFirst();
$last = $parser->getLast();
if (array_key_exists('AffiliationInfo', $a)) {
if (array_key_exists('Affiliation', $a['AffiliationInfo'])) {
$affi = $a['AffiliationInfo']['Affiliation'];
} else {
if (array_key_exists(0, $a['AffiliationInfo']) && array_key_exists('Affiliation', $a['AffiliationInfo'][0])) {
$affi = $a['AffiliationInfo'][0]['Affiliation'];
}
}
} else {
$affi = "";
}
$coauthorList[] = array("firstName" => $first, "lastName" => $last, "affiliation" => $affi);
}
$returnVal[] = array("title" => $title, "url" => $url, "people" => $coauthorList);
}
}
return $returnVal;
}