function import($reference)
{
global $year_found;
global $year_actual;
global $year_roman;
global $html;
print_r($reference);
if (!array_key_exists($reference->year, $year_actual)) {
$html[$reference->year] = '<html><body>';
$html[$reference->year] .= '<h1>' . $reference->year . '</h1>';
$year_actual[$reference->year] = 0;
$year_found[$reference->year] = 0;
$year_roman[$reference->year] = 0;
}
// Articles published this year
$year_actual[$reference->year]++;
if (is_numeric($reference->spage)) {
$openurl = reference2openurl($reference);
//echo $openurl . "\n";
$biostor_id = import_from_openurl($openurl);
if ($biostor_id != 0) {
$found = true;
// Articles found this year
$year_found[$reference->year]++;
$reference->url = 'http://biostor.org/reference/' . $biostor_id;
$url = $reference->url . '.json';
$json = get($url);
$j = json_decode($json);
//print_r($j);
//echo "---\n";
$PageID = $j->bhl_pages[0];
//echo $PageID . "\n";
$html[$reference->year] .= '<div>';
$html[$reference->year] .= '<a href="' . $reference->url . '" target="_new"><img src="http://biostor.org/bhl_image.php?PageID=' . $PageID . '&thumbnail" /></a><br/>';
$html[$reference->year] .= '<a href="http://www.biodiversitylibrary.org/page/' . $PageID . '" target="_new">' . $PageID . '</a><br/>';
$html[$reference->year] .= $reference->title . '<br/>';
$html[$reference->year] .= '</div>';
}
} else {
$year_roman[$reference->year]++;
}
}
function postprocess_citations($reference_id, $citations)
{
global $db;
$citation_count = 0;
// Avoid duplications
if ($reference_id != 0) {
$sql = 'DELETE FROM rdmp_reference_cites WHERE (reference_id=' . $reference_id . ')';
$result = $db->Execute($sql);
if ($result == false) {
die("failed [" . __FILE__ . ":" . __LINE__ . "]: " . $sql);
}
}
foreach ($citations as $citation) {
$citation_reference_id = 0;
if (isset($citation->genre)) {
// Lookup articles
if ($citation->genre == 'article') {
$citation_reference_id = 0;
// Do we have this already in BioStor?
$citation_reference_id = db_find_article($citation);
if ($citation_reference_id == 0) {
// Try BioStor OpenURL search
$citation_reference_id = import_from_openurl(reference_to_openurl($citation));
}
if ($citation_reference_id == 0) {
// Try bioGUID
if (bioguid_openurl_search($citation)) {
$citation_reference_id = db_store_article($citation);
}
}
}
}
// At this stage if citation_reference_id is 0 we haven't found it
// 1. store citation string (we are building a list of all such strings)
$citation_string_id = store_citation_string($citation);
// 2. If we've found it in BioStor, record link between citation string and reference id
if ($citation_reference_id != 0) {
$sql = 'DELETE FROM rdmp_reference_citation_string_joiner WHERE(reference_id=' . $citation_reference_id . ')
AND (citation_string_id=' . $citation_string_id . ')';
$result = $db->Execute($sql);
if ($result == false) {
die("failed [" . __FILE__ . ":" . __LINE__ . "]: " . $sql);
}
$sql = 'INSERT INTO rdmp_reference_citation_string_joiner (reference_id,citation_string_id)
VALUES(' . $citation_reference_id . ',' . $citation_string_id . ')';
$result = $db->Execute($sql);
if ($result == false) {
die("failed [" . __FILE__ . ":" . __LINE__ . "]: " . $sql);
}
}
// 3. Store link between source reference and citation string (which we will use to display literature cited)
if ($reference_id != 0) {
$sql = 'INSERT INTO rdmp_reference_cites (reference_id, citation_string_id, citation_order)
VALUES(' . $reference_id . ',' . $citation_string_id . ',' . $citation_count . ')';
$result = $db->Execute($sql);
if ($result == false) {
die("failed [" . __FILE__ . ":" . __LINE__ . "]: " . $sql);
}
$citation_count++;
}
}
}
while (!feof($file_handle)) {
$line = fgets($file_handle);
$parts = explode("\t", $line);
$parts[1] = trim($parts[1]);
if (preg_match('/(?<authorstring>.*)\\s*(?<year>[0-9]{4})\\.\\s+(?<title>.*)\\s+<em>(?<journal>.*)<\\/em>\\s+(?<series>.*)?\\s*<strong>(?<volume>\\d+)<\\/strong>(\\((?<issue>\\d+)\\))?:\\s+(?<spage>\\d+)([–|-](?<epage>\\d+))\\b/Uu', $parts[1], $m)) {
//print_r($m);
$matched = true;
$reference = reference_from_matches($m);
//print_r($reference);
$reference->id = $parts[0];
$reference->keywords[] = $parts[0];
$openurl = reference2openurl($reference);
//echo $openurl . "\n";
bioguid($reference);
if (isset($reference->epage)) {
$biostor_id = import_from_openurl($openurl);
if ($biostor_id != 0) {
$found = true;
$reference->url = 'http://biostor.org/reference/' . $biostor_id;
}
}
//print_r($reference);
echo reference2ris($reference);
} else {
/*
//echo $parts[1]. "\n";
// Book?
if (preg_match('/(?<authorstring>.*)\s+(?<year>[0-9]{4})\.\s+<em>(?<title>.*)<\/em>\.\s+(?<publoc>.*)\s*:(?<publisher>.*)\s*((?<frontmatter>[i|v|x]+)\s*)?(?<pages>\d+)\s+pp\.(\s*(?<extra>.*))?$/Uu', $parts[1], $m))
{
print_r($m);
