private function convertXerxesRecord(Xerxes_Record $xerxes)
{
$citation = new Xerxes_Citation_Data();
$map = array("type" => "format", "issued" => "year", "number" => "", "number_of_volumes" => "", "publisher_place" => "place", "event" => "", "event_place" => "", "page" => "startPage", "locator" => "", "genre" => "format");
foreach ($citation as $key => $value) {
if ($key == "names") {
foreach ($xerxes->getAuthors() as $author) {
$name = new Xerxes_Citation_Name();
foreach ($author as $key => $value) {
$name->{$key} = $value;
}
array_push($citation->names, $name);
}
}
$target = $key;
if (array_key_exists($key, $map)) {
$target = $map[$key];
}
$method = Xerxes_Framework_Parser::strtoupper(substr($target, 0, 1));
$method .= substr($target, 1);
$method = "get{$method}";
if (method_exists($xerxes, $method)) {
$citation->{$key} = $xerxes->{$method}();
}
}
print_r($citation);
}
public function getOpenURL($strResolver, $strReferer = null, $param_delimiter = "&")
{
$url = parent::getOpenURL($strResolver, $strReferer, $param_delimiter);
// always ignore dates for journals and books, since worldcat is describing
// the item as a whole, not any specific issue or part
return $url . "&sfx.ignore_date_threshold=1";
}
public function map()
{
$leader = $this->leader();
## source database
$sid = $this->datafield("SID");
$this->metalib_id = $sid->subfield("d")->__toString();
$this->record_number = $sid->subfield("j")->__toString();
$this->result_set = $sid->subfield("s")->__toString();
$this->database_name = $sid->subfield("t")->__toString();
$this->source = $sid->subfield("b")->__toString();
## metalib weirdness
// puts leader in control field
$strLeaderMetalib = $this->controlfield("LDR")->__toString();
if ($strLeaderMetalib != "") {
$leader->value = $strLeaderMetalib;
}
$demunge = array("1XX", "6XX");
// character entity references de-munging code -- thanks a lot metalib!
foreach ($demunge as $field) {
$got_one = true;
do {
$authors = $this->datafield($field);
$got_one = false;
// whether we found any in the list
for ($x = 0; $x < $authors->length(); $x++) {
$this_datafield = $authors->item($x);
$this_value = $this_datafield->subfield()->__toString();
// we found an un-terminated char entity ref
$matches = array();
if (preg_match('/\\&\\#\\d{3}$/', $this_value, $matches)) {
$got_one = true;
$new_value = "";
// grab the value out of the next field
$x++;
// hopefully we aren't at the end?
if ($x < $authors->length()) {
// nope, so grab the next field's value
$next_datafield = $authors->item($x);
$next_value = $next_datafield->subfield()->__toString();
// add back in the terminating semi-colon
$new_value = "{$this_value};{$next_value}";
// blank it so we don't re-process it
$next_datafield->tag = "XXX";
} else {
// yup, just add a terminating char to the value
$new_value = $this_value . ";";
}
// now create a new datafield composed of both old and new values
$fixed_datafield = new Xerxes_Marc_DataField();
$fixed_datafield->tag = $this_datafield->tag;
// we'll just assume this is |a
$fixed_subfield = new Xerxes_Marc_Subfield();
$fixed_subfield->code = "a";
$fixed_subfield->value = $new_value;
$fixed_datafield->addSubField($fixed_subfield);
// add it to the main record
$this->addDataField($fixed_datafield);
// now blank the old ones
$this_datafield->tag = "XXX";
} else {
// we need to shift this to the end to keep field order in tact
// (critical for authors) so the above code works right
$new_field = clone $this_datafield;
$this->addDataField($new_field);
$this_datafield->tag = "XXX";
}
}
// if we found one, cycle back again to see if our now-combined
// field(s) *also* have un-terminated references since there may have
// been more than one broken char reference for a single author, e.g.
} while ($got_one == true);
}
// z3950/sutrs and some screen-scrapers have multiple authors in repeating 100 fields;
// invalid marc, so switch all but first to 700
$authors = $this->datafield("100");
if ($authors->length() > 1) {
for ($x = 1; $x < $authors->length(); $x++) {
$author = $authors->item($x);
$author->tag = "700";
}
}
// there are often multiple 773's, just combine them into one so we don't
// have to iterate over all of them in other code
$obj773 = new Xerxes_Marc_DataField();
$obj773->tag = "773";
foreach ($this->datafield("773") as $linked_entry) {
// add all of its subfields to the new one
foreach ($linked_entry->subfield() as $linked_entry_subfield) {
$obj773->addSubField($linked_entry_subfield);
}
// now blank this one to take it out of the mix
$linked_entry->tag = "XXX";
}
// add our new one to the document
$this->addDataField($obj773);
## ebsco format
if (strstr($this->source, "EBSCO")) {
// leader appears to be hard-wired; useless
$leader->value = "";
// format
array_push($this->format_array, $this->datafield("656")->subfield("a")->__toString());
array_push($this->format_array, $this->datafield("514")->subfield("a")->__toString());
$strEbscoType = $this->datafield("072")->subfield("a")->__toString();
if (strstr($this->source, "EBSCO_PSY") || strstr($this->source, "EBSCO_PDH")) {
$strEbscoType = "";
}
array_push($this->format_array, $strEbscoType);
// ebsco book chapter
$strEbscoBookTitle = $this->datafield("771")->subfield("a")->__toString();
if ($strEbscoBookTitle != "") {
array_push($this->format_array, "Book Chapter");
}
}
// gale puts issn in 773b
if (strstr($this->source, 'GALE')) {
$strGaleIssn = $this->datafield("773")->subfield("b")->__toString();
if ($strGaleIssn != null) {
array_push($this->issns, $strGaleIssn);
}
}
// eric doc number
$this->eric_number = $this->datafield("ERI")->subfield("a")->__toString();
## full-text
// some databases have full-text but no 856
// will capture these here and add to links array
// pychcritiques -- no indicator of full-text either, assume all to be (9/5/07)
// no unique metalib config either, using psycinfo, so make determination based on name. yikes!
if (stristr($this->database_name, "psycCRITIQUES")) {
array_push($this->links, array("Full-Text in HTML", array("001" => $this->controlfield("001")->__toString()), "html"));
}
// factiva -- no indicator of full-text either, assume all to be (9/5/07)
if (stristr($this->source, "FACTIVA")) {
array_push($this->links, array("Full-Text Available", array("035_a" => $this->datafield("035")->subfield("a")->__toString()), "online"));
}
// eric -- document is recent enough to likely contain full-text;
// 340000 being a rough approximation of the document number after which they
// started digitizing
// EBSCO provides an indication of Full text available in 037 (jdwyn 01/12/2012)
if (strstr($this->source, "ERIC") && strlen($this->eric_number) > 3) {
$strEricType = substr($this->eric_number, 0, 2);
$strEricNumber = (int) substr($this->eric_number, 2);
$strEricAvailabilityNote = $this->datafield("037")->subfield("a")->__toString();
if ($strEricType == "ED") {
if ($strEricNumber >= 340000 || stristr($this->source, "EBSCO_ERIC") && $strEricAvailabilityNote != "Not available from ERIC" && $strEricAvailabilityNote != "Available on microfiche only") {
$strFullTextPdf = "http://www.eric.ed.gov/ERICWebPortal/contentdelivery/servlet/ERICServlet?accno=" . $this->eric_number;
array_push($this->links, array("Full-text at ERIC.gov", $strFullTextPdf, "pdf"));
}
}
}
// 7 Apr 09, jrochkind. Gale Biography Resource Center
// No 856 is included at all, but a full text link can be
// constructed from the 001 record id.
if (stristr($this->source, "GALE_ZBRC")) {
$url = "http://ic.galegroup.com/ic/bic1/ReferenceDetailsPage/ReferenceDetailsWindow?displayGroupName=K12-Reference&action=e&windowstate=normal&mode=view&documentId=GALE|" . $this->controlfield("001")->__toString();
array_push($this->links, array("Full-Text in HTML", $url, "html"));
}
// special handling of 856
$notes = $this->fieldArray("500", "a");
// needed for gale
foreach ($this->datafield("856") as $link) {
$strDisplay = $link->subfield("z")->__toString();
$strUrl = $link->subfield("u")->__toString();
// bad links
// records that have 856s, but are not always for full-text; in that case, specify them
// here as original records, and remove 856 so parent code doesn't process them as full-text links
//
// springer (metapress): does not distinguish between things in your subscription or not (9/16/08)
// cinahl (bzh): not only is 856 bad, but link missing http:// bah! thanks greg at upacific! (9/10/08)
// wilson: 901|t shows an indication of full-text (9/16/10)
// cabi: just point back to site (10/30/07)
// google scholar: just point back to site (3/26/07)
// amazon: just point back to site (3/20/08)
// abc-clio: just point back to site (7/30/07)
// engineering village (evii): has unreliable full-text links in a consortium environment (4/1/08)
// wiley interscience: wiley does not limit full-text links only to your subscription (4/29/08)
// oxford: only include the links that are free, otherwise just a link to abstract (5/7/08)
// gale: only has full-text if 'text available' note in 500 field (9/7/07) BUT: Not true of Gale virtual reference library (GALE_GVRL). 10/14/08 jrochkind.
// ieee xplore: does not distinguish between things in your subscription or not (2/13/09)
// elsevier links are not based on subscription (6/2/10)
// harvard business: these links in business source premiere are not part of your subscription (5/26/10)
// proquest (umi): these links are not full-text (thanks jerry @ uni) (5/26/10)
// proquest (gateway): there doesn't appear to be a general rule to this, so only doing it for fiaf (5/26/10)
// primo central: no actual links here
if (stristr($this->source, "METAPRESS_XML") || stristr($this->source, "EBSCO_RZH") || stristr($this->source, "WILSON_") && $this->datafield("901")->subfield("t")->__toString() == "" || stristr($this->source, "CABI") || stristr($this->source, "SCOPUS4") || stristr($this->source, "GOOGLE_SCH") || stristr($this->source, "AMAZON") || stristr($this->source, "ABCCLIO") || stristr($this->source, "EVII") || stristr($this->source, "WILEY_IS") || stristr($this->source, "OXFORD_JOU") && !strstr($strUrl, "content/full/") || strstr($this->source, "GALE") && !strstr($this->source, "GALE_GVRL") && !in_array("Text available", $notes) || stristr($this->source, "IEEE_XPLORE") || stristr($this->source, "ELSEVIER_SCOPUS") || stristr($this->source, "ELSEVIER_SCIRUS") || stristr($this->source, "EBSCO") && $strUrl != "" && !strstr($strUrl, "epnet") || strstr($strUrl, "proquest.umi.com") && strstr($strUrl, "Fmt=2") || strstr($strUrl, "gateway.proquest.com") && strstr($strUrl, "xri:fiaf:article") || stristr($this->source, "NEWPC")) {
// take it out so the parent class doesn't treat it as full-text
$link->tag = "XXX";
array_push($this->links, array($strDisplay, $strUrl, "original_record"));
} elseif (stristr($this->source, "EBSCO")) {
$strEbscoFullText = $link->subfield("i")->__toString();
$ebsco_fulltext_type = "";
// html
// there is (a) an indicator from ebsco that the record has full-text, or
// (b) an abberant 856 link that doesn't work, but the construct link will work,
// so we take that as something of a full-text indicator
if (strstr($strEbscoFullText, "T") || strstr($strDisplay, "View Full Text")) {
$ebsco_fulltext_type = "html";
} elseif (strstr($link->subfield("az")->__toString(), "PDF")) {
$ebsco_fulltext_type = "pdf";
}
if ($ebsco_fulltext_type != "") {
$str001 = $this->controlfield("001")->__toString();
$str016 = $this->datafield("016")->subfield("a")->__toString();
// see if the id number is 'dirty'
$bolAlpha001 = preg_match('/^\\W/', $str001);
// if so, and there is a 016, use that instead, if not go ahead and use
// the 001; if neither do nothing
if ($bolAlpha001 == true && $str016 != "") {
array_push($this->links, array($strDisplay, array("016" => $str016), $ebsco_fulltext_type));
} elseif ($bolAlpha001 == false) {
array_push($this->links, array($strDisplay, array("001" => $str001), $ebsco_fulltext_type));
}
$link->tag = "XXX";
array_push($this->links, array($strDisplay, $strUrl, "original_record"));
}
}
}
// Gale title clean-up, because for some reason unknown to man they put weird
// notes and junk at the end of the title. so remove them here and add them to notes.
if (strstr($this->source, 'GALE_')) {
$arrMatches = array();
$strGaleRegExp = '/\\(([^)]*)\\)/';
$title = $this->datafield("245");
$title_main = $title->subfield("a");
$title_sub = $title->subfield("b");
$note_field = new Xerxes_Marc_DataField();
$note_field->tag = "500";
if ($title_main != null) {
if (preg_match_all($strGaleRegExp, $title_main->value, $arrMatches) != 0) {
$title_main->value = preg_replace($strGaleRegExp, "", $title_main->value);
}
foreach ($arrMatches[1] as $strMatch) {
$subfield = new Xerxes_Marc_Subfield();
$subfield->code = "a";
$subfield->value = "From title: " . $strMatch;
$note_field->addSubField($subfield);
}
}
// sub title is only these wacky notes
if ($title_sub != null) {
$subfield = new Xerxes_Marc_Subfield();
$subfield->code = "a";
$subfield->value = "From title: " . $title_sub->value;
$note_field->addSubField($subfield);
$title_sub->value = "";
}
if ($note_field->subfield("a")->length() > 0) {
$this->addDataField($note_field);
}
}
// psycinfo and related databases
if (strstr($this->source, "EBSCO_PDH") || strstr($this->source, "EBSCO_PSYH") || strstr($this->source, "EBSCO_LOH")) {
// includes a 502 that is not a thesis note -- bonkers!
// need to make this a basic note, otherwise xerxes will assume this is a thesis
foreach ($this->datafield("502") as $thesis) {
$thesis->tag = "500";
}
}
######## PARENT MAPPING ###########
parent::map();
###################################
// metalib's own year, issue, volume fields
$year = $this->datafield("YR ")->subfield("a")->__toString();
if ($year != "") {
$this->year = $year;
}
if ($this->issue == null) {
$this->issue = $this->datafield("ISS")->subfield("a")->__toString();
}
if ($this->volume == null) {
$this->volume = $this->datafield("VOL")->subfield("a")->__toString();
}
// book chapters
if ($this->journal_title != "" && count($this->isbns) > 0 && count($this->issns) == 0) {
$this->book_title = $this->journal_title;
$this->journal_title = "";
$this->format = "Book Chapter";
}
## ebsco 77X weirdness
if (strstr($this->source, "EBSCO")) {
// pages in $p (abbreviated title)
$pages = $this->datafield("773")->subfield('p')->__toString();
if ($pages != "") {
$this->short_title = "";
}
// book chapter
$btitle = $this->datafield("771")->subfield('a')->__toString();
if ($btitle != "") {
$this->book_title = $btitle;
$this->format = "Book Chapter";
}
}
## oclc dissertation abstracts
// (HACK) 10/1/2007 this assumes that the diss abs record includes the 904, which means
// there needs to be a local search config that performs an 'add new' action rather than
// the 'remove' action that the parser uses by default
if (strstr($this->source, "OCLC_DABS")) {
$this->degree = $this->datafield("904")->subfield("j")->__toString();
$this->institution = $this->datafield("904")->subfield("h")->__toString();
$this->journal_title = $this->datafield("904")->subfield("c")->__toString();
$this->journal = $this->journal_title . " " . $this->journal;
if ($this->journal_title == "MAI") {
$this->format = "Thesis";
} else {
$this->format = "Dissertation";
}
}
// random format related changes
if (strstr($this->source, 'ERIC') && strstr($this->eric_number, 'ED') && !stristr($this->title, "proceeding")) {
$this->format = "Report";
} elseif (strstr($this->source, 'ERIC') && !strstr($this->eric_number, 'ED')) {
$this->format = "Article";
} elseif (strstr($this->source, 'OCLC_PAPERS')) {
$this->format = "Conference Paper";
} elseif (strstr($this->source, 'PCF1')) {
$this->format = "Conference Proceeding";
} elseif (stristr($this->source, "GOOGLE_B")) {
$this->format = "Book";
} elseif (strstr($this->source, "EBSCO_LOH")) {
$this->format = "Tests & Measures";
} elseif (strstr($this->source, "OXFORD_MUSIC_ONLINE")) {
$this->format = "Article";
} elseif (strstr($this->source, "ESPACENET")) {
$this->format = "Patent";
} elseif (strstr($this->source, "WIPO_PCT")) {
$this->format = "Patent";
} elseif (strstr($this->source, "USPA")) {
$this->format = "Patent";
} elseif (strstr($this->source, "DEPATIS")) {
$this->format = "Patent";
} elseif (strstr($this->source, "DART")) {
$this->format = "Thesis";
} elseif (strstr($this->source, "DDI")) {
$this->format = "Thesis";
} elseif (strstr($this->source, "ETHOS")) {
$this->format = "Thesis";
} elseif (strstr($this->source, "DIVA_EXTR")) {
$this->format = "Thesis";
} elseif (strstr($this->source, "UNION_NDLTD")) {
$this->format = "Thesis";
}
// JSTOR book review correction: title is meaningless, but subjects
// contain the title of the books, so we'll swap them to the title here
if (strstr($this->source, 'JSTOR') && $this->title == "Review") {
$this->title = "";
$this->sub_title = "";
foreach ($this->subjects as $subject) {
$this->title .= " " . $subject->value;
}
$this->title = trim($this->title);
$this->subjects = null;
$this->format = "Book Review";
}
// jstor links are all pdfs
if (strstr($this->source, 'JSTOR')) {
for ($x = 0; $x < count($this->links); $x++) {
$link = $this->links[$x];
$link[2] = "pdf";
$this->links[$x] = $link;
}
}
// CSA subject term clean-up,
// since they put an asterick in front of each term (2009-09-30)
if (strstr($this->source, 'CSA_')) {
for ($x = 0; $x < count($this->subjects); $x++) {
$subject_object = $this->subjects[$x];
$subject_object->value = str_replace("*", "", $subject_object->value);
$this->subjects[$x] = $subject_object;
}
}
// demote links based on config
$objConfig = Xerxes_Framework_Registry::getInstance();
$configIgnoreFullText = $objConfig->getConfig("FULLTEXT_IGNORE_SOURCES", false);
$configIgnoreFullText = str_replace(" ", "", $configIgnoreFullText);
$arrIgnore = explode(",", $configIgnoreFullText);
for ($x = 0; $x < count($this->links); $x++) {
$link = $this->links[$x];
if (in_array($this->source, $arrIgnore) || in_array($this->metalib_id, $arrIgnore)) {
$link[2] = "original_record";
}
$this->links[$x] = $link;
}
}
/**
* Add a record to the user's saved record space. $objXerxesRecord will be
* updated with internal db id and original id..
*
* @param string $username username to save the record under
* @param string $source name of the source database
* @param string $id identifier for the record
* @param Xerxes_Record $objXerxesRecord xerxes record object to save
* @return int status
*/
public function addRecord($username, $source, $id, Xerxes_Record $objXerxesRecord)
{
$arrValues = array();
$iRefereed = 0;
$iYear = (int) $objXerxesRecord->getYear();
$strTitle = $objXerxesRecord->getMainTitle();
$strSubTitle = $objXerxesRecord->getSubTitle();
if ($strSubTitle != "") {
$strTitle .= ": " . $strSubTitle;
}
// peer-reviwed look-up
if ($objXerxesRecord->getISSN() != null) {
$arrResults = $this->getRefereed($objXerxesRecord->getISSN());
if (count($arrResults) > 0) {
$iRefereed = 1;
}
}
$strSQL = "INSERT INTO xerxes_records \r\n\t\t\t( source, original_id, timestamp, username, nonsort, title, author, year, format, refereed, record_type, marc )\r\n\t\t\tVALUES \r\n\t\t\t( :source, :original_id, :timestamp, :username, :nonsort, :title, :author, :year, :format, :refereed, :record_type, :marc)";
$arrValues[":source"] = $source;
$arrValues[":original_id"] = $id;
$arrValues[":timestamp"] = date("Y-m-d H:i:s");
$arrValues[":username"] = $username;
$arrValues[":nonsort"] = $objXerxesRecord->getNonSort();
$arrValues[":title"] = substr($strTitle, 0, 90);
$arrValues[":author"] = $objXerxesRecord->getPrimaryAuthor(true);
$arrValues[":year"] = $iYear;
$arrValues[":format"] = $objXerxesRecord->getFormat();
$arrValues[":refereed"] = $iRefereed;
$arrValues[":marc"] = serialize($objXerxesRecord);
$arrValues[":record_type"] = "xerxes_record";
$status = $this->insert($strSQL, $arrValues);
// get the internal xerxes record id for the saved record, and fill record
// with it, so caller can use.
$getIDSql = "SELECT id FROM xerxes_records WHERE original_id = :original_id";
$getIDParam = array(":original_id" => $id);
$getIDResults = $this->select($getIDSql, $getIDParam);
$objXerxesRecord->id = $getIDResults[0]["id"];
$objXerxesRecord->original_id = $id;
return $status;
}
/**
* URL for the sms feature
*
* @param Xerxes_Record $result
* @return string url
*/
protected function linkSMS($result)
{
$arrParams = array("base" => $this->request->getProperty("base"), "action" => "sms", "id" => $result->getRecordID());
return $this->request->url_for($arrParams);
}
public function map()
{
parent::map();
}
