One Hat Cyber Team

View File Name : getit-issues

#!/usr/bin/php

<?php

$startyear = "1995";
$endyear = "1995";
$startissue = "1"; // issue and year are independent
$endissue = "4";

$outFile = $startyear . '-' . $endyear . '_' . $startissue . '-' . $endissue . '.xml';

$m = mysqli_connect('localhost','mfat','yumber','mfat');
mysqli_query($m, "set character_set_results='utf8'");
mysqli_query($m, "SET NAMES 'utf8'");
mysqli_query($m, "SET CHARACTER SET utf8");
mysqli_query($m, "SET SESSION collation_connection = 'utf8_unicode_ci'");

$sql = "select distinct article_year, article_item from articles";
$sql = $sql . " where article_print='Y' and article_year >= $startyear and article_year <= $endyear";
$sql = $sql . " and article_item >= $startissue and article_item <= $endissue";
$sql = $sql . " order by article_year, article_item";

// echo $sql;

$res = mysqli_query($m, $sql);


$out = new DOMDocument('1.0');
$out->preserveWhiteSpace = false;
$out->formatOutput = true;

$lang = [ 1 => "uk", 2 => "ru", 3 => "en"];

$issueId = 0;
$articleId = 0;
$authId = 0;

$rootNode = $out->createElementNS('http://pkp.sfu.ca', "issues");

// for each year and issue
while ($row = mysqli_fetch_assoc($res)) {
	$year = $row['article_year'];
	$issue = $row['article_item'];
	$volume = $year - 1994;
	if ($year == 1995) $month = 12;
	if ($year == 1996) $month = 4*$issue;
	else $month = 3*$issue;
	if ($month<10) $month="0" . $month;
	$datePublished = $year . "-" . $month . "-25";
	$dateModified = date("Y-m-d");
	$dateSubmitted = $datePublished;
	$issueId++;

	// Issue
	$issueNode = $out->createElementNS('http://pkp.sfu.ca', "issue");
	$issueNode->setAttribute("xmlns:xsi", 'http://www.w3.org/2001/XMLSchema-instance');
	$issueNode->setAttribute("published", "1");
	$issueNode->setAttribute("current", "0");
	$issueNode->setAttribute("access_status", "1");
	
	// Issue Id
	$issueIdNode = $out->createElement("id", $issueId);
	$issueIdNode->setAttribute("type", "internal");
	$issueIdNode->setAttribute("advice", "ignore");
	$issueNode->appendChild($issueIdNode);
	
	// Issue identification
	$issueIdentificationNode = $out->createElement("issue_identification");
	$volumeNode = $out->createElement("volume", $volume);
	$numberNode = $out->createElement("number", $issue);
	$yearNode = $out->createElement("year", $year);
	$issueIdentificationNode->appendChild($volumeNode);
	$issueIdentificationNode->appendChild($numberNode);
	$issueIdentificationNode->appendChild($yearNode);
	$issueNode->appendChild($issueIdentificationNode);
	
	// Dates
	$dateNode = $out->createElement("date_published", $datePublished);
	$issueNode->appendChild($dateNode);

	$lastModifiedNode = $out->createElement("last_modified", $dateModified);
	$issueNode->appendChild($lastModifiedNode);

	// Sections
	$sectionsNode = $out->createElement("sections");
	    $sectionNode = $out->createElement("section");
	    $sectionNode->setAttribute("ref", "ART");
	    $sectionNode->setAttribute("seq", "1");
	    $sectionNode->setAttribute("editor_restricted", "0");
	    $sectionNode->setAttribute("meta_indexed", "1");
	    $sectionNode->setAttribute("meta_reviewed", "1");
	    $sectionNode->setAttribute("abstracts_not_required", "0");
	    $sectionNode->setAttribute("hide_title", "0");
	    $sectionNode->setAttribute("hide_author", "0");
	    $sectionNode->setAttribute("abstract_word_count", "0");

		$sectionSubNode = $out->createElement("id", "1");
		$sectionSubNode->setAttribute("type", "internal");
		$sectionSubNode->setAttribute("advice", "ignore");
		$sectionNode->appendChild($sectionSubNode);

		$sectionSubNode = $out->createElement("abbrev", "ART");
		$sectionSubNode->setAttribute("locale", "en_US");
		$sectionNode->appendChild($sectionSubNode);

		$sectionSubNode = $out->createElement("title", "Articles");
		$sectionSubNode->setAttribute("locale", "en_US");
		$sectionNode->appendChild($sectionSubNode);

	    $sectionsNode->appendChild($sectionNode);
	$issueNode->appendChild($sectionsNode);

	// Issue coverpage
	
	// Issue galleys
	$issueGalleysNode = $out->createElementNS('http://pkp.sfu.ca', "issue_galleys");
	$issueGalleysNode->setAttribute("xmlns:xsi", 'http://www.w3.org/2001/XMLSchema-instance');
	$issueGalleysNode->setAttribute('xsi:schemaLocation', "http://pkp.sfu.ca native.xsd");
	$issueNode->appendChild($issueGalleysNode);

	// Articles
	$articlesNode = $out->createElementNS('http://pkp.sfu.ca', "articles");
	$articlesNode->setAttribute("xmlns:xsi", 'http://www.w3.org/2001/XMLSchema-instance');
	$articlesNode->setAttribute('xsi:schemaLocation', "http://pkp.sfu.ca native.xsd");

	$sql = "SELECT * FROM articles";
	$sql = $sql . " WHERE article_print='Y' AND article_year = " . $row['article_year'] ;
	$sql = $sql . " AND article_item = " . $row['article_item'];
	$sql = $sql . " ORDER BY CAST(SUBSTRING_INDEX(article_pages, '-', 1) AS signed)";

	// Article
	$resArt = mysqli_query($m, $sql);
	$artSeq = 0;

	while ($rowArt = mysqli_fetch_assoc($resArt)) {


		echo $rowArt['article_id'] .' ';
		$artSeq++;
		$articleLocale = "en_US";
		
		$dateSubmitted = $rowArt['article_received_text'];
		
		$dateSubmitted = preg_replace("|Received (\d\d).(\d\d).(\d\d\d\d).*|", "$3-$2-$1", $dateSubmitted);
		if (!$dateSubmitted) $dateSubmitted = $datePublished;

		// Create article
		$articleNode = $out->createElementNS('http://pkp.sfu.ca', "article");
		$articleNode->setAttribute("xmlns:xsi", 'http://www.w3.org/2001/XMLSchema-instance');
    		$articleNode->setAttribute("date_submitted", $dateSubmitted);
    		$articleNode->setAttribute("status", "3");
    		$articleNode->setAttribute("submission_progress", "0");
    		$articleNode->setAttribute("current_publication_id", $rowArt['article_id']);
    		$articleNode->setAttribute("stage", "production");

		// Article id
		$articleIdNode = $out->createElement('id', $rowArt['article_id']);
    		$articleIdNode->setAttribute("type", "internal");
    		$articleIdNode->setAttribute("advice", "ignore");
		$articleNode->appendChild($articleIdNode);

		// Submission files

		$basepath = '/home/mfat/public_html/article/';
		$subdir = $rowArt['article_item'];
		if ($subdir < 10) $subdir = '0' . $subdir;
		$dirpath = $basepath . $year . '/' . $subdir . '/';
		if($rowArt['article_pdf']){
			$path = $dirpath . $rowArt['article_pdf'];
			$filename = basename($path);
			$size = filesize($path);
			$contents = file_get_contents($path);
	
			$submissionNode = $out->createElementNS('http://pkp.sfu.ca', 'submission_file');
			$submissionNode->setAttribute("xmlns:xsi", 'http://www.w3.org/2001/XMLSchema-instance');
    			$submissionNode->setAttribute('xsi:schemaLocation', "http://pkp.sfu.ca native.xsd");
			$submissionNode->setAttribute('stage', 'proof');
			$submissionNode->setAttribute('id', $rowArt['article_id']);
		
			$revisionNode = $out->createElement('revision');
			$revisionNode->setAttribute('number', '1');
			$revisionNode->setAttribute('genre', 'Article text (PDF)');
			$revisionNode->setAttribute('filename', $rowArt['article_pdf']);
			$revisionNode->setAttribute('viewable', 'true');
			$revisionNode->setAttribute('date_uploaded', date("Y-m-d"));
			$revisionNode->setAttribute('date_modified', date("Y-m-d"));
			$revisionNode->setAttribute('filesize', $size);
			$revisionNode->setAttribute('filetype', 'application/pdf');
			$revisionNode->setAttribute('uploader', 'admin');

			$nameNode = $out->createElement('name', 'Imported: ' . $filename);
			$nameNode->setAttribute('locale', 'en_US');
			$revisionNode->appendChild($nameNode);
	
			$embedNode = $out->createElement('embed');
			$embedNode->setAttribute('encoding', 'base64');
			$embedNode->nodeValue = base64_encode($contents);
			$revisionNode->appendChild($embedNode);
			
			$submissionNode->appendChild($revisionNode);

			$articleNode->appendChild($submissionNode);
		}

		// Create publication
		$publicationNode = $out->createElementNS('http://pkp.sfu.ca', "publication");
		$publicationNode->setAttribute("xmlns:xsi", 'http://www.w3.org/2001/XMLSchema-instance');
    		$publicationNode->setAttribute('xsi:schemaLocation', "http://pkp.sfu.ca native.xsd");
    		$publicationNode->setAttribute("locale", $articleLocale);
    		$publicationNode->setAttribute("version", "1");
    		$publicationNode->setAttribute("status", "3");
    		$publicationNode->setAttribute("primary_contact_id", $authId + 1);
    		$publicationNode->setAttribute("seq", $artSeq);
    		$publicationNode->setAttribute("date_submitted", $dateSubmitted);
    		$publicationNode->setAttribute("date_published", $datePublished);
    		$publicationNode->setAttribute("section_ref", 'ART');
    		$publicationNode->setAttribute("access_status", "0");

		// Publication id
		$publicationIdNode = $out->createElement('id', $rowArt['article_id']);
    		$publicationIdNode->setAttribute("type", "internal");
    		$publicationIdNode->setAttribute("advice", "ignore");
		$publicationNode->appendChild($publicationIdNode);

		// Title
		$data = $rowArt['article_title'];
		$articleTitleNode = $out->createElement('title');
		$cdata = $out->createCDATASection($data);
		$articleTitleNode->appendChild($cdata);
    		$articleTitleNode->setAttribute("locale", "en_US");
		$publicationNode->appendChild($articleTitleNode);

		// Abstract
		$data = $rowArt['article_abstract'];
		$data = preg_replace('/<[\/]*a[^>]*>/i', '', $data);
		$articleAbstractNode = $out->createElement('abstract');
		$cdata = $out->createCDATASection($data);
		$articleAbstractNode->appendChild($cdata);
    		$articleAbstractNode->setAttribute("locale", "en_US");
		$publicationNode->appendChild($articleAbstractNode);

		// Copyright
		$articleCopyrightHolderEnNode = $out->createElement('copyrightHolder');
    		$articleCopyrightHolderEnNode->setAttribute("locale", "en_US");
		$publicationNode->appendChild($articleCopyrightHolderEnNode);

		$articleCopyrightYearNode = $out->createElement('copyrightYear', $year);
		$publicationNode->appendChild($articleCopyrightYearNode);

		// Keywords
		$data = $rowArt['article_key_words'];
		$data = str_replace("\r", "",  $data);
		$data = str_replace("\n", "",  $data);
		$keyWords = preg_split('/,/', $data);
//		print_r($keyWords);
		$keywordsNode = $out->createElement("keywords");
		$keywordsNode->setAttribute("locale", "en_US");
		foreach($keyWords as $keyword){
			$keyword = trim($keyword, " .,");
			$keywordNode = $out->createElement("keyword", $keyword);
			$keywordsNode->appendChild($keywordNode);
		}
		$publicationNode->appendChild($keywordsNode);
		
		// Authors
		$authorsNode = $out->createElementNS('http://pkp.sfu.ca', "authors");
		$authorsNode->setAttribute("xmlns:xsi", 'http://www.w3.org/2001/XMLSchema-instance');
    		$authorsNode->setAttribute('xsi:schemaLocation', "http://pkp.sfu.ca native.xsd");
    		$authSeq = 0;

		$sql = "SELECT distinct a.author_name as name,  a.author_institute as author_institute, a.author_email as author_email, a.author_orcid as author_orcid, b.author_order as author_order
                                       FROM authors a, authors_to_article b
                                       WHERE b.article_id='" . $rowArt["article_id"] . "' and b.author_id=a.author_id
                                       ORDER BY author_order";
		$resAuth = mysqli_query($m, $sql);
		if (mysqli_num_rows($resAuth) == 0) {
			echo "\nNo authors for " . $rowArt["article_id"] . "!\n";
			die;
		}
		$authorsEn = [];

		$affil = $rowArt['article_affiliation'];
		$affil = str_replace("\r", "",  $affil);
		$affil = str_replace("\n", "",  $affil);
//		$affil = html_entity_decode($affil);
//		echo $affil;
		$aaArray = preg_split ("|</p><p>|" , $affil);
		echo  "\n\n";
		$aus = [];
		$afs = [];
		foreach ($aaArray as $auAndAf){
//		    echo $auAndAf;
		    $auAndAf = preg_replace("|(<p>)?(.+?)(</p>)?|", "$2",  $auAndAf);
		    $arr = preg_split("|<br />|", $auAndAf);
		    $aus[] = strip_tags($arr[0]);
		    $afs[] =  $arr[1];
		}
		
		while ($rowAuth = mysqli_fetch_assoc($resAuth)) {
			$authSeq++;
			$authId++;
			$nameEn = $rowAuth['name'];
			array_push($authorsEn, $nameEn);
			
			$givenNameEn = "";

			$nameParts = explode(",", $nameEn);
			$familyNameEn = array_shift($nameParts);
			if($nameParts[0] == "Board") $familyNameEn = $familyNameEn . array_shift($nameParts);
			$givenNameEn .= implode(",", $nameParts);
			$givenNameEn = trim($givenNameEn);
			if ($givenNameEn == ""){
				$givenNameEn = $familyNameEn;
				$familyNameEn = "";
			}
			if ($givenNameEn == ""){
				echo "No English name for " . $rowArt['article_id'];
				die;
			}
			$authorNode = $out->createElement('author');
//			$authorNode->setAttribute('primary_contact', 'true');
			$authorNode->setAttribute('include_in_browse', 'true');
			$authorNode->setAttribute('user_group_ref', 'Author');
			$authorNode->setAttribute('seq', $authSeq);
			$authorNode->setAttribute('id', $authId);

			$givenNameNode = $out->createElement('givenname', $givenNameEn);
			$givenNameNode->setAttribute('locale','en_US');
			$authorNode->appendChild($givenNameNode);
			$familyNameNode = $out->createElement('familyname', $familyNameEn);
			$familyNameNode->setAttribute('locale','en_US');
			$authorNode->appendChild($familyNameNode);



			// Affiliation
			echo $givenNameEn . " " . $familyNameEn . " = " . array_shift($aus) . "\n";
			$affiliation = array_shift($afs);
			$affiliationNode = $out->createElement('affiliation');
			$cdata = $out->createCDATASection($affiliation);
			$affiliationNode->appendChild($cdata);
			$affiliationNode->setAttribute('locale','en_US');
			$authorNode->appendChild($affiliationNode);
			
			// Country
//			$countryNode = $out->createElement('country', 'UA');
//			$authorNode->appendChild($countryNode);
			
			// Email
			$emailNode = $out->createElement('email', $rowAuth['author_email']);
			$authorNode->appendChild($emailNode);

			$emailNode = $out->createElement('orcid', 'https://orcid.org/'.$rowAuth['author_orcid']);
			if ($rowAuth['author_orcid']) $authorNode->appendChild($emailNode);

			$authorsNode->appendChild($authorNode);

		}
		
		$articleCopyrightHolderEnNode->nodeValue = implode("; ", $authorsEn);
		$publicationNode->appendChild($authorsNode);

		
		// Article galleys
		$galleyNode = $out->createElementNS('http://pkp.sfu.ca', 'article_galley');
		$galleyNode->setAttribute("xmlns:xsi", 'http://www.w3.org/2001/XMLSchema-instance');
    		$galleyNode->setAttribute('xsi:schemaLocation', "http://pkp.sfu.ca native.xsd");
		$galleyNode->setAttribute('approved', 'true');
		
		$galleyIdNode = $out->createElement('id', $rowArt['article_id'] . '.0');
		$galleyIdNode->setAttribute('type', 'internal');
		$galleyIdNode->setAttribute('advice', 'ignore');
		$galleyNode->appendChild($galleyIdNode);

		$galleyNameNode = $out->createElement('name', 'PDF');
		$galleyNameNode->setAttribute('locale', $articleLocale);
		$galleyNode->appendChild($galleyNameNode);
		
		$galleySeqNode = $out->createElement('seq', '0');
		$galleyNode->appendChild($galleySeqNode);
		
		$galleySubNode = $out->createElement('submission_file_ref');
		$galleySubNode->setAttribute('id', $rowArt['article_id']);
		$galleySubNode->setAttribute('revision', '1');
		$galleyNode->appendChild($galleySubNode);
    		
    		if($rowArt['article_pdf']) $publicationNode->appendChild($galleyNode);
		

		// Pages
		$pagesNode = $out->createElement('pages', $rowArt['article_pages']);
		$publicationNode->appendChild($pagesNode);



		$articleNode->appendChild($publicationNode);
		$articlesNode->appendChild($articleNode);
		$articlesNode->appendChild($articleNode);

		// Citations
		$data = $rowArt['article_references'];
		$data = str_replace("\r", "",  $data);
		$data = str_replace("\n", "",  $data);
		$cite = new DOMDocument();
		if($data){
			$cite->loadHTML($data);
			$xpath = new DOMXpath($cite);
			$citations = $xpath->query("//*/cite");
//			print_r($citations);
			$citationsNode = $out->createElement("citations");
			foreach($citations as $citation){
				$data = $citation->C14N();
				$data = preg_replace("|<cite[^>]*>(.*?)</cite>|", "$1", $data);
				$cdata = $out->createCDATASection($data);
				$citationNode = $out->createElement("citation");
				$citationNode->appendChild($cdata);
				$citationsNode->appendChild($citationNode);
			}
			$publicationNode->appendChild($citationsNode);
		}


}


	$issueNode->appendChild($articlesNode);

	
	$rootNode->appendChild($issueNode);
}

$rootNode->setAttributeNS('http://www.w3.org/2000/xmlns/', "xmlns:xsi" ,  'http://www.w3.org/2001/XMLSchema-instance');
$rootNode->setAttribute('xsi:schemaLocation', "http://pkp.sfu.ca native.xsd");
$out->appendChild($rootNode);

$out->save($outFile);
// echo $out->saveXML();

die;

?>