#20 extract Mendeley references from LibreOffice's OOXML

src/docx2jats/DOCXArchive.php

@@ -34,6 +34,7 @@ public function __construct(string $filepath) {
 			$styles = $this->transformToXml("word/styles.xml");
 			$this->mediaFiles = $this->extractMediaFiles();
 			$numbering = $this->transformToXml("word/numbering.xml");
+			$docPropsCustom = $this->transformToXml("docProps/custom.xml");
 			$this->close();
 
 			// construct as an array
@@ -42,10 +43,9 @@ public function __construct(string $filepath) {
 			$params["ooxmlDocument"] = $this->ooxmlDocument;
 
 			if ($relationships) $params["relationships"] = $relationships;
-
 			if ($styles) $params["styles"] = $styles;
-
 			if ($numbering) $params["numbering"] = $numbering;
+			if ($docPropsCustom) $params["docPropsCustom"] = $docPropsCustom;
 
 			$document = new Document($params);
 

src/docx2jats/jats/Par.php

@@ -31,7 +31,7 @@ public function setContent() {
 					foreach ($content->getRefIds() as $key => $id) {
 						$refEl = $this->ownerDocument->createElement('xref', $id);
 						$refEl->setAttribute('ref-type', 'bibr');
-						$refEl->setAttribute('rid', 'bib' . $id);
+						$refEl->setAttribute('rid', Reference::JATS_REF_ID_PREFIX . $id);
 						$this->appendChild($refEl);
 						if ($key !== $lastKey) {
 							$refEl = $this->ownerDocument->createTextNode(' ');

src/docx2jats/jats/Table.php

@@ -6,7 +6,7 @@
  * Copyright (c) 2018-2020 Vitalii Bezsheiko
  * Distributed under the GNU GPL v3.
  *
- * @brief represent JATS XML table
+ * @brief representі JATS XML table
  */
 
 use docx2jats\objectModel\DataObject;

src/docx2jats/objectModel/Document.php

@@ -41,6 +41,13 @@ class Document {
 	private $numbering;
 	static $numberingXpath;
 
+	/**
+	 * @var $docPropsCustom \DOMDocument represents custom properties of the document,
+	 * e.g., Mendeley plugin for LibreOffice Writer exports CSL in this file
+	 */
+	private $docPropsCustom;
+	static $docPropsCustomXpath;
+
 	private $references = array();
 	private $refCount = 0;
 
@@ -53,10 +60,15 @@ class Document {
 	 * @brief Key numbers of paragraphs that contain bookmarks inside the content
 	 * is used to speed up a search
 	 */
-	private $elsHaveBookmarks = array();
+	private $elsHavefldCharRefs = array();
 	private $elsAreTables = array();
 	private $elsAreFigures = array();
 
+	/**
+	 * @var array bookmark id => name
+	 */
+	public $bookMarks = array();
+
 	public function __construct(array $params) {
 		if (array_key_exists("relationships", $params)) {
 			$this->relationships = $params["relationships"];
@@ -73,7 +85,13 @@ public function __construct(array $params) {
 			self::$numberingXpath = new \DOMXPath($this->numbering);
 		}
 
+		if (array_key_exists("docPropsCustom", $params)) {
+			$this->docPropsCustom = $params["docPropsCustom"];
+			self::$docPropsCustomXpath = new \DOMXPath($this->docPropsCustom);
+		}
+
 		self::$xpath = new \DOMXPath($params["ooxmlDocument"]);
+		$this->findBookmarks();
 
 		$childNodes = self::$xpath->query("//w:body/child::node()");
 
@@ -141,7 +159,7 @@ public function __construct(array $params) {
 						}
 
 						if ($par->hasBookmarks) {
-							$this->elsHaveBookmarks[] = count($content)-1;
+							$this->elsHavefldCharRefs[] = count($content)-1;
 						}
 					}
 					break;
@@ -337,29 +355,29 @@ public function getLastReference() : ?Reference {
 	 * it's slightly faster than looping over the whole content
 	 */
 	private function setInternalRefs(): void {
-		if (empty($this->elsHaveBookmarks)) return;
+		if (empty($this->elsHavefldCharRefs)) return;
 
 		// Find and map tables' and figures' bookmarks
 		$refTableMap = $this->getBookmarkCaptionMapping($this->elsAreTables);
 		$refFigureMap = $this->getBookmarkCaptionMapping($this->elsAreFigures);
 
 		// Find bookmark refs
-		foreach ($this->elsHaveBookmarks as $parKeyWithBookmark) {
+		foreach ($this->elsHavefldCharRefs as $parKeyWithBookmark) {
 			$par = $this->getContent()[$parKeyWithBookmark]; /* @var $par Par */
-			foreach ($par->bookmarkPos as $fieldKeyWithBookmark) {
+			foreach ($par->fldCharRefPos as $fieldKeyWithBookmark) {
 				$field = $par->getContent()[$fieldKeyWithBookmark]; /* @var $field \docx2jats\objectModel\body\Field */
 
 				// Set links to tables
 				foreach ($refTableMap as $tableId => $tableRefs) {
-					if (in_array($field->getBookmarkId(), $tableRefs)) {
+					if (in_array($field->getFldCharRefId(), $tableRefs)) {
 						$field->tableIdRef = $tableId;
 					}
 				}
 
 				// Set links to Figures
 				foreach ($refFigureMap as $figureId => $figureRefs) {
-					if (in_array($field->getBookmarkId(), $figureRefs)) {
-						$field->figureIdRef = $tableId;
+					if (in_array($field->getFldCharRefId(), $figureRefs)) {
+						$field->figureIdRef = $figureId;
 					}
 				}
 			}
@@ -368,7 +386,7 @@ private function setInternalRefs(): void {
 
 	/**
 	 * @return array
-	 * @brief (or not so brief) Map OOXML bookmark refs inside table and figures with correspondent table/figure IDs.
+	 * @brief (or not so brief) Map OOXML bookmark refs inside tables and figures with correspondent table/figure IDs.
 	 * In OOXML those bookmarks are stored inside captions
 	 * This is used to set right link to these objects from the text
 	 * Keep in mind that bookmarks also may be stored in an external file, e.g., Mendeley plugin for LibreOffice Writer
@@ -384,4 +402,18 @@ function getBookmarkCaptionMapping(array $keysInContent): array {
 
 		return $refMap;
 	}
+
+	/**
+	 * Find and retrieve id and name from all bookmarks in the main document part
+	 */
+	private function findBookmarks(): void {
+		$bookmarkEls = self::$xpath->query('//w:bookmarkStart');
+		foreach ($bookmarkEls as $bookmarkEl) {
+			$this->bookMarks[$bookmarkEl->getAttribute('w:id')] = $bookmarkEl->getAttribute('w:name');
+		}
+	}
+
+	public function docPropsCustom() {
+		return $this->docPropsCustom;
+	}
 }

src/docx2jats/objectModel/body/Field.php

@@ -26,7 +26,7 @@ class Field extends DataObject {
 	private $rawRuns = array();
 	/** @var array contains instructions to be processed as strings, e.g., CSL citations as a JSON string */
 	private $instructions = array();
-	private $bookmarkId;
+	private $fldCharRefId;
 
 	/**
 	 * @var $tableIdRef int
@@ -112,11 +112,11 @@ private function processRuns() {
 					elseif (strpos($instructionString, 'REF') !== false) {
 						$this->getParent()->hasBookmarks = true;
 						$this->type = self::DOCX_FIELD_BOOKMARK_REF;
-						$this->bookmarkId = $this->extractRefID($instructionString);
+						$this->fldCharRefId = $this->extractRefID($instructionString);
 					}
 				}
 			} else {
-				$this->content[] = new Text($run);
+				$this->content[] = new Text($run, $this->getOwnerDocument());
 			}
 		}
 	}
@@ -129,15 +129,16 @@ public function getType(): int
 		return $this->type;
 	}
 
-	private function extractRawCSL(string $instruction) {
-		$instruction = trim($instruction);
-		$pos = strpos($instruction, '{');
-		$instructionsRawPart = substr($instruction, 0, $pos);
+	/**
+	 * @param string $instruction
+	 * @return string containing raw CSL
+	 * @brief extract CSL as a string and determine its type (Zotero or Mendeley)
+	 */
+	private function extractRawCSL(string $instruction): string {
+		list($instructionsRawPart, $rawCSL) = Reference::extractRawCSL($instruction);
 		if (strpos($instructionsRawPart, 'ZOTERO_ITEM') !== false) {
 			$this->isZoteroCSL = true;
 		}
-
-		$rawCSL = substr($instruction, $pos);
 		return $rawCSL;
 	}
 
@@ -171,9 +172,9 @@ public function isZoteroCSL(): bool {
 	/**
 	 * @return mixed
 	 */
-	public function getBookmarkId() {
+	public function getFldCharRefId() {
 		if ($this->type === self::DOCX_FIELD_BOOKMARK_REF) {
-			return $this->bookmarkId;
+			return $this->fldCharRefId;
 		}
 
 		return null;

src/docx2jats/objectModel/body/Par.php

@@ -54,7 +54,7 @@ class Par extends DataObject {
 	static $numberingUnorderedMarkers = array("bullet", "none", "");
 
 	public $hasBookmarks = false;
-	public $bookmarkPos = array(); // position of bookmarks in the content
+	public $fldCharRefPos = array(); // position of bookmarks in the content
 
 	public function __construct(\DOMElement $domElement, Document $ownerDocument, DataObject $parent = null) {
 		parent::__construct($domElement, $ownerDocument, $parent);
@@ -112,16 +112,16 @@ protected function setContent(string $xpathExpression) {
 					$field->addContent($contentNode);
 
 					// record a position of field with a bookmark in an array
-					if ($field->getBookmarkId()) $this->bookmarkPos[] = count($content)-1;
+					if ($field->getFldCharRefId()) $this->fldCharRefPos[] = count($content)-1;
 					$field = null;
 				} else {
-					$text = new Text($contentNode);
+					$text = new Text($contentNode, $this->getOwnerDocument());
 					$content[] = $text;
 				}
 			} elseif ($contentNode->nodeName === "w:hyperlink") {
 				$children = $this->getXpath()->query('child::node()', $contentNode);
 				foreach ($children as $child) {
-					$href = new Text($child);
+					$href = new Text($child, $this->getOwnerDocument());
 					$href->addType($href::DOCX_TEXT_EXTLINK);
 					$href->setLink();
 					$content[] = $href;

src/docx2jats/objectModel/body/Reference.php

@@ -33,6 +33,18 @@ public function getCslId(): string
 		return $this->cslId;
 	}
 
+	/**
+	 * @param string $instruction
+	 * @return array [instructions command, CSL as a string]
+	 */
+	public static function extractRawCSL(string $instruction): array {
+		$instruction = trim($instruction);
+		$pos = strpos($instruction, '{');
+		$instructionsRawPart = substr($instruction, 0, $pos);
+		$rawCSL = substr($instruction, $pos);
+		return array($instructionsRawPart, $rawCSL);
+	}
+
 	public static function findRefsCSL(string $rawCSL) : array {
 		$citations = [];
 		$json = json_decode($rawCSL);
@@ -71,6 +83,14 @@ public static function findPlainCit(string $rawCSL): ?string {
 		// Mendeley
 		if (property_exists($json, 'mendeley')) {
 			$mendeley = $json->{'mendeley'};
+			if ($props && property_exists($mendeley, 'formattedCitation')) {
+				return $mendeley->{'formattedCitation'};
+			}
+
+			if (property_exists($mendeley, 'previouslyFormattedCitation')) {
+				return $mendeley->{'plainTextFormattedCitation'};
+			}
+
 			if (property_exists($mendeley, 'previouslyFormattedCitation')) {
 				return $mendeley->{'previouslyFormattedCitation'};
 			}