Merge lp:~sebastian-meyer/goobi-presentation/performance-optimization into lp:~sebastian-meyer/goobi-presentation/old-trunk

[Sebastian Meyer]

Fix Bug #733338: Loading of large documents is very slow.

[Sebastian Meyer (sebastian-meyer)]

review: Approve

=== modified file 'dlf/common/class.tx_dlf_document.php'
--- dlf/common/class.tx_dlf_document.php	2011-04-01 15:53:16 +0000
+++ dlf/common/class.tx_dlf_document.php	2011-05-31 10:41:13 +0000
@@ -208,23 +208,6 @@
 	protected static $registry = array ();
 
 	/**
-	 * This holds the smLinks of the document
-	 *
-	 * @var	array
-	 * @access protected
-	 */
-	protected $smLinks = array ();
-
-	/**
-	 * Are the METS file's smLinks loaded?
-	 * @see $smLinks
-	 *
-	 * @var	boolean
-	 * @access protected
-	 */
-	protected $smLinksLoaded = FALSE;
-
-	/**
 	 * This holds the PID for the structure definitions
 	 *
 	 * @var	integer
@@ -320,7 +303,7 @@
 	 *
 	 * @param	mixed		$uid: The unique identifier of the document to parse or URL of XML file
 	 * @param	integer		$pid: If > 0, then only document with this PID gets loaded
-	 * @param	boolean		$forceReload: Force reloading the document
+	 * @param	boolean		$forceReload: Force reloading the document instead of returning the cached instance
 	 *
 	 * @return	tx_dlf_document		Instance of this class
 	 */
@@ -396,8 +379,6 @@
 			// Load physical pages and smLinks.
 			$this->_getPhysicalPages();
 
-			$this->_getSmLinks();
-
 			$_struct = $_div[0];
 
 			$this->registerNamespaces($_struct);
@@ -415,44 +396,28 @@
 
 			$_details['type'] = (string) $_struct['TYPE'];
 
-			// Get the physical pages or external file this structure element is pointing at.
-			$_details['points'] = array ();
+			// Get the physical page or external file this structure element is pointing at.
+			$_details['points'] = '';
 
 			// Is there a mptr node?
 			if (($_mptr = $_struct->xpath('./mets:mptr[@LOCTYPE="URL"]'))) {
 
 				// Yes. Get the file reference.
-				$_details['points'][] = (string) $_mptr[0]->attributes('http://www.w3.org/1999/xlink')->href;
+				$_details['points'] = (string) $_mptr[0]->attributes('http://www.w3.org/1999/xlink')->href;
 
 			// Are there any physical pages?
 			} elseif ($this->physicalPages) {
 
-				// Yes. Get all physical pages related to this structure element.
-				if (!empty($this->smLinks[(string) $_struct['ID']])) {
-
-					foreach ($this->smLinks[(string) $_struct['ID']] as $_smLink) {
-
-						$_key = tx_dlf_helper::array_search_recursive($_smLink, $this->physicalPages);
-
-						// Check if smLink points to the "physSequence" element.
-						if ($_key === 0) {
-
-							$_keys = array_keys($this->physicalPages);
-
-							array_shift($_keys);
-
-							$_details['points'] = $_keys;
-
-						} else {
-
-							$_details['points'][] = $_key;
-
-						}
-
-					}
+				// Yes. Get first physical page related to this structure element.
+				if (($_smLink = $this->mets->xpath('./mets:structLink/mets:smLink[@xlink:from="'.(string) $_struct['ID'].'"]'))) {
+
+					$_details['points'] = tx_dlf_helper::array_search_recursive($_smLink[0]->attributes('http://www.w3.org/1999/xlink')->to, $this->physicalPages);
+
+					// Check if smLink points to the "physSequence" element (in which case it should point to the first image).
+					$_details['points'] = max(intval($_details['points']), 1);
 
 					// Get page number of the first page related to this structure element.
-					$_details['pagination'] = $this->physicalPages[$_details['points'][0]]['label'];
+					$_details['pagination'] = $this->physicalPages[$_details['points']]['label'];
 
 				} else {
 
@@ -665,20 +630,20 @@
 		// Get all logical structure nodes with metadata.
 		if (($_divs = $this->mets->xpath('./mets:structMap[@TYPE="LOGICAL"]//mets:div[@DMDID]'))) {
 
-			$this->_getSmLinks();
-
 			foreach ($_divs as $_div) {
 
+				$_id = (string) $_div['ID'];
+
 				// Are there physical structure nodes for this logical structure?
-				if (!empty($this->smLinks[(string) $_div['ID']])) {
+				if ($this->mets->xpath('./mets:structLink/mets:smLink[@xlink:from="'.$_id.'"]')) {
 
 					// Yes. That's what we're looking for.
-					return (string) $_div['ID'];
+					return $_id;
 
 				} elseif (!$id) {
 
 					// No. Remember this anyway, but keep looking for a better one.
-					$id = (string) $_div['ID'];
+					$id = $_id;
 
 				}
 
@@ -1041,9 +1006,9 @@
 		// Get UID of superior document.
 		$partof = 0;
 
-		if (!empty($this->tableOfContents[0]['points'][0]) && !t3lib_div::testInt($this->tableOfContents[0]['points'][0])) {
+		if (!empty($this->tableOfContents[0]['points']) && !t3lib_div::testInt($this->tableOfContents[0]['points'])) {
 
-			$superior = tx_dlf_document::getInstance($this->tableOfContents[0]['points'][0]);
+			$superior = tx_dlf_document::getInstance($this->tableOfContents[0]['points']);
 
 			if ($superior->pid != $pid) {
 
@@ -1417,37 +1382,6 @@
 	}
 
 	/**
-	 * This builds an array of all smLinks of the document
-	 *
-	 * @access	protected
-	 *
-	 * @return	array		The smLinks of the document
-	 */
-	protected function _getSmLinks() {
-
-		if (!$this->smLinksLoaded) {
-
-			$_smLinks = $this->mets->xpath('./mets:structLink/mets:smLink');
-
-			if ($_smLinks) {
-
-				foreach ($_smLinks as $_smLink) {
-
-					$this->smLinks[(string) $_smLink->attributes('http://www.w3.org/1999/xlink')->from][] = (string) $_smLink->attributes('http://www.w3.org/1999/xlink')->to;
-
-				}
-
-			}
-
-			$this->smLinksLoaded = TRUE;
-
-		}
-
-		return $this->smLinks;
-
-	}
-
-	/**
 	 * This returns $this->sPid via __get()
 	 *
 	 * @access	protected

=== modified file 'dlf/common/class.tx_dlf_indexing.php'
--- dlf/common/class.tx_dlf_indexing.php	2011-04-01 15:53:16 +0000
+++ dlf/common/class.tx_dlf_indexing.php	2011-05-31 10:41:13 +0000
@@ -354,7 +354,7 @@
 
 			$solrDoc->setField('pid', $doc->pid);
 
-			$solrDoc->setField('page', $logicalUnit['points'][0]);
+			$solrDoc->setField('page', $logicalUnit['points']);
 
 			$solrDoc->setField('partof', $doc->parentid);
 

=== modified file 'dlf/plugins/metadata/class.tx_dlf_metadata.php'
--- dlf/plugins/metadata/class.tx_dlf_metadata.php	2011-04-01 15:53:16 +0000
+++ dlf/plugins/metadata/class.tx_dlf_metadata.php	2011-05-31 10:41:13 +0000
@@ -226,7 +226,7 @@
 
 							}
 
-							$_value = $this->pi_linkTP(htmlspecialchars($_value), array ($this->prefixId => array ('id' => $this->doc->uid, 'page' => (!empty($details['points']) ? intval($details['points'][0]) : 1))), TRUE, $this->conf['targetPid']);
+							$_value = $this->pi_linkTP(htmlspecialchars($_value), array ($this->prefixId => array ('id' => $this->doc->uid, 'page' => (!empty($details['points']) ? intval($details['points']) : 1))), TRUE, $this->conf['targetPid']);
 
 						// Translate name of holding library.
 						} elseif ($_index_name == 'owner') {

=== modified file 'dlf/plugins/toc/class.tx_dlf_toc.php'
--- dlf/plugins/toc/class.tx_dlf_toc.php	2011-04-01 15:53:16 +0000
+++ dlf/plugins/toc/class.tx_dlf_toc.php	2011-05-31 10:41:13 +0000
@@ -40,6 +40,14 @@
 	public $scriptRelPath = 'plugins/toc/class.tx_dlf_toc.php';
 
 	/**
+	 * This holds the active entries according to the currently selected page
+	 *
+	 * @var	array
+	 * @access protected
+	 */
+	protected $activeEntries = array ();
+
+	/**
 	 * This builds an array for one menu entry
 	 *
 	 * @access	protected
@@ -67,15 +75,15 @@
 		$entryArray['ITEM_STATE'] = 'NO';
 
 		// Build menu links based on the $entry['points'] array.
-		if (!empty($entry['points'][0]) && t3lib_div::testInt($entry['points'][0])) {
-
-			$entryArray['_OVERRIDE_HREF'] = $this->pi_linkTP_keepPIvars_url(array ('page' => $entry['points'][0]), TRUE, FALSE, $this->conf['targetPid']);
-
-		} elseif (!empty($entry['points'][0]) && is_string($entry['points'][0])) {
-
-			$_doc = tx_dlf_document::getInstance($entry['points'][0], ($this->conf['excludeOther'] ? $this->conf['pages'] : 0));
-
-			$entryArray['_OVERRIDE_HREF'] = $this->pi_linkTP_keepPIvars_url(array ('id' => ($_doc->pid ? $_doc->uid : $entry['points'][0]), 'page' => 1), TRUE, FALSE, $this->conf['targetPid']);
+		if (!empty($entry['points']) && t3lib_div::testInt($entry['points'])) {
+
+			$entryArray['_OVERRIDE_HREF'] = $this->pi_linkTP_keepPIvars_url(array ('page' => $entry['points']), TRUE, FALSE, $this->conf['targetPid']);
+
+		} elseif (!empty($entry['points']) && is_string($entry['points'])) {
+
+			$_doc = tx_dlf_document::getInstance($entry['points'], ($this->conf['excludeOther'] ? $this->conf['pages'] : 0));
+
+			$entryArray['_OVERRIDE_HREF'] = $this->pi_linkTP_keepPIvars_url(array ('id' => ($_doc->pid ? $_doc->uid : $entry['points']), 'page' => 1), TRUE, FALSE, $this->conf['targetPid']);
 
 		} elseif (!empty($entry['points']['doc'])) {
 
@@ -88,7 +96,7 @@
 		}
 
 		// Set "ITEM_STATE" to "CUR" if this entry points to current page.
-		if (!empty($this->piVars['page']) && in_array($this->piVars['page'], $entry['points'])) {
+		if (in_array($entry['id'], $this->activeEntries)) {
 
 			$entryArray['ITEM_STATE'] = 'CUR';
 
@@ -102,14 +110,14 @@
 			// 2. Current menu node is in rootline
 			// 3. Current menu node points to another file
 			// 4. There are no physical pages in the current METS file
-			if (!empty($this->conf['menuConf.']['expAll']) || $entryArray['ITEM_STATE'] == 'CUR' || is_string($entry['points'][0]) || !$this->doc->physicalPages) {
+			if (!empty($this->conf['menuConf.']['expAll']) || $entryArray['ITEM_STATE'] == 'CUR' || is_string($entry['points']) || !$this->doc->physicalPages) {
 
 				$entryArray['_SUB_MENU'] = array ();
 
 				foreach ($entry['children'] as $_child) {
 
 					// Set "ITEM_STATE" to "ACT" if this entry points to current page and has sub-entries pointing to the same page.
-					if (!empty($this->piVars['page']) && in_array($this->piVars['page'], $_child['points'])) {
+					if (in_array($_child['id'], $this->activeEntries)) {
 
 						$entryArray['ITEM_STATE'] = 'ACT';
 
@@ -216,6 +224,17 @@
 		// Does the document have physical pages or is it an external file?
 		if ($this->doc->physicalPages || !t3lib_div::testInt($this->doc->uid)) {
 
+			// Get all logical units the current page is a part of.
+			if (!empty($this->piVars['page']) && $this->doc->physicalPages) {
+
+				foreach ($this->doc->mets->xpath('./mets:structLink/mets:smLink[@xlink:to="'.$this->doc->physicalPages[$this->piVars['page']]['id'].'"]') as $_logId) {
+
+					$this->activeEntries[] = (string) $_logId->attributes('http://www.w3.org/1999/xlink')->from;
+
+				}
+
+			}
+
 			// Go through table of contents and create all menu entries.
 			foreach ($this->doc->tableOfContents as $_entry) {