Merged in from DEV trunk...

KTS-3233 "If you use a (-) in a lookup fieldset name, the browse by lookup value returns all documents associated to the fieldset." Fixed. The (-) was being replaced by an (_). Committed by: Megan Watson Reviewed by: Conrad Vermeulen KTS-3165 "Remove background task permissions problems by making web server user deal with background tasks" Fixed. Added some debug logging Committed By: Conrad Vermeulen Reviewed By: Kevin Fourie KTS-3235 "Indexer unable to index a MS PPT 2003 document" Fixed. Committed By: Conrad Vermeulen Reviewed By: Kevin Fourie KTS-3236 "Indexing process should skip temporary files as the indexer process is likely to fail resulting in users being confused." Fixed. Committed By: Conrad Vermeulen Reviewed By: Kevin Fourie KTS-3237 "Document Indexer dashlet shows the number of documents in the queue to include the problem documents in the queue" Fixed. Committed By: Conrad Vermeulen Reviewed By: Kevin Fourie git-svn-id: https://kt-dms.svn.sourceforge.net/svnroot/kt-dms/STABLE/trunk@8348 c91229c3-7414-0410-bfa2-8a42b809f60b

Merged in from DEV trunk...
KTS-3233 "If you use a (-) in a lookup fieldset name, the browse by lookup value returns all documents associated to the fieldset." Fixed. The (-) was being replaced by an (_). Committed by: Megan Watson Reviewed by: Conrad Vermeulen KTS-3165 "Remove background task permissions problems by making web server user deal with background tasks" Fixed. Added some debug logging Committed By: Conrad Vermeulen Reviewed By: Kevin Fourie KTS-3235 "Indexer unable to index a MS PPT 2003 document" Fixed. Committed By: Conrad Vermeulen Reviewed By: Kevin Fourie KTS-3236 "Indexing process should skip temporary files as the indexer process is likely to fail resulting in users being confused." Fixed. Committed By: Conrad Vermeulen Reviewed By: Kevin Fourie KTS-3237 "Document Indexer dashlet shows the number of documents in the queue to include the problem documents in the queue" Fixed. Committed By: Conrad Vermeulen Reviewed By: Kevin Fourie git-svn-id: https://kt-dms.svn.sourceforge.net/svnroot/kt-dms/STABLE/trunk@8348 c91229c3-7414-0410-bfa2-8a42b809f60b
kevin_fourie
1 parent 39180d5f
Showing 5 changed files with 35 additions and 16 deletions
lib/browse/Criteria.inc
lib/util/ktutil.inc
plugins/search2/LuceneStatisticsDashlet.php
search2/indexing/extractorCore.inc.php
search2/indexing/indexerCore.inc.php
@@ -242,7 +242,8 @@ class BrowseCriterion {
     }
     function getWidgetBase () {
-        return strtr($this->getNamespace(), '-', '_');
+        //return strtr($this->getNamespace(), '-', '_');
+        return $this->getNamespace();
     }
     function getSearchTable() {
@@ -731,21 +732,21 @@ class SizeCriterion extends BrowseCriterion {
 			'M'=>'Megabytes');
     var $aTypeAssocs = array('B' => 1, 'KB' => 1024, 'M' => 1048576);
-    
+
     var $aCmps = array();
     var $aCmpAssocs = array();
     function SizeCriterion() {
 		$this->sDisplay = _kt('File Size');
-		
+
 		$this->aCmps = array('LT'=>_kt('Less than'),
 		       'GT'=>_kt('Greater than'),
 		       'EQ'=>_kt('Equal to'),
 		       'NEQ'=>_kt('Not equal to'));
-		       
+
 		$this->aCmpAssocs = array('LT' => '<', 'GT' => '>', 'EQ' => '=', 'NEQ' => '!=');
-		
+
     }
     function documentDisplay ($oDocument) {
@@ -928,7 +929,7 @@ class DateCreatedDeltaCriterion extends DateCreatedCriterion {
     function DateCreatedDeltaCriterion() {
 		$this->sDisplay = _kt('Date Created Delta');
-		
+
 		$this->aTypes = array('MINUTE'=>_kt('Minutes'),
 			'HOUR'=>_kt('Hours'),
 			'DAY'=>_kt('Days'),
@@ -107,19 +107,20 @@ class KTUtil {
 		$cacheDir = $config->get('cache/cacheDirectory');
 		$base_url = @file_get_contents($cacheDir . '/' . KTUtil::SERVER_NAME_FILE);
+		global $default;
 		if (false == $base_url)
 		{
-			global $default;
 			$default->log->info("call_page: $path - cannot call script until user logs in for the first time!");
 			return;
 		}
-
 		$kt_url = KTUtil::kt_url();
 		$full_url = $kt_url . '/' . $path;
+		$default->log->debug("call_page: calling $full_url");
+
 		$ch = curl_init($full_url);
 		curl_setopt($ch,CURLOPT_SSL_VERIFYPEER, false);
 		curl_setopt($ch,CURLOPT_SSL_VERIFYHOST, false);
@@ -101,7 +101,7 @@ class LuceneStatisticsDashlet extends KTBaseDashlet
 			$docsInIndex = $index->getDocumentsInIndex();
 			// we are only interested in documents that are active
-			$docsInQueue = $index->getIndexingQueue();
+			$docsInQueue = $index->getIndexingQueue(false);
 			$docsInQueue = count($docsInQueue);
 			$errorsInQueue = $index->getIndexingQueue(true);
@@ -631,21 +631,29 @@ abstract class CompositeExtractor extends DocumentExtractor
 		$this->sourceExtractor->setSourceFile($this->sourcefile);
 		$this->sourceExtractor->setTargetFile($intermediateFile);
+		$this->sourceExtractor->setDocument($this->getDocument());
 		$this->sourceExtractor->setMimeType($this->mimetype);
 		$this->sourceExtractor->setExtension($this->extension);
 		if (!$this->sourceExtractor->extractTextContent())
 		{
+			$this->output = $this->sourceExtractor->output;
+			@unlink($intermediateFile);
 			return false;
 		}
 		$intermediateFile = $this->sourceExtractor->getTargetFile();
 		$this->targetExtractor->setSourceFile($intermediateFile);
 		$this->targetExtractor->setTargetFile($this->targetfile);
+		$this->targetExtractor->setDocument($this->getDocument());
 		$this->targetExtractor->setMimeType($this->targetMimeType);
 		$this->targetExtractor->setExtension($this->targetExtension);
 		$result = $this->targetExtractor->extractTextContent();
+		if (!$result)
+		{
+			$this->output = $this->targetExtractor->output;
+		}
-		unlink(@$intermediateFile);
+		@unlink($intermediateFile);
 		return $result;
 	}
@@ -1012,6 +1012,20 @@ abstract class Indexer
         		$this->logPendingDocumentInfoStatus($docId, sprintf(_kt("Processing docid: %d.\n"),$docId), 'info');
         	}
+        	$document = Document::get($docId);
+        	if (PEAR::isError($document))
+        	{
+        		Indexer::unqueueDocument($docId,sprintf(_kt("indexDocuments: Cannot resolve document id %d: %s."),$docId, $document->getMessage()), 'error');
+        		continue;
+        	}
+
+        	$filename = $document->getFileName();
+        	if (substr($filename,0,1) == '~')
+        	{
+        		Indexer::unqueueDocument($docId,sprintf(_kt("indexDocuments: Filename for document id %d starts with a tilde (~). This is assumed to be a temporary file. This is ignored."),$docId), 'error');
+        		continue;
+        	}
+
         	$removeFromQueue = true;
         	if ($indexDocument)
         	{
@@ -1030,12 +1044,7 @@ abstract class Indexer
 					continue;
 				}
-        		$document = Document::get($docId);
-        		if (PEAR::isError($document))
-        		{
-        			Indexer::unqueueDocument($docId,sprintf(_kt("indexDocuments: Cannot resolve document id %d: %s."),$docId, $document->getMessage()), 'error');
-        			continue;
-        		}
+
         		$version = $document->getMajorVersionNumber() . '.' . $document->getMinorVersionNumber();
         		$sourceFile = $storageManager->temporaryFile($document);