Fix bug #49386 - avoid NPE when extracting OOXML file properties which are dates

Also tidy up POIXMLPropertiesTextExtractor to exclude properties which are missing


git-svn-id: https://svn.apache.org/repos/asf/poi/trunk@951384 13f79535-47bb-0310-9956-ffa450edef68

5 files changed, 87 insertions, 41 deletions

diff --git a/src/documentation/content/xdocs/status.xml b/src/documentation/content/xdocs/status.xml
index 4ec3ca5ba3..c29e9f9ee4 100644
--- a/src/documentation/content/xdocs/status.xml
+++ b/src/documentation/content/xdocs/status.xml
@@ -34,6 +34,7 @@
 
     <changes>
         <release version="3.7-SNAPSHOT" date="2010-??-??">
+           <action dev="POI-DEVELOPERS" type="fix">49386 - avoid NPE when extracting OOXML file properties which are dates</action>
            <action dev="POI-DEVELOPERS" type="fix">49377 - only call DecimalFormat.setRoundingMode on Java 1.6 - it's needed to match excel's rendering of numbers</action>
            <action dev="POI-DEVELOPERS" type="fix">49378 - correct 1.6ism</action>
            <action dev="POI-DEVELOPERS" type="add">Parse the HSMF headers chunk if present, and use it to find Dates in text extraction if needed</action>
diff --git a/src/ooxml/java/org/apache/poi/POIXMLPropertiesTextExtractor.java b/src/ooxml/java/org/apache/poi/POIXMLPropertiesTextExtractor.java
index 47fe05ed91..6d31340557 100644
--- a/src/ooxml/java/org/apache/poi/POIXMLPropertiesTextExtractor.java
+++ b/src/ooxml/java/org/apache/poi/POIXMLPropertiesTextExtractor.java
@@ -20,6 +20,8 @@ package org.apache.poi;
 import org.apache.poi.openxml4j.opc.internal.PackagePropertiesPart;
 import org.openxmlformats.schemas.officeDocument.x2006.customProperties.CTProperty;
 
+import java.util.Date;
+
 /**
  * A {@link POITextExtractor} for returning the textual
  *  content of the OOXML file properties, eg author
@@ -41,6 +43,24 @@ public class POIXMLPropertiesTextExtractor extends POIXMLTextExtractor {
 	public POIXMLPropertiesTextExtractor(POIXMLTextExtractor otherExtractor) {
 		super(otherExtractor.getDocument());
 	}
+	
+   private void appendIfPresent(StringBuffer text, String thing, boolean value) {
+      appendIfPresent(text, thing, Boolean.toString(value));
+   }
+   private void appendIfPresent(StringBuffer text, String thing, int value) {
+      appendIfPresent(text, thing, Integer.toString(value));
+   }
+   private void appendIfPresent(StringBuffer text, String thing, Date value) {
+      if(value == null) { return; }
+      appendIfPresent(text, thing, value.toString());
+   }
+	private void appendIfPresent(StringBuffer text, String thing, String value) {
+	   if(value == null) { return; }
+	   text.append(thing);
+	   text.append(" = ");
+	   text.append(value);
+	   text.append("\n");
+	}
 
 	/**
 	 * Returns the core document properties, eg author
@@ -50,25 +70,26 @@ public class POIXMLPropertiesTextExtractor extends POIXMLTextExtractor {
 		PackagePropertiesPart props =
 			getDocument().getProperties().getCoreProperties().getUnderlyingProperties();
 
-		text.append("Category = " + props.getCategoryProperty().getValue() + "\n");
-		text.append("ContentStatus = " + props.getContentStatusProperty().getValue() + "\n");
-		text.append("ContentType = " + props.getContentTypeProperty().getValue() + "\n");
-		text.append("Created = " + props.getCreatedProperty().getValue() + "\n");
-		text.append("CreatedString = " + props.getCreatedPropertyString() + "\n");
-		text.append("Creator = " + props.getCreatorProperty().getValue() + "\n");
-		text.append("Description = " + props.getDescriptionProperty().getValue() + "\n");
-		text.append("Identifier = " + props.getIdentifierProperty().getValue() + "\n");
-		text.append("Keywords = " + props.getKeywordsProperty().getValue() + "\n");
-		text.append("Language = " + props.getLanguageProperty().getValue() + "\n");
-		text.append("LastModifiedBy = " + props.getLastModifiedByProperty().getValue() + "\n");
-		text.append("LastPrinted = " + props.getLastPrintedProperty().getValue() + "\n");
-		text.append("LastPrintedString = " + props.getLastPrintedPropertyString() + "\n");
-		text.append("Modified = " + props.getModifiedProperty().getValue() + "\n");
-		text.append("ModifiedString = " + props.getModifiedPropertyString() + "\n");
-		text.append("Revision = " + props.getRevisionProperty().getValue() + "\n");
-		text.append("Subject = " + props.getSubjectProperty().getValue() + "\n");
-		text.append("Title = " + props.getTitleProperty().getValue() + "\n");
-		text.append("Version = " + props.getVersionProperty().getValue() + "\n");
+		appendIfPresent(text, "Category", props.getCategoryProperty().getValue());
+		appendIfPresent(text, "Category", props.getCategoryProperty().getValue());
+		appendIfPresent(text, "ContentStatus", props.getContentStatusProperty().getValue());
+		appendIfPresent(text, "ContentType", props.getContentTypeProperty().getValue());
+		appendIfPresent(text, "Created", props.getCreatedProperty().getValue());
+		appendIfPresent(text, "CreatedString", props.getCreatedPropertyString());
+		appendIfPresent(text, "Creator", props.getCreatorProperty().getValue());
+		appendIfPresent(text, "Description", props.getDescriptionProperty().getValue());
+		appendIfPresent(text, "Identifier", props.getIdentifierProperty().getValue());
+		appendIfPresent(text, "Keywords", props.getKeywordsProperty().getValue());
+		appendIfPresent(text, "Language", props.getLanguageProperty().getValue());
+		appendIfPresent(text, "LastModifiedBy", props.getLastModifiedByProperty().getValue());
+		appendIfPresent(text, "LastPrinted", props.getLastPrintedProperty().getValue());
+		appendIfPresent(text, "LastPrintedString", props.getLastPrintedPropertyString());
+		appendIfPresent(text, "Modified", props.getModifiedProperty().getValue());
+		appendIfPresent(text, "ModifiedString", props.getModifiedPropertyString());
+		appendIfPresent(text, "Revision", props.getRevisionProperty().getValue());
+		appendIfPresent(text, "Subject", props.getSubjectProperty().getValue());
+		appendIfPresent(text, "Title", props.getTitleProperty().getValue());
+		appendIfPresent(text, "Version", props.getVersionProperty().getValue());
 
 		return text.toString();
 	}
@@ -81,21 +102,21 @@ public class POIXMLPropertiesTextExtractor extends POIXMLTextExtractor {
 		org.openxmlformats.schemas.officeDocument.x2006.extendedProperties.CTProperties
 			props = getDocument().getProperties().getExtendedProperties().getUnderlyingProperties();
 
-		text.append("Application = " + props.getApplication() + "\n");
-		text.append("AppVersion = " + props.getAppVersion() + "\n");
-		text.append("Characters = " + props.getCharacters() + "\n");
-		text.append("CharactersWithSpaces = " + props.getCharactersWithSpaces() + "\n");
-		text.append("Company = " + props.getCompany() + "\n");
-		text.append("HyperlinkBase = " + props.getHyperlinkBase() + "\n");
-		text.append("HyperlinksChanged = " + props.getHyperlinksChanged() + "\n");
-		text.append("Lines = " + props.getLines() + "\n");
-		text.append("LinksUpToDate = " + props.getLinksUpToDate() + "\n");
-		text.append("Manager = " + props.getManager() + "\n");
-		text.append("Pages = " + props.getPages() + "\n");
-		text.append("Paragraphs = " + props.getParagraphs() + "\n");
-		text.append("PresentationFormat = " + props.getPresentationFormat() + "\n");
-		text.append("Template = " + props.getTemplate() + "\n");
-		text.append("TotalTime = " + props.getTotalTime() + "\n");
+		appendIfPresent(text, "Application", props.getApplication());
+		appendIfPresent(text, "AppVersion", props.getAppVersion());
+		appendIfPresent(text, "Characters", props.getCharacters());
+		appendIfPresent(text, "CharactersWithSpaces", props.getCharactersWithSpaces());
+		appendIfPresent(text, "Company", props.getCompany());
+		appendIfPresent(text, "HyperlinkBase", props.getHyperlinkBase());
+		appendIfPresent(text, "HyperlinksChanged", props.getHyperlinksChanged());
+		appendIfPresent(text, "Lines", props.getLines());
+		appendIfPresent(text, "LinksUpToDate", props.getLinksUpToDate());
+		appendIfPresent(text, "Manager", props.getManager());
+		appendIfPresent(text, "Pages", props.getPages());
+		appendIfPresent(text, "Paragraphs", props.getParagraphs());
+		appendIfPresent(text, "PresentationFormat", props.getPresentationFormat());
+		appendIfPresent(text, "Template", props.getTemplate());
+		appendIfPresent(text, "TotalTime", props.getTotalTime());
 
 		return text.toString();
 	}
diff --git a/src/ooxml/java/org/apache/poi/openxml4j/opc/internal/PackagePropertiesPart.java b/src/ooxml/java/org/apache/poi/openxml4j/opc/internal/PackagePropertiesPart.java
index 800895df6c..af40d361aa 100644
--- a/src/ooxml/java/org/apache/poi/openxml4j/opc/internal/PackagePropertiesPart.java
+++ b/src/ooxml/java/org/apache/poi/openxml4j/opc/internal/PackagePropertiesPart.java
@@ -583,10 +583,15 @@ public final class PackagePropertiesPart extends PackagePart implements
 		if (d == null) {
 			return "";
 		}
+		Date date = d.getValue();
+		if (date == null) {
+		   return "";
+		}
+		
 		SimpleDateFormat df = new SimpleDateFormat(
 				"yyyy-MM-dd'T'HH:mm:ss'Z'");
-        df.setTimeZone(TimeZone.getTimeZone("UTC"));
-		return df.format(d.getValue());
+		df.setTimeZone(TimeZone.getTimeZone("UTC"));
+		return df.format(date);
 	}
 
 	@Override
diff --git a/src/ooxml/testcases/org/apache/poi/TestXMLPropertiesTextExtractor.java b/src/ooxml/testcases/org/apache/poi/TestXMLPropertiesTextExtractor.java
index 303946a113..b19477f04c 100644
--- a/src/ooxml/testcases/org/apache/poi/TestXMLPropertiesTextExtractor.java
+++ b/src/ooxml/testcases/org/apache/poi/TestXMLPropertiesTextExtractor.java
@@ -16,17 +16,17 @@
 ==================================================================== */
 package org.apache.poi;
 
-import java.io.File;
+import junit.framework.TestCase;
 
-import org.apache.poi.xssf.extractor.XSSFExcelExtractor;
-import org.apache.poi.xssf.usermodel.XSSFWorkbook;
 import org.apache.poi.openxml4j.opc.OPCPackage;
 import org.apache.poi.util.PackageHelper;
-
-import junit.framework.TestCase;
+import org.apache.poi.xslf.XSLFSlideShow;
+import org.apache.poi.xssf.extractor.XSSFExcelExtractor;
+import org.apache.poi.xssf.usermodel.XSSFWorkbook;
 
 public final class TestXMLPropertiesTextExtractor extends TestCase {
     private static final POIDataSamples _ssSamples = POIDataSamples.getSpreadSheetInstance();
+    private static final POIDataSamples _slSamples = POIDataSamples.getSlideShowInstance();
 
 	public void testGetFromMainExtractor() throws Exception {
 		OPCPackage pkg = PackageHelper.open(_ssSamples.openResourceAsStream("ExcelWithAttachments.xlsm"));
@@ -87,4 +87,23 @@ public final class TestXMLPropertiesTextExtractor extends TestCase {
 	public void testCustom() {
 		// TODO!
 	}
+	
+	/**
+	 * Bug #49386 - some properties, especially
+	 *  dates can be null
+	 */
+	public void testWithSomeNulls() throws Exception {
+      OPCPackage pkg = OPCPackage.open(
+            _slSamples.openResourceAsStream("49386-null_dates.pptx")
+      );
+      XSLFSlideShow sl = new XSLFSlideShow(pkg);
+   
+      POIXMLPropertiesTextExtractor ext = new POIXMLPropertiesTextExtractor(sl);
+      ext.getText();
+      
+      String text = ext.getText();
+      assertFalse(text.contains("Created =")); // With date is null
+      assertTrue(text.contains("CreatedString = ")); // Via string is blank
+      assertTrue(text.contains("LastModifiedBy = IT Client Services"));
+	}
 }
diff --git a/test-data/slideshow/49386-null_dates.pptx b/test-data/slideshow/49386-null_dates.pptx
new file mode 100644
index 0000000000..77fc378ce9
--- /dev/null
+++ b/test-data/slideshow/49386-null_dates.pptx