From d9df038f4a5e9c08c96845288f2c69b2843161ac Mon Sep 17 00:00:00 2001
From: Dominik Stadler <centic@apache.org>
Date: Sun, 7 May 2017 21:03:46 +0000
Subject: [PATCH] Bug 60352: Text extraction: Don't include the text "null" for
 empty cells

git-svn-id: https://svn.apache.org/repos/asf/poi/trunk@1794260 13f79535-47bb-0310-9956-ffa450edef68
---
 .../poi/xssf/extractor/XSSFExcelExtractor.java       | 12 +++++++-----
 .../poi/xssf/extractor/TestXSSFExcelExtractor.java   | 12 +++---------
 2 files changed, 10 insertions(+), 14 deletions(-)

diff --git a/src/ooxml/java/org/apache/poi/xssf/extractor/XSSFExcelExtractor.java b/src/ooxml/java/org/apache/poi/xssf/extractor/XSSFExcelExtractor.java
index 192d60d71..f479ef3b4 100644
--- a/src/ooxml/java/org/apache/poi/xssf/extractor/XSSFExcelExtractor.java
+++ b/src/ooxml/java/org/apache/poi/xssf/extractor/XSSFExcelExtractor.java
@@ -55,8 +55,8 @@ public class XSSFExcelExtractor extends POIXMLTextExtractor
     private Locale locale;
     private XSSFWorkbook workbook;
     private boolean includeSheetNames = true;
-    private boolean formulasNotResults = false;
-    private boolean includeCellComments = false;
+    private boolean formulasNotResults;
+    private boolean includeCellComments;
     private boolean includeHeadersFooters = true;
     private boolean includeTextBoxes = true;
 
@@ -110,7 +110,7 @@ public class XSSFExcelExtractor extends POIXMLTextExtractor
     }
     /**
      * Should text within textboxes be included? Default is true
-     * @param includeTextBoxes
+     * @param includeTextBoxes True if textboxes should be included, false if not.
      */
     public void setIncludeTextBoxes(boolean includeTextBoxes){
         this.includeTextBoxes = includeTextBoxes;
@@ -255,8 +255,10 @@ public class XSSFExcelExtractor extends POIXMLTextExtractor
 
         // No supported styling applies to this cell
         String contents = ((XSSFCell)cell).getRawValue();
-        checkMaxTextSize(text, contents);
-        text.append( contents );
+        if (contents != null) {
+            checkMaxTextSize(text, contents);
+            text.append(contents);
+        }
     }
 
     private String extractHeaderFooter(HeaderFooter hf) {
diff --git a/src/ooxml/testcases/org/apache/poi/xssf/extractor/TestXSSFExcelExtractor.java b/src/ooxml/testcases/org/apache/poi/xssf/extractor/TestXSSFExcelExtractor.java
index 3fc281808..f4450c010 100644
--- a/src/ooxml/testcases/org/apache/poi/xssf/extractor/TestXSSFExcelExtractor.java
+++ b/src/ooxml/testcases/org/apache/poi/xssf/extractor/TestXSSFExcelExtractor.java
@@ -45,7 +45,6 @@ public class TestXSSFExcelExtractor extends TestCase {
 
 	/**
 	 * Get text out of the simple file
-	 * @throws IOException 
 	 */
 	public void testGetSimpleText() throws IOException {
 		// a very simple file
@@ -105,7 +104,7 @@ public class TestXSSFExcelExtractor extends TestCase {
 		
 		extractor.close();
 	}
-	
+
 	public void testGetComplexText() throws IOException {
 		// A fairly complex file
 		XSSFExcelExtractor extractor = getExtractor("AverageTaxRates.xlsx");
@@ -114,10 +113,9 @@ public class TestXSSFExcelExtractor extends TestCase {
 		assertTrue(text.length() > 0);
 		
 		// Might not have all formatting it should do!
-		// TODO decide if we should really have the "null" in there
-		assertStartsWith(text, 
+		assertStartsWith(text,
 						"Avgtxfull\n" +
-						"null\t(iii) AVERAGE TAX RATES ON ANNUAL"	
+						"\t(iii) AVERAGE TAX RATES ON ANNUAL"
 		);
 		
 		extractor.close();
@@ -127,7 +125,6 @@ public class TestXSSFExcelExtractor extends TestCase {
 	 * Test that we return pretty much the same as
 	 *  ExcelExtractor does, when we're both passed
 	 *  the same file, just saved as xls and xlsx
-	 * @throws IOException 
 	 */
 	public void testComparedToOLE2() throws IOException {
 		// A fairly simple file - ooxml
@@ -156,7 +153,6 @@ public class TestXSSFExcelExtractor extends TestCase {
 	
 	/**
 	 * From bug #45540
-	 * @throws IOException 
 	 */
 	public void testHeaderFooter() throws IOException {
 		String[] files = new String[] {
@@ -176,7 +172,6 @@ public class TestXSSFExcelExtractor extends TestCase {
 
 	/**
 	 * From bug #45544
-	 * @throws IOException 
 	 */
 	public void testComments() throws IOException {
 		XSSFExcelExtractor extractor = getExtractor("45544.xlsx");
@@ -220,7 +215,6 @@ public class TestXSSFExcelExtractor extends TestCase {
 	}
 	/**
 	 * Simple test for text box text
-	 * @throws IOException
 	 */
 	public void testTextBoxes() throws IOException {
 	    XSSFExcelExtractor extractor = getExtractor("WithTextBox.xlsx");