From: Nick Burch <nick@apache.org>
Date: Thu, 29 Mar 2007 11:50:50 +0000 (+0000)
Subject: Don't NPE if we have a reference to a note, where there's no core record to go with... 
X-Git-Tag: REL_3_0_RC3~4
X-Git-Url: https://source.dussan.org/?a=commitdiff_plain;h=3ad3a22fd533582b63d82c43f5cb945a6927d59c;p=poi.git

Don't NPE if we have a reference to a note, where there's no core record to go with that note's RefID. Instead, just log it, and pretend the note wasn't there.

git-svn-id: https://svn.apache.org/repos/asf/jakarta/poi/trunk@523678 13f79535-47bb-0310-9956-ffa450edef68
---

diff --git a/src/scratchpad/src/org/apache/poi/hslf/usermodel/SlideShow.java b/src/scratchpad/src/org/apache/poi/hslf/usermodel/SlideShow.java
index 400eecdc33..fdecb2bb04 100644
--- a/src/scratchpad/src/org/apache/poi/hslf/usermodel/SlideShow.java
+++ b/src/scratchpad/src/org/apache/poi/hslf/usermodel/SlideShow.java
@@ -273,8 +273,13 @@ public class SlideShow
 	private Record getCoreRecordForRefID(int refID) {
 		Integer coreRecordId = (Integer)
 			_sheetIdToCoreRecordsLookup.get(new Integer(refID));
-		Record r = _mostRecentCoreRecords[coreRecordId.intValue()];
-		return r;
+		if(coreRecordId != null) {
+			Record r = _mostRecentCoreRecords[coreRecordId.intValue()];
+			return r;
+		} else {
+			logger.log(POILogger.ERROR, "We tried to look up a reference to a core record, but there was no core ID for reference ID " + refID);
+			return null;
+		}
 	}
 
   /**
@@ -352,23 +357,26 @@ public class SlideShow
 	} else {
 		// Match up the records and the SlideAtomSets
 		notesSets = notesSLWT.getSlideAtomsSets();
-		notesRecords = new org.apache.poi.hslf.record.Notes[notesSets.length];
+		ArrayList notesRecordsL = new ArrayList();
 		for(int i=0; i<notesSets.length; i++) {
 			// Get the right core record
 			Record r = getCoreRecordForSAS(notesSets[i]);
 			
 			// Ensure it really is a notes record
-			if(r instanceof org.apache.poi.hslf.record.Notes) {
-				notesRecords[i] = (org.apache.poi.hslf.record.Notes)r;
+			if(r != null && r instanceof org.apache.poi.hslf.record.Notes) {
+				notesRecordsL.add( (org.apache.poi.hslf.record.Notes)r );
+				
+				// Record the match between slide id and these notes
+				SlidePersistAtom spa = notesSets[i].getSlidePersistAtom();
+				Integer slideId = new Integer(spa.getSlideIdentifier());
+				slideIdToNotes.put(slideId, new Integer(i));
 			} else {
 				logger.log(POILogger.ERROR, "A Notes SlideAtomSet at " + i + " said its record was at refID " + notesSets[i].getSlidePersistAtom().getRefID() + ", but that was actually a " + r);
 			}
-			
-			// Record the match between slide id and these notes
-			SlidePersistAtom spa = notesSets[i].getSlidePersistAtom();
-			Integer slideId = new Integer(spa.getSlideIdentifier());
-			slideIdToNotes.put(slideId, new Integer(i));
 		}
+		notesRecords = new org.apache.poi.hslf.record.Notes[notesRecordsL.size()];
+		notesRecords = (org.apache.poi.hslf.record.Notes[])
+			notesRecordsL.toArray(notesRecords);
 	}
 	
 	// Now, do the same thing for our slides
diff --git a/src/scratchpad/testcases/org/apache/poi/hslf/data/missing_core_records.ppt b/src/scratchpad/testcases/org/apache/poi/hslf/data/missing_core_records.ppt
new file mode 100644
index 0000000000..67325ebcf6
Binary files /dev/null and b/src/scratchpad/testcases/org/apache/poi/hslf/data/missing_core_records.ppt differ
diff --git a/src/scratchpad/testcases/org/apache/poi/hslf/extractor/TextExtractor.java b/src/scratchpad/testcases/org/apache/poi/hslf/extractor/TextExtractor.java
index 3ca3a19952..e0e0318e77 100644
--- a/src/scratchpad/testcases/org/apache/poi/hslf/extractor/TextExtractor.java
+++ b/src/scratchpad/testcases/org/apache/poi/hslf/extractor/TextExtractor.java
@@ -29,13 +29,15 @@ import junit.framework.TestCase;
  * @author Nick Burch (nick at torchbox dot com)
  */
 public class TextExtractor extends TestCase {
-	// Extractor primed on the 2 page basic test data
+	/** Extractor primed on the 2 page basic test data */
 	private PowerPointExtractor ppe;
-	// Extractor primed on the 1 page but text-box'd test data
+	/** Extractor primed on the 1 page but text-box'd test data */
 	private PowerPointExtractor ppe2;
+	/** Where to go looking for our test files */
+	private String dirname;
 
     public TextExtractor() throws Exception {
-		String dirname = System.getProperty("HSLF.testdata.path");
+		dirname = System.getProperty("HSLF.testdata.path");
 		String filename = dirname + "/basic_test_ppt_file.ppt";
 		ppe = new PowerPointExtractor(filename);
 		String filename2 = dirname + "/with_textbox.ppt";
@@ -70,6 +72,28 @@ public class TextExtractor extends TestCase {
 		
 		ensureTwoStringsTheSame(expectText, notesText);
 	}
+
+	/**
+	 * Test that when presented with a PPT file missing the odd
+	 *  core record, we can still get the rest of the text out
+	 * @throws Exception
+	 */
+	public void testMissingCoreRecords() throws Exception {
+		String filename = dirname + "/missing_core_records.ppt";
+		ppe = new PowerPointExtractor(filename);
+		
+		String text = ppe.getText(true, false);
+		String nText = ppe.getNotes();
+
+		assertNotNull(text);
+		assertNotNull(nText);
+		
+		// Notes record were corrupt, so don't expect any
+		assertEquals(nText.length(), 0);
+		
+		// Slide records were fine
+		assertTrue(text.startsWith("Using Disease Surveillance and Response"));
+	}
 	
     private void ensureTwoStringsTheSame(String exp, String act) throws Exception {
 		assertEquals(exp.length(),act.length());