Make the ooxml and ole2 powerpoint extractors more in keeping with the others, and with each other

git-svn-id: https://svn.apache.org/repos/asf/poi/trunk@608830 13f79535-47bb-0310-9956-ffa450edef68

src/scratchpad/ooxml-src/org/apache/poi/hslf/extractor/HXFPowerPointExtractor.java

@@ -37,6 +37,8 @@ import org.openxmlformats.schemas.presentationml.x2006.main.CTSlideIdListEntry;
 
 public class HXFPowerPointExtractor extends POIXMLTextExtractor {
 	private HSLFXMLSlideShow slideshow;
+	private boolean slidesByDefault = true;
+	private boolean notesByDefault = false;
 	
 	public HXFPowerPointExtractor(Package container) throws XmlException, OpenXML4JException, IOException {
 		this(new HSLFXMLSlideShow(
@@ -62,10 +64,25 @@ public class HXFPowerPointExtractor extends POIXMLTextExtractor {
 	}
 
 	/**
-	 * Gets the slide and notes text
+	 * Should a call to getText() return slide text?
+	 * Default is yes
+	 */
+	public void setSlidesByDefault(boolean slidesByDefault) {
+		this.slidesByDefault = slidesByDefault;
+	}
+	/**
+	 * Should a call to getText() return notes text?
+	 * Default is no
+	 */
+	public void setNotesByDefault(boolean notesByDefault) {
+		this.notesByDefault = notesByDefault;
+	}
+	
+	/**
+	 * Gets the slide text, but not the notes text
 	 */
 	public String getText() {
-		return getText(true, true);
+		return getText(slidesByDefault, notesByDefault);
 	}
 	
 	/**

src/scratchpad/ooxml-testcases/org/apache/poi/hslf/extractor/TestHXFPowerPointExtractor.java

@@ -60,7 +60,7 @@ public class TestHXFPowerPointExtractor extends TestCase {
 		
 		// Check Basics
 		assertTrue(text.startsWith("Lorem ipsum dolor sit amet\n"));
-		assertTrue(text.endsWith("amet\n\n\n\n"));
+		assertTrue(text.endsWith("amet\n\n"));
 		
 		// Just slides, no notes
 		text = extractor.getText(true, false);
@@ -97,5 +97,13 @@ public class TestHXFPowerPointExtractor extends TestCase {
 				"amet\n" +
 				"\n\n\n", text
 		);
+		
+		// Via set defaults
+		extractor.setSlidesByDefault(false);
+		extractor.setNotesByDefault(true);
+		text = extractor.getText();
+		assertEquals(
+				"\n\n\n\n", text
+		);
 	}
 }

src/scratchpad/src/org/apache/poi/hslf/extractor/PowerPointExtractor.java

@@ -41,7 +41,9 @@ public class PowerPointExtractor extends POITextExtractor
 	private HSLFSlideShow _hslfshow;
 	private SlideShow _show;
 	private Slide[] _slides;
-	private Notes[] _notes;
+	
+	private boolean slidesByDefault = true;
+	private boolean notesByDefault = false;
 
   /**
    * Basic extractor. Returns all the text, and optionally all the notes
@@ -99,7 +101,6 @@ public class PowerPointExtractor extends POITextExtractor
 		_hslfshow = ss;
 		_show = new SlideShow(_hslfshow);
 		_slides = _show.getSlides();
-		_notes = _show.getNotes();
 	}
 
 	/**
@@ -110,15 +111,31 @@ public class PowerPointExtractor extends POITextExtractor
 		_hslfshow = null;
 		_show = null;
 		_slides = null;
-		_notes = null;
 	}
 
+	/**
+	 * Should a call to getText() return slide text?
+	 * Default is yes
+	 */
+	public void setSlidesByDefault(boolean slidesByDefault) {
+		this.slidesByDefault = slidesByDefault;
+	}
+	/**
+	 * Should a call to getText() return notes text?
+	 * Default is no
+	 */
+	public void setNotesByDefault(boolean notesByDefault) {
+		this.notesByDefault = notesByDefault;
+	}
 
 	/**
-   * Fetches all the slide text from the slideshow, but not the notes
+	 * Fetches all the slide text from the slideshow, 
+	 *  but not the notes, unless you've called
+	 *  setSlidesByDefault() and setNotesByDefault()
+	 *  to change this
 	 */
 	public String getText() {
-	return getText(true,false);
+		return getText(slidesByDefault,notesByDefault);
 	}
 
 	/**
@@ -154,7 +171,7 @@ public class PowerPointExtractor extends POITextExtractor
 			}
 		}
 		if(getNoteText) {
-			ret.append(" ");
+			ret.append("\n");
 		}
 	}
 

src/scratchpad/testcases/org/apache/poi/hslf/extractor/TextExtractor.java

@@ -73,6 +73,29 @@ public class TextExtractor extends TestCase {
 		ensureTwoStringsTheSame(expectText, notesText);
 	}
 	
+	public void testReadBoth() throws Exception {
+		String[] slText = new String[] {
+				"This is a test title\nThis is a test subtitle\nThis is on page 1\n",
+				"This is the title on page 2\nThis is page two\nIt has several blocks of text\nNone of them have formatting\n"
+		};
+		String[] ntText = new String[] {
+				"These are the notes for page 1\n",
+				"These are the notes on page two, again lacking formatting\n"
+		};
+		
+		ppe.setSlidesByDefault(true);
+		ppe.setNotesByDefault(false);
+		assertEquals(slText[0]+slText[1], ppe.getText());
+		
+		ppe.setSlidesByDefault(false);
+		ppe.setNotesByDefault(true);
+		assertEquals(ntText[0]+ntText[1], ppe.getText());
+		
+		ppe.setSlidesByDefault(true);
+		ppe.setNotesByDefault(true);
+		assertEquals(slText[0]+slText[1]+"\n"+ntText[0]+ntText[1], ppe.getText());
+	}
+
 	/**
 	 * Test that when presented with a PPT file missing the odd
 	 *  core record, we can still get the rest of the text out