Tweaked patch from bug #46287 - Control of header and footer extraction in ExcelExtractor / XSSFExcelExtractor

git-svn-id: https://svn.apache.org/repos/asf/poi/trunk@775738 13f79535-47bb-0310-9956-ffa450edef68

src/documentation/content/xdocs/changes.xml

@@ -37,6 +37,7 @@
 
 		<!-- Don't forget to update status.xml too! -->
         <release version="3.5-beta6" date="2009-??-??">
+           <action dev="POI-DEVELOPERS" type="add">46287 - Control of header and footer extraction in ExcelExtractor / XSSFExcelExtractor</action>
            <action dev="POI-DEVELOPERS" type="add">46554 - New ant target "jar-examples"</action>
            <action dev="POI-DEVELOPERS" type="add">46161 - Support in XSSF for setGroupColumnCollapsed and setGroupRowCollapsed</action>
            <action dev="POI-DEVELOPERS" type="add">46806 - Allow columns greater than 255 and rows greater than 0x100000 in XSSF formulas</action>

src/documentation/content/xdocs/status.xml

@@ -34,6 +34,7 @@
 	<!-- Don't forget to update changes.xml too! -->
     <changes>
         <release version="3.5-beta6" date="2009-??-??">
+           <action dev="POI-DEVELOPERS" type="add">46287 - Control of header and footer extraction in ExcelExtractor / XSSFExcelExtractor</action>
            <action dev="POI-DEVELOPERS" type="add">46554 - New ant target "jar-examples"</action>
            <action dev="POI-DEVELOPERS" type="add">46161 - Support in XSSF for setGroupColumnCollapsed and setGroupRowCollapsed</action>
            <action dev="POI-DEVELOPERS" type="add">46806 - Allow columns greater than 255 and rows greater than 0x100000 in XSSF formulas</action>

src/java/org/apache/poi/hssf/extractor/ExcelExtractor.java

@@ -55,6 +55,7 @@ public class ExcelExtractor extends POIOLE2TextExtractor implements org.apache.p
 	private boolean _shouldEvaluateFormulas = true;
 	private boolean _includeCellComments = false;
 	private boolean _includeBlankCells = false;
+	private boolean _includeHeadersFooters = true;
 	
 	public ExcelExtractor(HSSFWorkbook wb) {
 		super(wb);
@@ -79,6 +80,7 @@ public class ExcelExtractor extends POIOLE2TextExtractor implements org.apache.p
 		private final boolean _evaluateFormulas;
 		private final boolean _showCellComments;
 		private final boolean _showBlankCells;
+		private final boolean _headersFooters;
 		public CommandArgs(String[] args) throws CommandParseException {
 			int nArgs = args.length;
 			File inputFile = null;
@@ -87,6 +89,7 @@ public class ExcelExtractor extends POIOLE2TextExtractor implements org.apache.p
 			boolean evaluateFormulas = true;
 			boolean showCellComments = false;
 			boolean showBlankCells = false;
+			boolean headersFooters = true;
 			for (int i=0; i<nArgs; i++) {
 				String arg = args[i];
 				if ("-help".equalsIgnoreCase(arg)) {
@@ -127,6 +130,10 @@ public class ExcelExtractor extends POIOLE2TextExtractor implements org.apache.p
 					showBlankCells = parseBoolArg(args, ++i);
 					continue;
 				}
+				if ("--headers-footers".equals(arg)) {
+					headersFooters = parseBoolArg(args, ++i);
+					continue;
+				}
 				throw new CommandParseException("Invalid argument '" + arg + "'");
 			}
 			_requestHelp = requestHelp;
@@ -135,6 +142,7 @@ public class ExcelExtractor extends POIOLE2TextExtractor implements org.apache.p
 			_evaluateFormulas = evaluateFormulas;
 			_showCellComments = showCellComments;
 			_showBlankCells = showBlankCells;
+			_headersFooters = headersFooters;
 		}
 		private static boolean parseBoolArg(String[] args, int i) throws CommandParseException {
 			if (i >= args.length) {
@@ -167,7 +175,9 @@ public class ExcelExtractor extends POIOLE2TextExtractor implements org.apache.p
 		public boolean shouldShowBlankCells() {
 			return _showBlankCells;
 		}
-		
+		public boolean shouldIncludeHeadersFooters() {
+			return _headersFooters;
+		}
 	}
 	
 	private static void printUsageMessage(PrintStream ps) {
@@ -180,6 +190,7 @@ public class ExcelExtractor extends POIOLE2TextExtractor implements org.apache.p
 		ps.println("       --evaluate-formulas Y");
 		ps.println("       --show-comments     N");
 		ps.println("       --show-blanks       Y");
+		ps.println("       --headers-footers   Y");
 	}
 
 	/**
@@ -216,6 +227,7 @@ public class ExcelExtractor extends POIOLE2TextExtractor implements org.apache.p
 			extractor.setFormulasNotResults(!cmdArgs.shouldEvaluateFormulas());
 			extractor.setIncludeCellComments(cmdArgs.shouldShowCellComments());
 			extractor.setIncludeBlankCells(cmdArgs.shouldShowBlankCells());
+			extractor.setIncludeHeadersFooters(cmdArgs.shouldIncludeHeadersFooters());
 			System.out.println(extractor.getText());
 		} catch (Exception e) {
 			e.printStackTrace();
@@ -249,6 +261,13 @@ public class ExcelExtractor extends POIOLE2TextExtractor implements org.apache.p
 	public void setIncludeBlankCells(boolean includeBlankCells) {
 		_includeBlankCells = includeBlankCells;
 	}
+	/**
+	 * Should headers and footers be included in the output?
+	 * Default is to include them.
+	 */
+	public void setIncludeHeadersFooters(boolean includeHeadersFooters) {
+		_includeHeadersFooters = includeHeadersFooters; 
+	}
 	
 	/**
 	 * Retrieves the text contents of the file
@@ -274,7 +293,7 @@ public class ExcelExtractor extends POIOLE2TextExtractor implements org.apache.p
 			}
 			
 			// Header text, if there is any
-			if(sheet.getHeader() != null) {
+			if(_includeHeadersFooters && sheet.getHeader() != null) {
 				text.append(
 						_extractHeaderFooter(sheet.getHeader())
 				);
@@ -364,7 +383,7 @@ public class ExcelExtractor extends POIOLE2TextExtractor implements org.apache.p
 			}
 			
 			// Finally Feader text, if there is any
-			if(sheet.getFooter() != null) {
+			if(_includeHeadersFooters && sheet.getFooter() != null) {
 				text.append(
 						_extractHeaderFooter(sheet.getFooter())
 				);

src/ooxml/java/org/apache/poi/xssf/extractor/XSSFExcelExtractor.java

@@ -40,6 +40,7 @@ public class XSSFExcelExtractor extends POIXMLTextExtractor implements org.apach
 	private boolean includeSheetNames = true;
 	private boolean formulasNotResults = false;
 	private boolean includeCellComments = false;
+	private boolean includeHeadersFooters = true;
 	
 	public XSSFExcelExtractor(String path) throws XmlException, OpenXML4JException, IOException {
 		this(new XSSFWorkbook(path));
@@ -82,6 +83,12 @@ public class XSSFExcelExtractor extends POIXMLTextExtractor implements org.apach
     public void setIncludeCellComments(boolean includeCellComments) {
         this.includeCellComments = includeCellComments;
     }
+	/**
+     * Should headers and footers be included? Default is true
+     */
+    public void setIncludeHeadersFooters(boolean includeHeadersFooters) {
+        this.includeHeadersFooters = includeHeadersFooters;
+    }
 	
 	/**
 	 * Retreives the text contents of the file
@@ -96,6 +103,7 @@ public class XSSFExcelExtractor extends POIXMLTextExtractor implements org.apach
 			}
 			
 			// Header(s), if present
+			if(includeHeadersFooters) {
 				text.append(
 						extractHeaderFooter(sheet.getFirstHeader())
 				);
@@ -105,6 +113,7 @@ public class XSSFExcelExtractor extends POIXMLTextExtractor implements org.apach
 				text.append(
 						extractHeaderFooter(sheet.getEvenHeader())
 				);
+			}
 
 			// Rows and cells
 			for (Object rawR : sheet) {
@@ -138,6 +147,7 @@ public class XSSFExcelExtractor extends POIXMLTextExtractor implements org.apach
 			}
 			
 			// Finally footer(s), if present
+			if(includeHeadersFooters) {
 				text.append(
 						extractHeaderFooter(sheet.getFirstFooter())
 				);
@@ -148,6 +158,7 @@ public class XSSFExcelExtractor extends POIXMLTextExtractor implements org.apach
 						extractHeaderFooter(sheet.getEvenFooter())
 				);
 			}
+		}
 		
 		return text.toString();
 	}