Begin adding Excel 5 support to OldExcelExtractor for TIKA-1490

git-svn-id: https://svn.apache.org/repos/asf/poi/trunk@1642548 13f79535-47bb-0310-9956-ffa450edef68

2 files changed, 36 insertions, 4 deletions

diff --git a/src/testcases/org/apache/poi/hssf/dev/TestBiffViewer.java b/src/testcases/org/apache/poi/hssf/dev/TestBiffViewer.java
index b7013c1503..e5062a0241 100644
--- a/src/testcases/org/apache/poi/hssf/dev/TestBiffViewer.java
+++ b/src/testcases/org/apache/poi/hssf/dev/TestBiffViewer.java
@@ -38,7 +38,9 @@ public class TestBiffViewer extends BaseXLSIteratingTest {
 		SILENT_EXCLUDED.add("46904.xls");
         SILENT_EXCLUDED.add("35897-type4.xls"); // unsupported crypto api header 
 		SILENT_EXCLUDED.add("xor-encryption-abc.xls"); // unsupported XOR-encryption
-        SILENT_EXCLUDED.add("testEXCEL_4.xls"); // Biff 4 / Excel 4, pre-OLE2
+        SILENT_EXCLUDED.add("testEXCEL_4.xls");  // Biff 4 / Excel 4, pre-OLE2
+        SILENT_EXCLUDED.add("testEXCEL_5.xls");  // Biff 5 / Excel 5
+        SILENT_EXCLUDED.add("testEXCEL_95.xls"); // Biff 5 / Excel 95
 	}
 
 	@Override
diff --git a/src/testcases/org/apache/poi/hssf/extractor/TestOldExcelExtractor.java b/src/testcases/org/apache/poi/hssf/extractor/TestOldExcelExtractor.java
index f6c36e16c1..8c58f9e4f3 100644
--- a/src/testcases/org/apache/poi/hssf/extractor/TestOldExcelExtractor.java
+++ b/src/testcases/org/apache/poi/hssf/extractor/TestOldExcelExtractor.java
@@ -24,7 +24,8 @@ import junit.framework.TestCase;
 import org.apache.poi.hssf.HSSFTestDataSamples;
 
 /**
- * Unit tests for the Excel 4 (and older) text extractor
+ * Unit tests for the Excel 5/95 and Excel 4 (and older) text 
+ *  extractor
  */
 public final class TestOldExcelExtractor extends TestCase {
     private static OldExcelExtractor createExtractor(String sampleFileName) {
@@ -37,7 +38,7 @@ public final class TestOldExcelExtractor extends TestCase {
         }
     }
 
-    public void testSimple() {
+    public void testSimpleExcel4() {
         OldExcelExtractor extractor = createExtractor("testEXCEL_4.xls");
 
         // Check we can call getText without error
@@ -51,6 +52,22 @@ public final class TestOldExcelExtractor extends TestCase {
         assertTrue(text, text.contains("11"));
         assertTrue(text, text.contains("784"));
     }
+    public void DISABLEDtestSimpleExcel5() {
+        for (String ver : new String[] {"5", "95"}) {
+            OldExcelExtractor extractor = createExtractor("testEXCEL_"+ver+".xls");
+    
+            // Check we can call getText without error
+            String text = extractor.getText();
+    
+            // Check we find a few words we expect in there
+            assertTrue(text, text.contains("Sample Excel"));
+            assertTrue(text, text.contains("Written and saved"));
+            
+            // Check we find a few numbers we expect in there
+            assertTrue(text, text.contains("15"));
+            assertTrue(text, text.contains("169"));
+        }
+    }
 
     public void testStrings() {
         OldExcelExtractor extractor = createExtractor("testEXCEL_4.xls");
@@ -71,7 +88,7 @@ public final class TestOldExcelExtractor extends TestCase {
         // TODO Find some then test
     }
 
-    public void testFormattedNumbers() {
+    public void testFormattedNumbersExcel4() {
         OldExcelExtractor extractor = createExtractor("testEXCEL_4.xls");
         String text = extractor.getText();
 
@@ -88,4 +105,17 @@ public final class TestOldExcelExtractor extends TestCase {
 //      assertTrue(text, text.contains("55,624"));
 //      assertTrue(text, text.contains("11,743,477"));
     }
+    public void DISABLEDtestFormattedNumbersExcel5() {
+        for (String ver : new String[] {"5", "95"}) {
+            OldExcelExtractor extractor = createExtractor("testEXCEL_"+ver+".xls");
+            String text = extractor.getText();
+            
+            // Simple numbers
+            assertTrue(text, text.contains("1"));
+            
+            // Numbers which come from formulas
+            assertTrue(text, text.contains("13"));
+            assertTrue(text, text.contains("169"));
+        }
+    }
 }