Fix TIKA-1515 - Handle Excel 3 files with a 0x8001 codepage

git-svn-id: https://svn.apache.org/repos/asf/poi/trunk@1651517 13f79535-47bb-0310-9956-ffa450edef68

diff --git a/src/java/org/apache/poi/hssf/record/OldStringRecord.java b/src/java/org/apache/poi/hssf/record/OldStringRecord.java
index 0a5b4922389476c2e01261fffe49d6b52fef25ee..4964f84803bea829d7348b2f87fc585fac70a405 100644 (file)
--- a/src/java/org/apache/poi/hssf/record/OldStringRecord.java
+++ b/src/java/org/apache/poi/hssf/record/OldStringRecord.java
@@ -75,7 +75,7 @@ public final class OldStringRecord {
     protected static String getString(byte[] data, CodepageRecord codepage) {
         int cp = CodePageUtil.CP_ISO_8859_1;
         if (codepage != null) {
-            cp = codepage.getCodepage();
+            cp = codepage.getCodepage() & 0xffff;
         }
         try {
             return CodePageUtil.getStringFromCodePage(data, cp);

diff --git a/src/java/org/apache/poi/util/CodePageUtil.java b/src/java/org/apache/poi/util/CodePageUtil.java
index 02e0ed3c0c6a42cb41b4c1869d05ba5b56f87caa..145929182a63d3608b1c7cc96acc502089bac7db 100644 (file)
--- a/src/java/org/apache/poi/util/CodePageUtil.java
+++ b/src/java/org/apache/poi/util/CodePageUtil.java
@@ -53,6 +53,7 @@ public class CodePageUtil
 
     /** <p>Codepage for Windows 1252</p> */
     public static final int CP_WINDOWS_1252 = 1252;
+    public static final int CP_WINDOWS_1252_BIFF23 = 32769;
 
     /** <p>Codepage for Windows 1253</p> */
     public static final int CP_WINDOWS_1253 = 1253;
@@ -77,6 +78,7 @@ public class CodePageUtil
 
     /** <p>Codepage for Macintosh Roman (Java: MacRoman)</p> */
     public static final int CP_MAC_ROMAN = 10000;
+    public static final int CP_MAC_ROMAN_BIFF23 = 32768;
 
     /** <p>Codepage for Macintosh Japan (Java: unknown - use SJIS, cp942 or
      * cp943)</p> */
@@ -298,6 +300,7 @@ public class CodePageUtil
                 else
                     return "windows-1251";
             case CP_WINDOWS_1252:
+            case CP_WINDOWS_1252_BIFF23:
                 if (javaLangFormat)
                     return "Cp1252";
                 else
@@ -335,6 +338,7 @@ public class CodePageUtil
             case CP_JOHAB:
                 return "johab";
             case CP_MAC_ROMAN:
+            case CP_MAC_ROMAN_BIFF23:
                 return "MacRoman";
             case CP_MAC_JAPAN:
                 return "SJIS";

diff --git a/src/testcases/org/apache/poi/hssf/extractor/TestOldExcelExtractor.java b/src/testcases/org/apache/poi/hssf/extractor/TestOldExcelExtractor.java
index 92235e70a7c8ecb2f7a25344f39fb37bc1a0ce4d..bb2cfea66f92781014025771eb03408935a33bdb 100644 (file)
--- a/src/testcases/org/apache/poi/hssf/extractor/TestOldExcelExtractor.java
+++ b/src/testcases/org/apache/poi/hssf/extractor/TestOldExcelExtractor.java
@@ -38,7 +38,7 @@ public final class TestOldExcelExtractor extends POITestCase {
         }
     }
     
-    public void DISABLEDtestSimpleExcel3() {
+    public void testSimpleExcel3() {
         OldExcelExtractor extractor = createExtractor("testEXCEL_3.xls");
 
         // Check we can call getText without error