TIKA-2873 -- workaround for newly re-discovered bug in POI's ChunkedCipherInputStream - bug 63431

Author: tballison

tika-parsers/src/main/java/org/apache/tika/parser/microsoft/OfficeParser.java

@@ -224,10 +224,11 @@ protected void parse(
                     // Decrypt the OLE2 stream, and delegate the resulting OOXML
                     //  file to the regular OOXML parser for normal handling
                     OOXMLParser parser = new OOXMLParser();
-
-                    parser.parse(d.getDataStream(root), new EmbeddedContentHandler(
-                                    new BodyContentHandler(xhtml)),
-                            metadata, context);
+                    try (TikaInputStream tis = TikaInputStream.get(d.getDataStream(root))) {
+                        parser.parse(tis, new EmbeddedContentHandler(
+                                        new BodyContentHandler(xhtml)),
+                                metadata, context);
+                    }
                 } catch (GeneralSecurityException ex) {
                     throw new EncryptedDocumentException(ex);
                 }

tika-parsers/src/test/java/org/apache/tika/parser/microsoft/ooxml/OOXMLParserTest.java

@@ -1215,12 +1215,18 @@ public void testPPTXThumbnail() throws Exception {
     @Test
     public void testEncrypted() throws Exception {
         Map<String, String> tests = new HashMap<String, String>();
+        //the first three contain javax.crypto.CipherInputStream
         tests.put("testWORD_protected_passtika.docx",
                 "This is an encrypted Word 2007 File");
         tests.put("testPPT_protected_passtika.pptx",
                 "This is an encrypted PowerPoint 2007 slide.");
         tests.put("testEXCEL_protected_passtika.xlsx",
                 "This is an Encrypted Excel spreadsheet.");
+        //TIKA-2873 this one contains a ChunkedCipherInputStream
+        //that is buggy at the POI level...can unwrap TikaInputStream in OfficeParser
+        //once https://bz.apache.org/bugzilla/show_bug.cgi?id=63431 is fixed.
+        tests.put("testEXCEL_protected_passtika_2.xlsx",
+                "This is an Encrypted Excel spreadsheet with a ChunkedCipherInputStream.");
 
         Parser parser = new AutoDetectParser();
         Metadata m = new Metadata();