Bug 66425: Avoid exceptions found via poi-fuzz

Avoid a ClassCastException and trigger some more
code in integration-testing.

Fixes https://bugs.chromium.org/p/oss-fuzz/issues/detail?id=63358

git-svn-id: https://svn.apache.org/repos/asf/poi/trunk@1914404 13f79535-47bb-0310-9956-ffa450edef68

6 files changed, 33 insertions, 2 deletions

diff --git a/poi-integration/src/test/java/org/apache/poi/stress/AbstractFileHandler.java b/poi-integration/src/test/java/org/apache/poi/stress/AbstractFileHandler.java
index c413fb0733..c1c96aaa12 100644
--- a/poi-integration/src/test/java/org/apache/poi/stress/AbstractFileHandler.java
+++ b/poi-integration/src/test/java/org/apache/poi/stress/AbstractFileHandler.java
@@ -109,13 +109,38 @@ public abstract class AbstractFileHandler implements FileHandler {
                 assertEquals(modified, file.lastModified(), "File should not be modified by extractor");
 
                 if (extractor instanceof POIOLE2TextExtractor) {
-                    try (HPSFPropertiesExtractor hpsfExtractor = new HPSFPropertiesExtractor((POIOLE2TextExtractor) extractor)) {
+                    POIOLE2TextExtractor ole2Extractor = (POIOLE2TextExtractor) extractor;
+                    ole2Extractor.getRoot();
+                    if (!(ole2Extractor instanceof EventBasedExcelExtractor)) {
+                        ole2Extractor.getSummaryInformation();
+                        ole2Extractor.getDocSummaryInformation();
+                    }
+
+                    try (HPSFPropertiesExtractor hpsfExtractor = new HPSFPropertiesExtractor(ole2Extractor)) {
                         assertNotNull(hpsfExtractor.getDocumentSummaryInformationText());
                         assertNotNull(hpsfExtractor.getSummaryInformationText());
                         String text = hpsfExtractor.getText();
                         //System.out.println(text);
                         assertNotNull(text);
                     }
+
+                    if (ole2Extractor.getRoot() != null && !Boolean.getBoolean("scratchpad.ignore")) {
+                        POITextExtractor[] embedded = ExtractorFactory.getEmbeddedDocsTextExtractors(ole2Extractor);
+                        try {
+                            for (POITextExtractor poiTextExtractor : embedded) {
+                                poiTextExtractor.getText();
+                                poiTextExtractor.getDocument();
+                                poiTextExtractor.getFilesystem();
+                                POITextExtractor metaData = poiTextExtractor.getMetadataTextExtractor();
+                                metaData.getFilesystem();
+                                metaData.getText();
+                            }
+                        } finally {
+                            for (POITextExtractor embeddedExtractor : embedded) {
+                                embeddedExtractor.close();
+                            }
+                        }
+                    }
                 }
 
                 // test again with including formulas and cell-comments as this caused some bugs
@@ -166,6 +191,8 @@ public abstract class AbstractFileHandler implements FileHandler {
                 assertNotNull(streamExtractor);
 
                 assertNotNull(streamExtractor.getText());
+
+                assertNotNull(streamExtractor.getMetadataTextExtractor());
             }
         }
     }
diff --git a/poi/src/main/java/org/apache/poi/extractor/ExtractorFactory.java b/poi/src/main/java/org/apache/poi/extractor/ExtractorFactory.java
index 21e61e572a..f784842274 100644
--- a/poi/src/main/java/org/apache/poi/extractor/ExtractorFactory.java
+++ b/poi/src/main/java/org/apache/poi/extractor/ExtractorFactory.java
@@ -373,7 +373,9 @@ public final class ExtractorFactory {
 
         ArrayList<POITextExtractor> textExtractors = new ArrayList<>();
         for (Entry dir : dirs) {
-            textExtractors.add(createExtractor((DirectoryNode) dir));
+            if (dir instanceof DirectoryNode) {
+                textExtractors.add(createExtractor((DirectoryNode) dir));
+            }
         }
         for (InputStream stream : nonPOIFS) {
             try {
diff --git a/poi/src/test/java/org/apache/poi/hssf/dev/TestBiffViewer.java b/poi/src/test/java/org/apache/poi/hssf/dev/TestBiffViewer.java
index 218ecb93f6..4655517e08 100644
--- a/poi/src/test/java/org/apache/poi/hssf/dev/TestBiffViewer.java
+++ b/poi/src/test/java/org/apache/poi/hssf/dev/TestBiffViewer.java
@@ -44,6 +44,7 @@ class TestBiffViewer extends BaseTestIteratingXLS {
         excludes.put("protected_66115.xls", RecordFormatException.class);
         excludes.put("clusterfuzz-testcase-minimized-POIHSSFFuzzer-5786329142919168.xls", IllegalStateException.class);
         excludes.put("clusterfuzz-testcase-minimized-POIHSSFFuzzer-5889658057523200.xls", IndexOutOfBoundsException.class);
+        excludes.put("clusterfuzz-testcase-minimized-POIHSSFFuzzer-5175219985448960.xls", IndexOutOfBoundsException.class);
 
         return excludes;
     }
diff --git a/poi/src/test/java/org/apache/poi/hssf/dev/TestRecordLister.java b/poi/src/test/java/org/apache/poi/hssf/dev/TestRecordLister.java
index 7f2f85dcbb..c04d994a43 100644
--- a/poi/src/test/java/org/apache/poi/hssf/dev/TestRecordLister.java
+++ b/poi/src/test/java/org/apache/poi/hssf/dev/TestRecordLister.java
@@ -47,6 +47,7 @@ class TestRecordLister extends BaseTestIteratingXLS {
         Map<String, Class<? extends Throwable>> excludes = super.getExcludes();
         excludes.put("clusterfuzz-testcase-minimized-POIHSSFFuzzer-5786329142919168.xls", RecordFormatException.class);
         excludes.put("clusterfuzz-testcase-minimized-POIHSSFFuzzer-5889658057523200.xls", IndexOutOfBoundsException.class);
+        excludes.put("clusterfuzz-testcase-minimized-POIHSSFFuzzer-5175219985448960.xls", RecordFormatException.class);
         return excludes;
     }
 
diff --git a/test-data/spreadsheet/clusterfuzz-testcase-minimized-POIHSSFFuzzer-5175219985448960.xls b/test-data/spreadsheet/clusterfuzz-testcase-minimized-POIHSSFFuzzer-5175219985448960.xls
new file mode 100644
index 0000000000..8a982d7961
--- /dev/null
+++ b/test-data/spreadsheet/clusterfuzz-testcase-minimized-POIHSSFFuzzer-5175219985448960.xls
diff --git a/test-data/spreadsheet/stress.xls b/test-data/spreadsheet/stress.xls
index 31a8b184cd..275bc91d8e 100644
--- a/test-data/spreadsheet/stress.xls
+++ b/test-data/spreadsheet/stress.xls