Bug47950 -- make stream/directory name lookup in OLE2 case insensitive

git-svn-id: https://svn.apache.org/repos/asf/poi/trunk@1912438 13f79535-47bb-0310-9956-ffa450edef68

8 files changed, 57 insertions, 43 deletions

diff --git a/poi-scratchpad/src/test/java/org/apache/poi/hslf/TestReWrite.java b/poi-scratchpad/src/test/java/org/apache/poi/hslf/TestReWrite.java
index 48e2666aa1..0606c28950 100644
--- a/poi-scratchpad/src/test/java/org/apache/poi/hslf/TestReWrite.java
+++ b/poi-scratchpad/src/test/java/org/apache/poi/hslf/TestReWrite.java
@@ -92,20 +92,20 @@ public final class TestReWrite {
             assertSlideShowWritesOutTheSame(hssC, pfsC);
 
             // Currently has a Macros stream
-            assertNotNull(pfsC.getRoot().getEntry("Macros"));
+            assertNotNull(pfsC.getRoot().getEntryCaseInsensitive("Macros"));
 
             // Write out normally, will loose the macro stream
             UnsynchronizedByteArrayOutputStream baos = UnsynchronizedByteArrayOutputStream.builder().get();
             hssC.write(baos);
             try (POIFSFileSystem pfsNew = new POIFSFileSystem(baos.toInputStream())) {
-                assertFalse(pfsNew.getRoot().hasEntry("Macros"));
+                assertFalse(pfsNew.getRoot().hasEntryCaseInsensitive("Macros"));
             }
 
             // But if we write out with nodes preserved, will be there
             baos.reset();
             hssC.write(baos, true);
             try (POIFSFileSystem pfsNew = new POIFSFileSystem(baos.toInputStream())) {
-                assertTrue(pfsNew.getRoot().hasEntry("Macros"));
+                assertTrue(pfsNew.getRoot().hasEntryCaseInsensitive("Macros"));
             }
         }
     }
@@ -145,8 +145,8 @@ public final class TestReWrite {
 
     private void assertSame(POIFSFileSystem origPFS, POIFSFileSystem newPFS) throws IOException {
         // Check that the "PowerPoint Document" sections have the same size
-        DocumentEntry oProps = (DocumentEntry) origPFS.getRoot().getEntry(POWERPOINT_DOCUMENT);
-        DocumentEntry nProps = (DocumentEntry) newPFS.getRoot().getEntry(POWERPOINT_DOCUMENT);
+        DocumentEntry oProps = (DocumentEntry) origPFS.getRoot().getEntryCaseInsensitive(POWERPOINT_DOCUMENT);
+        DocumentEntry nProps = (DocumentEntry) newPFS.getRoot().getEntryCaseInsensitive(POWERPOINT_DOCUMENT);
         assertEquals(oProps.getSize(), nProps.getSize());
 
 
diff --git a/poi-scratchpad/src/test/java/org/apache/poi/hslf/extractor/TestExtractor.java b/poi-scratchpad/src/test/java/org/apache/poi/hslf/extractor/TestExtractor.java
index 6b8a08ff2d..4d70257650 100644
--- a/poi-scratchpad/src/test/java/org/apache/poi/hslf/extractor/TestExtractor.java
+++ b/poi-scratchpad/src/test/java/org/apache/poi/hslf/extractor/TestExtractor.java
@@ -185,8 +185,8 @@ public final class TestExtractor {
             };
 
             for (int i=0; i<TEST_SET.length; i+=2) {
-                DirectoryNode dir = (DirectoryNode)root.getEntry(TEST_SET[i]);
-                assertTrue(dir.hasEntry(HSLFSlideShow.POWERPOINT_DOCUMENT));
+                DirectoryNode dir = (DirectoryNode)root.getEntryCaseInsensitive(TEST_SET[i]);
+                assertTrue(dir.hasEntryCaseInsensitive(HSLFSlideShow.POWERPOINT_DOCUMENT));
 
                 try (final SlideShow<?,?> ppt = SlideShowFactory.create(dir);
                      final SlideShowExtractor<?,?> ppe = new SlideShowExtractor<>(ppt)) {
diff --git a/poi-scratchpad/src/test/java/org/apache/poi/hslf/model/TestTextRunReWrite.java b/poi-scratchpad/src/test/java/org/apache/poi/hslf/model/TestTextRunReWrite.java
index b38ae3ba51..658e2df370 100644
--- a/poi-scratchpad/src/test/java/org/apache/poi/hslf/model/TestTextRunReWrite.java
+++ b/poi-scratchpad/src/test/java/org/apache/poi/hslf/model/TestTextRunReWrite.java
@@ -79,8 +79,8 @@ public final class TestTextRunReWrite {
                 // Check that the "PowerPoint Document" sections have the same size
                 DirectoryNode oDir = ppt1.getSlideShowImpl().getDirectory();
 
-                DocumentEntry oProps = (DocumentEntry) oDir.getEntry(HSLFSlideShow.POWERPOINT_DOCUMENT);
-                DocumentEntry nProps = (DocumentEntry) npfs.getRoot().getEntry(HSLFSlideShow.POWERPOINT_DOCUMENT);
+                DocumentEntry oProps = (DocumentEntry) oDir.getEntryCaseInsensitive(HSLFSlideShow.POWERPOINT_DOCUMENT);
+                DocumentEntry nProps = (DocumentEntry) npfs.getRoot().getEntryCaseInsensitive(HSLFSlideShow.POWERPOINT_DOCUMENT);
                 assertEquals(oProps.getSize(), nProps.getSize());
 
                 // Check that they contain the same data
@@ -130,8 +130,8 @@ public final class TestTextRunReWrite {
                 // Check that the "PowerPoint Document" sections have the same size
                 DirectoryNode oDir = ppt1.getSlideShowImpl().getDirectory();
 
-                DocumentEntry oProps = (DocumentEntry) oDir.getEntry(HSLFSlideShow.POWERPOINT_DOCUMENT);
-                DocumentEntry nProps = (DocumentEntry) npfs.getRoot().getEntry(HSLFSlideShow.POWERPOINT_DOCUMENT);
+                DocumentEntry oProps = (DocumentEntry) oDir.getEntryCaseInsensitive(HSLFSlideShow.POWERPOINT_DOCUMENT);
+                DocumentEntry nProps = (DocumentEntry) npfs.getRoot().getEntryCaseInsensitive(HSLFSlideShow.POWERPOINT_DOCUMENT);
                 assertEquals(oProps.getSize(), nProps.getSize());
 
                 // Check that they contain the same data
diff --git a/poi-scratchpad/src/test/java/org/apache/poi/hslf/record/TestCurrentUserAtom.java b/poi-scratchpad/src/test/java/org/apache/poi/hslf/record/TestCurrentUserAtom.java
index a1ee955fd3..780e36500a 100644
--- a/poi-scratchpad/src/test/java/org/apache/poi/hslf/record/TestCurrentUserAtom.java
+++ b/poi-scratchpad/src/test/java/org/apache/poi/hslf/record/TestCurrentUserAtom.java
@@ -76,7 +76,7 @@ public final class TestCurrentUserAtom {
         // Get raw contents from a known file
         byte[] contents;
         try (POIFSFileSystem fs = new POIFSFileSystem(_slTests.getFile(normalFile))) {
-            DocumentEntry docProps = (DocumentEntry) fs.getRoot().getEntry("Current User");
+            DocumentEntry docProps = (DocumentEntry) fs.getRoot().getEntryCaseInsensitive("Current User");
             contents = new byte[docProps.getSize()];
             try (InputStream in = fs.getRoot().createDocumentInputStream("Current User")) {
                 in.read(contents);
diff --git a/poi-scratchpad/src/test/java/org/apache/poi/hslf/record/TestExOleObjStg.java b/poi-scratchpad/src/test/java/org/apache/poi/hslf/record/TestExOleObjStg.java
index 46d607e729..8563bf7c5e 100644
--- a/poi-scratchpad/src/test/java/org/apache/poi/hslf/record/TestExOleObjStg.java
+++ b/poi-scratchpad/src/test/java/org/apache/poi/hslf/record/TestExOleObjStg.java
@@ -62,7 +62,7 @@ public final class TestExOleObjStg {
         assertEquals(len, oledata.length);
 
         try (POIFSFileSystem fs = new POIFSFileSystem(record.getData())) {
-            DocumentEntry doc = (DocumentEntry) fs.getRoot().getEntry("Contents");
+            DocumentEntry doc = (DocumentEntry) fs.getRoot().getEntryCaseInsensitive("Contents");
             assertNotNull(doc);
         }
     }
diff --git a/poi-scratchpad/src/test/java/org/apache/poi/hsmf/parsers/TestPOIFSChunkParser.java b/poi-scratchpad/src/test/java/org/apache/poi/hsmf/parsers/TestPOIFSChunkParser.java
index f93ff19f0f..17d58effea 100644
--- a/poi-scratchpad/src/test/java/org/apache/poi/hsmf/parsers/TestPOIFSChunkParser.java
+++ b/poi-scratchpad/src/test/java/org/apache/poi/hsmf/parsers/TestPOIFSChunkParser.java
@@ -56,13 +56,13 @@ public final class TestPOIFSChunkParser {
     void testFindsCore() throws IOException, ChunkNotFoundException {
         try (POIFSFileSystem simple = new POIFSFileSystem(samples.getFile("quick.msg"), true)) {
 
-            // Check a few core things are present
-            simple.getRoot().getEntry(
-                    (new StringChunk(MAPIProperty.SUBJECT.id, Types.ASCII_STRING)).getEntryName()
-            );
-            simple.getRoot().getEntry(
-                    (new StringChunk(MAPIProperty.SENDER_NAME.id, Types.ASCII_STRING)).getEntryName()
-            );
+      // Check a few core things are present
+      simple.getRoot().getEntryCaseInsensitive(
+            (new StringChunk(MAPIProperty.SUBJECT.id, Types.ASCII_STRING)).getEntryName()
+      );
+      simple.getRoot().getEntryCaseInsensitive(
+            (new StringChunk(MAPIProperty.SENDER_NAME.id, Types.ASCII_STRING)).getEntryName()
+      );
 
             // Now load the file
             try (MAPIMessage msg = new MAPIMessage(simple)) {
@@ -82,7 +82,7 @@ public final class TestPOIFSChunkParser {
     void testFindsRecips() throws IOException, ChunkNotFoundException {
         try (POIFSFileSystem simple = new POIFSFileSystem(samples.getFile("quick.msg"), true)) {
 
-            simple.getRoot().getEntry("__recip_version1.0_#00000000");
+      simple.getRoot().getEntryCaseInsensitive("__recip_version1.0_#00000000");
 
             ChunkGroup[] groups = POIFSChunkParser.parse(simple.getRoot());
             assertEquals(3, groups.length);
@@ -136,12 +136,12 @@ public final class TestPOIFSChunkParser {
     void testFindsMultipleRecipients() throws IOException, ChunkNotFoundException {
         try (POIFSFileSystem multiple = new POIFSFileSystem(samples.getFile("example_received_unicode.msg"), true)) {
 
-            multiple.getRoot().getEntry("__recip_version1.0_#00000000");
-            multiple.getRoot().getEntry("__recip_version1.0_#00000001");
-            multiple.getRoot().getEntry("__recip_version1.0_#00000002");
-            multiple.getRoot().getEntry("__recip_version1.0_#00000003");
-            multiple.getRoot().getEntry("__recip_version1.0_#00000004");
-            multiple.getRoot().getEntry("__recip_version1.0_#00000005");
+      multiple.getRoot().getEntryCaseInsensitive("__recip_version1.0_#00000000");
+      multiple.getRoot().getEntryCaseInsensitive("__recip_version1.0_#00000001");
+      multiple.getRoot().getEntryCaseInsensitive("__recip_version1.0_#00000002");
+      multiple.getRoot().getEntryCaseInsensitive("__recip_version1.0_#00000003");
+      multiple.getRoot().getEntryCaseInsensitive("__recip_version1.0_#00000004");
+      multiple.getRoot().getEntryCaseInsensitive("__recip_version1.0_#00000005");
 
             ChunkGroup[] groups = POIFSChunkParser.parse(multiple.getRoot());
             assertEquals(9, groups.length);
@@ -228,7 +228,7 @@ public final class TestPOIFSChunkParser {
     void testFindsNameId() throws IOException {
         try (POIFSFileSystem simple = new POIFSFileSystem(samples.getFile("quick.msg"), true)) {
 
-            simple.getRoot().getEntry("__nameid_version1.0");
+      simple.getRoot().getEntryCaseInsensitive("__nameid_version1.0");
 
             ChunkGroup[] groups = POIFSChunkParser.parse(simple.getRoot());
             assertEquals(3, groups.length);
@@ -253,10 +253,11 @@ public final class TestPOIFSChunkParser {
                 POIFSFileSystem without = new POIFSFileSystem(samples.getFile("quick.msg"), true)) {
             AttachmentChunks attachment;
 
-            // Check raw details on the one with
-            with.getRoot().getEntry("__attach_version1.0_#00000000");
-            with.getRoot().getEntry("__attach_version1.0_#00000001");
-            POIFSChunkParser.parse(with.getRoot());
+
+      // Check raw details on the one with
+      with.getRoot().getEntryCaseInsensitive("__attach_version1.0_#00000000");
+      with.getRoot().getEntryCaseInsensitive("__attach_version1.0_#00000001");
+      POIFSChunkParser.parse(with.getRoot());
 
             ChunkGroup[] groups = POIFSChunkParser.parse(with.getRoot());
             assertEquals(5, groups.length);
@@ -276,9 +277,10 @@ public final class TestPOIFSChunkParser {
             assertEquals("pj1.txt", attachment.getAttachLongFileName().toString());
             assertEquals(89, attachment.getAttachData().getValue().length);
 
-            // Check raw details on one without
-            assertFalse(without.getRoot().hasEntry("__attach_version1.0_#00000000"));
-            assertFalse(without.getRoot().hasEntry("__attach_version1.0_#00000001"));
+
+      // Check raw details on one without
+      assertFalse(without.getRoot().hasEntryCaseInsensitive("__attach_version1.0_#00000000"));
+      assertFalse(without.getRoot().hasEntryCaseInsensitive("__attach_version1.0_#00000001"));
 
             // One with, from the top
             try (MAPIMessage msgWith = new MAPIMessage(with)) {
diff --git a/poi-scratchpad/src/test/java/org/apache/poi/hwpf/HWPFDocFixture.java b/poi-scratchpad/src/test/java/org/apache/poi/hwpf/HWPFDocFixture.java
index ea2a93cc12..54052acc58 100644
--- a/poi-scratchpad/src/test/java/org/apache/poi/hwpf/HWPFDocFixture.java
+++ b/poi-scratchpad/src/test/java/org/apache/poi/hwpf/HWPFDocFixture.java
@@ -44,7 +44,7 @@ public final class HWPFDocFixture
               POIDataSamples.getDocumentInstance().openResourceAsStream(_testFile));
 
       DocumentEntry documentProps =
-        (DocumentEntry) filesystem.getRoot().getEntry("WordDocument");
+        (DocumentEntry) filesystem.getRoot().getEntryCaseInsensitive("WordDocument");
       _mainStream = new byte[documentProps.getSize()];
       filesystem.createDocumentInputStream("WordDocument").read(_mainStream);
 
@@ -59,7 +59,7 @@ public final class HWPFDocFixture
 
       // read in the table stream.
       DocumentEntry tableProps =
-        (DocumentEntry) filesystem.getRoot().getEntry(name);
+        (DocumentEntry) filesystem.getRoot().getEntryCaseInsensitive(name);
       _tableStream = new byte[tableProps.getSize()];
       filesystem.createDocumentInputStream(name).read(_tableStream);
 
diff --git a/poi-scratchpad/src/test/java/org/apache/poi/hwpf/extractor/TestWordExtractor.java b/poi-scratchpad/src/test/java/org/apache/poi/hwpf/extractor/TestWordExtractor.java
index 4ef67c44c8..a53fefd7fe 100644
--- a/poi-scratchpad/src/test/java/org/apache/poi/hwpf/extractor/TestWordExtractor.java
+++ b/poi-scratchpad/src/test/java/org/apache/poi/hwpf/extractor/TestWordExtractor.java
@@ -142,15 +142,15 @@ public final class TestWordExtractor {
         POIFSFileSystem fs = new POIFSFileSystem(is);
         is.close();
 
-        DirectoryNode dirA = (DirectoryNode) fs.getRoot().getEntry("MBD0000A3B7");
-        DirectoryNode dirB = (DirectoryNode) fs.getRoot().getEntry("MBD0000A3B2");
+        DirectoryNode dirA = (DirectoryNode) fs.getRoot().getEntryCaseInsensitive("MBD0000A3B7");
+        DirectoryNode dirB = (DirectoryNode) fs.getRoot().getEntryCaseInsensitive("MBD0000A3B2");
 
         // Should have WordDocument and 1Table
-        assertNotNull(dirA.getEntry("1Table"));
-        assertNotNull(dirA.getEntry("WordDocument"));
+        assertNotNull(dirA.getEntryCaseInsensitive("1Table"));
+        assertNotNull(dirA.getEntryCaseInsensitive("WordDocument"));
 
-        assertNotNull(dirB.getEntry("1Table"));
-        assertNotNull(dirB.getEntry("WordDocument"));
+        assertNotNull(dirB.getEntryCaseInsensitive("1Table"));
+        assertNotNull(dirB.getEntryCaseInsensitive("WordDocument"));
 
         // Check each in turn
         HWPFDocument docA = new HWPFDocument(dirA);
@@ -403,6 +403,18 @@ public final class TestWordExtractor {
     }
 
     @Test
+    public void testCaseInsensitiveOLENames() throws Exception {
+        //test files are from Ross Johnson on TIKA-4091
+        for (String n : new String[]{"normal", "lower", "upper"}) {
+            try (InputStream is = docTests.openResourceAsStream("47950_" + n + ".doc");
+                 POIFSFileSystem fs = new POIFSFileSystem(is);
+                 WordExtractor wExt = new WordExtractor(fs)) {
+                assertContains(wExt.getText(), "This is a sample Word document");
+            }
+        }
+    }
+
+    @Test
     void testCapitalized() throws Exception {
         try (WordExtractor wExt = openExtractor("capitalized.doc")) {
             String text = wExt.getText().trim();