Merge pull request #46 from levigo/feature/JS-2500_iwork_office_types_1

Feature/js 2500 iwork office types 1

Author: welschsn

pom.xml

@@ -52,29 +52,29 @@
     <zip4j.version>2.11.5</zip4j.version>
     <slf4j.version>1.7.36</slf4j.version>
 
-    <poi.version>5.2.5</poi.version>
+    <poi.version>5.3.0</poi.version>
     <rtfparserkit.version>1.16.0</rtfparserkit.version>
-    <commons.io.version>2.16.1</commons.io.version>
-    <commons-compress.version>1.27.0</commons-compress.version>
-    <opennlp-tools.version>1.9.3</opennlp-tools.version>
+    <commons.io.version>2.18.0</commons.io.version>
+    <commons-compress.version>1.27.1</commons-compress.version>
+    <opennlp-tools.version>1.9.4</opennlp-tools.version>
 
-    <pdfbox.version>2.0.31</pdfbox.version>
+    <pdfbox.version>2.0.32</pdfbox.version>
 
     <jaxb-api.version>2.3.1</jaxb-api.version>
-    <jaxb-runtime.version>2.3.6</jaxb-runtime.version>
+    <jaxb-runtime.version>2.3.9</jaxb-runtime.version>
 
     <!-- testing -->
-    <junit.version>5.11.0</junit.version>
+    <junit.version>5.11.3</junit.version>
     <hamcrest.version>3.0</hamcrest.version>
     <jmock-junit5.version>2.13.1</jmock-junit5.version>
-    <log4j.version>2.23.1</log4j.version>
+    <log4j.version>2.24.2</log4j.version>
 
     <!-- sonarcloud -->
     <sonar.projectKey>levigo_filetype-analyzer</sonar.projectKey>
     <sonar.organization>levigo</sonar.organization>
     <sonar.host.url>https://sonarcloud.io</sonar.host.url>
 
-    <bouncycastle.version>1.78.1</bouncycastle.version>
+    <bouncycastle.version>1.79</bouncycastle.version>
   </properties>
 
   <build>

src/main/java/org/jadice/filetype/matchers/OfficeOpenXMLMatcher.java

@@ -191,7 +191,10 @@ public enum OfficeOpenType {
         "application/vnd.ms-visio.template.main+xml"),
     VISIO_TEMPLATE_MACRO_ENABLED("application/vnd.ms-visio.template.macroEnabled.12",
         "Microsoft Visio 2013 macro-enabled Template", "vstm",
-        "application/vnd.ms-visio.template.macroEnabled.main+xml");
+        "application/vnd.ms-visio.template.macroEnabled.main+xml"),
+    APPLE_KEYNOTE("application/vnd.apple.keynote","Apple Keynote Document", "key")    ,
+    APPLE_NUMBERS("application/vnd.apple.numbers", "Apple Numbers Document", "numbers"),
+    APPLE_PAGES("application/vnd.apple.pages", "Apple Pages Document", "pages"),;
 
 
     final String mimeType;
@@ -258,6 +261,12 @@ private void detect(final Context ctx, final ZipFile archive) throws IOException
     final SortedSet<Relationship> rels = getRelationships(archive);
     final List<ContentType> contentTypes = getContentTypes(archive);
     if (rels.isEmpty() || contentTypes.isEmpty()) {
+      final OfficeOpenType type = findIWorkDocumentType(ctx, archive);
+      if (type != null) {
+        ctx.setProperty(MimeTypeAction.KEY, type.mimeType);
+        ctx.setProperty(DescriptionAction.KEY, type.description);
+        ctx.setProperty(ExtensionAction.KEY, type.extension);
+      }
       return;
     }
 
@@ -283,6 +292,27 @@ private void detect(final Context ctx, final ZipFile archive) throws IOException
 
   }
 
+  private OfficeOpenType findIWorkDocumentType(final Context ctx, final ZipFile archive) throws IOException {
+    final InputStream indexDocument = getSafeInputStream("Index/Document.iwa", archive);
+    final InputStream metadataDocumentIdentifier = getSafeInputStream("Metadata/DocumentIdentifier", archive);
+    OfficeOpenType mimeType = null;
+    if (indexDocument != null && metadataDocumentIdentifier != null) {
+      switch (ctx.getStatedExtension()){
+        case "pages" :
+          mimeType = OfficeOpenType.APPLE_PAGES;
+          break;
+        case "key" :
+          mimeType = OfficeOpenType.APPLE_KEYNOTE;
+          break;
+        case "numbers" :
+          mimeType = OfficeOpenType.APPLE_NUMBERS;
+          break;
+        default:
+      }
+    }
+    return mimeType;
+  }
+
   private OfficeOpenType findDocumentType(final SortedSet<Relationship> rels, final List<ContentType> contentTypes) {
 
     final Relationship mainDocRel = getMainDocument(rels);
@@ -354,7 +384,7 @@ private Map<String, Object> readXMLMetaData(final InputStream inputStream) throw
       if (n instanceof Element) {
         String key = n.getNodeName();
         String val = DOMUtil.getNodeText(n);
-        if (val != null && val.length() > 0) {
+        if (val != null && !val.isEmpty()) {
           // Ignore empty information
           result.put(key, val);
         }

src/test/java/TestVariousTypes.java

@@ -64,7 +64,32 @@ void testEmptyStream() throws Exception {
     assertEquals("Binary data, ASCII Text Document", results.get(DescriptionAction.KEY));
   }
 
-  public static Stream<Arguments> dataProvider() {
+  public static Stream<Arguments> dataProviderVarious() {
+    return Stream.of(
+        arguments("/various_types/iworks_pages_file.pages", "application/vnd.apple.pages", "Apple Pages Document", "pages"),
+        arguments("/various_types/iworks_numbers_file.numbers", "application/vnd.apple.numbers", "Apple Numbers Document", "numbers"),
+        arguments("/various_types/iworks_keynote_file.key", "application/vnd.apple.keynote", "Apple Keynote Document", "key")
+    );
+  }
+
+  @ParameterizedTest
+  @MethodSource("dataProviderVarious")
+  void testVariousTypesExplicitly(String resource, String expectedMimeType, String expectedDescription, String expectedExtension) throws Exception {
+    final URL url = getClass().getResource(resource);
+    assertNotNull(url);
+    final File file = new File(url.toURI());
+    final Map<String, Object> results = analyzer.analyze(file);
+    printResult(results);
+    assertNotNull(results, file + " could not be analyzed");
+    assertNotNull(results.get(MimeTypeAction.KEY), "mimeType missing");
+    assertEquals(expectedMimeType, results.get(MimeTypeAction.KEY), "wrong mimeType");
+    assertNotNull(results.get(DescriptionAction.KEY), "description missing");
+    assertEquals(expectedDescription, results.get(DescriptionAction.KEY), "wrong description");
+    assertNotNull(results.get(ExtensionAction.KEY), "could not be analyzed");
+    assertEquals(expectedExtension, results.get(ExtensionAction.KEY), "wrong extension");
+  }
+
+  public static Stream<Arguments> dataProviderXRechnung() {
     return Stream.of(
         arguments("/various_types/BASIC_Einfach.pdf", "application/pdf"),
         arguments("/various_types/EN16931_Einfach.pdf", "application/pdf"),
@@ -75,7 +100,7 @@ public static Stream<Arguments> dataProvider() {
   }
 
   @ParameterizedTest
-  @MethodSource("dataProvider")
+  @MethodSource("dataProviderXRechnung")
   void testXRechnung(String resource, String expectedMimeType) throws Exception {
     final URL url = getClass().getResource(resource);
     assertNotNull(url);
@@ -109,7 +134,7 @@ private void ensureXRechnungIsTrue(final Map<String, Object> results, final Stri
   }
 
 
-  private static void printResult(final Map<String, Object> results) {
+  public static void printResult(final Map<String, Object> results) {
     for (final Map.Entry<String, Object> e : results.entrySet()) {
       LOGGER.info("   {}={}", e.getKey(), e.getValue());
     }