Merge pull request #45 from levigo/feature/JS-2500_iwork_office_types

feat(JS-2500): recognize iWork office types

Author: welschsn

pom.xml

@@ -52,30 +52,30 @@
     <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>1.26.2</commons-compress.version>
+    <commons.io.version>2.18.0</commons.io.version>
+    <commons-compress.version>1.27.1</commons-compress.version>
     <opennlp-tools.version>1.9.3</opennlp-tools.version>
 
-    <pdfbox.version>2.0.31</pdfbox.version>
+    <pdfbox.version>2.0.32</pdfbox.version>
 
     <jaxb-api.version>4.0.2</jaxb-api.version>
     <jaxb-core.version>4.0.5</jaxb-core.version>
     <jaxb-impl.version>4.0.5</jaxb-impl.version>
 
     <!-- testing -->
-    <junit.version>5.10.3</junit.version>
+    <junit.version>5.11.3</junit.version>
     <hamcrest.version>2.2</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

@@ -70,7 +70,10 @@ public static Stream<Arguments> dataProviderVarious() {
         arguments("/various_types/File-PNG_8pbc_GRAY.png", "image/png", "PNG image data, grayscale,", "png"),
         arguments("/various_types/File-PNG_8bpc_RGBA.png", "image/png", "PNG image data, \\b/color RGBA,", "png"),
         arguments("/various_types/File-PNG_16bpc_RGBA.png", "image/png", "PNG image data, \\b/color RGBA,", "png"),
-        arguments("/various_types/sample-rtf-files-sample2.rtf", "text/rtf", "Rich Text Format data", "rtf")
+        arguments("/various_types/sample-rtf-files-sample2.rtf", "text/rtf", "Rich Text Format data", "rtf"),
+        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")
     );
   }