Merge branch 'feature/git-detect-media-type-better' of npcode/yobi

from pull request 977

app/controllers/CodeApp.java

@@ -24,6 +24,7 @@
 import controllers.annotation.IsAllowed;
 import models.Project;
 import models.enumeration.Operation;
+import org.apache.commons.io.FilenameUtils;
 import org.apache.tika.Tika;
 import org.codehaus.jackson.node.ObjectNode;
 import org.eclipse.jgit.api.errors.GitAPIException;
@@ -35,6 +36,7 @@
 import playRepository.PlayRepository;
 import playRepository.RepositoryService;
 import utils.ErrorViews;
+import utils.FileUtil;
 import views.html.code.nohead;
 import views.html.code.nohead_svn;
 import views.html.code.view;
@@ -254,6 +256,6 @@ public static Result openFile(String userName, String projectName, String revisi
             return notFound(ErrorViews.NotFound.render("error.notfound"));
         }
 
-        return ok(raw).as(tika.detect(raw));
+        return ok(raw).as(FileUtil.detectMediaType(raw, FilenameUtils.getName(path)));
     }
 }

app/models/Attachment.java

@@ -278,18 +278,7 @@ public boolean store(File file, String name, Resource container) throws IOExcept
         }
 
         if (this.mimeType == null) {
-            Metadata meta = new Metadata();
-            meta.add(Metadata.RESOURCE_NAME_KEY, this.name);
-            MediaType mediaType = new Tika().getDetector().detect(
-                    new BufferedInputStream(new FileInputStream(file)), meta);
-            this.mimeType = mediaType.toString();
-            if (mediaType.getType().toLowerCase().equals("text")) {
-                this.mimeType += "; charset=" + FileUtil.detectCharset(new FileInputStream(file));
-            } else if (mediaType.equals(MediaType.audio("ogg"))
-                    && FilenameUtils.getExtension(name).toLowerCase().equals("ogv")) {
-                // This fixes Tika's misjudge of media type for ogg videos.
-                this.mimeType = "video/ogg";
-            }
+            this.mimeType = FileUtil.detectMediaType(file, name);
         }
 
         // the size must be set before it is moved.

app/utils/FileUtil.java

@@ -20,12 +20,14 @@
  */
 package utils;
 
+import org.apache.commons.io.FilenameUtils;
 import org.apache.commons.lang.StringUtils;
+import org.apache.tika.Tika;
+import org.apache.tika.metadata.Metadata;
+import org.apache.tika.mime.MediaType;
 import org.mozilla.universalchardet.UniversalDetector;
 
-import java.io.File;
-import java.io.IOException;
-import java.io.InputStream;
+import java.io.*;
 
 public class FileUtil {
 
@@ -91,4 +93,42 @@ public static String detectCharset(InputStream is) throws IOException {
 
         return or(detector.getDetectedCharset(), "UTF-8");
     }
+
+    public static String detectMediaType(File file, String name) throws IOException {
+        return detectMediaType(new FileInputStream(file), name);
+    }
+
+    public static String detectMediaType(byte[] bytes, String name) throws IOException {
+        return detectMediaType(new ByteArrayInputStream(bytes), name);
+    }
+
+    /**
+     * Detects media type of the given resource, by using Apache Tika.
+     *
+     * This method does following additional tasks besides Tika:
+     * 1. Adds a charset parameter for text resources.
+     * 2. Fixes Tika's misjudge of media type for ogg videos
+     *
+     * @param is    the input stream to read the resource
+     * @param name  the filename of the resource
+     * @return the detected media type which optionally includes a charset parameter
+     *         e.g. "text/plain; charset=utf-8"
+     * @throws IOException
+     */
+    public static String detectMediaType(InputStream is, String name) throws IOException {
+        Metadata meta = new Metadata();
+        meta.add(Metadata.RESOURCE_NAME_KEY, name);
+        MediaType mediaType = new Tika().getDetector().detect(
+                new BufferedInputStream(is), meta);
+        String mimeType = mediaType.toString();
+        if (mediaType.getType().toLowerCase().equals("text")) {
+            mimeType += "; charset=" + FileUtil.detectCharset(is);
+        } else if (mediaType.equals(MediaType.audio("ogg"))
+                && FilenameUtils.getExtension(name).toLowerCase().equals("ogv")) {
+            // This fixes Tika's misjudge of media type for ogg videos.
+            mimeType = "video/ogg";
+        }
+
+        return mimeType;
+    }
 }