refactor: optimize file type detection and support decision-making based on file name (#6675)

#### What type of PR is this?
/kind improvement
/area core
/milestone 2.20.x

#### What this PR does / why we need it:
优化文件类型检测并支持根据文件名作为决策依据

#### Does this PR introduce a user-facing change?
```release-note
优化文件类型检测并支持根据文件名作为决策依据
```

api/src/main/java/run/halo/app/infra/utils/FileTypeDetectUtils.java

@@ -1,29 +1,52 @@
 package run.halo.app.infra.utils;
 
+import java.io.BufferedInputStream;
 import java.io.IOException;
 import java.io.InputStream;
 import lombok.experimental.UtilityClass;
-import org.apache.tika.Tika;
+import org.apache.tika.detect.DefaultDetector;
+import org.apache.tika.detect.Detector;
+import org.apache.tika.metadata.Metadata;
+import org.apache.tika.metadata.TikaCoreProperties;
 import org.apache.tika.mime.MimeTypeException;
 import org.apache.tika.mime.MimeTypes;
+import org.springframework.util.Assert;
 
 @UtilityClass
 public class FileTypeDetectUtils {
 
-    private static final Tika tika = new Tika();
+    private static final Detector detector = new DefaultDetector();
+
+    /**
+     * <p>Detects the media type of the given document.</p>
+     * <p>The type detection is based on the content of the given document stream and the name of
+     * the document.</p>
+     *
+     * @param inputStream the document stream must not be null
+     * @throws IOException if the stream can not be read
+     */
+    public static String detectMimeType(InputStream inputStream, String name) throws IOException {
+        Assert.notNull(name, "The name of the document must not be null");
+        var metadata = new Metadata();
+        metadata.set(TikaCoreProperties.RESOURCE_NAME_KEY, name);
+        return doDetectMimeType(inputStream, metadata);
+    }
 
     /**
      * Detect mime type.
      *
-     * @param inputStream input stream will be closed after detection.
+     * @param inputStream input stream will be closed after detection, must not be null
      */
     public static String detectMimeType(InputStream inputStream) throws IOException {
-        try {
-            return tika.detect(inputStream);
-        } finally {
-            if (inputStream != null) {
-                inputStream.close();
+        return doDetectMimeType(inputStream, new Metadata());
     }
+
+    private static String doDetectMimeType(InputStream inputStream, Metadata metadata)
+        throws IOException {
+        Assert.notNull(inputStream, "The inputStream must not be null");
+        try (var stream = (!inputStream.markSupported()
+            ? new BufferedInputStream(inputStream) : inputStream)) {
+            return detector.detect(stream, metadata).toString();
         }
     }
 

application/src/main/java/run/halo/app/core/attachment/endpoint/LocalAttachmentUploadHandler.java

@@ -158,7 +158,7 @@ class LocalAttachmentUploadHandler implements AttachmentHandler {
             var typeValidator = file.content()
                 .next()
                 .handle((dataBuffer, sink) -> {
-                    var mimeType = detectMimeType(dataBuffer.asInputStream());
+                    var mimeType = detectMimeType(dataBuffer.asInputStream(), file.name());
                     var isAllow = setting.getAllowedFileTypes()
                         .stream()
                         .map(FileCategoryMatcher::of)
@@ -178,9 +178,9 @@ class LocalAttachmentUploadHandler implements AttachmentHandler {
     }
 
     @NonNull
-    private String detectMimeType(InputStream inputStream) {
+    private String detectMimeType(InputStream inputStream, String name) {
         try {
-            return FileTypeDetectUtils.detectMimeType(inputStream);
+            return FileTypeDetectUtils.detectMimeType(inputStream, name);
         } catch (IOException e) {
             log.warn("Failed to detect file type", e);
             return "Unknown";

application/src/test/java/run/halo/app/infra/utils/FileTypeDetectUtilsTest.java

@@ -3,6 +3,7 @@ package run.halo.app.infra.utils;
 import static org.assertj.core.api.Assertions.assertThat;
 
 import java.io.IOException;
+import java.io.InputStream;
 import java.nio.file.Files;
 import org.apache.tika.mime.MimeTypeException;
 import org.junit.jupiter.api.Test;
@@ -31,6 +32,60 @@ class FileTypeDetectUtilsTest {
         assertThat(mimeType).isEqualTo("application/zip");
     }
 
+    @Test
+    void detectMimeTypeWithNameTest() throws IOException {
+        var stream = getFileInputStream("classpath:file-type-detect/index.js");
+        String mimeType = FileTypeDetectUtils.detectMimeType(stream, "index.js");
+        assertThat(mimeType).isEqualTo("application/javascript");
+
+        stream = getFileInputStream("classpath:file-type-detect/index.html");
+        mimeType =
+            FileTypeDetectUtils.detectMimeType(stream, "index.html");
+        assertThat(mimeType).isEqualTo("text/html");
+
+        stream = getFileInputStream("classpath:file-type-detect/test.json");
+        mimeType = FileTypeDetectUtils.detectMimeType(stream, "test.json");
+        assertThat(mimeType).isEqualTo("application/json");
+
+        stream = getFileInputStream("classpath:file-type-detect/other.xlsx");
+        mimeType = FileTypeDetectUtils.detectMimeType(stream, "other.xlsx");
+        assertThat(mimeType).isEqualTo(
+            "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet");
+
+        // other.xlsx detect without name
+        stream = getFileInputStream("classpath:file-type-detect/other.xlsx");
+        mimeType = FileTypeDetectUtils.detectMimeType(stream);
+        assertThat(mimeType).isEqualTo("application/zip");
+
+        // other.xlsx detect with wrong name
+        stream = getFileInputStream("classpath:file-type-detect/other.xlsx");
+        mimeType = FileTypeDetectUtils.detectMimeType(stream, "other.txt");
+        assertThat(mimeType).isEqualTo("application/zip");
+
+        stream = getFileInputStream("classpath:file-type-detect/test.docx");
+        mimeType = FileTypeDetectUtils.detectMimeType(stream, "test.docx");
+        assertThat(mimeType).isEqualTo(
+            "application/vnd.openxmlformats-officedocument.wordprocessingml.document");
+
+        // docx detect without file name
+        stream = getFileInputStream("classpath:file-type-detect/test.docx");
+        mimeType = FileTypeDetectUtils.detectMimeType(stream);
+        assertThat(mimeType).isEqualTo("application/zip");
+
+        stream = getFileInputStream("classpath:file-type-detect/test.svg");
+        mimeType = FileTypeDetectUtils.detectMimeType(stream, "test.svg");
+        assertThat(mimeType).isEqualTo("image/svg+xml");
+
+        stream = getFileInputStream("classpath:file-type-detect/test.png");
+        mimeType = FileTypeDetectUtils.detectMimeType(stream, "test.png");
+        assertThat(mimeType).isEqualTo("image/png");
+    }
+
+    private static InputStream getFileInputStream(String location) throws IOException {
+        var file = ResourceUtils.getFile(location);
+        return Files.newInputStream(file.toPath());
+    }
+
     @Test
     void detectFileExtensionTest() throws MimeTypeException {
         var ext = FileTypeDetectUtils.detectFileExtension("application/x-x509-key; format=pem");

application/src/test/resources/file-type-detect/test.svg

@@ -0,0 +1 @@
+<svg xmlns="http://www.w3.org/2000/svg" width="32" height="32" viewBox="0 0 24 24"><path fill="currentColor" d="M12 5q-.425 0-.712-.288Q11 4.425 11 4t.288-.713Q11.575 3 12 3t.713.287Q13 3.575 13 4t-.287.712Q12.425 5 12 5Zm0-3q-.425 0-.712-.288Q11 1.425 11 1t.288-.713Q11.575 0 12 0t.713.287Q13 .575 13 1t-.287.712Q12.425 2 12 2ZM8.4 22q-1.6-.95-2.5-2.538Q5 17.875 5 16q0-2.025 1.075-3.738Q7.15 10.55 9 9.675V7q0-.425.288-.713Q9.575 6 10 6h4q.425 0 .713.287Q15 6.575 15 7v2.675q1.825.875 2.913 2.587Q19 13.975 19 16q0 1.85-.913 3.45q-.912 1.6-2.487 2.55Zm.6-2h6q.95-.725 1.475-1.763Q17 17.2 17 16q0-1.45-.762-2.663q-.763-1.212-2.088-1.837L13 10.95V8h-2v2.95l-1.15.55q-1.325.625-2.087 1.837Q7 14.55 7 16q0 1.2.525 2.237Q8.05 19.275 9 20ZM7 3q-.425 0-.713-.288Q6 2.425 6 2t.287-.713Q6.575 1 7 1t.713.287Q8 1.575 8 2t-.287.712Q7.425 3 7 3Zm10 0q-.425 0-.712-.288Q16 2.425 16 2t.288-.713Q16.575 1 17 1t.712.287Q18 1.575 18 2t-.288.712Q17.425 3 17 3ZM9 5q-.425 0-.712-.288Q8 4.425 8 4t.288-.713Q8.575 3 9 3t.713.287Q10 3.575 10 4t-.287.712Q9.425 5 9 5Zm6 0q-.425 0-.712-.288Q14 4.425 14 4t.288-.713Q14.575 3 15 3t.713.287Q16 3.575 16 4t-.287.712Q15.425 5 15 5Zm-3 9Z"/></svg>