Simplify MimeType class and parse mimetypes in a more lenient way

modules/files/src/main/scala/docspell/files/TikaMimetype.scala

@@ -31,10 +31,10 @@ object TikaMimetype {
   private def convert(mt: MediaType): MimeType =
     Option(mt) match {
       case Some(_) =>
-        val params = mt.getParameters.asScala.toMap
+        val cs = mt.getParameters.asScala.toMap.get("charset").getOrElse("unknown")
         val primary = mt.getType
         val sub = mt.getSubtype
-        normalize(MimeType(primary, sub, params))
+        normalize(MimeType(primary, sub, None).withCharsetName(cs))
       case None =>
         MimeType.octetStream
     }
@@ -48,8 +48,8 @@ object TikaMimetype {
 
   private def normalize(in: MimeType): MimeType =
     in match {
-      case MimeType(_, sub, p) if sub contains "xhtml" =>
-        MimeType.html.copy(params = p)
+      case MimeType(_, sub, cs) if sub contains "xhtml" =>
+        MimeType.html.copy(charset = cs)
       case _ => in
     }
 
@@ -86,7 +86,7 @@ object TikaMimetype {
   def resolve[F[_]: Sync](dt: DataType, data: Stream[F, Byte]): F[MimeType] =
     dt match {
       case DataType.Exact(mt) =>
-        mt.resolveCharset match {
+        mt.charset match {
           case None if mt.primary == "text" =>
             detectCharset[F](data, MimeTypeHint.advertised(mt))
               .map {

modules/files/src/test/scala/docspell/files/TikaMimetypeTest.scala

@@ -0,0 +1,72 @@
+/*
+ * Copyright 2020 Eike K. & Contributors
+ *
+ * SPDX-License-Identifier: AGPL-3.0-or-later
+ */
+
+package docspell.files
+
+import docspell.common.{MimeType, MimeTypeHint}
+
+import munit.FunSuite
+import scodec.bits.ByteVector
+
+class TikaMimetypeTest extends FunSuite {
+
+  private def detect(bv: ByteVector, hint: MimeTypeHint): MimeType =
+    TikaMimetype.detect(bv, hint)
+
+  test("detect text/plain") {
+    val mt = detect(ByteVector.view("hello world".getBytes), MimeTypeHint.none)
+    assertEquals(mt.baseType, MimeType.plain)
+  }
+
+  test("detect image/jpeg") {
+    val mt = detect(
+      ByteVector.fromValidBase64("/9j/4AAQSkZJRgABAgAAZABkAAA="),
+      MimeTypeHint.none
+    )
+    assertEquals(mt, MimeType.jpeg)
+  }
+
+  test("detect image/png") {
+    val mt = detect(
+      ByteVector.fromValidBase64("iVBORw0KGgoAAAANSUhEUgAAA2I="),
+      MimeTypeHint.none
+    )
+    assertEquals(mt, MimeType.png)
+  }
+
+  test("detect application/json") {
+    val mt =
+      detect(
+        ByteVector.view("""{"name":"me"}""".getBytes),
+        MimeTypeHint.filename("me.json")
+      )
+    assertEquals(mt, MimeType.json)
+  }
+
+  test("detect application/json") {
+    val mt = detect(
+      ByteVector.view("""{"name":"me"}""".getBytes),
+      MimeTypeHint.advertised("application/json")
+    )
+    assertEquals(mt, MimeType.json)
+  }
+
+  test("detect image/jpeg wrong advertised") {
+    val mt = detect(
+      ByteVector.fromValidBase64("/9j/4AAQSkZJRgABAgAAZABkAAA="),
+      MimeTypeHint.advertised("image/png")
+    )
+    assertEquals(mt, MimeType.jpeg)
+  }
+
+  test("just filename") {
+    assertEquals(
+      detect(ByteVector.empty, MimeTypeHint.filename("doc.pdf")),
+      MimeType.pdf
+    )
+  }
+
+}