chore(search): move audio metadata extraction in tika to dedicated method

2023-12-06 15:13:09 +01:00
parent 1aa6112d1e
commit 07edd1e23a
1 changed files with 75 additions and 59 deletions
@@ -90,65 +90,7 @@ func (t Tika) Extract(ctx context.Context, ri *provider.ResourceInfo) (Document,
 		}

 		if contentType, err := getFirstValue(meta, "Content-Type"); err == nil && strings.HasPrefix(contentType, "audio/") {
-			audio := libregraph.Audio{}
-
-			if v, err := getFirstValue(meta, "xmpDM:album"); err == nil {
-				audio.SetAlbum(v)
-			}
-
-			if v, err := getFirstValue(meta, "xmpDM:albumArtist"); err == nil {
-				audio.SetAlbumArtist(v)
-			}
-
-			if v, err := getFirstValue(meta, "xmpDM:artist"); err == nil {
-				audio.SetArtist(v)
-			}
-
-			// TODO: audio.Bitrate: not provided by tika
-			// TODO: audio.Composers: not provided by tika
-			// TODO: audio.Copyright: not provided by tika for audio files?
-
-			if v, err := getFirstValue(meta, "xmpDM:discNumber"); err == nil {
-				if i, err := strconv.ParseInt(v, 10, 32); err == nil {
-					audio.SetDisc(int32(i))
-				}
-
-			}
-
-			//  TODO: audio.DiscCount: not provided by tika
-
-			if v, err := getFirstValue(meta, "xmpDM:duration"); err == nil {
-				if i, err := strconv.ParseInt(v, 10, 64); err == nil {
-					audio.SetDuration(i * 1000)
-				}
-			}
-
-			if v, err := getFirstValue(meta, "xmpDM:genre"); err == nil {
-				audio.SetGenre(v)
-			}
-
-			// TODO: audio.HasDrm: not provided by tika
-			// TODO: audio.IsVariableBitrate: not provided by tika
-
-			if v, err := getFirstValue(meta, "dc:title"); err == nil {
-				audio.SetTitle(v)
-			}
-
-			if v, err := getFirstValue(meta, "xmpDM:trackNumber"); err == nil {
-				if i, err := strconv.ParseInt(v, 10, 32); err == nil {
-					audio.SetTrack(int32(i))
-				}
-			}
-
-			// TODO: audio.TrackCount: not provided by tika
-
-			if v, err := getFirstValue(meta, "xmpDM:releaseDate"); err == nil {
-				if i, err := strconv.ParseInt(v, 10, 32); err == nil {
-					audio.SetYear(int32(i))
-				}
-			}
-
-			doc.Audio = &audio
+			doc.Audio = t.getAudio(meta)
 		}
 	}

@@ -158,3 +100,77 @@ func (t Tika) Extract(ctx context.Context, ri *provider.ResourceInfo) (Document,

 	return doc, nil
 }
+
+func (t Tika) getAudio(meta map[string][]string) *libregraph.Audio {
+	var audio *libregraph.Audio
+	initAudio := func() {
+		audio = libregraph.NewAudio()
+	}
+
+	if v, err := getFirstValue(meta, "xmpDM:album"); err == nil {
+		initAudio()
+		audio.SetAlbum(v)
+	}
+
+	if v, err := getFirstValue(meta, "xmpDM:albumArtist"); err == nil {
+		initAudio()
+		audio.SetAlbumArtist(v)
+	}
+
+	if v, err := getFirstValue(meta, "xmpDM:artist"); err == nil {
+		initAudio()
+		audio.SetArtist(v)
+	}
+
+	// TODO: audio.Bitrate: not provided by tika
+	// TODO: audio.Composers: not provided by tika
+	// TODO: audio.Copyright: not provided by tika for audio files?
+
+	if v, err := getFirstValue(meta, "xmpDM:discNumber"); err == nil {
+		if i, err := strconv.ParseInt(v, 10, 32); err == nil {
+			initAudio()
+			audio.SetDisc(int32(i))
+		}
+
+	}
+
+	//  TODO: audio.DiscCount: not provided by tika
+
+	if v, err := getFirstValue(meta, "xmpDM:duration"); err == nil {
+		if i, err := strconv.ParseInt(v, 10, 64); err == nil {
+			initAudio()
+			audio.SetDuration(i * 1000)
+		}
+	}
+
+	if v, err := getFirstValue(meta, "xmpDM:genre"); err == nil {
+		initAudio()
+		audio.SetGenre(v)
+	}
+
+	// TODO: audio.HasDrm: not provided by tika
+	// TODO: audio.IsVariableBitrate: not provided by tika
+
+	if v, err := getFirstValue(meta, "dc:title"); err == nil {
+		initAudio()
+		audio.SetTitle(v)
+	}
+
+	if v, err := getFirstValue(meta, "xmpDM:trackNumber"); err == nil {
+		if i, err := strconv.ParseInt(v, 10, 32); err == nil {
+			initAudio()
+			audio.SetTrack(int32(i))
+		}
+	}
+
+	// TODO: audio.TrackCount: not provided by tika
+
+	if v, err := getFirstValue(meta, "xmpDM:releaseDate"); err == nil {
+		if i, err := strconv.ParseInt(v, 10, 32); err == nil {
+			initAudio()
+			audio.SetYear(int32(i))
+		}
+	}
+
+	return audio
+}