TIKA-4801: Extract cover art from MP3, MP4, Vorbis and FLAC audio

tika-parsers/tika-parsers-standard/tika-parsers-standard-modules/tika-parser-audiovideo-module/src/main/java/org/apache/tika/parser/mp3/ID3Tags.java

@@ -16,6 +16,7 @@
  */
 package org.apache.tika.parser.mp3;
 
+import java.util.Collections;
 import java.util.List;
 
 /**
@@ -163,6 +164,36 @@ public interface ID3Tags {
             /* 125 */ "Dance Hall",
             /* sentinel */ ""};
 
+    /**
+     * List of predefined picture types for embedded pictures, indexed
+     * by the type byte of the ID3v2 APIC / PIC frames. The FLAC picture
+     * block, as also embedded in Vorbis comments, reuses the same list.
+     * <p>
+     * See <a href="https://id3.org/id3v2.4.0-frames">http://id3.org/id3v2.4.0-frames</a>
+     */
+    String[] PICTURE_TYPES = new String[]{
+            /*  0 */ "Other",
+            /*  1 */ "32x32 pixels 'file icon' (PNG only)",
+            /*  2 */ "Other file icon",
+            /*  3 */ "Cover (front)",
+            /*  4 */ "Cover (back)",
+            /*  5 */ "Leaflet page",
+            /*  6 */ "Media (e.g. label side of CD)",
+            /*  7 */ "Lead artist/lead performer/soloist",
+            /*  8 */ "Artist/performer",
+            /*  9 */ "Conductor",
+            /* 10 */ "Band/Orchestra",
+            /* 11 */ "Composer",
+            /* 12 */ "Lyricist/text writer",
+            /* 13 */ "Recording Location",
+            /* 14 */ "During recording",
+            /* 15 */ "During performance",
+            /* 16 */ "Movie/video screen capture",
+            /* 17 */ "A bright coloured fish",
+            /* 18 */ "Illustration",
+            /* 19 */ "Band/artist logotype",
+            /* 20 */ "Publisher/Studio logotype"};
+
     /**
      * Does the file contain this kind of tags?
      */
@@ -198,6 +229,15 @@ public interface ID3Tags {
      */
     List<ID3Comment> getComments();
 
+    /**
+     * Retrieves the embedded pictures (e.g. cover art), if any.
+     * Only ID3v2 tags can carry pictures, so this defaults to
+     * an empty list.
+     */
+    default List<ID3Picture> getPictures() {
+        return Collections.emptyList();
+    }
+
     String getGenre();
 
     String getYear();
@@ -258,4 +298,50 @@ public String getText() {
             return text;
         }
     }
+
+    /**
+     * Represents an embedded picture in ID3 v2, such as cover art,
+     * as carried by the APIC (v2.3/v2.4) and PIC (v2.2) frames
+     */
+    class ID3Picture {
+        private final String mimeType;
+        private final String description;
+        private final int pictureType;
+        private final byte[] data;
+
+        public ID3Picture(String mimeType, String description, int pictureType, byte[] data) {
+            this.mimeType = mimeType;
+            this.description = description;
+            this.pictureType = pictureType;
+            this.data = data;
+        }
+
+        /**
+         * Gets the declared mime type, if present
+         */
+        public String getMimeType() {
+            return mimeType;
+        }
+
+        /**
+         * Gets the description, if present
+         */
+        public String getDescription() {
+            return description;
+        }
+
+        /**
+         * Gets the picture type byte, which indexes {@link #PICTURE_TYPES}
+         */
+        public int getPictureType() {
+            return pictureType;
+        }
+
+        /**
+         * Gets the raw picture data
+         */
+        public byte[] getData() {
+            return data;
+        }
+    }
 }

tika-parsers/tika-parsers-standard/tika-parsers-standard-modules/tika-parser-audiovideo-module/src/main/java/org/apache/tika/parser/mp3/ID3v22Handler.java

@@ -44,6 +44,7 @@ public class ID3v22Handler implements ID3Tags {
     private String albumArtist;
     private String disc;
     private List<ID3Comment> comments = new ArrayList<>();
+    private List<ID3Picture> pictures = new ArrayList<>();
 
     public ID3v22Handler(ID3v2Frame frame) throws IOException, SAXException, TikaException {
         RawTagIterator tags = new RawV22TagIterator(frame);
@@ -78,6 +79,13 @@ public ID3v22Handler(ID3v2Frame frame) throws IOException, SAXException, TikaExc
                     }
                     break;
                 }
+                case "PIC": {
+                    ID3Picture picture = ID3v2Frame.getV22Picture(tag.data, 0, tag.data.length);
+                    if (picture != null) {
+                        pictures.add(picture);
+                    }
+                    break;
+                }
                 case "TRK":
                     trackNumber = getTagString(tag.data, 0, tag.data.length);
                     break;
@@ -149,6 +157,10 @@ public List<ID3Comment> getComments() {
         return comments;
     }
 
+    public List<ID3Picture> getPictures() {
+        return pictures;
+    }
+
     public String getGenre() {
         return genre;
     }

tika-parsers/tika-parsers-standard/tika-parsers-standard-modules/tika-parser-audiovideo-module/src/main/java/org/apache/tika/parser/mp3/ID3v23Handler.java

@@ -45,6 +45,7 @@ public class ID3v23Handler implements ID3Tags {
     private String disc;
     private String compilation;
     private List<ID3Comment> comments = new ArrayList<>();
+    private List<ID3Picture> pictures = new ArrayList<>();
 
     public ID3v23Handler(ID3v2Frame frame) throws IOException, SAXException, TikaException {
         RawTagIterator tags = new RawV23TagIterator(frame);
@@ -79,6 +80,13 @@ public ID3v23Handler(ID3v2Frame frame) throws IOException, SAXException, TikaExc
                     }
                     break;
                 }
+                case "APIC": {
+                    ID3Picture picture = ID3v2Frame.getPicture(tag.data, 0, tag.data.length);
+                    if (picture != null) {
+                        pictures.add(picture);
+                    }
+                    break;
+                }
                 case "TRCK":
                     trackNumber = getTagString(tag.data, 0, tag.data.length);
                     break;
@@ -135,6 +143,10 @@ public List<ID3Comment> getComments() {
         return comments;
     }
 
+    public List<ID3Picture> getPictures() {
+        return pictures;
+    }
+
     public String getGenre() {
         return genre;
     }

tika-parsers/tika-parsers-standard/tika-parsers-standard-modules/tika-parser-audiovideo-module/src/main/java/org/apache/tika/parser/mp3/ID3v24Handler.java

@@ -46,6 +46,7 @@ public class ID3v24Handler implements ID3Tags {
     private String disc;
     private String compilation;
     private List<ID3Comment> comments = new ArrayList<>();
+    private List<ID3Picture> pictures = new ArrayList<>();
 
     public ID3v24Handler(ID3v2Frame frame) throws IOException, SAXException, TikaException {
         RawTagIterator tags = new RawV24TagIterator(frame);
@@ -85,6 +86,13 @@ public ID3v24Handler(ID3v2Frame frame) throws IOException, SAXException, TikaExc
                     }
                     break;
                 }
+                case "APIC": {
+                    ID3Picture picture = ID3v2Frame.getPicture(tag.data, 0, tag.data.length);
+                    if (picture != null) {
+                        pictures.add(picture);
+                    }
+                    break;
+                }
                 case "TRCK":
                     trackNumber = getTagString(tag.data, 0, tag.data.length);
                     break;
@@ -141,6 +149,10 @@ public List<ID3Comment> getComments() {
         return comments;
     }
 
+    public List<ID3Picture> getPictures() {
+        return pictures;
+    }
+
     public String getGenre() {
         return genre;
     }
@@ -163,7 +175,8 @@ public String getCompilation() {
 
     private class RawV24TagIterator extends RawTagIterator {
         private RawV24TagIterator(ID3v2Frame frame) {
-            frame.super(4, 4, 1, 2);
+            //v2.4 frame sizes are synchsafe integers
+            frame.super(4, 4, 1, 2, true);
         }
     }
 }