Changeset 3479623

Timestamp:

03/10/2026 11:54:41 PM (2 weeks ago)

Author:

dglingren

Message:

Fix XMP metadata extraction for MP3 files with embedded images.

Location:

media-library-assistant/trunk

Files:

• includes/class-mla-core.php (modified) (1 diff)
• includes/class-mla-data.php (modified) (5 diffs)
• index.php (modified) (1 diff)
• readme.txt (modified) (1 diff)

media-library-assistant/trunk/includes/class-mla-core.php

@@ -31 +31 @@
      * @var string
      */
-    const MLA_DEVELOPMENT_VERSION = '20260307';
+    const MLA_DEVELOPMENT_VERSION = '20260310';
 
     /**

media-library-assistant/trunk/includes/class-mla-data.php

@@ -2829 +2829 @@
 //error_log( __LINE__ . " MLAData::mla_parse_xmp_metadata( {$new_offset} ) ", 0 );
                 $xmp_chunk = file_get_contents( $file_name, true, NULL, $new_offset, $chunksize );
-//error_log( __LINE__ . " MLAData::mla_parse_xmp_metadata( {$new_offset} ) chunk = \r\n" . MLAData::mla_hex_dump( $xmp_chunk ), 0 );
+//error_log( __LINE__ . " MLAData::mla_parse_xmp_metadata( {$new_offset} ) chunk = \r\n" . MLAData::mla_hex_dump( $xmp_chunk ), 0, 32, 0 );
             } // while not found
         } else { // if not found
@@ -2835 +2835 @@
         }
 
-//error_log( __LINE__ . " MLAData::mla_parse_xmp_metadata( {$start_tag} ) ", 0 );
+//error_log( __LINE__ . " MLAData::mla_parse_xmp_metadata( offset {$new_offset}, start {$start_tag} ) ", 0 );
         if ( false === $start_tag ) {
             return NULL;
         }
 
-        // If necessary and possible, expand the $xmp_chunk until it contains the start tag
-        if ( false === ( $end_tag = strpos( $xmp_chunk, '</x:xmpmeta>', $start_tag ) ) && ( $chunksize == strlen( $xmp_chunk ) ) ) {
+        // If necessary and possible, expand the $xmp_chunk until it contains the end tag
+        if ( false === ( $end_tag = strpos( $xmp_chunk, '</x:xmpmeta>', $start_tag ) ) && ( $chunksize === strlen( $xmp_chunk ) ) ) {
             $new_offset = $new_offset + $start_tag;
             $start_tag = 0;
@@ -2856 +2856 @@
         }
 
-        $xmp_string = "<?xml version='1.0'?>\n" . substr( $xmp_chunk, $start_tag, ( $end_tag + 12 ) - $start_tag );
+        $length =  ( $end_tag + 12 ) - $start_tag;
+//error_log( __LINE__ . " MLAData::mla_parse_xmp_metadata( offset {$new_offset}, start {$start_tag}, end {$end_tag}, length {$length} ) ", 0 );
+        $xmp_string = "<?xml version='1.0'?>\n" . substr( $xmp_chunk, $start_tag, $length );
 //error_log( __LINE__ . " MLAData::mla_parse_xmp_metadata xmp_string = " . var_export( $xmp_string, true ), 0 );
 //error_log( __LINE__ . "  MLAData::mla_parse_xmp_metadata xmp_string = \r\n" . MLAData::mla_hex_dump( $xmp_string ), 0 );
 
         $results = MLAData::mla_parse_xml_string( $xmp_string );
+        if ( is_array( $results ) ) {
+            $results['xmp_offset'] = $new_offset + $start_tag;
+            $results['xmp_length'] = $length;
 //error_log( __LINE__ . " MLAData::mla_parse_xmp_metadata results = " . var_export( $results, true ), 0 );
-        return $results;
+
+            //$more_results = self::mla_parse_xmp_metadata( $file_name, $new_offset + $start_tag + $length );
+//error_log( __LINE__ . " MLAData::mla_parse_xmp_metadata more_results = " . var_export( $results, true ), 0 );
+        }
+
+return $results;
     }
 
@@ -4279 +4289 @@
         $id3_metadata = self::mla_fetch_attachment_id3_metadata( $post_id );
         if ( ! empty( $id3_metadata ) && !isset( $id3_metadata['error'] ) ) {
+            // Replace image blobs with a text message
+            if ( isset( $id3_metadata['comments']['picture'] ) && is_array( $id3_metadata['comments']['picture'] ) ) {
+                foreach ( $id3_metadata['comments']['picture'] as $key => $picture ) {
+                    if ( isset( $picture['data'] ) ) {
+                        $image_size = strlen( $picture['data'] );
+                        $id3_metadata['comments']['picture'][ $key ]['data'] = '( ' . $image_size . ' ' . __( 'bytes of BINARY DATA', 'media-library-assistant' ) . ' )';
+                    }
+                }
+
+            }
+
+            if ( isset( $id3_metadata['image']['data'] ) ) {
+                $image_size = strlen( $id3_metadata['image']['data'] );
+                $id3_metadata['image']['data'] = '( ' . $image_size . ' ' . __( 'bytes of BINARY DATA', 'media-library-assistant' ) . ' )';
+            }
+
             $text .= self::_compose_metadata_array( $id3_metadata, 'id3', ':' );
         }
 
         return $text; // var_export( $metadata, true );
+    }
+
+    /**
+     * Fetch XMP metadata for an MP3 audio file.
+     * 
+     * @since 3.35
+     *
+     * @param   string  Path to the MP3 file.
+     *
+     * @return  array   XMP Meta data variables, if present, or an empty string
+     */
+    public static function mla_fetch_mp3_xmp_metadata( $path ) {
+        $results = array(
+            'mla_id3_xmp_metadata' => array(),
+            'mla_id3_xmp_errors' => array(),
+        );
+
+        $data = file_get_contents( $path );
+        if ( false === $data ) {
+            $results['mla_id3_xmp_errors'][] = "Unable to read MP3 file";
+            return $results;
+        }
+
+        if ( 'ID3' !== substr( $data, 0, 3 ) ) {
+            $results['mla_id3_xmp_errors'][] = "Invalid MP3 file; no ID3 tag found";
+            return $results;
+        }
+
+        // Parse synchsafe tag size
+        $tagSize = (
+            (ord($data[6]) & 0x7F) << 21 |
+            (ord($data[7]) & 0x7F) << 14 |
+            (ord($data[8]) & 0x7F) << 7  |
+            (ord($data[9]) & 0x7F)
+        );
+
+        $tagData = substr( $data, 10, $tagSize );
+        $offset = 0;
+        $len = strlen( $tagData );
+        $xmp_frame = '';
+
+        while ($offset + 10 <= $len) {
+            $frameId   = substr( $tagData, $offset, 4) ;
+            $frameSize = unpack( "N", substr( $tagData, $offset + 4, 4 ) )[1];
+
+            if ( $frameSize <= 0 || $offset + 10 + $frameSize > $len ) {
+                $results['mla_id3_xmp_errors'][] = "Invalid MP3 file; bad frame size at offset {$offset}";
+                break;
+            }
+
+//error_log( __LINE__ . " MLAData::mla_fetch_mp3_xmp_metadata( $offset, $frameId, $frameSize )", 0 );
+            $frameData = substr( $tagData, $offset + 10, $frameSize );
+
+            // XMP FRAME (XMP , XMP\0, XMP_)
+            if ( str_starts_with( $frameId, "XMP" ) ) {
+                $xmp_frame = $frameData;
+                break;
+            }
+
+            // PRIV FRAME
+            if ( $frameId === "PRIV" ) {
+
+                // Owner identifier (null-terminated)
+                $nullPos = strpos( $frameData, "\x00" );
+
+                if ( false !== $nullPos ) {
+                    if ( 'XMP' === substr($frameData, 0, $nullPos) ) {
+                        $xmp_frame = substr($frameData, $nullPos + 1);
+                        break;
+                    }
+                }
+            }
+
+            $offset += 10 + $frameSize;
+        }
+
+        if ( ! empty( $xmp_frame ) ) {
+//error_log( __LINE__ . " MLAData::mla_fetch_mp3_xmp_metadata( {$path} ) XMP FRAME = " . var_export( $xmp_frame, true ), 0 );
+            $start = strpos( $xmp_frame, '<x:xmpmeta' );
+            if ( false !== $start ) {
+                $end = strpos( $xmp_frame, '</x:xmpmeta>' );
+                if ( false !== $end ) {
+                    $xmp_frame = substr( $xmp_frame, $start, $end - $start + 12 );
+                    $results['mla_id3_xmp_metadata'] = MLAData::mla_parse_xml_string( $xmp_frame );
+                    if ( empty( $results['mla_id3_xmp_metadata'] ) ) {
+                        $results['mla_id3_xmp_errors'][] = "XMP metadata found but could not be parsed";
+                    }
+                } else {
+                    $results['mla_id3_xmp_errors'][] = "XMP metadata found but end tag not found";
+                }
+            } else {
+                $results['mla_id3_xmp_errors'][] = "XMP metadata frame found but start tag not found";
+            }
+        }
+
+        return $results;
     }
 
@@ -4518 +4640 @@
             } // image/webp
 
-            $results['mla_xmp_metadata'] = self::mla_parse_xmp_metadata( $path, 0 );
-            if ( NULL === $results['mla_xmp_metadata'] ) {
-                $results['mla_xmp_metadata'] = array();
+            if ( false !== strpos( $filetype['type'], 'audio/mpeg' ) ) {
+                $id3_metadata = self::mla_fetch_mp3_xmp_metadata( $path );
+
+                if ( ! empty( $id3_metadata['mla_id3_xmp_metadata'] ) ) {
+                    $results['mla_xmp_metadata'] = $id3_metadata['mla_id3_xmp_metadata'];
+                    
+                    if ( ! empty( $id3_metadata['mla_id3_xmp_errors'] ) ) {
+                    $results['mla_xmp_errors'] = $id3_metadata['mla_id3_xmp_errors'];
+                    }
+                }
+            } // audio/mpeg
+
+            if ( empty( $results['mla_xmp_metadata'] ) ) {
+                $results['mla_xmp_metadata'] = self::mla_parse_xmp_metadata( $path, 0 );
+
+                if ( NULL === $results['mla_xmp_metadata'] ) {
+                    $results['mla_xmp_metadata'] = array();
+                }
             }
 

media-library-assistant/trunk/index.php

@@ -16 +16 @@
 Plugin Name: Media Library Assistant
 Plugin URI: http://davidlingren.com/#two
-Description: 20260307 Enhances the Media Library; powerful [mla_gallery] [mla_tag_cloud] [mla_term_list], [mla_custom_list], [mla_archive_list], taxonomy support, IPTC/EXIF/XMP/PDF processing, bulk/quick edit.
+Description: 20260310 Enhances the Media Library; powerful [mla_gallery] [mla_tag_cloud] [mla_term_list], [mla_custom_list], [mla_archive_list], taxonomy support, IPTC/EXIF/XMP/PDF processing, bulk/quick edit.
 Version: 3.34
 Requires at least: 4.7

media-library-assistant/trunk/readme.txt

@@ -199 +199 @@
 
 = 3.35 =
+* Fix: For MP3 files, XMP metadata is now extracted from XML and PRIV frames, bypassing embedded image files.
 * Fix: For shortcodes that generate pagination links, a defect in handling URL fragments has been corrected.