Clean up vainfo somewhat

Handling of unknown tags was wrong (we should just pass them through and
hope for the best, not do some weird-ass id3v2 thing).  Add matroska to
the list of known tags. Switch the video types to a generic cleanup
function and move the several-times-duplicated general information
gathering into the general information gathering function where it
belongs.  Treat the general information as another tag type instead of
special-casing it as a complete override through array_merge. Drop the
useless iconv stuff (getID3 already translates tags to our requested
charset). Rename some functions to more closely match their purpose.
Fix some private functions that were marked public. Fix encoding
detection, which has been completely broken for a while.

1 files changed, 233 insertions, 528 deletions

diff --git a/lib/class/vainfo.class.php b/lib/class/vainfo.class.php
index 1244245c..3b2d6add 100644
--- a/lib/class/vainfo.class.php
+++ b/lib/class/vainfo.class.php
@@ -32,17 +32,16 @@ class vainfo {
     public $encoding_id3v1 = '';
     public $encoding_id3v2 = '';
 
-    public $filename    = '';
-    public $type        = '';
-    public $tags        = array();
+    public $filename = '';
+    public $type = '';
+    public $tags = array();
 
-    protected $_raw         = array();
-    protected $_getID3     = '';
-    protected $_iconv        = false;
+    protected $_raw = array();
+    protected $_getID3 = '';
 
-    protected $_file_encoding    = '';
-    protected $_file_pattern    = '';
-    protected $_dir_pattern    = '';
+    protected $_file_encoding = '';
+    protected $_file_pattern = '';
+    protected $_dir_pattern = '';
 
     private $_pathinfo;
     private $_broken = false;
@@ -50,40 +49,20 @@ class vainfo {
     /**
      * Constructor
      *
-     * This function just sets up the class, it doesn't
-     * actually pull the information
+     * This function just sets up the class, it doesn't pull the information.
      *
-     * @param    string    $file    filename
-     * @param    string    $encoding    Target encode character set
-     * @param    string    $encoding_id3v1    Default id3v1 encode character set
-     * @param    string    $encoding_iv3v2    Default id3v2 encode character set
-     * @param    string    $dir_pattern    Directory pattern
-     * @param    string    $file_pattern    File pattern
-     * @return    mixed    If can't analyze file, return false. default return: void
      */
     public function __construct($file, $encoding = null, $encoding_id3v1 = null, $encoding_id3v2 = null, $dir_pattern, $file_pattern) {
 
-        /* Check for ICONV */
-        if (function_exists('iconv') && Config::get('use_iconv') == "1") {
-            $this->_iconv = true;
-        }
-        else {
-            $this->_iconv = false;
-        }
-
         $this->filename = $file;
-        if ($encoding) {
-            $this->encoding = $encoding;
-        }
-        else {
-            $this->encoding = Config::get('site_charset');
-        }
+        $this->encoding = $encoding ?: Config::get('site_charset');
 
         /* These are needed for the filename mojo */
         $this->_file_pattern = $file_pattern;
         $this->_dir_pattern  = $dir_pattern;
 
-        if(strtoupper(substr(PHP_OS,0,3)) == 'WIN') {
+        // FIXME: This looks ugly and probably wrong
+        if(strtoupper(substr(PHP_OS, 0, 3)) == 'WIN') {
             $this->_pathinfo = str_replace('%3A', ':', urlencode($this->filename));
             $this->_pathinfo = pathinfo(str_replace('%5C', '\\', $this->_pathinfo));
         }
@@ -95,25 +74,24 @@ class vainfo {
         // Initialize getID3 engine
         $this->_getID3 = new getID3();
 
-        $this->_getID3->option_md5_data        = false;
-        $this->_getID3->option_md5_data_source    = false;
-        $this->_getID3->option_tags_html    = false;
-        $this->_getID3->option_extra_info    = true;
-        $this->_getID3->option_tag_lyrics3    = true;
-        $this->_getID3->option_tags_process    = true;
-        $this->_getID3->encoding        = $this->encoding;
+        $this->_getID3->option_md5_data = false;
+        $this->_getID3->option_md5_data_source = false;
+        $this->_getID3->option_tags_html = false;
+        $this->_getID3->option_extra_info = true;
+        $this->_getID3->option_tag_lyrics3 = true;
+        $this->_getID3->option_tags_process = true;
+        $this->_getID3->encoding = $this->encoding;
 
         // get id3tag encoding (try to work around off-spec id3v1 tags)
         try {
             $this->_raw = $this->_getID3->analyze($file);
         }
         catch (Exception $error) {
-            debug_event('Getid3()',"Broken file detected $file - " . $error->message,'1');
+            debug_event('getID3', "Broken file detected: $file: " . $error->message, 1);
             $this->_broken = true;
             return false;
         }
 
-        /* Use default mb_detect_order in php.ini or not */
         if (Config::get('mb_detect_order')) {
             $mb_order = Config::get('mb_detect_order');
         }
@@ -140,6 +118,7 @@ class vainfo {
         }
 
         if (Config::get('getid3_detect_id3v2_encoding')) {
+            // The user has told us to be moronic, so let's do that thing
             foreach ($test_tags as $tag) {
                 if ($value = $this->_raw['id3v2']['comments'][$tag]) {
                     $tags[$tag] = $value;
@@ -151,18 +130,17 @@ class vainfo {
         }
 
         $this->_getID3->encoding_id3v1 = $this->encoding_id3v1;
-
-    } // vainfo
+    }
 
     /**
      * _detect_encoding
+     *
      * Takes an array of tags and attempts to automatically detect their
      * encoding.
      */
     private static function _detect_encoding($tags, $mb_order) {
         if (function_exists('mb_detect_encoding')) {
             $encodings = array();
-            $tags = array('artist', 'album', 'genre', 'title');
             foreach ($tags as $tag) {
                 $encodings[mb_detect_encoding($tag, $mb_order, true)]++;
             }
@@ -189,10 +167,10 @@ class vainfo {
 
     /**
      * get_info
+     *
      * This function runs the various steps to gathering the metadata
      */
     public function get_info() {
-
         // If this is broken, don't waste time figuring it out a second
         // time, just return their rotting carcass of a media file.
         if ($this->_broken) {
@@ -200,12 +178,11 @@ class vainfo {
             return true;
         }
 
-        /* Get the Raw file information */
         try {
             $this->_raw = $this->_getID3->analyze($this->filename);
         }
         catch (Exception $error) {
-            debug_event('Getid3()',"Unable to catalog file:" . $error->message,'1');
+            debug_event('getID2', 'Unable to catalog file: ' . $error->message, 1);
         }
 
         /* Figure out what type of file we are dealing with */
@@ -227,49 +204,47 @@ class vainfo {
 
     /**
      * get_tag_type
+     *
      * This takes the result set and the tag_order defined in your config
      * file and tries to figure out which tag type(s) it should use. If your
      * tag_order doesn't match anything then it throws up its hands and uses
-     * everything.
+     * everything in random order.
      */
     public static function get_tag_type($results, $config_key = 'metadata_order') {
-
         $order = (array)Config::get($config_key);
 
-        /* Iterate through the defined key order adding them to an
-         * ordered array as we go.
-         */
-
+        // Iterate through the defined key order adding them to an ordered array.
         foreach($order as $key) {
             if ($results[$key]) {
                 $returned_keys[] = $key;
             }
         }
 
-        /* If we didn't find anything then default to everything.
-         */
+        // If we didn't find anything then default to everything.
         if (!isset($returned_keys)) {
             $returned_keys = array_keys($results);
             $returned_keys = sort($returned_keys);
         }
 
-        return $returned_keys;
+        // Unless they explicitly set it, add bitrate/mode/mime/etc.
+        if (!in_array('general', $returned_keys)) {
+            $returned_keys[] = 'general';
+        }
 
-    } // get_tag_type
+        return $returned_keys;
+    }
 
     /**
      * clean_tag_info
+     *
      * This function takes the array from vainfo along with the
      * key we've decided on and the filename and returns it in a
      * sanitized format that ampache can actually use
      */
     public static function clean_tag_info($results, $keys, $filename = null) {
-
         $info = array();
 
-        if ($filename) {
-            $info['file'] = $filename;
-        }
+        $info['file'] = $filename;
 
         // Iteration!
         foreach ($keys as $key) {
@@ -286,14 +261,13 @@ class vainfo {
             $info['time'] = $info['time'] ?: intval($tags['time']);
             $info['channels'] = $info['channels'] ?: $tags['channels'];
 
-            /* These are used to generate the correct IDs later */
             $info['title'] = $info['title'] ?: stripslashes(trim($tags['title']));
 
             $info['year'] = $info['year'] ?: intval($tags['year']);
 
             $info['disk'] = $info['disk'] ?: intval($tags['disk']);
 
-            $info['totaldiscs'] = $info['totaldiscs'] ?: intval($tags['totaldiscs']);
+            $info['totaldisks'] = $info['totaldisks'] ?: intval($tags['totaldisks']);
 
             $info['artist']    = $info['artist'] ?: trim($tags['artist']);
 
@@ -317,7 +291,6 @@ class vainfo {
             $info['mb_albumid'] = $info['mb_albumid'] ?: trim($tags['mb_albumid']);
             $info['mb_artistid'] = $info['mb_artistid'] ?: trim($tags['mb_artistid']);
 
-            /* @TODO language doesn't import from id3tag. @momo-i */
             $info['language'] = $info['language'] ?: Dba::escape($tags['language']);
 
             $info['lyrics']    = $info['lyrics']
@@ -333,128 +306,118 @@ class vainfo {
             $info['video_codec'] = $info['video_codec'] ?: Dba::escape($tags['video_codec']);
         }
 
-        if ($info['totaldiscs'] == 1 && $info['disk'] == 1) {
+        // Some things set the disk number even though there aren't multiple
+        if ($info['totaldisks'] == 1 && $info['disk'] == 1) {
             unset($info['disk']);
-            unset($info['totaldiscs']);
+            unset($info['totaldisks']);
         }
 
         return $info;
-
-    } // clean_tag_info
+    }
 
     /**
      * _get_type
-     * This function takes the raw information and figures out
-     * what type of file we are dealing with for use by the tag
-     * function
+     *
+     * This function takes the raw information and figures out what type of
+     * file we are dealing with.
      */
-    public function _get_type() {
-
-        /* There are a few places that the file type can
-         * come from, in the end we trust the encoding
-         * type
-         */
+    private function _get_type() {
+        // There are a few places that the file type can come from, in the end
+        // we trust the encoding type.
         if ($type = $this->_raw['video']['dataformat']) {
-            $type = $this->_clean_type($type);
-            return $type;
+            return $this->_clean_type($type);
         }
         if ($type = $this->_raw['audio']['streams']['0']['dataformat']) {
-            $type = $this->_clean_type($type);
-            return $type;
+            return $this->_clean_type($type);
         }
         if ($type = $this->_raw['audio']['dataformat']) {
-            $type = $this->_clean_type($type);
-            return $type;
+            return $this->_clean_type($type);
         }
         if ($type = $this->_raw['fileformat']) {
-            $type = $this->_clean_type($type);
-            return $type;
+            return $this->_clean_type($type);
         }
 
         return false;
-
-    } // _get_type
+    }
 
 
     /**
      * _get_tags
-     * This function takes the raw information and the type and
-     * attempts to gather the tags and then normalize them into
-     * ['tag_name']['var'] = value
+     *
+     * This processes the raw getID3 output and bakes it.
      */
-    public function _get_tags() {
+    private function _get_tags() {
 
         $results = array();
 
-        /* The tags can come in many different shapes and colors
-         * depending on the encoding time of day and phase of the moon.
-         */
+        // The tags can come in many different shapes and colors
+        // depending on the encoding time of day and phase of the moon.
+
         foreach ($this->_raw['tags'] as $key => $tag_array) {
             switch ($key) {
+                case 'ape':
+                case 'avi':
+                case 'flv':
+                case 'matroska':
+                    debug_event('vainfo', 'Cleaning ' . $key, 5);
+                    $parsed = $this->_cleanup_generic($tag_array);
+                break;
                 case 'vorbiscomment':
-                    debug_event('_get_tags', 'Parsing vorbis', '5');
-                    $results[$key] = $this->_parse_vorbiscomment($tag_array);
+                    debug_event('vainfo', 'Cleaning vorbis', 5);
+                    $parsed = $this->_cleanup_vorbiscomment($tag_array);
                 break;
                 case 'id3v1':
-                    debug_event('_get_tags', 'Parsing id3v1', '5');
-                    $results[$key] = $this->_parse_id3v1($tag_array);
+                    debug_event('vainfo', 'Cleaning id3v1', 5);
+                    $parsed = $this->_cleanup_id3v1($tag_array);
                 break;
                 case 'id3v2':
-                    debug_event('_get_tags', 'Parsing id3v2', '5');
-                    $results[$key] = $this->_parse_id3v2($tag_array);
-                break;
-                case 'ape':
-                    debug_event('_get_tags', 'Parsing ape', '5');
-                    $results[$key] = $this->_parse_ape($tag_array);
+                    debug_event('vainfo', 'Cleaning id3v2', 5);
+                    $parsed = $this->_cleanup_id3v2($tag_array);
                 break;
                 case 'quicktime':
-                    debug_event('_get_tags', 'Parsing quicktime', '5');
-                    $results[$key] = $this->_parse_quicktime($tag_array);
+                    debug_event('vainfo', 'Cleaning quicktime', 5);
+                    $parsed = $this->_cleanup_quicktime($tag_array);
                 break;
                 case 'riff':
-                    debug_event('_get_tags', 'Parsing riff', '5');
-                    $results[$key] = $this->_parse_riff($tag_array);
-                break;
-                case 'flv':
-                    debug_event('_get_tags', 'Parsing flv', '5');
-                    $results[$key] = $this->_parse_flv($this->_raw);
+                    debug_event('vainfo', 'Cleaning riff', 5);
+                    $parsed = $this->_cleanup_riff($tag_array);
                 break;
                 case 'mpg':
                 case 'mpeg':
-                    debug_event('_get_tags', 'Parsing MPEG', '5');
-                    $results['mpeg'] = $this->_parse_mpg($this->_raw);
+                    $key = 'mpeg';
+                    debug_event('vainfo', 'Cleaning MPEG', 5);
+                    $parsed = $this->_cleanup_generic($tag_array);
                 break;
                 case 'asf':
                 case 'wmv':
-                    debug_event('_get_tags', 'Parsing WMV/WMA/ASF', '5');
-                    $results['asf'] = $this->_parse_wmv($this->_raw);
-                break;
-                case 'avi':
-                    debug_event('_get_tags', 'Parsing avi', '5');
-                    $results[$key] = $this->_parse_avi($this->_raw);
+                    $key = 'asf';
+                    debug_event('vainfo', 'Cleaning WMV/WMA/ASF', 5);
+                    $parsed = $this->_cleanup_generic($tag_array);
                 break;
                 case 'lyrics3':
-                    debug_event('_get_tags', 'Parsing lyrics3', '5');
-                    $results[$key] = $this->_parse_lyrics($tag_array);
+                    debug_event('vainfo', 'Cleaning lyrics3', 5);
+                    $parsed = $this->_cleanup_lyrics($tag_array);
                 break;
                 default:
-                    debug_event('vainfo','Error: Unable to determine tag type of ' . $key . ' for file ' . $this->filename . ' Assuming id3v2','5');
-                    $results[$key] = $this->_parse_id3v2($this->_raw['id3v2']['comments']);
+                    debug_event('vainfo', 'Cleaning unrecognised tag type ' . $key . ' for file ' . $this->filename, 5);
+                    $parsed = $this->_cleanup_generic($tag_array);
                 break;
-            } // end switch
-        } // end foreach
+            }
+
+            $results[$key] = $parsed;
+        }
 
+        $results['general'] = $this->_parse_general($this->_raw);
 
         $cleaned = self::clean_tag_info($results, self::get_tag_type($results, 'getid3_tag_order'), $this->filename);
-        $cleaned = array_merge($cleaned, $this->_get_info());
         $cleaned['raw'] = $results;
 
         return $cleaned;
-
-    } // _get_tags
+    }
 
     /**
      * _get_plugin_tags
+     *
      * Get additional metadata from plugins
      */
     private function _get_plugin_tags() {
@@ -464,7 +427,7 @@ class vainfo {
         }
 
         $plugin_names = Plugin::get_plugins('get_metadata');
-        foreach ($tag_order as $key => $tag_source) {
+        foreach ($tag_order as $tag_source) {
             if (in_array($tag_source, $plugin_names)) {
                 $plugin = new Plugin($tag_source);
                 if ($plugin->load()) {
@@ -472,51 +435,36 @@ class vainfo {
                 }
             }
         }
-    } // _get_plugin_tags
+    }
 
     /**
-     * _get_info
-     * Gather and return the general information about a song (vbr/cbr,
+     * _parse_general
+     *
+     * Gather and return the general information about a file (vbr/cbr,
      * sample rate, channels, etc.)
      */
-    private function _get_info() {
-
-        $array = array();
-
-        /* Try to pull the information directly from
-         * the audio array
-         */
-        if ($this->_raw['audio']['bitrate_mode']) {
-            $array['mode']     = $this->_raw['audio']['bitrate_mode'];
-            if ($array['mode'] == 'con') {
-                $array['mode'] = 'cbr';
-            }
-        }
-        if ($this->_raw['audio']['bitrate']) {
-            $array['bitrate']    = $this->_raw['audio']['bitrate'];
-        }
-        if ($this->_raw['audio']['channels']) {
-            $array['channels']     = intval($this->_raw['audio']['channels']);
-        }
-        if ($this->_raw['audio']['sample_rate']) {
-            $array['rate']    = intval($this->_raw['audio']['sample_rate']);
-        }
-        if ($this->_raw['filesize']) {
-            $array['size']    = intval($this->_raw['filesize']);
-        }
-        if ($this->_raw['encoding']) {
-            $array['encoding']    = $this->_raw['encoding'];
-        }
-        if ($this->_raw['mime_type']) {
-            $array['mime']        = $this->_raw['mime_type'];
-        }
-        if ($this->_raw['playtime_seconds']) {
-            $array['time']    = $this->_raw['playtime_seconds'];
-        }
-
-        return $array;
-
-    } // _get_info
+    private function _parse_general($tags) {
+        $parsed = array();
+
+        $parsed['title'] = urldecode($this->_pathinfo['filename']);
+        $parsed['mode'] = $tags['audio']['bitrate_mode'];
+        if ($parsed['mode'] == 'con') {
+            $parsed['mode'] = 'cbr';
+        }
+        $parsed['bitrate'] = $tags['audio']['bitrate'];
+        $parsed['channels'] = intval($tags['audio']['channels']);
+        $parsed['rate'] = intval($tags['audio']['sample_rate']);
+        $parsed['size'] = intval($tags['filesize']);
+        $parsed['encoding'] = $tags['encoding'];
+        $parsed['mime'] = $tags['mime_type'];
+        $parsed['time'] = $tags['playtime_seconds'];
+        $parsed['video_codec'] = $tags['video']['fourcc'];
+        $parsed['audio_codec'] = $tags['audio']['dataformat'];
+        $parsed['resolution_x'] = $tags['video']['resolution_x'];
+        $parsed['resolution_y'] = $tags['video']['resolution_y'];
+
+        return $parsed;
+    }
 
     /**
      * _clean_type
@@ -547,150 +495,139 @@ class vainfo {
                 debug_event('vainfo','Unable to determine file type from ' . $type . ' on file ' . $this->filename,'5');
                 return $type;
             break;
-        } // end switch on type
-
-    } // _clean_type
+        }
+    } 
 
     /**
-     * _parse_lyrics
-     * This function takes a lyrics3 from getid3()
-     * nothing to do?
+     * _cleanup_generic
+     *
+     * This does generic cleanup.
      */
-    private function _parse_lyrics($tags) {
+    private function _cleanup_generic($tags) {
+        $parsed = array();
+        foreach ($tags as $tagname => $data) {
+            switch ($tagname) {
+                case 'genre':
+                    // Pass the array through
+                    $parsed[$tagname] = $data;
+                break;
+                default:
+                    $parsed[$tagname] = $data[0];
+                break;
+            }
+        }
 
-        /* Results array */
-        $array = array();
+        return $parsed;
+    }
+
+    /**
+     * _cleanup_lyrics
+     *
+     * This is supposed to handle lyrics3. FIXME: does it?
+     */
+    private function _cleanup_lyrics($tags) {
+        $parsed = array();
 
-        /* go through them all! */
         foreach ($tags as $tag => $data) {
             if ($tag == 'unsynchedlyrics' || $tag == 'unsynchronised lyric') {
                 $tag = 'lyrics';
             }
-            $array[$tag] = $this->_clean_tag($data['0']);
-
-        } // end foreach
-
-        return $array;
-
-    } // _parse_lyrics
+            $parsed[$tag] = $data[0];
+        } 
+        return $parsed;
+    }
 
     /**
-     * _parse_vorbiscomment
-     * This function takes a vorbiscomment from getid3() and then
-     * returns the elements translated using iconv if needed in a
-     * pretty little format
+     * _cleanup_vorbiscomment
+     *
+     * Standardises tag names from vorbis.  
      */
-    private function _parse_vorbiscomment($tags) {
-
-        /* Results array */
-        $array = array();
+    private function _cleanup_vorbiscomment($tags) {
+        $parsed = array();
 
-        /* go through them all! */
-        foreach ($tags as $tag=>$data) {
-
-            /* We need to translate a few of these tags */
+        foreach ($tags as $tag => $data) {
             switch ($tag) {
-
                 case 'genre':
-                    // multiple genre support
-                    foreach($data as $foo) {
-                        $array['genre'][] = $this->_clean_tag($foo,'');
-                    }
+                    // Pass the array through
+                    $parsed[$tag] = $data;
                 break;
                 case 'tracknumber':
-                    $array['track']    = $this->_clean_tag($data['0']);
+                    $parsed['track'] = $data[0];
                 break;
                 case 'discnumber':
-                    $el = explode('/', $data['0']);
-                    $array['disk']     = $el[0];
+                    $elements = explode('/', $data[0]);
+                    $parsed['disk'] = $elements[0];
+                    $parsed['totaldisks'] = $elements[1];
                 break;
                 case 'date':
-                    $array['year']    = $this->_clean_tag($data['0']);
+                    $parsed['year'] = $data[0];
                 break;
                 default:
-                    $array[$tag] = $this->_clean_tag($data['0']);
+                    $parsed[$tag] = $data[0];
                 break;
-            } // end switch
-
-        } // end foreach
-
-        return $array;
+            } 
+        }
 
-    } // _parse_vorbiscomment
+        return $parsed;
+    }
 
     /**
-     * _parse_id3v1
-     * This function takes an id3v1 tag set from getid3() and then
-     * returns the elements (translated using iconv if needed) in a
-     * pretty little format.
+     * _cleanup_id3v1
+     *
+     * Doesn't do much.
      */
-    private function _parse_id3v1($tags) {
+    private function _cleanup_id3v1($tags) {
+        $parsed = array();
 
-        $array = array();
-
-        /* Go through all the tags */
-        foreach ($tags as $tag=>$data) {
-
-            /* This is our baseline for naming
-             * so no translation needed
-             */
-            $array[$tag]    = $this->_clean_tag($data['0']);
-
-        } // end foreach
-
-        return $array;
+        foreach ($tags as $tag => $data) {
+            // This is our baseline for naming so everything's already right,
+            // we just need to shuffle off the array.
+            $parsed[$tag] = $data[0];
+        }
 
-    } // _parse_id3v1
+        return $parsed;
+    }
 
     /**
-     * _parse_id3v2
-     * This function takes an id3v2 tag set from getid3() and then
-     * returns the elements (translated using iconv if needed) in a
-     * pretty little format.
+     * _cleanup_id3v2
+     *
+     * Whee, v2!
      */
-    private function _parse_id3v2($tags) {
-
-        $array = array();
+    private function _cleanup_id3v2($tags) {
+        $parsed = array();
 
-        /* Go through the tags */
-        foreach ($tags as $tag=>$data) {
+        foreach ($tags as $tag => $data) {
 
-            /**
-             * the new getid3 handles this differently
-             * so we now need to account for it :(
-             */
             switch ($tag) {
                 case 'genre':
-                    // multiple genre support
-                    foreach($data as $genre) {
-                        $array['genre'][] = $this->_clean_tag($genre);
-                    }
+                    // Pass the array through
+                    $parsed[$tag] = $data;
                 break;
                 case 'part_of_a_set':
-                    $el = explode('/', $data['0']);
-                    $array['disk'] = $el[0];
-                    $array['totaldiscs'] = $el[1];
+                    $elements = explode('/', $data[0]);
+                    $parsed['disk'] = $elements[0];
+                    $parsed['totaldisks'] = $elements[1];
                 break;
                 case 'track_number':
-                    $array['track'] = $this->_clean_tag($data['0']);
+                    $parsed['track'] = $data[0];
                 break;
                 case 'comments':
-                    $array['comment'] = $this->_clean_tag($data['0']);
+                    $parsed['comment'] = $data[0];
                 break;
                 default:
-                    $array[$tag]    = $this->_clean_tag($data['0']);
+                    $parsed[$tag] = $data[0];
                 break;
-            } // end switch on tag
-
-        } // end foreach
+            }
+        }
 
+        // getID3 doesn't do all the parsing we need, so grab the raw data
         $id3v2 = $this->_raw['id3v2'];
 
         if(!empty($id3v2['UFID'])) {
             // Find the MBID for the track
             foreach ($id3v2['UFID'] as $ufid) {
                 if ($ufid['ownerid'] == 'http://musicbrainz.org') {
-                    $array['mb_trackid'] = $this->_clean_tag($ufid['data']);
+                    $parsed['mb_trackid'] = $ufid['data'];
                 }
             }
 
@@ -698,10 +635,10 @@ class vainfo {
             foreach ($id3v2['TXXX'] as $txxx) {
                 switch ($txxx['description']) {
                     case 'MusicBrainz Album Id':
-                        $array['mb_albumid'] = $this->_clean_tag($txxx['data']);
+                        $parsed['mb_albumid'] = $txxx['data'];
                     break;
                     case 'MusicBrainz Artist Id':
-                        $array['mb_artistid'] = $this->_clean_tag($txxx['data']);
+                        $parsed['mb_artistid'] = $txxx['data'];
                     break;
                 }
             }
@@ -713,225 +650,72 @@ class vainfo {
                 if (array_key_exists('email', $popm) &&
                     $user = User::get_from_email($popm['email'])) {
                     // Ratings are out of 255; scale it
-                    $array['rating'][$user->id] = $popm['rating'] / 255 * 5;
+                    $parsed['rating'][$user->id] = $popm['rating'] / 255 * 5;
                 }
             }
         }
 
-        return $array;
-
-    } // _parse_id3v2
-
-    /**
-     * _parse_ape
-     * This function takes ape tags set by getid3() and then
-     * returns the elements translated using iconv if needed in a
-     * pretty little format
-     */
-    private function _parse_ape($tags) {
-
-        foreach ($tags as $tag=>$data) {
-            switch ($tag) {
-
-                case 'genre':
-                    // multiple genre support
-                    foreach($data as $genre) {
-                        $array['genre'][] = $this->_clean_tag($genre);
-                    }
-                break;
-
-                default:
-                    $array[$tag] = $this->_clean_tag($data['0'], $this->_file_encoding);
-                break;
-            } // end switch on tag
-
-        } // end foreach tags
-
-        return $array;
-
-    } // _parse_ape
+        return $parsed;
+    }
 
     /**
-     * _parse_riff
-     * this function takes the riff take information passed by getid3() and
-     * then reformats it so that it matches the other formats. May require iconv
+     * _cleanup_riff
      */
-    private function _parse_riff($tags) {
-
-        foreach ($tags as $tag=>$data) {
+    private function _cleanup_riff($tags) {
+        $parsed = array();
 
+        foreach ($tags as $tag => $data) {
             switch ($tag) {
                 case 'product':
-                    $array['album'] = $this->_clean_tag($data['0'], $this->_file_encoding);
+                    $parsed['album'] = $data[0];
                 break;
                 default:
-                    $array[$tag] = $this->_clean_tag($data['0'], $this->_file_encoding);
+                    $parsed[$tag] = $data[0];
                 break;
-            } // end switch on tag
-
-        } // foreach tags
-
-        return $array;
+            }
+        }
 
-    } // _parse_riff
+        return $parsed;
+    }
 
     /**
-     * _parse_quicktime
-     * this function takes the quicktime tags set by getid3() and then
-     * returns the elements translated using iconv if needed in a
-     * pretty little format
+     * _cleanup_quicktime
      */
-    private function _parse_quicktime($tags) {
+    private function _cleanup_quicktime($tags) {
+        $parsed = array();
 
-        /* Results array */
-        $array = array();
-
-        /* go through them all! */
-        foreach ($tags as $tag=>$data) {
-
-            /* We need to translate a few of these tags */
+        foreach ($tags as $tag => $data) {
             switch ($tag) {
                 case 'creation_date':
                     if (strlen($data['0']) > 4) {
-                        /* Weird Date format, attempt to normalize */
-                        $data['0'] = date("Y",strtotime($data['0']));
+                        // Weird date format, attempt to normalize it
+                        $data[0] = date('Y', strtotime($data[0]));
                     }
-                    $array['year']  = $this->_clean_tag($data['0']);
+                    $parsed['year'] = $data[0];
                 break;
                 case 'MusicBrainz Track Id':
-                    $array['mb_trackid'] = $this->_clean_tag($data['0']);
+                    $parsed['mb_trackid'] = $data[0];
                 break;
                 case 'MusicBrainz Album Id':
-                    $array['mb_albumid'] = $this->_clean_tag($data['0']);
+                    $parsed['mb_albumid'] = $data[0];
                 break;
                 case 'MusicBrainz Artist Id':
-                    $array['mb_artistid'] = $this->_clean_tag($data['0']);
+                    $parsed['mb_artistid'] = $data[0];
                 break;
-            } // end switch
-
-            $array[$tag] = $this->_clean_tag($data['0']);
-
-        } // end foreach
-
-        // Also add in any video related stuff we might find
-        if (strpos($this->_raw['mime_type'], 'video') !== false) {
-            $info = $this->_parse_avi($this->_raw);
-            $info['video_codec'] = $this->_raw['quicktime']['ftyp']['fourcc'];
-            $array = array_merge($info, $array);
-        }
-
-        return $array;
-
-    } // _parse_quicktime
-
-    /**
-     * _parse_avi
-     * This attempts to parse our the information on an avi file and present it in some
-     * kind of sane format, this is a little hard as these files don't have tags
-     */
-    private function _parse_avi($tags) {
-
-        $array = array();
-
-        $array['title']         = urldecode($this->_pathinfo['filename']);
-        $array['video_codec']         = $tags['video']['fourcc'];
-        $array['audio_codec']         = $tags['audio']['dataformat'];
-        $array['resolution_x']        = $tags['video']['resolution_x'];
-        $array['resolution_y']        = $tags['video']['resolution_y'];
-        $array['mime']            = $tags['mime_type'];
-        $array['comment']        = $tags['video']['codec'];
-
-        return $array;
-
-    } // _parse_avi
-
-    /**
-     * _parse_mpg
-     * This attempts to parse our the information on a mpg file and present it in some
-     * kind of sane format, this is a little hard as these files don't have tags
-     */
-    private function _parse_mpg($tags) {
-
-        $array = array();
-
-        $array['title']            = urldecode($this->_pathinfo['filename']);
-        $array['video_codec']        = $tags['video']['codec'];
-        $array['audio_codec']        = $tags['audio']['dataformat'];
-        $array['resolution_x']        = $tags['video']['resolution_x'];
-        $array['resolution_y']        = $tags['video']['resolution_y'];
-        $array['mime']            = $tags['mime_type'];
-        $array['comment']        = $tags['video']['codec'];
-
-        return $array;
-
-    } // _parse_mpg
-
-    /**
-     * _parse_wmv
-     * This attempts to parse our the information on a asf/wmv file and present it in some
-     * kind of sane format, this is a little hard as these files don't have tags
-     */
-    private function _parse_wmv($tags) {
-
-        $array = array();
-
-        $array['mime']        = $tags['mime_type'];
-
-        switch($array['mime']) {
-            default:
-            case 'video/x-ms-wmv':
-                if(isset($tags['tags']['asf']['title']['0'])) {
-                    $array['title']        = $tags['tags']['asf']['title']['0'];
-                }
-                else {
-                    $array['title']        = urldecode($this->_pathinfo['filename']);
-                }
-                $array['video_codec']        = $tags['video']['streams']['2']['codec'];
-                $array['audio_codec']        = $tags['audio']['streams']['1']['codec'];
-                $array['resolution_x']        = $tags['video']['streams']['2']['resolution_x'];
-                $array['resolution_y']        = $tags['video']['streams']['2']['resolution_y'];
-                $array['comment']        = $tags['tags']['asf']['title']['1'];
-            break;
-        }
-
-        foreach($tags['tags']['asf'] as $tag => $data) {
-            $array[$tag] = $this->_clean_tag($data['0']);
+                default:
+                    $parsed[$tag] = $data[0];
+                break;
+            }
         }
 
-        return $array;
-
-    } // _parse_wmv
-
-    /**
-     * _parse_flv
-     * This attempts to parse our the information on an flv file and present it in some
-     * kind of sane format, this is a little hard as these files don't have tags
-     */
-    private function _parse_flv($tags) {
-
-        $array = array();
-
-        $array['title']            = urldecode($this->_pathinfo['filename']);
-        $array['video_codec']        = $tags['video']['codec'];
-        $array['audio_codec']        = $tags['audio']['dataformat'];
-        $array['resolution_x']        = $tags['video']['resolution_x'];
-        $array['resolution_y']        = $tags['video']['resolution_y'];
-        $array['mime']            = $tags['mime_type'];
-        $array['comment']        = $tags['video']['codec'];
-
-        return $array;
-
-    } // _parse_flv
+        return $parsed;
+    }
 
     /**
      * _parse_filename
      *
-     * This function uses the passed file and dir patterns
-     * To pull out extra tag information and populate it into
-     * its own array
-     *
-     * @param    string    $filename    Filename that want to parse 
-     * @return    array    Parsed results
+     * This function uses the file and directory patterns to pull out extra tag
+     * information.
      */
     private function _parse_filename($filename) {
 
@@ -980,85 +764,6 @@ class vainfo {
     }
 
     /**
-     * _id3v2_tag_to_frame
-     *
-     * This translates the tag name to a frame, if there a many it returns the first
-     * one if finds that exists in the raw
-     *
-     * @param    string    $tag_name    Tag name
-     * @return    mixed    If found id3v2 frame, return frame. If not found, return false.
-     */
-    private function _id3v2_tag_to_frame($tag_name) {
-
-        static $map = array(
-            'comment'=>array('COM','COMM'),
-            'cd_ident'=>array('MCDI','MCI'),
-            'album'=>array('TAL','TALB'),
-            'language'=>array('TLA','TLAN'),
-            'mood'=>array('TMOO'),
-            'artist'=>array('TPE1'),
-            'year'=>array('TDRC'));
-
-        foreach ($map[$tag_name] as $frame) {
-            if (isset($this->_raw['id3v2'][$frame])) {
-                return $frame;
-            }
-        }
-
-        return false;
-
-    } // _id3v2_tag_to_frame
-
-    /**
-     * _clean_tag
-     *
-     * This function cleans up the tag that it's passed using Iconv
-     * if we've got it. It also takes an optional encoding param
-     * for the cases where we know what the tags source encoding
-     * is, and or if it's different then the encoding recorded
-     * in the file
-     *
-     * @param    string    $tag    Encoding string
-     * @param    string    $encoding    Encode charset
-     * @return    string    Return encoded string
-     */
-    private function _clean_tag($tag, $encoding = null) {
-
-        // Default to getID3's native encoding
-        if (!$encoding) {
-            $encoding = $this->_getID3->encoding;
-        }
-        // If we've got iconv then go ahead and clear her up
-        if ($encoding == $this->encoding) {
-            debug_event('vainfo', "\$encoding -> ${encoding}, \$this->encoding -> {$this->encoding}", 5);
-            return $tag;
-        }
-        if ($this->_iconv) {
-            debug_event('vainfo', 'Use iconv()',5);
-
-            // Try GNU iconv //TRANSLIT extension first
-            $new_encoding = $this->encoding . '//TRANSLIT';
-            $clean = iconv($encoding, $new_encoding, $tag);
-
-            // If that fails, do a plain conversion
-            if(strcmp($clean, '') == 0) {
-                $clean = iconv($encoding, $this->encoding, $tag);
-            }
-        }
-        elseif (function_exists('mb_convert_encoding')) {
-            debug_event('vainfo', 'Use mbstring',5);
-            debug_event('vainfo', "Try to convert from {$this->encoding} to $encoding", 5);
-            $clean = mb_convert_encoding($tag, $encoding, $this->encoding);
-        }
-        else {
-            $clean = $tag;
-        }
-
-        return $clean;
-
-    } // _clean_tag
-
-    /**
      * set_broken
      *
      * This fills all tag types with Unknown (Broken)