Changeset 62087 for trunk/src/wp-includes/ID3/module.audio-video.matroska.php

Timestamp:

03/21/2026 11:47:15 PM (5 months ago)

Author:

SergeyBiryukov

Message:

External Libraries: Update getID3 to version 1.9.25.

The latest version includes various bug fixes, as well as improvements for PHP 8.5 support.

References:

• ​getID3 1.9.25 release notes
• ​Full list of changes in getID3 1.9.25

Follow-up to [47601], [48278], [52254], [54376], [56975], [61253].

Props Presskopp, SergeyBiryukov.
Fixes #64914.

File:

• trunk/src/wp-includes/ID3/module.audio-video.matroska.php (modified) (8 diffs)

trunk/src/wp-includes/ID3/module.audio-video.matroska.php

@@ -212 +212 @@
 
 /**
+ * Matroska constants
+ */
+define('MATROSKA_DEFAULT_TIMECODESCALE', 1000000);
+
+/**
+ * Matroska scan modes are internal state flags for how much of the file we are scanning
+ */
+define('MATROSKA_SCAN_HEADER', 0);
+define('MATROSKA_SCAN_WHOLE_FILE', 1);
+define('MATROSKA_SCAN_FIRST_CLUSTER', 2);
+define('MATROSKA_SCAN_LAST_CLUSTER', 3);
+
+/**
 * @tutorial http://www.matroska.org/technical/specs/index.html
 *
@@ -242 +255 @@
         private $current_offset    = 0;
         private $unuseful_elements = array(EBML_ID_CRC32, EBML_ID_VOID);
+        private $scan_mode = MATROSKA_SCAN_HEADER;
 
         /**
@@ -249 +263 @@
         {
                 $info = &$this->getid3->info;
+                $this->scan_mode = $this->parse_whole_file ? MATROSKA_SCAN_WHOLE_FILE : MATROSKA_SCAN_HEADER;
 
                 // parse container
@@ -257 +272 @@
                 }
 
-                // calculate playtime
-                if (isset($info['matroska']['info']) && is_array($info['matroska']['info'])) {
-                        foreach ($info['matroska']['info'] as $key => $infoarray) {
-                                if (isset($infoarray['Duration'])) {
-                                        // TimecodeScale is how many nanoseconds each Duration unit is
-                                        $info['playtime_seconds'] = $infoarray['Duration'] * ((isset($infoarray['TimecodeScale']) ? $infoarray['TimecodeScale'] : 1000000) / 1000000000);
-                                        break;
-                                }
+                $this->playtimeFromMetadata($info);
+
+                // If there was no duration metadata, this might be an incomplete file or a streaming file
+                // We need Cluster information so we can use their timecodes to estimate playtime.
+                if (!isset($info['playtime_seconds']) && !$this->parse_whole_file) {
+                        // Scan the start and end of file for Clusters to estimate duration
+                        $this->scanStartEndForClusters($info);
+                }
+
+                if (isset($info['matroska']['cluster']) && is_array($info['matroska']['cluster'])) {
+                        if (!isset($info['playtime_seconds']) && !empty($info['matroska']['cluster'])) {
+                                // estimate playtime using clusters if not yet known
+                                $this->calculatePlaytimeFromClusters($info);
+                        }
+
+                        // Remove cluster information from output if hide_clusters is true
+                        // These could have been set during scanStartEndForClusters()
+                        if ($this->hide_clusters) {
+                                unset($info['matroska']['cluster']);
                         }
                 }
@@ -333 +359 @@
                                                 }
 
-                                                $info['video']['streams'][$trackarray['TrackUID']] = $track_info;
+                                                if (isset($trackarray['TrackUID'])) {
+                                                        $info['video']['streams'][$trackarray['TrackUID']] = $track_info;
+                                                } else {
+                                                        $this->warning('Missing mandatory TrackUID for video track');
+                                                }
                                                 break;
 
@@ -481 +511 @@
                                                                 break;
                                                 }
-
-                                                $info['audio']['streams'][$trackarray['TrackUID']] = $track_info;
+                                                if (isset($trackarray['TrackUID'])) {
+                                                        $info['audio']['streams'][$trackarray['TrackUID']] = $track_info;
+                                                } else {
+                                                        $this->warning('Missing mandatory TrackUID for audio track');
+                                                }
                                                 break;
                                 }
@@ -1247 +1280 @@
                                                                         $this->current_offset = $subelement['end'];
                                                                 }
-                                                                if (!$this->hide_clusters) {
+
+                                                                if (!$this->hide_clusters || $this->playtimeFromMetadata($info) === false) {
                                                                         $info['matroska']['cluster'][] = $cluster_entry;
                                                                 }
+                                                                if ($this->scan_mode === MATROSKA_SCAN_FIRST_CLUSTER) {
+                                                                        // Stop parsing after finding first cluster
+                                                                        return;
+                                                                }
 
                                                                 // check to see if all the data we need exists already, if so, break out of the loop
-                                                                if (!$this->parse_whole_file) {
+                                                                if ($this->scan_mode === MATROSKA_SCAN_HEADER) {
                                                                         if (isset($info['matroska']['info']) && is_array($info['matroska']['info'])) {
                                                                                 if (isset($info['matroska']['tracks']['tracks']) && is_array($info['matroska']['tracks']['tracks'])) {
@@ -1920 +1958 @@
         }
 
+        /**
+         * @param array $info
+         *
+         * @return float|bool Duration when present in metadata or false
+         */
+        private function playtimeFromMetadata(&$info) {
+                if (isset($info['matroska']['info']) && is_array($info['matroska']['info'])) {
+                        foreach ($info['matroska']['info'] as $infoarray) {
+                                if (isset($infoarray['Duration'])) {
+                                        // TimecodeScale is how many nanoseconds each Duration unit is
+                                        $info['playtime_seconds'] = $infoarray['Duration'] * ((isset($infoarray['TimecodeScale']) ? $infoarray['TimecodeScale'] : MATROSKA_DEFAULT_TIMECODESCALE) / 1000000000);
+                                        return $info['playtime_seconds'];
+                                }
+                        }
+                }
+                return false;
+        }
+
+        /**
+         * @param int $offset New starting offset for the buffer
+         *
+         * @return void
+         */
+        private function resetParserBuffer($offset) {
+                $this->current_offset = $offset;
+                $this->EBMLbuffer = '';
+                $this->EBMLbuffer_offset = 0;
+                $this->EBMLbuffer_length = 0;
+        }
+
+        /**
+         * Scan start and end of file for cluster information when Duration is missing
+         * Only use this if no Duration was found in the Info element and we are not in parse_whole_file mode
+         *
+         * @param array $info
+         *
+         * @return void
+         */
+        private function scanStartEndForClusters(&$info) {
+                // Scan beginning of file for first cluster
+                $this->resetParserBuffer($info['avdataoffset']);
+                $this->scan_mode = MATROSKA_SCAN_FIRST_CLUSTER;
+
+                try {
+                        $this->parseEBML($info);
+                } catch (Exception $e) {
+                        $this->error('EBML parser (start of file): '.$e->getMessage());
+                }
+
+                // Scan end of file for last cluster
+                if (is_array($info['matroska']['cluster']) && !empty($info['matroska']['cluster'])) {
+                        // Scan maximum 1MB window before EOF
+                        $this->resetParserBuffer(max(0, $info['avdataend'] - (1024 * 1024)));
+                        $this->scan_mode = MATROSKA_SCAN_LAST_CLUSTER;
+
+                        try {
+                                $this->parseEBML($info);
+                        } catch (Exception $e) {
+                                $this->error('EBML parser (end of file): '.$e->getMessage());
+                        }
+                }
+
+                // Reset to header parsing mode (this method is only called during header-only parsing)
+                $this->scan_mode = MATROSKA_SCAN_HEADER;
+        }
+
+        /**
+         * Fetch TimecodeScale from Info element
+         *
+         * @param array $info
+         *
+         * @return int TimecodeScale value
+         */
+        private function getTimecodeScale(&$info) {
+                $timecodeScale = MATROSKA_DEFAULT_TIMECODESCALE;
+                if (isset($info['matroska']['info']) && is_array($info['matroska']['info'])) {
+                        foreach ($info['matroska']['info'] as $infoarray) {
+                                if (isset($infoarray['TimecodeScale'])) {
+                                        $timecodeScale = $infoarray['TimecodeScale'];
+                                        break;
+                                }
+                        }
+                }
+                return $timecodeScale;
+        }
+
+        /**
+         * Calculate duration from scanned cluster timecodes
+         *
+         * @param array $info
+         *
+         * @return void
+         */
+        private function calculatePlaytimeFromClusters(&$info) {
+                $minTimecode = null;
+                $maxTimecode = null;
+                if (isset($info['matroska']['cluster']) && is_array($info['matroska']['cluster'])) {
+                        foreach ($info['matroska']['cluster'] as $cluster) {
+                                if (isset($cluster['ClusterTimecode'])) {
+                                        if ($minTimecode === null || $cluster['ClusterTimecode'] < $minTimecode) {
+                                                $minTimecode = $cluster['ClusterTimecode'];
+                                        }
+                                        if ($maxTimecode === null || $cluster['ClusterTimecode'] > $maxTimecode) {
+                                                $maxTimecode = $cluster['ClusterTimecode'];
+                                        }
+                                }
+                        }
+                }
+                if ($maxTimecode !== null && $minTimecode !== null && $maxTimecode > $minTimecode) {
+                        $info['playtime_seconds'] = ($maxTimecode - $minTimecode) * ($this->getTimecodeScale($info) / 1000000000);
+                }
+        }
 }