Warning: chmod(): Operation not permitted in /var/www/hopeinstoughton_www/wp-content/plugins/simple-universal-google-analytics/main.php on line 8
[0] in function chmod in /var/www/hopeinstoughton_www/wp-content/plugins/simple-universal-google-analytics/main.php on line 8
[1] in function include_once in /var/www/hopeinstoughton_www/wp-settings.php on line 560
[2] in function require_once in /var/www/hopeinstoughton_www/wp-config.php on line 85
[3] in function require_once in /var/www/hopeinstoughton_www/wp-load.php on line 50
[4] in function require_once in /var/www/hopeinstoughton_www/wp-blog-header.php on line 13
[5] in function require in /var/www/hopeinstoughton_www/index.php on line 17

<?php

// process an audio file

class pai_audio extends Model {
	
	private $source;
	private $targetfolder;
	private $target;
	private $mainfile;
	
	const FFMPEG = '/usr/bin/ffmpeg';
	
	public function __construct() {
		parent::__construct();
		$this->targetfolder = HOME . '/storage/patientapps/audio';
		@mkdir($this->targetfolder, 0777, true); 
	}
	
	// source filename
	public function Source() {
		if (func_num_args()) {
			$this->source = func_get_arg(0);
			return $this;
		}
		return $this->source;
	}

	// Target Procedure Abbreviation
	public function Target() {
		if (func_num_args()) {
			$this->target = func_get_arg(0);
			return $this;
		}
		return $this->target;
	}
	
	public function Process() {
        $elements = pathinfo($this->source);
        $ext = $elements['extension'];
        $base = $elements['dirname'];
		$this->mainfile = $this->target . '.' . $ext;
        if (file_exists($base . '/final.flac')) {
            $basename = $elements['basename'];
            $dtarget = escapeshellarg($base . '/' . $basename);
            $dsource = escapeshellarg($base . '/final.flac');
			$command = self::FFMPEG . " -i {$dsource} -ab 192k -ar 22050 {$dtarget}";
			`$command`;

            $this->source = $base . '/final.flac';
			$this->prepareMP3();			// mp3. tag it?
            
            $target = $this->targetfolder . '/' . $this->target . '.cc';
            if (!file_exists($target)) {
                $content = file_get_contents($base . '/final.cc');
                $content = $this->_trimContent($content);   // remove trailing spaces
                file_put_contents($target, $content);
            }
        } elseif (file_exists($this->source)) {
			$this->prepareMainFile();		// copy wav to storage
			$this->prepareMP3();			// mp3. tag it?
			$this->prepareFlac();			// flac
			$this->Transcribe();			// create CC file, if missing. 
			$this->Timings();				// edit the generated file manually, and make a copy of it
			// to do:  use ffmpeg to get duration?
		}
       
		return $this;
	}
    
    private function _trimContent($content) {
        $output = '';
        foreach(explode("\n", $content) as $line) {
            $line = rtrim($line);
            $output .= "$line\n";
        }
        return $output;
    }
	
	public function prepareMP3() {
		$target = $this->targetfolder . '/' . $this->target . '.mp3';
		$dsource = escapeshellarg($this->source);
		$dtarget = escapeshellarg($target);
		
		if (!file_exists($target)) {
			$command = self::FFMPEG . " -i {$dsource} -ab 192k -ar 22050 {$dtarget}";
			`$command`;
		}
	}

	public function prepareFlac() {		// used by speech to text
		$target = $this->targetfolder . '/' . $this->target . '.flac';
		$dsource = escapeshellarg($this->source);
		$dtarget = escapeshellarg($target);
		
		if (!file_exists($target)) {
			$command = self::FFMPEG . " -i {$dsource} -acodec flac -b:a 32k -ac 1 -ar 16000 -compression_level 10 {$dtarget}";
			`$command`;
			$target = $this->targetfolder . '/' . $this->target . '.serial';
			@unlink($target);
		}
	}

	public function Transcribe() {
		$target = $this->targetfolder . '/' . $this->target . '.serial';
		if (!file_exists($target)) {

			$creds = file_get_contents(__DIR__ . '/credentials.serial');
			$creds = unserialize($creds);

			$source = $this->targetfolder . '/' . $this->target . '.flac';
			$output = $this->Watson_ProcessAudio($creds, $source);
			file_put_contents($target, serialize($output));
		}
	}
	
	public function Timings() {
		$target = $this->targetfolder . '/' . $this->target . '.cc';
		if (!file_exists($target)) {
			$source = $this->targetfolder . '/' . $this->target . '.serial';
			$result = $this->_transcribe_simplify($source);
			file_put_contents($target, $result);
		}
	}

	private function prepareMainFile() {
		$dest = $this->targetfolder . '/' . $this->mainfile;
		if (!file_exists($dest)) {
			copy($this->source, $dest);
		}
	}

	private function Watson_ProcessAudio($info, $flac_file) {
		global $outfile;
		$dir = __DIR__;
		set_time_limit(600);
		
		$url = 'https://stream.watsonplatform.net/speech-to-text/api/v1/recognize';
		$url .= '?model=en-US_NarrowbandModel';
		$url .= '&timestamps=true';
		$ch = curl_init($url);
		
		$headers = array();
		$headers[] = 'Content-Type: audio/flac';
		$headers[] = "Transfer-Encoding: chunked";
		$headers[] = 'X-Watson-Learning-Opt-Out: true';
		curl_setopt($ch, CURLOPT_HTTPHEADER, $headers);
		curl_setopt($ch, CURLOPT_BINARYTRANSFER, TRUE);
		// send a file
		curl_setopt($ch, CURLOPT_POST, true);
		$data = file_get_contents($flac_file);
		curl_setopt($ch, CURLOPT_POSTFIELDS, $data );
		curl_setopt($ch, CURLOPT_INFILESIZE, strlen($data));

		curl_setopt($ch, CURLOPT_HTTPAUTH, CURLAUTH_ANY);
		curl_setopt($ch, CURLOPT_USERPWD, "apikey:{$info['apikey']}");
//		curl_setopt($ch, CURLOPT_HEADER, 1);
		// output the response
		curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
		$info = curl_exec($ch);
//        var_dump($info);
//        exit;
		$info = json_decode($info, TRUE);

// close the session
		curl_close($ch);
		return $info;
	}
    
/*
HTTP/1.1 100 Continue
X-EdgeConnect-MidMile-RTT: 1
X-EdgeConnect-Origin-MEX-Latency: 45

HTTP/1.1 500 Internal Server Error
Server: AkamaiGHost
Mime-Version: 1.0
Content-Type: text/html
Content-Length: 177
Expires: Fri, 12 Mar 2021 19:43:03 GMT
X-Reference-Error: 179.573a2f17.1615578183.5bdd5467
Date: Fri, 12 Mar 2021 19:43:03 GMT
Connection: close

<HTML><HEAD><TITLE>Error</TITLE></HEAD><BODY>
An error occurred while processing your request.<p>
Reference&#32;&#35;179&#46;573a2f17&#46;1615578183&#46;5bdd5467
</BODY></HTML>
*/
	
	public function AsArray() {
		$target = $this->targetfolder . '/' . $this->target . '.cc';
		$cc =  file_get_contents($target);
		return [
			'filename' => [
				$this->mainfile,
				$this->target . '.mp3',
				$this->target . '.flac'
			],
			'cc' => $cc,
		];
	}
	
	// word, from/to timeperiod (in ms)
	// also total length (in ms)
	private function _transcribe_simplify($target) {
		$data = unserialize(file_get_contents($target));
		$output = [];
		if (isset($data['results'])) {
			foreach($data['results'] as $segment) {
				$this->_processAlternatives($output, $segment['alternatives'][0]['timestamps']);
			}
			// look for breaks
		}
		$result = $this->_breakintoPhrases($output);
		$output = $this->_asText($result);
		return $output;
	}
	
	private function _breakintoPhrases($words) {
		$phrases = [];
		$previous = 0;
		$idx = 0;
		foreach($words as $word) {
			$gap = $word['start'] - $previous;
			if ($gap >= 300) {
				$idx++;
			}
			$phrases[$idx][] = $word;
			$previous = $word['end'];
		}
		// join words in phrases together
		$output = [];
		foreach($phrases as $phrase) {
			$c = count($phrase);
			$current = $phrase[0];
			for($i = 1; $i < $c; $i++) {
				if (strlen($current['phrase'] . ' ' . $phrase[$i]['phrase']) > 30) {
					$output[] = $current;
					$current = $phrase[$i];
				} else {
					$current['phrase'] .= ' ' . $phrase[$i]['phrase'];
					$current['end'] = $phrase[$i]['end'];
				}
			}
			$output[] = $current;
		}
		return $output;
	}
	
	private function _asText($words) {
		$output = '';
		$previous = 0;
		foreach($words as $word) {
			$output .= "{$word['start']}-{$word['end']}={$word['phrase']}\n";
		}
		return $output;
	}
	
	
	private function _processAlternatives(&$output, $words) {
//		$result = '';
		foreach($words as $word) {
			$start = intval($word[1] * 1000);
			$end = intval($word[2] * 1000);
			$output[] = [ 'start' => $start, 'end' => $end, 'phrase' => $word[0] ];
//			$result .= "{$start}-{$end}={$word[0]}\n";
		}
//		return $result;
	}
	
	function filesize_formatted($size)
	{
		$units = array( 'B', 'KB', 'MB', 'GB', 'TB', 'PB', 'EB', 'ZB', 'YB');
		$power = $size > 0 ? floor(log($size, 1024)) : 0;
		return number_format($size / pow(1024, $power), 2, '.', ',') . ' ' . $units[$power];
	}
	
	private function _getJsonItemSize($filename, $entry) {
		$content = json_decode(file_get_contents($filename), TRUE);
		$content = $content[$entry] ?? '';
		return strlen($content);
	}
	
	public function GetFilelist($subdomain, $fullhash) {
		$result = [];
		$files = glob($this->targetfolder . '/' . $subdomain . '.*');
		foreach($files as $filename) {
			$ext = pathinfo($filename, PATHINFO_EXTENSION);
			$name = basename($filename);
			if ($ext == 'serial') continue;			// internal use only - placement of individual words in the audio file
			if ($ext == 'flac') continue;			// internal use only - used for speech to text

			$stat = stat($filename);
			$date = $stat['mtime'];
			$size = $stat['size'];
			$node = [
				'name' => $name,
				'date' => date('Y-m-d H:i', $date),
				'fs' => $size,
				'size' => $this->filesize_formatted($size),
			];
			switch($ext) {
				case 'cc':
					$node['action'] = 'edit captions';
					break;
				case 'sql3':
					$node['action'] = 'sqlite3 database';
					break;
				
				case 'json';
					// product json is the file containing page layouts for the app. 
					// the large json file created here contains this product json, and all the images.
					$itemsize = $this->_getJsonItemSize($filename, 'data.json');
					$node['action'] = '<a href="/patientapps/product-json/' . $fullhash . '" target="_blank" title="the product json is a subset of this json file - it does not contain the images">view product json</a>'.
								' [<span title="This has a limit of 5MB">' . $this->filesize_formatted($itemsize) . '</span>]' ;
//					$node['action'] = '<a href="/patientapps/product-toc/' . rawurlencode($name) . '" target="_blank">view json tree</a>';
					$x = rawurlencode(str_replace('.json', '', $name));
					$node['link'] = 'https://beta.patientapps.net/api/content/' . $x;
					break;
			}
			$result[] = $node;
		}
		return !count($result) ? FALSE : $result;
	}

}