univ-pau-ics/build/service/CalendarExtractor.php

<?php

	namespace service;

	use \lightdb\core\lightdb;
	use \service\Tesseract;


	class CalendarExtractor{


		/* [1] Attributes
		=========================================================*/
		private $start_d = null;  // 1st day date
		private $img_url = null;  // image url
		private $d_uid   = null;  // diplome uid
		private $img_res = null;  // image resource
		private $event   = [];    // events

		/* [2] Constants
		=========================================================*/
		public static $start_y = 79;   // start y
		public static $stop_y  = 699;  // stop y
		public static $start_h = 8;    // start hour
		public static $stop_h  = 20;   // stop hour



		/* (1) Builds a calendar extractor
		*
		* @d_uid<String>							UID of the diplome
		* @img_url<String>							URL of the string to extract from
		* @start_d<String>							Date of the first day
		*
		* @return instance<CalendarExtractor>		Instance
		*
		---------------------------------------------------------*/
		public function __construct($d_uid=null, $img_url=null, $start_d=null){

			/* [1] Check arguments
			=========================================================*/ {

				/* (1) Check type */
				if( !is_string($img_url) || !is_string($start_d) )
					throw new \Exception("CalendarExtractor.__construct(<String>, <String>, <String>) expected but CalendarExtractor.__construct(<".gettype($d_uid).">, <".gettype($img_url).">, <".gettype($start_d).">) received");

				/* (2) Check @img_url link availability */
				if( !($img_head=@get_headers($img_url)) )
					throw new \Exception("CalendarExtractor.__construct(<String>, <URL>, <String>) received but cannot reach <URL>");

				if( !preg_match('@HTTP.+200@', $img_head[0]) )
					throw new \Exception("CalendarExtractor.__construct(<String>, <URL>, <String>) received but cannot reach <URL>");

				/* (3) Check @start_d format */
				if( !preg_match("@^\d{1,2}-\d{1,2}-\d{3,}$@", $start_d) )
					throw new \Exception("CalendarExtractor.__construct(<String>, <String>, <DATE>) received <DATE> has not the correct format");

				/* (4) Check @d_uid format */
				if( !preg_match("@^T\d+$@", $d_uid) )
					throw new \Exception("CalendarExtractor.__construct(<UID>, <String>, <String>) received <UID> has not the correct format");


			}


			/* [2] Fetch file
			=========================================================*/ {

				/* (1) Try to open file with GD */
				$this->img_res = @imagecreatefrompng($img_url);

				/* (2) Manage error */
				if( !$this->img_res )
					throw new \Exception("URL is not a JPEG image, or is unreachable");

				/* (3) Register data */
				$this->img_url = $img_url;
				$this->d_uid   = $d_uid;
				$this->start_d = $start_d;

			}

		}




		/* (2) Extracts calendar data from image
		*
		* @return error<bool>			FALSE on error
		*
		---------------------------------------------------------*/
		public function process(){

			/* [1] Global variables
			=========================================================*/ {

				/* (1) Image size */ {

					// {1} Request for size //
					$img_siz = @getimagesize($this->img_url);

					// {2} If error //
					if( !$img_siz )
						throw new \Exception("Cannot get image size");

					$img_siz = [
						'w' => $img_siz[0],
						'h' => $img_siz[1]
					];


				/* (2) Light database
				---------------------------------------------------------*/

					/* (1) Create/Open database */
					try{
						$ldb = new lightdb('config');

					/* (2) Manage error */
					}catch(\Exception $e){
						throw new \Exception("Cannot access database");
					}

					/* (3) Fetch diplome data */
					$d_db = $ldb->fetch($this->d_uid);

					/* (4) Manage error */
					if( $d_db == false )
						$d_db = [];



				}

			}




			/* [2] Extract day limits
			=========================================================*/ {

				/* (1) Will contain the column+1 w index */
				$col_ind = [];

				/* (2) Extract column indexes */
				for( $x = 0 ; $x < $img_siz['w'] ; $x++ ){

					if( $this->getColor($x, 62) <= 10 )
						$col_ind[] = $x+1;

				}


			}



			/* [3] For each day -> get events
			=========================================================*/ {

				$uid = 0;

				/* (1) For each day */
				for( $day_n = 0 ; $day_n < count($col_ind)-1 ; $day_n++ ){
					$col_x = $col_ind[$day_n];


					/* (2) For each y pixel -> exctract event */
					for( $y = self::$start_y ; $y < self::$stop_y ; $y++ ){

						/* (3) Get current color + next */
						$p  = $this->getColor($col_x, $y);
						$p1 = $this->getColor($col_x, $y+1);
						$p1e = '#'.dechex($p1);

						/* (4) If on black pixel and next not white */
						if( $p == 0 && $p1 != 0xffffff ){

							// {1} calculate time //
							$start_y = $y;
							$time = $this->yToTime($day_n, $start_y);

							// {2} Incr uid //
							$uid++;

							// {3} Store event start //
							$this->event[$uid][$time] = [];

							// {4} Seek end of event //
							$y++;
							while( $y < self::$stop_y && $this->getColor($col_x, $y) != 0 )
								$y++;


							// {5} If end reached //
							$this->event[$uid][$time] = [ $this->yToTime($day_n, $y) ];

							// {6} Exctract event's image //
							$ev = $this->extractEvent("$time-$uid", [$col_x, $start_y+1], [$col_ind[$day_n+1]-1, $y]);
							$this->event[$uid][$time][1] = $ev[0];
							$this->event[$uid][$time][2] = $ev[1];


						}



					}

				}


			}


			/* [4] Save db changes
			=========================================================*/
			/* (1) Update data */
			$ldb->delete($this->d_uid);
			$ldb->insert($this->d_uid, $d_db);

			/* (2) Close connection */
			$ldb->close();


		}





		/* (3) Exctracts an event using OCR
		*
		* @uid<String>							Image uid
		* @start<Array>							Start rect (x, y)
		* @stop<Array>							Stop rect (x, y)
		*
		---------------------------------------------------------*/
		public function extractEvent($uid, $start=null, $stop=null){

			$link = __ROOT__."/tmp/$uid.jpeg";
			$width  = $stop[0]-$start[0];
			$height = $stop[1]-$start[1];
			$resize_factor = 2;

			/* [1] Get the right clip
			=========================================================*/ {

				/* (1) Create clipped copy */
				$clip = \imagecreatetruecolor($width*$resize_factor, $height*$resize_factor);

				$copied = \imagecopyresized(
					$clip,                  // destin img
					$this->img_res,         // source img
					0,                      // dest x
					0,                      // dest y
					$start[0],              // src x
					$start[1],              // src y
					$width*$resize_factor,  // dest w
					$height*$resize_factor, // dest h
					$width,                 // src w
					$height                 // src h
					);

				/* (2) Manage copy error */
				if( !$copied )
					return [ null, null ];

				/* (3) Save to jpeg */
				\imagesavealpha($clip, true);

				// ob_start();
				\imagejpeg($clip, $link);
				// $image_data = \base64_encode(ob_get_contents());
				// ob_end_clean();

			}



			/* [2] Apply Tesseract
			=========================================================*/ {

				/* (1) Load image with tesseract */
				try{

					$tesseract = new Tesseract($link);
					$read = $tesseract->read();

				/* (2) Manage error */
				}catch(\Exception $e){

					$read = [ null, null ];

				}


			}



			/* [3] End procedure
			=========================================================*/
			/* (1) Remove file */
			unlink($link);

			/* (2) Return read value */
			return $read;

		}


		/* (3) Get a pixel's color
		*
		* @x<int>							X coordinate
		* @y<int>							Y coordinate
		*
		* @return color<u_int32>			Raw color
		*
		---------------------------------------------------------*/
		private function getColor($x, $y){
			return imagecolorat($this->img_res, $x, $y);
		}




		/* (4) Get time from a y-coordinate
		*
		* @day_n<int>						Day relative index
		* @y<int>							y Coordinate
		*
		* @return time<String>				Formatted time (HH:mm)
		*
		---------------------------------------------------------*/
		private function yToTime($day_n, $y){

			/* [1] Get the date
			=========================================================*/ {

				/* (1) Get the day's date */
				$day_ts = strtotime($this->start_d." + $day_n days");

				/* (2) Format it */
				$day = date('Ymd', $day_ts);

			}


			/* [2] Get the time
			=========================================================*/{

				/* (1) Get the time */
				$time = self::$start_h + (self::$stop_h-self::$start_h) * ($y-self::$start_y) / (self::$stop_y-self::$start_y);

				/* (2) Calculate hour form time */
				$hour = floor($time);

				/* (3) Calculate min from time */
				$min = round( 60 * ($time-$hour) );

				/* (4) Round minutes to 10min */
				$min = round($min/10)*10;

				/* (5) Format to 2-digit */
				$hour = ( $hour < 10 ) ? "0$hour" : $hour;
				$min  = ( $min  < 10 ) ? "0$min"  : $min ;

			}


			/* [3] Convert to GMT (UTC+0)
			=========================================================*/
			/* (1) Set fixed timezone offset */
			$tz_offset = +2;

			/* (2) Get GMT (UTC+0) timestamp */
			$ts = strtotime("${day} $hour:$min:00") - (3600*$tz_offset);

			/* (3) Return GMT date */
			return date("Ymd\THis\Z", $ts);

		}






		/* (5) Generate ICS output
		*
		* @return ics<String>			ICS Representation of the events
		*
		---------------------------------------------------------*/
		public function toIcs(){
			$RAW = "";



			/* [1] Get light-database
			=========================================================*/
			/* (1) Create/Open database */
			try{
				$ldb = new lightdb('config');

			/* (2) Manage error */
			}catch(\Exception $e){
				throw new \Exception("Cannot access database");
			}

			/* (3) Get color association if available */
			$col_assoc = [];
			if( !is_null($ldb->index($this->d_uid)) )
				$col_assoc = $ldb->fetch($this->d_uid);

			// var_dump($col_assoc);
			// die(1);

			/* [2] For each event
			=========================================================*/
			foreach($this->event as $event_col=>$events){

				/* (2) For each event of each type
				---------------------------------------------------------*/
				foreach($events as $start_t=>$data){
					$RAW .= "BEGIN:VEVENT\n";
					$RAW .= "DTSTAMP:".gmdate("Ymd\THis\Z", time())."\n"; // required
					$RAW .= "DTSTART:${start_t}\n";
					$RAW .= "DTEND:${data[0]}\n";
					$RAW .= "UID:$start_t-univ-pau-ics\n";     // required
					if( !is_null($data[1]) )
						$RAW .= "SUMMARY:${data[1]}\n";
					else
						$RAW .= "SUMMARY:?\n";
					if( !is_null($data[2]) ){
						$RAW .= "LOCATION:${data[2]}\n";
						$RAW .= "DESCRIPTION:${data[2]}\n";
					}
					$RAW .= "CATEGORIES: UPPA Calendar\n";
					$RAW .= "END:VEVENT\n";
				}

			}

			return $RAW;

		}


	}