OCR v0.1

build/router/controller/ics.php

@@ -39,11 +39,13 @@
 			/* [2] Display file
 			=========================================================*/
 			/* (1) Headers */
-			header('Content-Type: text/calendar; charset=utf-8');
+			// header('Content-Type: text/calendar; charset=utf-8');
-			header('Content-Disposition: attachment; filename='.$this->diplome_id.'.ics');
+			// header('Content-Disposition: attachment; filename='.$this->diplome_id.'.ics');
 
 			/* (2) Body */
+			echo "<pre>";
 			readfile($file_name);
+			echo "</pre>";
 
 
 		}

build/service/CalendarExtractor.php

@@ -3,6 +3,7 @@
 	namespace service;
 
 	use \lightdb\core\lightdb;
+	use \service\Tesseract;
 
 
 	class CalendarExtractor{
@@ -197,7 +198,9 @@
 							$this->event[$uid][$time] = [ $this->yToTime($day_n, $y) ];
 
 							// {6} Exctract event's image //
-							$this->event[$uid][$time][1] = $this->extractEvent("$time-$uid", [$col_x, $start_y+1], [$col_ind[$day_n+1]-1, $y]);
+							$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];
 
 
 						}
@@ -240,11 +243,13 @@
 			$link = __ROOT__."/tmp/$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, $height);
+				$clip = \imagecreatetruecolor($width*$resize_factor, $height*$resize_factor);
 
 				$copied = \imagecopyresized(
 					$clip,                  // destin img
@@ -253,8 +258,8 @@
 					0,                      // dest y
 					$start[0],              // src x
 					$start[1],              // src y
-					$width, // dest w
+					$width*$resize_factor,  // dest w
-					$height, // dest h
+					$height*$resize_factor, // dest h
 					$width,                 // src w
 					$height                 // src h
 					);
@@ -263,19 +268,47 @@
 				if( !$copied )
 					throw new \Exception("Cannot clip image");
 
-				/* (3) Save to base64 */
+				/* (3) Save to jpeg */
 				\imagesavealpha($clip, true);
 
-				ob_start();
+				// ob_start();
-				\imagejpeg($clip);
+				\imagejpeg($clip, $link);
-				$image_data = \base64_encode(ob_get_contents());
+				// $image_data = \base64_encode(ob_get_contents());
-				ob_end_clean();
+				// 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 = [ 'unkown', 'unknown' ];
+
+				}
 
 
 			}
 
 
-			return $image_data;
+
+			/* [3] End procedure
+			=========================================================*/
+			/* (1) Remove file */
+			unlink($link);
+
+			/* (2) Return read value */
+			return $read;
+
 		}
 
 
@@ -389,11 +422,6 @@
 			=========================================================*/
 			foreach($this->event as $event_col=>$events){
 
-				$type = "unknown";
-
-				if( isset($col_assoc[$event_col]) )
-					$type = $col_assoc[$event_col];
-
 				/* (2) For each event of each type
 				---------------------------------------------------------*/
 				foreach($events as $start_t=>$data){
@@ -402,8 +430,9 @@
 					$RAW .= "DTSTART:${start_t}\n";
 					$RAW .= "DTEND:${data[0]}\n";
 					$RAW .= "UID:$start_t-univ-pau-ics\n";     // required
-					$RAW .= "SUMMARY:$type\n";
+					$RAW .= "SUMMARY:${data[1]}\n";
-					$RAW .= "ATTACH;ENCODING=BASE64;VALUE=BINARY;FILENAME=att.jpg:${data[1]}\n";
+					$RAW .= "LOCATION:${data[2]}\n";
+					// $RAW .= "ATTACH;ENCODING=BASE64;VALUE=BINARY;FILENAME=att.jpg:${data[1]}\n";
 					$RAW .= "CATEGORIES: UPPA Calendar\n";
 					$RAW .= "END:VEVENT\n";
 				}

build/service/Tesseract.php

@@ -6,85 +6,80 @@
 
 		/* [1] Attributes
 		=========================================================*/
-		private $filename = null;
+		private $fname    = null;
-		private $content = null;
-		private $course = null;
-        private $teacher = null;
-        private $room = null;
 
 
 		/* (1) Constructs and initialise a readed file
 		*
+		* @fname<String>						Path of the image
+		*
 		* @return instance<Tesseract>			New Tesseract
 		*
 		---------------------------------------------------------*/
-		public function __construct($filename){
+		public function __construct($fname=null){
-		    $this->filename = $filename;
+
+			/* [1] Check argument
+			=========================================================*/ {
+
+				/* (1) Check type */
+				if( !is_string($fname) )
+					throw new \Exception("Tesseract.__construct(<String>) expected but Tesseract.__construct(<".gettype($fname).">) received");
+
+				/* (2) Check file validity */
+				if( !file_exists($fname) )
+					throw new \Exception("Tesseract.__construct(<PATH>) but <PATH> is not valid");
+
+			}
+
+
+			/* [2] Store as attribute
+			=========================================================*/
+		    $this->fname = $fname;
 
-		    return $this;
 		}
 
 
 
 		/* (2) Read the image file
 		*
-		* @return this
+		* @return read<String>					The read content
 		*
 		---------------------------------------------------------*/
 		public function read(){
 
 			/* [1] Record the text from the image
 			=========================================================*/
-			$filename = $this->filename;
+			/* (1) Process tesseract */
-            $this->content = shell_exec("tesseract $filename stdout -l fra");
+			$read = shell_exec("tesseract ".$this->fname." stdout -l fra --user-words ".__ROOT__."/config/edt.user-words -c language_model_penalty_non_freq_dict_word=0.1 -c language_model_penalty_non_dict_word=.15 -c tessedit_char_whitelist=abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789 2>/dev/null");
 
-            $lists = explode(chr(10), $this->content);
+			/* (2) If empty */
+			if( is_null($read) )
+				throw new \Exception("Nothing read");
+
+			/* (3) Split by lines */
+			$by_line = explode("\n", $read);
+
+			/* (4) Get first line (title) */
+			$title = $by_line[0];
+
+			/* (5) Get last non-empty line */
+			for( $i = count($by_line)-1 ; $i > 0 ; $i-- ){
+
+				// {1} Check not empty //
+				if( empty($by_line[$i]) )
+					continue;
+
+				// {2} Matches //
+				if( preg_match('@^amphi@i', $by_line[$i]) || // 'amphi A', 'amphi 600 droit'
+					preg_match('@^S\d+@i', $by_line[$i])     // 'S10', 'S22'
+				)
+					return [ $title, $by_line[$i] ];
 
-            if (count($lists) < 3) {
-                throw new \Exception('Result not interpreted');
 			}
 
-            $this->course   = $lists[0];
-            $this->teacher  = $lists[1];
-            $this->room     = $lists[2];
 
-			return $this;
+			return [ $title, 'unknown' ];
-		}
 
-        /* (3) Return the text readed by the Tesseract OCR
-		*
-		* @return $this->content
-		*
-		---------------------------------------------------------*/
-		public function getContent() {
-		    return $this->content;
-        }
-
-        /* (4) Return the course readed by the Tesseract OCR
-		*
-		* @return $this->course
-		*
-		---------------------------------------------------------*/
-        public function getCourse() {
-            return $this->course;
-        }
-
-        /* (5) Return the teacher readed by the Tesseract OCR
-		*
-		* @return $this->teacher
-		*
-		---------------------------------------------------------*/
-        public function getTeacher() {
-		    return $this->teacher;
-        }
-
-        /* (6) Return the room class readed by the Tesseract OCR
-		*
-		* @return $this->room
-		*
-		---------------------------------------------------------*/
-        public function getRoom() {
-		    return $this->room;
 		}
 
 

config/edt.user-words

@@ -0,0 +1,19 @@
+intro
+introduction
+diversite
+evolution
+mathematiques
+general
+Gestion
+gestion
+Molécules
+cellulaire
+biologie
+physique
+amphi
+CTD
+droit
+gauche
+grammaire
+methodologie
+Cours

install.sh

@@ -2,3 +2,4 @@
 
 sudo apt-get install php7.0-gd;
 sudo service apache2 restart;
+sudo apt-get install tesseract-ocr tesseract-ocr-fra;