REPERAGE D'INFORMATIONS PERSONNELLES AVEC REGEXP

<?php
echo '<meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1">';

//pratique pour les regexp
function accent($chaine)
{
   $accent  ="ÀÁÂÃÄÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖØÙÚÛÜÝÞßàáâãäåæçèéêëìíîïðñòóôõöøùúûýýþÿ";
   $noaccent="aaaaaaaceeeeiiiidnoooooouuuuybsaaaaaaaceeeeiiiidnoooooouuuyyby";
   return strtr(trim($chaine),$accent,$noaccent);
} 

if(isset($_FILES["fichier"]["name"])) {

        //cette partie doit etre grandement amelioree niveau securite

	if(empty($_FILES["fichier"]["name"])) echo "Veuillez indiquer le chemin du fichier";

	$dir="Upload\\";
	$fichier=$dir . basename($_FILES["fichier"]["name"]);
	$tmp=$_FILES["fichier"]["tmp_name"];
	$size=$_FILES["fichier"]["size"];
	
	if(!is_uploaded_file($tmp)) { echo " fichier absent"; };
	if(!move_uploaded_file($tmp, $fichier)) { echo " pb deplacement "; };

	echo "$fichier <br>";
	echo "$size <br>";

	$com = 'antiword -t "'.$fichier.'" > temp.txt';
	echo "$com <br>";
	
	if(exec($com)) echo "conversion reussie"; 

	if($handle = fopen("temp.txt", "rb")) {
	echo "ouverture reussie <br>";
	}
	
        //pre-affichage

	while (!feof($handle)) {
		$buffer = fgets($handle, 4096);
		$buffer = preg_replace(array('#[|]#s','#^-#','#^—#','#^[(]#','#^[ ]#'), '', trim($buffer));
		$buffer = preg_replace('#([ ][ ]+)#s', ' ', trim($buffer));
		$buffer = accent($buffer);
		echo "ligne : $buffer<br>";
	}
   fclose($handle);
   
   if(false !== ($handle = fopen('temp.txt', 'rb'))) {
	echo "ouverture reussie <br>";
	}
		
	$j=0;
	$c=0;
	$d=0;
	$k=0;
	$z=0;
	$f=0;
	$societe="";
	$prenom="";
	$nom="";
	$adresse="";
	$ville="";
	$cp="";
	$divers="";
	$taille=0;
	$age=0;
	$l=false;
	$testVille=false;
	$testNom=false;
	$testCiv=false;
	$testEmail=false;
	$testAdresse=false;
	$testFixe=false;
	$testPort=false;
	$testCp=false;
	$testAge=false;
	$testExp=false;
	$testForm=false;
		
	while (feof($handle)===false) {
		if($testForm==true && $testExp==false && $z<=250){
		$z++;
		}else{
		$buffer = fgets($handle, 4096);
                //on vire les saloperies de debut de ligne
		$buffer = preg_replace(array('#[|]#s','#^-#','#^—#','#^[(]#','#^[ ]#'), '', trim($buffer));
                //on remplace les espaces mutliples
		$buffer = preg_replace('#([ ][ ]+)#s', ' ', trim($buffer));
		$buffer = accent($buffer);
		}
		
                //Si vous avez du mal a voir a quoi sert la regexp, regardez le nom du booleen

		if(preg_match('#(M(onsieur|adame|mademoiselle|r|d|lle))#',$buffer,$out) !==0 && $testCiv===false) { 
			$civilite = $out[0];
			print_r($out);
			echo "<br>";
			echo $buffer = str_replace($civilite, '', $buffer); 
			echo "<br>civilite : $civilite<br>";
			$testCiv=true;
			echo"je suis passe dans la boucle civilite<br>";
		}
			
		if(preg_match('#[A-Z][[:alpha:]]+(?:\s)?[-—]?(?:\s)?([A-Z][[:alpha:]]+)?(?:\s)?([A-Z][[:alpha:]]+)?(?:\s)?[-—]?(?:\s)?[A-Z][[:alpha:]]+\s*#',$buffer,$out) !==0 && $testNom===false) {
			$res = preg_split('# #', $buffer, -1, PREG_SPLIT_NO_EMPTY);
			$testNom=true;
			switch (count($res)) {
				case (2) : $prenom = $res[0]; $nom = $res[1];
					break;
					
				case (3) : $prenom = $res[0]; $nom = $res[1].' '.$res[2];
					break;
					
				case (4) : 
					if ( $res[1]=="-" || $res[1]=="—") { 
						$prenom = $res[0].' '.$res[1].' '.$res[2];
						$nom = $res[3];
						break; }
					if ( $res[2]=="-" || $res[2]=="—") {
						$prenom = $res[0];
						$nom = $res[1].' '.$res[2].' '.$res[3]; 
						break; 
						}else{
							$prenom = $res[0];
							$nom = $res[1].' '.$res[2].' '.$res[3];
							break;
						}
							
				case (5) :
					if ($res[1]=="-" || $res[1]=="—") {
						$prenom = $res[0].' '.$res[1].' '.$res[2];
						$nom = $res[3].' '.$res[4];
						break;
						}
					if ( $res[3]=="-" || $res[3]=="—") {
						$prenom = $res[0];
						$nom = $res[1].' '.$res[2].' '.$res[3].' '.$res[4];
						break;
						}
							
					case (6) :
						$prenom = $res[0].' '.$res[1].' '.$res[2];
						$nom = $res[3].' '.$res[4].' '.$res[5];
						break;
			}
			echo "nom : $nom<br>";
			echo "prenom : $prenom<br>";
		}
			
		if(preg_match('#([0-9]{1,3})[, ]?(?:[a-zA-Z]{1,3})?[, ]?(rue|avenue|boulevard|impasse|cours)(?:,)?(?:\s)?([-—\'0-9[:alpha:]](?:\s)?)+\s*#i', $buffer, $out) !==0 && $testAdresse===false) {
			$adresse = $out[0];
			echo "adresse : $adresse<br>";
			$testAdresse=true;
		}
			
		if(preg_match('#(([0-9](?:\s)?){5})\s([-—[:alpha:]](?:\s)?)+\s*#', $buffer, $out) !==0 && $testCp===false) {
			$res = preg_split('# #', $buffer, -1, PREG_SPLIT_NO_EMPTY);
			$testCp=true;
			$pm = sizeof($res);
			for ($i=0; $i<=sizeof($res)-1; $i++) {
				if (preg_match('#[0-9]{2}|[0-9]{3}#', $res[$i], $out) !==0 && $j<=2) {
					$cp .= $res[$i];
					if(preg_match('#[0-9]{5}#', $cp, $out)) {
						$j=3;
					}
				$j++;
				}
				
				if (preg_match('#(([[:alpha:]](?:\s)?)[-—]?(?:\s)?)+#', $res[$i], $out) !==0 && $testVille===false) {
					$ville .= $out[0].' ';
				}
			}
			echo "ville : $ville<br>";
			echo "cp : $cp<br>";
		}
		
		if(preg_match('#[a-z0-9]([-_.]?[a-z0-9])+@[a-z0-9]([-_.]?[a-z0-9])+\.[a-z]{2,5}$#ixm', $buffer, $out) !==0 && $testEmail===false) {
			$email = $out[0];
			echo " email : $email<br>";
			$testEmail = true;
		}
			
		if(preg_match('#[0][1234589][- \.—]?([0-9][0-9][- \.—]?){4}#', $buffer, $out) !==0 && $testFixe===false) {
			$telFixe = $out[0];
			echo "fix : $telFixe<br>";
			$buffer = str_replace($telFixe, '', trim($buffer));
			$testFixe=true;
		}
			
		if(preg_match('#[0][6][- \.—]?([0-9][0-9][- \.—]?){4}#', $buffer, $out) !==0 && $testPort===false) {
			$telPort = $out[0];
			echo "portable : $telPort<br>";
			$buffer = str_replace($telPort, '', trim($buffer));
			$testPort=true;
		}
			
		if(preg_match('#(^(age :[ ]*))?(^([0-9]{2}))[ ][a][n][s][ ]*#i',$buffer, $out) !==0 && $testAge===false) {
			if(preg_match('#exp(é|e)rience#', $buffer, $out)){
				}else{
			echo "bufAge : $buffer<br>";
			print_r($out);
			$testAge=true;
			$res = preg_split('# #', $buffer, -1, PREG_SPLIT_NO_EMPTY);
			$age = date('Y') - $res[0];
			echo "<br>age : $age<br>";
				}
			
		}
		
		if(preg_match('#((Né)[e]?)?[ ](le|:|en)?[ ]([0-2][0-9]|[3][0-1]|[0-9])[-—/ ]([0][1-9]|[1][0-2])?([[:alpha:]]+)?[-—/ ]([0-9]{2,4})#ixm', $buffer, $out) !==0 && $testAge===false) {
			$buffer = strtolower($buffer);
			$buffer = accent($buffer);
			$testAge=true;
			$taille=sizeof($out);
			$pos = 3;
			for ($i=0; $i<=$taille-1; $i++) {
				if(preg_match('#[[:alpha:]]#i', $out[$i], $res)) {
					$out[$i] = strtolower($out[$i]);
					$out[$i] = accent($out[$i]);
                                        //un preg_replace a venir
					switch($out[$i]) {
						case('janvier') : { $out = str_replace($out[$i], '01', $out);$pos = 2;break; }
						case('fevrier') : { $out = str_replace($out[$i], '02', $out);/*echo "<br>apres : ";print_r($out);echo"<br>"*/;$pos = 2;break; }
						case('mars') : { $out = str_replace($out[$i], '03', $out);$pos = 2;break; }
						case('avril') : { $out = str_replace($out[$i], '04', $out);$pos = 2;break; }
						case('mai') : { $out = str_replace($out[$i], '05', $out);$pos = 2;break; }
						case('juin') : { $out = str_replace($out[$i], '06', $out);$pos = 2;break; }
						case('juillet') : { $out = str_replace($out[$i], '07', $out);$pos = 2;break; }
						case('aout') : { $out = str_replace($out[$i], '08', $out);$pos = 2;break; }
						case('septembre') : { $out = str_replace($out[$i], '09', $out);$pos = 2;break; }
						case('octobre') : { $out = str_replace($out[$i], '10', $out);$pos = 2;break; }
						case('novembre') : { $out = str_replace($out[$i], '11', $out);$pos = 2;break; }
						case('decembre') : { $out = str_replace($out[$i], '12', $out);$pos = 2;break; }
					}
				}
			}
						
			if(preg_match('#\b[0-9]\b#', $out[$taille-4], $res)) $out[$taille-4] = '0'.$out[$taille-4];
			
			if(preg_match('#([0-9]{4})#',$out[$taille-1], $res)) {
				$age = $out[$taille-1].'-'.$out[$taille-$pos].'-'.$out[$taille-4];
				$buffer = str_replace($out,'',trim($buffer));
					}else{
						if(preg_match('#([0-9]{2})#',$out[$taille-1], $res)) {
						$age = '19'.$out[$taille-1].'-'.$out[$taille-$pos].'-'.$out[$taille-4];
						$buffer = str_replace($out,'',trim($buffer));
						}
					}
				echo "age : $age<br>";
		}
		
		$buffer = accent($buffer);
		
		if(preg_match('#((experience[s]?|parcours)[ ](professionnel(le)?[s]?))#i', $buffer, $out) && $testExp==false ) {
			while(preg_match('#^([ ]*[F](ormation|ORMATION))|([L](angues|ANGUES))|^([D](ivers|IVERS))\b|([L](oisirs|OISIRS))|([D](iplome|IPLOME)[sS]?)#', $buffer, $out) == 0 ) {
				$buffer = fgets($handle, 4096);
				$buffer = preg_replace(array('#[|]#s','#^-#','#^—#','#^[(]#','#^[ ]#'), '', trim($buffer));
				$buffer = preg_replace('#([ ][ ]+)#s', ' ', trim($buffer));
				$buffer = accent($buffer);
					
				if(feof($handle)) break; // quand je met ca dans le while ca ne marche jamais -> ???
					
				if(preg_match('#(([0-2][0-9]|[3][0-1]|[0-9])?[-—/]?([0][1-9]|[1][0-2])?([[:alpha:]]+)?[-—/]?([1-2]([9]|[0])[0-9]{2})|(([0-2][0-9]|[3][0-1]|[0-9])?[-—/]?([0][1-9]|[1][0-2])[-—/]?([0-9]{2})))#ixm', $buffer, $out) !==0 && $c<=2) {
				
					$testExp=true;
					$c++;
					$xdeb = preg_replace('#[/aà— ]#', '-', trim($out[0]));
					$buffer = str_replace($out[0], '', trim($buffer));
															
					if(preg_match('#(([0-2][0-9]|[3][0-1]|[0-9])?[-—/ ]?([0][1-9]|[1][0-2])?([[:alpha:]]+)?[-—aà/ ]?([1-2]([9]|[0])[0-9]{2})|(([0-2][0-9]|[3][0-1]|[0-9])?[-—/]?([0][1-9]|[1][0-2])[-—/]?([0-9]{2})))#ixm', $buffer, $out) !==0 ) {
						
						$xfin = preg_replace('#(a|à)#', '', trim($out[0]));
						$xfin = preg_replace('#[/aà—]#', '-', trim($out[0]));
						$buffer = str_replace(trim($out[0]), '', trim($buffer));
						}else{
							if(preg_match('#((a|à)(\s)(ce)(\s)(jour))|(depuis)#i', $buffer, $out) || $c==1) {
								$xfin = Date('m').'-'.Date('Y');
								if(isset($out[0])) 
									$buffer = str_replace($out[0], '', trim($buffer));
									}else{
										$xfin=$exp[$c-1][1];
									}
						}
						$xfin=preg_replace('#^[-—.a:]#', '', $xfin);
						$exp[$c][1]=$xdeb;
						$exp[$c][2]=$xfin;
						
						while(preg_match('#([0-2][0-9]|[3][0-1]|[0-9])?[-—/ ]?([0][1-9]|[1][0-2])?([[:alpha:]]+)?[-—aà/ ]?([1-2]([9]|[0])[0-9]{2})#ixm', $buffer, $out) == 0 && $k<=2 && $l===false ) {
							if(preg_match('#[A-Z]([[:alpha:]\']+[ ]?)+#i', $buffer, $out) !==0) {
								$poste[$k] = $out[0];
								$buffer = str_replace($out[0], '', trim($buffer));
								$k++;
								
							}
							
							if(preg_match('#[A-Z]([[:alpha:]]*[.\']?[ ]?[:]?[ ]?[[:alpha:]]+[.\']?[ ]?)+#', $buffer, $out) !==0) {
								$societe[$k] = $out[0];
								$buffer = str_replace($out[0], '', trim($buffer));
								$l=true;
							}else $l=true;
						}
						$l=false;
						
				}
			}
			if(isset($poste)){
			echo"<br>postes : ";
			print_r($poste);
			}
			if(isset($societe)){
			echo"<br>societe : ";
			print_r($societe);
			}
			if(isset($exp)){
			echo"<br>dates : ";
			print_r($exp);
			echo"<br>";
			}
		}
			
			$k=0;
			$l=false;
			
		if(preg_match('#(^[ ]?[F](ormation|ORMATION))|(^[D](iplome)[s]?|IPLOME[S]?)#', $buffer, $out) || preg_match('#DIPLOME#', $buffer, $out) && $testForm==false) {
			while((preg_match('#(langue[s]?)|((expérience[s]?|parcours)[ ](professionnel(le)?[s]?))|\b(divers)\b|(loisirs)|(informatique)|((langue)[s]?)#i', $buffer, $out) == 0) ) {
				
				if(feof($handle)) break;
				
				$buffer = fgets($handle, 4096);
				$buffer = preg_replace(array('#[|]#s','#^-#','#^—#','#^[(]#','#^[ ]#'), '', trim($buffer));
				$buffer = preg_replace('#([ ][ ]+)#s', ' ', trim($buffer));
				$buffer = accent($buffer);
								
				if(preg_match('#(([0-2][0-9]|[3][0-1]|[0-9])[-—/])?([0][1-9]|[1][0-2])?([[:alpha:]]+)?[-—/a]?([1-2]([9]|[0])[0-9]{2})#ixm', $buffer, $out) !==0 && $d<=2) {
					$testForm=true;
					$d++;
					$fdeb = preg_replace('#[/aà— ]#', '-', trim($out[0]));
					$buffer = str_replace($out[0], '', trim($buffer));
										
					if(preg_match('#([0-2][0-9]|[3][0-1]|[0-9])?[-—/ ]?([0][1-9]|[1][0-2])?([[:alpha:]]+)?[-—aà/ ]?[1-2]([9]|[0])[0-9]{2}#ixm', $buffer, $out) !==0 ) {
						$ffin = preg_replace('#(a|à)#', '', trim($out[0]));
						$ffin = preg_replace('#[/aà—]#', '-', trim($out[0]));
						$buffer = preg_replace('#^[ ](a|à)[ ]#', '', $buffer);
						$buffer = str_replace(trim($out[0]), '', trim($buffer));
						}else{
							if(isset($out[0])) $buffer = str_replace($out[0], '', trim($buffer));
						}
							
					if(isset($ffin)) $ffin=preg_replace('#^[-—.a:]#', '', $ffin);
						$form[$d][1]=$fdeb;
						
						if(isset($ffin)) $form[$d][2]=$ffin;
						
						$fdeb="";
						unset($ffin);
							while(preg_match('#([0-2][0-9]|[3][0-1]|[0-9])?[-—/ ]?([0][1-9]|[1][0-2])?([[:alpha:]]+)?[-—aà/ ]?([1-2]([9]|[0])[0-9]{2})#ixm', $buffer, $out) == 0 && $k<=2 && $l===false ) {
								if(preg_match('#([A-Z][.]?[[:alpha:]\']+[.]?[ ]?[:]?[ ]?[[:alpha:]\']+[.]?[ ]?)+#i', $buffer, $out) !==0) {
									$dipl[$k] = $out[0];
									$buffer = str_replace($out[0], '', trim($buffer));
									$k++;
								}
								if(preg_match('#[A-Z]([[:alpha:]]*[.\']?[ ]?[:]?[ ]?[[:alpha:]]+[.\']?[ ]?)+#', $buffer, $out) !==0) {
									$etab[$k] = $out[0];
									$buffer = str_replace($out[0], '', trim($buffer));
									$l=true;
									}else break;
							}
						$l=false;
				}
		}
		if(isset($dipl)) {
		echo"<br>diplomes : ";
		print_r($dipl);
		}else echo "<br>no diplomes found<br>";
		
		if(isset($etab)) {
		echo"<br>Etablissements : ";
		print_r($etab);
		}else echo "<br>no etablissement found<br>";
				
		if(isset($form)) {
		echo"<br>dates : ";
		print_r($form);
		}else echo "<br>no dates diplomes found<br>";
		}
		
		if(preg_match('#anglais|allemand|italien|espagnol|portugais|arabe|chinois#i', $buffer, $out)) { 
			$f++;
			$langue[$f]=$buffer;
			}
		
		if(preg_match('#([D](ivers|IVERS))\b#', $buffer, $out)) {
			while(preg_match('#.+#', $buffer, $out)==true){
				if(feof($handle)) break;
				$divers.="$buffer<br>";
				$buffer = fgets($handle, 4096);
			}
		}
	}
	if(isset($langue)){
		echo"langues : ";
		print_r($langue);
		echo"<br>";
	}
	
	if(isset($divers)) echo "<br>$divers";
	
	fclose($handle);
}

?>
<html>
<head><title>Formulaire CV
</title></head>

<body>
<br>

<form method="POST" action="test.php" name="formulaire" enctype="multipart/form-data"> 
<br>

<div class="block">
<h2><span>Envoyez votre CV</span></h2>
	<div class="box">
		<table>
		<tr><td>
<input type="file" name="fichier"> <br>
<input type="submit" name="valider"> <br>
		<tr><td>
		</table>
	</div>
</div>
</form>
</body></html>