WebSVN – eFlore/Projets.eflore-projets – Diff – /trunk/scripts/modules/bdtfx/Bdtfx.php

 <?php
 //declare(encoding='UTF-8');
 /**
  * Exemple de lancement du script : :
  * /opt/lampp/bin/php cli.php bdtfx -a chargerTous
+ *
  * @category	php 5.2
  * @package		eFlore/Scripts
  * @author		Jennifer DHÉ <jennifer@tela-botanica.org>
  * @author		Jean-Pascal MILCENT <jpm@tela-botanica.org>
  * @copyright	Copyright (c) 2011, Tela Botanica (accueil@tela-botanica.org)
  * @license		http://www.cecill.info/licences/Licence_CeCILL_V2-fr.txt Licence CECILL
  * @license		http://www.gnu.org/licenses/gpl.html Licence GNU-GPL
  * @version		$Id$
  */
 class Bdtfx extends EfloreScript {
 	private $table = null;
 	private $pasInsertion = 1000;
 	private $departInsertion = 0;
 	protected $parametres_autorises = array(
 		'-t' => array(false, false, 'Permet de tester le script sur un jeux réduit de données (indiquer le nombre de lignes).'));
 	public function executer() {
 		try {
 			$this->initialiserProjet('bdtfx');
 			// Lancement de l'action demandée
 			$cmd = $this->getParametre('a');
 			switch ($cmd) {
 				case 'chargerTous' :
 					$this->chargerStructureSql();
 					$this->chargerBdtfx();
 					$this->genererNomSciHtml();
 					$this->genererDonneesTestMultiVersion();
 					break;
 				case 'chargerStructureSql' :
 					$this->chargerStructureSql();
 					break;
 				case 'chargerBdtfx' :
 					$this->chargerBdtfx();
 					break;
 				case 'genererNomSciHtml' :
 					$this->genererNomSciHtml();
 					break;
 				case 'genererChpFamille' :
 					$this->genererChpFamille();
 					break;
 				case 'genererDonneesTestMultiVersion' :
 					$this->genererDonneesTestMultiVersion();
 					break;
 				case 'supprimerDonneesTestMultiVersion' :
 					$this->supprimerDonneesTestMultiVersion();
 					break;
 				case 'supprimerTous' :
 					$this->supprimerTous();
 					break;
 				default :
 					throw new Exception("Erreur : la commande '$cmd' n'existe pas!");
+			}
 		} catch (Exception $e) {
 			$this->traiterErreur($e->getMessage());
+		}
+	}
 	private function chargerBdtfx() {
 		$chemin = Config::get('chemins.bdtfx');
 		$table = Config::get('tables.bdtfx');
 		$requete = "LOAD DATA INFILE '$chemin' ".
 				"REPLACE INTO TABLE $table ".
 				'CHARACTER SET utf8 '.
 				'FIELDS '.
 				"	TERMINATED BY '\t' ".
 				"	ENCLOSED BY '' ".
 				"	ESCAPED BY '\\\' ".
 				'IGNORE 1 LINES';
 		$this->getBdd()->requeter($requete);
+	}
 	private function genererNomSciHtml() {
 		$this->initialiserGenerationChamps();
 		$this->preparerTablePrChpNomSciHtml();
 		$generateur = new GenerateurNomSciHtml();
 		$nbreTotal = $this->recupererNbTotalTuples();
 		$this->departInsertion = 0;
 		while ($this->departInsertion < $nbreTotal) {
 			$resultat = $this->recupererTuples();
 			$nomsSciEnHtml = $generateur->generer($resultat);
 			$this->lancerRequeteModification($nomsSciEnHtml);
 			$this->departInsertion += $this->pasInsertion;
 			$this->afficherAvancement("Insertion des noms scientifique au format HTML dans la base par paquet de {$this->pasInsertion} en cours");
 			if ($this->stopperLaBoucle($this->getParametre('t'))) break;
+		}
 		echo "\n";
+	}
 	private function initialiserGenerationChamps() {
 		$this->table = Config::get('tables.bdtfx');
+	}
 	private function preparerTablePrChpNomSciHtml() {
 		$requete = "SHOW COLUMNS FROM {$this->table} LIKE 'nom_sci_html' ";
 		$resultat = $this->getBdd()->recuperer($requete);
 		if ($resultat === false) {
 			$requete = 	"ALTER TABLE {$this->table} ".
 				'ADD nom_sci_html VARCHAR( 500 ) '.
 				'CHARACTER SET utf8 COLLATE utf8_general_ci NULL DEFAULT NULL ';
 			$this->getBdd()->requeter($requete);
+		}
+	}
 	private function recupererNbTotalTuples(){
 		$requete = "SELECT count(*) AS nb FROM {$this->table} ";
 		$resultat = $this->getBdd()->recuperer($requete);
 		return $resultat['nb'];
+	}
 	private function recupererTuples() {
 		$requete = 'SELECT 	num_nom, rang, nom_supra_generique, genre, epithete_infra_generique, '.
 			'	epithete_sp, type_epithete, epithete_infra_sp,cultivar_groupe, '.
 			'	nom_commercial, cultivar '.
 			"FROM {$this->table} ".
 			"LIMIT {$this->departInsertion},{$this->pasInsertion} ";
 		$resultat = $this->getBdd()->recupererTous($requete);
 		return $resultat;
+	}
 	private function lancerRequeteModification($nomsSciHtm) {
 		foreach ($nomsSciHtm as $id => $html) {
 			$html = $this->getBdd()->proteger($html);
 			$requete = "UPDATE {$this->table} ".
 				"SET nom_sci_html = $html ".
 				"WHERE num_nom = $id ";
 			$resultat = $this->getBdd()->requeter($requete);
 			if ($resultat === false) {
 				throw new Exception("Erreur d'insertion pour le tuple $id");
+			}
+		}
+	}
 	private function genererChpFamille() {
 		$this->initialiserGenerationChamps();
 		$this->preparerTablePrChpFamille();
 		$resultats = $this->recupererTuplesPrChpFamille();
 		$famille = array();
 		$noms = array();
 		$introuvables = array();
+		$introuvablesSyno = array();
-		foreach ($resultats as $nom) {
+		foreach ($resultats as $id => $nom) {
 			$nn = $nom['num_nom'];
+			$nnr = $nom['num_nom_retenu'];
 			$nts = $nom['num_tax_sup'];
 			$rg = $nom['rang'];
-			if ($rg == '180') {
+			if ($nnr != '') {
-				$famille[$nn] = $nom['nom_sci'];
-			} else {
-				if (isset($noms[$nts])) {
+				if ($rg == '180') {
-					$noms[$nn] = $noms[$nts];
+					$famille[$nn] = $nom['nom_sci'];
-				} else if (isset($famille[$nts])) {
-					$noms[$nn] = $famille[$nts];
 				} else {
+					if ($nn == $nnr) {// nom retenu
+						if (isset($noms[$nts])) {
+							$noms[$nn] = $noms[$nts];
+						} else if (isset($famille[$nts])) {
+							$noms[$nn] = $famille[$nts];
+						} else {
-					$introuvables[] = $nn;
+							$introuvables[] = $nn;
+						}
+					} else {// nom synonyme
+						if (isset($noms[$nnr])) {
+							$noms[$nn] = $noms[$nnr];
+						} else {
+							$introuvablesSyno[] = $nom;
+						}
+					}
+				}
+			}
+			unset($resultats[$id]);
 			$this->afficherAvancement("Attribution de leur famille aux noms en cours");
 			if ($this->stopperLaBoucle($this->getParametre('t'))) break;
+		}
 		echo "\n";
+		foreach ($introuvablesSyno as $id => $nom) {
+			$nn = $nom['num_nom'];
+			$nnr = $nom['num_nom_retenu'];
+			if (isset($noms[$nnr])) {
+				$noms[$nn] = $noms[$nnr];
+			} else {
+				$introuvables[] = $nn;
+			}
+			unset($introuvablesSyno[$id]);
+			$this->afficherAvancement("Attribution de leur famille aux synonymes en cours");
+		}
+		echo "\n";
-		if ($introuvables != 0) {
+		if (count($introuvables) != 0) {
+			$introuvablesNbre = count($introuvables);
+			echo "Famille introuvable pour $introuvablesNbre noms ! Voir le log.\n";
+			$logContenu = implode(", \n", $introuvables);
+			$logFichier = realpath(dirname(__FILE__)).'/log/famille_introuvable.log';
+			echo $logFichier."\n";
+			file_put_contents($logFichier, $logContenu);
+		}
+		if (count($introuvablesSyno) != 0) {
-			$introuvables = implode(', ', $introuvables);
+			//$introuvablesSyno = implode(', ', $introuvablesSyno);
+			$introuvablesSyno = count($introuvablesSyno);
-			echo "Parent introuvable pour les noms : $introuvables\n";
+			echo "Synonyme avec nom retenu introuvable : $introuvablesSyno\n";
+		}
 		$this->lancerRequeteModificationPrChpFamille($noms);
+	}
 	private function preparerTablePrChpFamille() {
 		$requete = "SHOW COLUMNS FROM {$this->table} LIKE 'famille' ";
 		$resultat = $this->getBdd()->recuperer($requete);
 		if ($resultat === false) {
 			$requete = 	"ALTER TABLE {$this->table} ".
 				'ADD famille VARCHAR(255) '.
 				'CHARACTER SET utf8 COLLATE utf8_general_ci NULL DEFAULT NULL ';
 			$this->getBdd()->requeter($requete);
+		}
+	}
 	private function recupererTuplesPrChpFamille() {
-		$requete = 'SELECT num_nom, num_tax_sup, rang, nom_sci '.
+		$requete = 'SELECT num_nom, num_nom_retenu, num_tax_sup, rang, nom_sci '.
 				"FROM {$this->table} ".
 				"WHERE rang >= 180 ".
-				"	AND num_tax_sup != '' ".
+				"ORDER BY rang ASC, num_tax_sup ASC , num_nom_retenu DESC ";
-				"ORDER BY rang ASC,num_tax_sup ASC ";
+		//echo $requete."\n";
 		$resultat = $this->getBdd()->recupererTous($requete);
 		return $resultat;
+	}
 	private function lancerRequeteModificationPrChpFamille($noms) {
 		foreach ($noms as $id => $famille) {
 			$famille = $this->getBdd()->proteger($famille);
 			$requete = "UPDATE {$this->table} ".
 					"SET famille = $famille ".
 					"WHERE num_nom = $id ";
 			$resultat = $this->getBdd()->requeter($requete);
 			if ($resultat === false) {
 				throw new Exception("Erreur d'insertion pour le tuple $id");
+			}
 			$this->afficherAvancement("Insertion des noms de famille dans la base en cours");
+		}
 		echo "\n";
+	}
 	private function genererDonneesTestMultiVersion() {
 		$contenuSql = $this->recupererContenu(Config::get('chemins.structureSqlTest'));
 		$this->executerScripSql($contenuSql);
 		$table = Config::get('tables.bdtfx');
 		$tableTest = Config::get('tables.bdtfxTest');
 		$requete = "INSERT INTO $tableTest SELECT * FROM $table";
 		$this->getBdd()->requeter($requete);
+	}
 	private function supprimerDonneesTestMultiVersion() {
 		$tableMeta = Config::get('tables.bdtfxMeta');
 		$requete = "DELETE FROM $tableMeta WHERE guid = 'urn:lsid:tela-botanica.org:bdtfx:1.02'";
 		$this->getBdd()->requeter($requete);
 		$tableTest = Config::get('tables.bdtfxTest');
 		$requete = "DROP TABLE $tableTest";
 		$this->getBdd()->requeter($requete);
+	}
 	private function supprimerTous() {
 		$requete = "DROP TABLE bdtfx_meta, bdtfx_v1_01, bdtfx_v1_02";
 		$this->getBdd()->requeter($requete);
+	}
+}
 ?>

Subversion Repositories eFlore/Projets.eflore-projets

(root)/trunk/scripts/modules/bdtfx/Bdtfx.php – Rev 336 → 339