Subversion Repositories eFlore/Projets.eflore-projets

Rev

Rev 1242 | Details | Compare with Previous | Last modification | View Log | RSS feed

Rev Author Line No. Line
11 jpm 1 
<?php
121 jpm 2 
//declare(encoding='UTF-8');
11 jpm 3 
/**
4 
 * Exemple de lancement du script : :
121 jpm 5 
 * /opt/lampp/bin/php cli.php bdtfx -a chargerTous
6 
 *
11 jpm 7 
 * @category	php 5.2
121 jpm 8 
 * @package		eFlore/Scripts
11 jpm 9 
 * @author		Jennifer DHÉ <jennifer@tela-botanica.org>
10 
 * @author		Jean-Pascal MILCENT <jpm@tela-botanica.org>
11 
 * @copyright	Copyright (c) 2011, Tela Botanica (accueil@tela-botanica.org)
12 
 * @license		http://www.cecill.info/licences/Licence_CeCILL_V2-fr.txt Licence CECILL
13 
 * @license		http://www.gnu.org/licenses/gpl.html Licence GNU-GPL
14 
 * @version		$Id$
15 
 */
68 jpm 16 
class Bdtfx extends EfloreScript {
46 jpm 17 
 
68 jpm 18 
	private $table = null;
19 
	private $pasInsertion = 1000;
20 
	private $departInsertion = 0;
11 jpm 21 
 
22 
	protected $parametres_autorises = array(
369 mathilde 23 
		'-t' => array(false, false, 'Permet de tester le script sur un jeu réduit de données (indiquer le nombre de lignes).'));
11 jpm 24 
 
25 
	public function executer() {
46 jpm 26 
		try {
68 jpm 27 
			$this->initialiserProjet('bdtfx');
46 jpm 28 
 
29 
			// Lancement de l'action demandée
30 
			$cmd = $this->getParametre('a');
31 
			switch ($cmd) {
32 
				case 'chargerTous' :
33 
					$this->chargerStructureSql();
34 
					$this->chargerBdtfx();
340 jpm 35 
					$this->genererChpNomSciHtml();
36 
					$this->genererChpFamille();
1279 delphine 37 
					//$this->genererDonneesTestMultiVersion();
699 aurelien 38 
					$this->genererChpHierarchie();
1242 delphine 39 
					$this->genererBDTFXReduit();
46 jpm 40 
					break;
41 
				case 'chargerStructureSql' :
42 
					$this->chargerStructureSql();
43 
					break;
44 
				case 'chargerBdtfx' :
45 
					$this->chargerBdtfx();
46 
					break;
47 
				case 'genererNomSciHtml' :
340 jpm 48 
					$this->genererChpNomSciHtml();
46 jpm 49 
					break;
335 jpm 50 
				case 'genererChpFamille' :
51 
					$this->genererChpFamille();
52 
					break;
699 aurelien 53 
				case 'genererChpHierarchie' :
54 
					$this->genererChpHierarchie();
55 
					break;
103 jpm 56 
				case 'genererDonneesTestMultiVersion' :
57 
					$this->genererDonneesTestMultiVersion();
58 
					break;
1242 delphine 59 
				case 'genererBDTFXR' :
60 
					$this->genererBDTFXReduit();
61 
					break;
103 jpm 62 
				case 'supprimerDonneesTestMultiVersion' :
63 
					$this->supprimerDonneesTestMultiVersion();
64 
					break;
130 jpm 65 
				case 'supprimerTous' :
66 
					$this->supprimerTous();
67 
					break;
46 jpm 68 
				default :
69 
					throw new Exception("Erreur : la commande '$cmd' n'existe pas!");
70 
			}
71 
		} catch (Exception $e) {
72 
			$this->traiterErreur($e->getMessage());
11 jpm 73 
		}
46 jpm 74 
	}
11 jpm 75 
 
46 jpm 76 
	private function chargerBdtfx() {
77 
		$chemin = Config::get('chemins.bdtfx');
78 
		$table = Config::get('tables.bdtfx');
1279 delphine 79 
		$requete = "LOAD DATA LOCAL INFILE '$chemin' ".
46 jpm 80 
				"REPLACE INTO TABLE $table ".
81 
				'CHARACTER SET utf8 '.
82 
				'FIELDS '.
83 
				"	TERMINATED BY '\t' ".
84 
				"	ENCLOSED BY '' ".
85 
				"	ESCAPED BY '\\\' ".
86 
				'IGNORE 1 LINES';
68 jpm 87 
		$this->getBdd()->requeter($requete);
46 jpm 88 
	}
89 
 
340 jpm 90 
	private function genererChpNomSciHtml() {
335 jpm 91 
		$this->initialiserGenerationChamps();
92 
		$this->preparerTablePrChpNomSciHtml();
46 jpm 93 
		$generateur = new GenerateurNomSciHtml();
94 
		$nbreTotal = $this->recupererNbTotalTuples();
661 jpm 95 
		$erreurs = array();
335 jpm 96 
		$this->departInsertion = 0;
46 jpm 97 
		while ($this->departInsertion < $nbreTotal) {
340 jpm 98 
			$resultat = $this->recupererTuplesPrChpNomSciHtml();
661 jpm 99 
 
100 
			try {
101 
				$nomsSciEnHtml = $generateur->generer($resultat);
102 
			} catch (Exception $e) {
103 
				$erreurs[] = $e->getMessage();
104 
			}
105 
 
340 jpm 106 
			$this->remplirChpNomSciHtm($nomsSciEnHtml);
46 jpm 107 
			$this->departInsertion += $this->pasInsertion;
108 
			$this->afficherAvancement("Insertion des noms scientifique au format HTML dans la base par paquet de {$this->pasInsertion} en cours");
68 jpm 109 
			if ($this->stopperLaBoucle($this->getParametre('t'))) break;
11 jpm 110 
		}
111 
		echo "\n";
661 jpm 112 
 
113 
		$this->creerFichierLog('Erreurs lors de la génération HTML des noms scientifiques', $erreurs, 'erreurs_noms_sci_html');
11 jpm 114 
	}
115 
 
335 jpm 116 
	private function initialiserGenerationChamps() {
46 jpm 117 
		$this->table = Config::get('tables.bdtfx');
11 jpm 118 
	}
119 
 
335 jpm 120 
	private function preparerTablePrChpNomSciHtml() {
11 jpm 121 
		$requete = "SHOW COLUMNS FROM {$this->table} LIKE 'nom_sci_html' ";
68 jpm 122 
		$resultat = $this->getBdd()->recuperer($requete);
11 jpm 123 
		if ($resultat === false) {
124 
			$requete = 	"ALTER TABLE {$this->table} ".
335 jpm 125 
				'ADD nom_sci_html VARCHAR( 500 ) '.
126 
				'CHARACTER SET utf8 COLLATE utf8_general_ci NULL DEFAULT NULL ';
68 jpm 127 
			$this->getBdd()->requeter($requete);
11 jpm 128 
		}
129 
	}
130 
 
131 
	private function recupererNbTotalTuples(){
46 jpm 132 
		$requete = "SELECT count(*) AS nb FROM {$this->table} ";
68 jpm 133 
		$resultat = $this->getBdd()->recuperer($requete);
46 jpm 134 
		return $resultat['nb'];
11 jpm 135 
	}
136 
 
340 jpm 137 
	private function recupererTuplesPrChpNomSciHtml() {
869 delphine 138 
		$requete = 'SELECT 	num_nom, rang, nom_sci, nom_supra_generique, genre, epithete_infra_generique, '.
335 jpm 139 
			'	epithete_sp, type_epithete, epithete_infra_sp,cultivar_groupe, '.
140 
			'	nom_commercial, cultivar '.
141 
			"FROM {$this->table} ".
142 
			"LIMIT {$this->departInsertion},{$this->pasInsertion} ";
68 jpm 143 
		$resultat = $this->getBdd()->recupererTous($requete);
11 jpm 144 
		return $resultat;
145 
	}
146 
 
340 jpm 147 
	private function remplirChpNomSciHtm($nomsSciHtm) {
46 jpm 148 
		foreach ($nomsSciHtm as $id => $html) {
68 jpm 149 
			$html = $this->getBdd()->proteger($html);
340 jpm 150 
			$requete = "UPDATE {$this->table} SET nom_sci_html = $html WHERE num_nom = $id ";
68 jpm 151 
			$resultat = $this->getBdd()->requeter($requete);
46 jpm 152 
			if ($resultat === false) {
153 
				throw new Exception("Erreur d'insertion pour le tuple $id");
11 jpm 154 
			}
155 
		}
156 
	}
339 jpm 157 
 
757 raphael 158 
	private function traiterResultatsFamille(&$resultats, &$noms, &$introuvables, &$introuvablesSyno) {
339 jpm 159 
		foreach ($resultats as $id => $nom) {
336 jpm 160 
			$nn = $nom['num_nom'];
339 jpm 161 
			$nnr = $nom['num_nom_retenu'];
336 jpm 162 
			$nts = $nom['num_tax_sup'];
163 
			$rg = $nom['rang'];
339 jpm 164 
			if ($nnr != '') {
165 
				if ($rg == '180') {
340 jpm 166 
					$noms[$nn] = $nom['nom_sci'];
336 jpm 167 
				} else {
339 jpm 168 
					if ($nn == $nnr) {// nom retenu
169 
						if (isset($noms[$nts])) {
757 raphael 170 
							// signifie que recupererTuplesPrChpFamille() devrait
171 
							// récupérer ce record *avant*
339 jpm 172 
							$noms[$nn] = $noms[$nts];
173 
						} else {
174 
							$introuvables[] = $nn;
175 
						}
176 
					} else {// nom synonyme
757 raphael 177 
						if (isset($noms[$nnr])) {
178 
							// signifie que recupererTuplesPrChpFamille() devrait
179 
							// récupérer ce record *avant*
180 
							$noms[$nn] = $noms[$nnr];
181 
						} else {
182 
							$introuvablesSyno[] = $nom;
339 jpm 183 
						}
184 
					}
336 jpm 185 
				}
186 
			}
339 jpm 187 
			unset($resultats[$id]);
336 jpm 188 
			$this->afficherAvancement("Attribution de leur famille aux noms en cours");
335 jpm 189 
			if ($this->stopperLaBoucle($this->getParametre('t'))) break;
190 
		}
757 raphael 191 
	}
339 jpm 192 
 
757 raphael 193 
	private function genererChpFamille() {
194 
		$this->initialiserGenerationChamps();
195 
		$this->preparerTablePrChpFamille();
196 
		$resultats = $this->recupererTuplesPrChpFamille();
197 
		$noms = array();
198 
		$introuvables = array();
199 
		$introuvablesSyno = array();
200 
		$i = 1;
201 
 
202 
		while(true) {
203 
			printf("passe n°%d:\n", $i);
204 
			$this->traiterResultatsFamille($resultats, $noms, $introuvables, $introuvablesSyno);
205 
			echo "\n\n";
206 
			// printf("noms: %d, introuvables: %d, introuvablesSyno: %d\n", count($noms), count($introuvables), count($introuvablesSyno));
207 
			// XXX, au 22/07/2013, 3 passes sont suffisantes
208 
			// TODO: MySQL procédure stockée !
209 
			if($i++ == 3) break;
210 
			$resultats = array_merge($resultats, $introuvables, $introuvablesSyno);
211 
			$introuvables = $introuvablesSyno = array();
212 
		}
213 
 
339 jpm 214 
		foreach ($introuvablesSyno as $id => $nom) {
215 
			$nn = $nom['num_nom'];
216 
			$nnr = $nom['num_nom_retenu'];
217 
			if (isset($noms[$nnr])) {
218 
				$noms[$nn] = $noms[$nnr];
219 
			} else {
220 
				$introuvables[] = $nn;
221 
			}
222 
			unset($introuvablesSyno[$id]);
223 
			$this->afficherAvancement("Attribution de leur famille aux synonymes en cours");
336 jpm 224 
		}
339 jpm 225 
		echo "\n";
226 
 
661 jpm 227 
		$msg = 'Plusieurs familles sont introuvables';
228 
		$this->creerFichierLog($msg, $introuvables, 'famille_introuvable');
339 jpm 229 
 
340 jpm 230 
		$this->remplirChpFamille($noms);
335 jpm 231 
	}
699 aurelien 232 
 
335 jpm 233 
	private function preparerTablePrChpFamille() {
234 
		$requete = "SHOW COLUMNS FROM {$this->table} LIKE 'famille' ";
235 
		$resultat = $this->getBdd()->recuperer($requete);
236 
		if ($resultat === false) {
237 
			$requete = 	"ALTER TABLE {$this->table} ".
238 
				'ADD famille VARCHAR(255) '.
239 
				'CHARACTER SET utf8 COLLATE utf8_general_ci NULL DEFAULT NULL ';
240 
			$this->getBdd()->requeter($requete);
241 
		}
242 
	}
339 jpm 243 
 
335 jpm 244 
	private function recupererTuplesPrChpFamille() {
339 jpm 245 
		$requete = 'SELECT num_nom, num_nom_retenu, num_tax_sup, rang, nom_sci '.
335 jpm 246 
				"FROM {$this->table} ".
247 
				"WHERE rang >= 180 ".
340 jpm 248 
				"ORDER BY rang ASC, num_tax_sup ASC, num_nom_retenu DESC ";
335 jpm 249 
		$resultat = $this->getBdd()->recupererTous($requete);
250 
		return $resultat;
251 
	}
339 jpm 252 
 
340 jpm 253 
	private function remplirChpFamille($noms) {
336 jpm 254 
		foreach ($noms as $id => $famille) {
255 
			$famille = $this->getBdd()->proteger($famille);
340 jpm 256 
			$requete = "UPDATE {$this->table} SET famille = $famille WHERE num_nom = $id ";
336 jpm 257 
			$resultat = $this->getBdd()->requeter($requete);
258 
			if ($resultat === false) {
259 
				throw new Exception("Erreur d'insertion pour le tuple $id");
260 
			}
261 
			$this->afficherAvancement("Insertion des noms de famille dans la base en cours");
262 
		}
263 
		echo "\n";
264 
	}
699 aurelien 265 
 
266 
	private function genererChpHierarchie() {
267 
		$this->initialiserGenerationChamps();
268 
		$this->preparerTablePrChpHierarchie();
269 
		$table = Config::get('tables.bdtfx');
270 
 
271 
		$requete = "UPDATE $table SET hierarchie = NULL ";
272 
		$mise_a_jour = $this->getBdd()->requeter($requete);
273 
 
274 
		$requete_hierarchie = "SELECT num_nom, num_nom_retenu, num_tax_sup FROM ".$table." ORDER BY rang DESC";
275 
 
276 
		$resultat = $this->getBdd()->recupererTous($requete_hierarchie);
277 
		$num_nom_a_num_sup = array();
278 
		foreach($resultat as &$taxon) {
279 
			$num_nom_a_num_sup[$taxon['num_nom']] = $taxon['num_tax_sup'];
280 
		}
281 
		$chemin_taxo = "";
282 
		foreach($resultat as &$taxon) {
283 
			$chemin_taxo = $this->traiterHierarchieNumTaxSup($taxon['num_nom_retenu'], $num_nom_a_num_sup).'-';
284 
			$requete = "UPDATE $table SET hierarchie = ".$this->getBdd()->proteger($chemin_taxo)." WHERE num_nom = ".$taxon['num_nom']." ";
285 
			$mise_a_jour = $this->getBdd()->requeter($requete);
286 
			$this->afficherAvancement("Insertion de la hierarchie taxonomique en cours");
287 
		}
288 
		echo "\n";
289 
	}
290 
 
291 
	private function traiterHierarchieNumTaxSup($num_nom_retenu, &$num_nom_a_num_sup) {
292 
		$chaine_hierarchie = "";
293 
		if(isset($num_nom_a_num_sup[$num_nom_retenu])) {
294 
			$num_tax_sup = $num_nom_a_num_sup[$num_nom_retenu];
295 
			$chaine_hierarchie = '-'.$num_tax_sup;
296 
			if($num_tax_sup != 0 && $num_tax_sup != '') {
297 
				$chaine_hierarchie = $this->traiterHierarchieNumTaxSup($num_tax_sup, $num_nom_a_num_sup).$chaine_hierarchie;
298 
			}
299 
		}
300 
		return $chaine_hierarchie;
301 
	}
302 
 
303 
	private function preparerTablePrChpHierarchie() {
304 
		$requete = "SHOW COLUMNS FROM {$this->table} LIKE 'hierarchie' ";
305 
		$resultat = $this->getBdd()->recuperer($requete);
306 
		if ($resultat === false) {
307 
			$requete = 	"ALTER TABLE {$this->table} ".
308 
					'ADD hierarchie VARCHAR(1000) '.
309 
					'CHARACTER SET utf8 COLLATE utf8_general_ci NULL DEFAULT NULL ';
310 
			$this->getBdd()->requeter($requete);
311 
		}
312 
	}
11 jpm 313 
 
103 jpm 314 
	private function genererDonneesTestMultiVersion() {
315 
		$contenuSql = $this->recupererContenu(Config::get('chemins.structureSqlTest'));
316 
		$this->executerScripSql($contenuSql);
68 jpm 317 
 
103 jpm 318 
		$table = Config::get('tables.bdtfx');
319 
		$tableTest = Config::get('tables.bdtfxTest');
320 
		$requete = "INSERT INTO $tableTest SELECT * FROM $table";
321 
		$this->getBdd()->requeter($requete);
322 
	}
323 
 
324 
	private function supprimerDonneesTestMultiVersion() {
325 
		$tableMeta = Config::get('tables.bdtfxMeta');
326 
		$requete = "DELETE FROM $tableMeta WHERE guid = 'urn:lsid:tela-botanica.org:bdtfx:1.02'";
327 
		$this->getBdd()->requeter($requete);
328 
 
329 
		$tableTest = Config::get('tables.bdtfxTest');
342 jpm 330 
		$requete = "DROP TABLE IF EXISTS $tableTest";
103 jpm 331 
		$this->getBdd()->requeter($requete);
332 
	}
1242 delphine 333 
 
334 
	private function genererBDTFXReduit() {
335 
		$table = Config::get('tables.bdtfx');
336 
		$version = Config::get('version');
337 
 
338 
		$requete = "create table bdtfxr_v".$version." as (SELECT *
339 
						FROM ".$table."
340 
						WHERE
341 
							(`presence` = 'P' AND num_nom = num_nom_retenu)
342 
						OR
343 
							(`flore_bonnier_num` != '' OR `flore_cnrs_num` != '' OR `flore_coste_num` != ''
344 
								OR `flore_fournier_num` != '' OR `flore_fg_num` != ''))";
345 
		$table = $this->getBdd()->requeter($requete);
346 
 
347 
		$requete_meta = "INSERT INTO `bdtfxr_meta`(`guid`, `langue_meta`, `code`, `version`, `titre`, `description`, `mots_cles`, `citation`, `url_tech`, `url_projet`, `source`, `createurs`, `editeur`, `contributeurs`, `droits`, `url_droits`, `langue`, `date_creation`, `date_validite`, `couverture_spatiale`, `couverture_temporelle`, `web_services`) ".
348 
				" SELECT * FROM `bdtfx_meta` WHERE `version` = '".str_replace('_', '.', $version)."'";
349 
		$meta = $this->getBdd()->requeter($requete_meta);
350 
	}
130 jpm 351 
 
352 
	private function supprimerTous() {
342 jpm 353 
		$requete = "DROP TABLE IF EXISTS bdtfx_meta, bdtfx_v1_01, bdtfx_v1_02";
130 jpm 354 
		$this->getBdd()->requeter($requete);
355 
	}
661 jpm 356 
 
357 
	private function creerFichierLog($message, $lignes, $nomFichier) {
358 
		$lignesNbre = count($lignes);
359 
		if ($lignesNbre != 0) {
360 
			echo "$message. Voir le log de $lignesNbre lignes :\n";
361 
 
362 
			$logContenu = implode(", \n", $lignes);
363 
			$logFichier = realpath(dirname(__FILE__))."/log/$nomFichier.log";
364 
			echo $logFichier."\n";
365 
					file_put_contents($logFichier, $logContenu);
366 
		}
367 
	}
11 jpm 368 
}
369 
?>