| 11 |
jpm |
1 |
<?php
|
|
|
2 |
/** Exemple lancement:
|
| 22 |
jpm |
3 |
* /opt/lampp/bin/php -d memory_limit=3500M ~/web/eflore-projets/scripts/cli.php nvjfl
|
| 28 |
jpm |
4 |
* -a chargerTous
|
| 11 |
jpm |
5 |
* Options :
|
| 28 |
jpm |
6 |
* -t : Permet de tester le script sur un jeux réduit de données (indiquer le nombre de lignes).
|
| 11 |
jpm |
7 |
*/
|
| 79 |
jpm |
8 |
class Nvjfl extends EfloreScript {
|
| 11 |
jpm |
9 |
|
| 22 |
jpm |
10 |
private $nomsIndex = array();
|
|
|
11 |
private $numeroIndex = 1;
|
|
|
12 |
|
| 11 |
jpm |
13 |
protected $parametres_autorises = array(
|
| 105 |
jpm |
14 |
'-t' => array(false, false, 'Permet de tester le script sur un jeux réduit de données (indiquer le nombre de lignes).'));
|
| 11 |
jpm |
15 |
|
|
|
16 |
public function executer() {
|
| 838 |
raphael |
17 |
// désactive le problématique gestionnaire d'erreur du framework
|
|
|
18 |
restore_error_handler();
|
|
|
19 |
restore_exception_handler();
|
|
|
20 |
ini_set('mysql.default_socket', '/home/raphael/web/mysql/mysql.sock');
|
| 11 |
jpm |
21 |
// Lancement de l'action demandée
|
| 22 |
jpm |
22 |
try {
|
| 79 |
jpm |
23 |
$this->initialiserProjet('nvjfl');
|
| 24 |
jpm |
24 |
|
| 22 |
jpm |
25 |
$cmd = $this->getParametre('a');
|
|
|
26 |
switch ($cmd) {
|
| 24 |
jpm |
27 |
case 'chargerTous' :
|
| 44 |
jpm |
28 |
$this->chargerStructureSql();
|
| 24 |
jpm |
29 |
$this->chargerNvjfl();
|
|
|
30 |
$this->chargerBiblio();
|
|
|
31 |
$this->chargerBiblioLien();
|
| 27 |
jpm |
32 |
$this->chargerOntologies();
|
| 24 |
jpm |
33 |
break;
|
| 53 |
jpm |
34 |
case 'chargerStructure' :
|
|
|
35 |
$this->chargerStructureSql();
|
|
|
36 |
break;
|
| 24 |
jpm |
37 |
case 'chargerNvjfl' :
|
|
|
38 |
$this->chargerNvjfl();
|
| 22 |
jpm |
39 |
break;
|
| 24 |
jpm |
40 |
case 'chargerBiblio' :
|
|
|
41 |
$this->chargerBiblio();
|
|
|
42 |
break;
|
|
|
43 |
case 'chargerBiblioLien' :
|
|
|
44 |
$this->chargerBiblioLien();
|
|
|
45 |
break;
|
| 27 |
jpm |
46 |
case 'chargerOntologies' :
|
|
|
47 |
$this->chargerOntologies();
|
|
|
48 |
break;
|
| 130 |
jpm |
49 |
case 'supprimerTous' :
|
|
|
50 |
$this->supprimerTous();
|
|
|
51 |
break;
|
| 22 |
jpm |
52 |
default :
|
| 79 |
jpm |
53 |
throw new Exception("Erreur : la commande '$cmd' n'existe pas!");
|
| 22 |
jpm |
54 |
}
|
|
|
55 |
} catch (Exception $e) {
|
|
|
56 |
$this->traiterErreur($e->getMessage());
|
| 11 |
jpm |
57 |
}
|
|
|
58 |
}
|
| 24 |
jpm |
59 |
|
| 22 |
jpm |
60 |
/**
|
| 24 |
jpm |
61 |
* Charge le fichier en créant un id pour chaque nom vernaculaire.
|
| 22 |
jpm |
62 |
*/
|
| 24 |
jpm |
63 |
private function chargerNvjfl() {
|
|
|
64 |
//Debug::printr(Config::get('fichiers'));
|
|
|
65 |
$fichierOuvert = $this->ouvrirFichier(Config::get('chemins.nvjfl'));
|
| 11 |
jpm |
66 |
|
| 838 |
raphael |
67 |
while(($donnees = $this->analyserFichier($fichierOuvert))) {
|
|
|
68 |
foreach ($donnees as $donnee) {
|
|
|
69 |
$requete = 'INSERT INTO '.Config::get('tables.nvjfl').' VALUES ('.implode(', ', $donnee).')';
|
|
|
70 |
var_dump($donnee);die;
|
|
|
71 |
$this->getBdd()->requeter($requete);
|
|
|
72 |
|
|
|
73 |
$this->afficherAvancement("Insertion des noms vernaculaires dans la base de données");
|
|
|
74 |
if ($this->stopperLaBoucle($this->getParametre('t'))) {
|
|
|
75 |
break;
|
|
|
76 |
}
|
| 22 |
jpm |
77 |
}
|
| 838 |
raphael |
78 |
echo "\n";
|
| 22 |
jpm |
79 |
}
|
| 838 |
raphael |
80 |
fclose($fichierOuvert);
|
| 22 |
jpm |
81 |
}
|
|
|
82 |
|
| 838 |
raphael |
83 |
private function analyserFichier($fichierOuvert, $limite = 30000) {
|
|
|
84 |
static $passe = 1;
|
|
|
85 |
if($passe == 1) $entetesCsv = fgets($fichierOuvert);
|
| 22 |
jpm |
86 |
$donnees = array();
|
| 838 |
raphael |
87 |
while (count($donnees) < $limite && ($ligneCsv = fgets($fichierOuvert))) {
|
| 22 |
jpm |
88 |
$champs = explode("\t", trim($ligneCsv));
|
| 24 |
jpm |
89 |
if (count($champs) > 0) {
|
| 105 |
jpm |
90 |
if (isset($champs[2])) {
|
|
|
91 |
$nomVernaculaire = $champs[2];
|
|
|
92 |
$indexCourrant = $this->getIndexNomVernaculaire($nomVernaculaire);
|
|
|
93 |
$champs = array_merge(array($indexCourrant), $champs);
|
|
|
94 |
$donnees[] = $this->protegerValeursDesChamps($champs);
|
|
|
95 |
}
|
| 24 |
jpm |
96 |
}
|
| 838 |
raphael |
97 |
$this->afficherAvancement("Analyse du fichier des noms vernaculaires (passe $passe)");
|
| 22 |
jpm |
98 |
if ($this->stopperLaBoucle()) {
|
|
|
99 |
break;
|
|
|
100 |
}
|
|
|
101 |
}
|
| 838 |
raphael |
102 |
$passe++;
|
| 22 |
jpm |
103 |
echo "\n";
|
|
|
104 |
return $donnees;
|
|
|
105 |
}
|
|
|
106 |
|
|
|
107 |
private function getIndexNomVernaculaire($nomVernaculaire) {
|
|
|
108 |
$indexCourrant = null;
|
|
|
109 |
if (array_key_exists($nomVernaculaire, $this->nomsIndex) == false) {
|
|
|
110 |
$this->nomsIndex[$nomVernaculaire] = $this->numeroIndex++;
|
|
|
111 |
}
|
|
|
112 |
$indexCourrant = $this->nomsIndex[$nomVernaculaire];
|
| 838 |
raphael |
113 |
echo "$nomVernaculaire : $indexCourrant\n";
|
| 22 |
jpm |
114 |
return $indexCourrant;
|
|
|
115 |
}
|
|
|
116 |
|
|
|
117 |
private function ouvrirFichier($chemin) {
|
|
|
118 |
$fichierOuvert = false;
|
|
|
119 |
if ($chemin) {
|
|
|
120 |
if (file_exists($chemin) === true) {
|
|
|
121 |
$fichierOuvert = fopen($chemin, 'r');
|
|
|
122 |
if ($fichierOuvert == false) {
|
|
|
123 |
throw new Exception("Le fichier $chemin n'a pas pu être ouvert.");
|
| 11 |
jpm |
124 |
}
|
|
|
125 |
} else {
|
| 22 |
jpm |
126 |
throw new Exception("Le fichier $chemin est introuvable.");
|
| 11 |
jpm |
127 |
}
|
| 22 |
jpm |
128 |
} else {
|
|
|
129 |
throw new Exception("Aucun chemin de fichier n'a été fourni.");
|
| 11 |
jpm |
130 |
}
|
| 22 |
jpm |
131 |
return $fichierOuvert;
|
| 11 |
jpm |
132 |
}
|
| 22 |
jpm |
133 |
|
|
|
134 |
private function protegerValeursDesChamps($champs) {
|
|
|
135 |
$champsProteges = array();
|
|
|
136 |
for ($i = 0; $i < 9; $i++) {
|
|
|
137 |
$valeur = isset($champs[$i]) ? $champs[$i] : '';
|
| 79 |
jpm |
138 |
$champsProteges[] = $this->getBdd()->proteger($valeur);
|
| 22 |
jpm |
139 |
}
|
|
|
140 |
return $champsProteges;
|
|
|
141 |
}
|
| 24 |
jpm |
142 |
|
|
|
143 |
private function chargerBiblio() {
|
|
|
144 |
$cheminsNvjflBiblio = Config::get('chemins.nvjflBiblio');
|
|
|
145 |
$tableNvjflBiblio = Config::get('tables.nvjflBiblio');
|
|
|
146 |
$requete = "LOAD DATA INFILE '$cheminsNvjflBiblio' ".
|
|
|
147 |
"REPLACE INTO TABLE $tableNvjflBiblio ".
|
|
|
148 |
'CHARACTER SET utf8 '.
|
|
|
149 |
'FIELDS '.
|
|
|
150 |
" TERMINATED BY '\t' ".
|
|
|
151 |
" ENCLOSED BY '' ".
|
|
|
152 |
" ESCAPED BY '\\\' ".
|
|
|
153 |
'IGNORE 1 LINES';
|
| 79 |
jpm |
154 |
$this->getBdd()->requeter($requete);
|
| 24 |
jpm |
155 |
}
|
|
|
156 |
|
|
|
157 |
private function chargerBiblioLien() {
|
|
|
158 |
$cheminNvjflLienBiblio = Config::get('chemins.nvjflLienBiblio');
|
|
|
159 |
$tableNvjflLienBiblio = Config::get('tables.nvjflLienBiblio');
|
|
|
160 |
$requete = "LOAD DATA INFILE '$cheminNvjflLienBiblio' ".
|
|
|
161 |
"REPLACE INTO TABLE $tableNvjflLienBiblio ".
|
|
|
162 |
'CHARACTER SET utf8 '.
|
|
|
163 |
'FIELDS '.
|
|
|
164 |
" TERMINATED BY '\t' ".
|
|
|
165 |
" ENCLOSED BY '' ".
|
|
|
166 |
" ESCAPED BY '\\\' ".
|
|
|
167 |
'IGNORE 1 LINES';
|
| 79 |
jpm |
168 |
$this->getBdd()->requeter($requete);
|
| 24 |
jpm |
169 |
}
|
| 27 |
jpm |
170 |
|
|
|
171 |
private function chargerOntologies() {
|
|
|
172 |
$cheminOntologies = Config::get('chemins.ontologies');
|
|
|
173 |
$tableOntologies = Config::get('tables.ontologies');
|
|
|
174 |
$requete = "LOAD DATA INFILE '$cheminOntologies' ".
|
|
|
175 |
"REPLACE INTO TABLE $tableOntologies ".
|
|
|
176 |
'CHARACTER SET utf8 '.
|
|
|
177 |
'FIELDS '.
|
|
|
178 |
" TERMINATED BY '\t' ".
|
|
|
179 |
" ENCLOSED BY '' ".
|
|
|
180 |
" ESCAPED BY '\\\' ".
|
|
|
181 |
'IGNORE 1 LINES';
|
| 79 |
jpm |
182 |
$this->getBdd()->requeter($requete);
|
| 27 |
jpm |
183 |
}
|
| 130 |
jpm |
184 |
|
|
|
185 |
private function supprimerTous() {
|
| 349 |
jpm |
186 |
$requete = "DROP TABLE IF EXISTS nvjfl_biblio_v2007, nvjfl_lien_biblio_v2007, nvjfl_meta, nvjfl_ontologies_v2007, nvjfl_v2007";
|
| 130 |
jpm |
187 |
$this->getBdd()->requeter($requete);
|
|
|
188 |
}
|
| 11 |
jpm |
189 |
}
|
|
|
190 |
?>
|