Subversion Repositories Applications.papyrus

Rev

Rev 2143 | Details | Compare with Previous | Last modification | View Log | RSS feed

Rev Author Line No. Line
2103 drzraf 1 
<?php
2140 jp_milcent 2 
/*vim: set expandtab tabstop=4 shiftwidth=4: */
3 
// +------------------------------------------------------------------------------------------------------+
4 
// | PHP version 4.1                                                                                      |
5 
// +------------------------------------------------------------------------------------------------------+
6 
// | Copyright (C) 2004 Tela Botanica (accueil@tela-botanica.org)                                         |
7 
// +------------------------------------------------------------------------------------------------------+
8 
// | This file is part of Papyrus.                                                                        |
9 
// |                                                                                                      |
10 
// | Foobar is free software; you can redistribute it and/or modify                                       |
11 
// | it under the terms of the GNU General Public License as published by                                 |
12 
// | the Free Software Foundation; either version 2 of the License, or                                    |
13 
// | (at your option) any later version.                                                                  |
14 
// |                                                                                                      |
15 
// | Foobar is distributed in the hope that it will be useful,                                            |
16 
// | but WITHOUT ANY WARRANTY; without even the implied warranty of                                       |
17 
// | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the                                        |
18 
// | GNU General Public License for more details.                                                         |
19 
// |                                                                                                      |
20 
// | You should have received a copy of the GNU General Public License                                    |
21 
// | along with Foobar; if not, write to the Free Software                                                |
22 
// | Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA                            |
23 
// +------------------------------------------------------------------------------------------------------+
24 
/**
2103 drzraf 25 
 * Moteur de recherche SPHINX
2140 jp_milcent 26 
 *
27 
 * Installation de sphinx :
28 
 * sudo urpmi lib64sphinxclient-devel
29 
 * sudo pecl install sphinx
30 
 *
31 
 * See also:
32 
 *  - http://www.ibm.com/developerworks/library/os-sphinx/
33 
 *  - http://sphinxsearch.com/docs/manual-2.0.7.html#extended-syntax
34 
 *
35 
 * TODO: http://sphinxsearch.com/blog/2010/08/17/how-sphinx-relevance-ranking-works/
36 
 * TODO: attention, projet,bazaar,spip et papyrus sont latin9, ainsi en est-il de l'input attendu du <form>
37 
 * mais pour coste... c'est full utf-8
38 
 * Structure retournée :
39 
 * 	'poids' => 0,
40 
 * 	'url' => '',
41 
 * 	'titre' => '',
42 
 * 	'hreflang' => '',
43 
 * 	'accesskey' => '',
44 
 * 	'title' => '', // balise 'title'
45 
 * 	'date_creation' => '',
46 
 * 	'description' => ''
47 
 *
48 
 *
49 
 * // http://www.php.net/manual/fr/sphinx.examples.php
50 
 * $s = new SphinxClient;
51 
 * $s->setServer("localhost", 9306);
52 
 * $s->setMatchMode(SPH_MATCH_ANY);
53 
 * $s->setMaxQueryTime(3);
54 
 * var_dump($s->query("test"));
55 
 *
56 
 * Note: conversion côté client SQL:
57 
 * mysql -h0 -P 9306 < <(iconv -f utf8 -t latin1 <<<"SELECT * FROM i_projet, i_spip, i_papyrus, i_bazar WHERE MATCH('journée');")
58 
 *
59 
 *
60 
 * Test :
61 
 * ddiff
62 
 * <(mysql -h0 -P 9306 <<<"SELECT main_id FROM i_projet, i_spip, i_papyrus, i_bazar, i_coste, i_nvjfl WHERE MATCH('test') LIMIT 50;"|awk '{print $3}'|sed 1d) \
63 
 * <(POST http://localhost/site:reseau<<<"more_motif=test&"|awk -F'=>' '{print $2}'|sed -e 's/ //g' -e '/^$/d')
64 
 *
65 
 * both should be equal.
66 
 * [ SELECT main_id, group_id FROM i_projet, i_spip, i_papyrus, i_bazar WHERE MATCH('test') LIMIT 50; ]
67 
 *
68 
 *
69 
 *@package Applette
70 
 *@subpackage Moteur_recherche
71 
 //Auteur original :
72 
 * @author		Raphaël Droz <raphael@tela-botanica.org
73 
 //Autres auteurs :
74 
 *@author		Jean-Pascal MILCENT <jpm@tela-botanica.org>
75 
 *@copyright	Tela-Botanica 2000-2013
76 
 *@version		$Revision$
77 
 // +------------------------------------------------------------------------------------------------------+
2103 drzraf 78 
 */
79 
 
2140 jp_milcent 80 
define('SPHINX_DSN', '193.54.123.216:9306');
2117 drzraf 81 
define('_MRS_SPHINX_BASEHOST', $_SERVER['HTTP_HOST']);
2105 drzraf 82 
// pour strftime()
83 
date_default_timezone_set('Europe/Paris');
84 
setlocale(LC_TIME, 'fr_FR');
85 
 
2140 jp_milcent 86 
function sphinx_search($q = NULL, $page = 1) {
87 
	if(!$q) return array();
88 
 
89 
	// quelques aliases pour faciliter l'usage sans passer par le full sphinxQL
90 
	// $q = preg_replace('/\<actu\>(.*)/', '\1 @group_id i_spip', $q);
91 
	// $q = preg_replace('/\<eflore\>(.*)/', '\1 @group_id i_bazar|i_coste', $q);
92 
 
93 
	$db = mysql_connect(SPHINX_DSN, NULL, NULL, TRUE);
94 
	// AFAICT, pas de réel risque de SQL-injection du côté de sphinx (au pire, $req = FALSE)
95 
	// et il serait dommage de devoir limiter la puissante syntaxe offerte à  l'utilisation
96 
	//$requeteTpl = "SELECT group_id, main_id FROM i_projet, i_spip, i_papyrus, i_bazar, i_coste, i_nvjfl WHERE MATCH('%s') LIMIT 50";
2143 jp_milcent 97 
	$requeteTpl = 'SELECT group_id, main_id, id, WEIGHT() AS poids '.
2141 jp_milcent 98 
		'FROM i_projet, i_spip, i_papyrus, i_bazar, i_coste, i_nvjfl '.
2140 jp_milcent 99 
		"WHERE MATCH('%s') ".
100 
		'LIMIT 50 ';
101 
	$requete = mysql_query(sprintf($requeteTpl, $q), $db);
102 
 
103 
	$retour = array();
104 
	if ($requete) {
105 
		$res = array('spip' => array(), 'bazar' => array(), 'projet' => array(), 'papyrus' => array(), 'coste' => array(), 'nvjfl' => array());
106 
		$ids_par_poids = array();
2143 jp_milcent 107 
		$poidsMax = 0;
108 
		while ($rec = mysql_fetch_array($requete, MYSQL_ASSOC)) {
109 
			$poidsMax = ($rec['poids'] > $poidsMax) ? $rec['poids'] : $poidsMax;
2140 jp_milcent 110 
			$res[$rec['group_id']][$rec['id']] = $rec;
111 
			$ids_par_poids[] = $rec['main_id'];
112 
		}
113 
 
114 
		$docs = array();
115 
		// spip
116 
		$spip = new MoteurRecherche_SPIP();
2143 jp_milcent 117 
		foreach ($spip->get(array_filter(array_keys($res['spip']), 'intval'), $q) as $v) {
2140 jp_milcent 118 
			unset($res['spip'][$v['id']]['group_id'],
2143 jp_milcent 119 
				$res['spip'][$v['id']]['main_id'],
120 
				$res['spip'][$v['id']]['id']);
2140 jp_milcent 121 
			$docs['spip-' . $v['id']] = array_merge($v,	$res['spip'][$v['id']]);
122 
		}
123 
 
124 
		// bazar
125 
		$bazar = new MoteurRecherche_BAZAR();
2143 jp_milcent 126 
		foreach ($bazar->get(array_filter(array_keys($res['bazar']), 'intval'), $q) as $v) {
2140 jp_milcent 127 
			unset($res['bazar'][$v['id']]['group_id'],
2143 jp_milcent 128 
				$res['bazar'][$v['id']]['main_id'],
129 
				$res['bazar'][$v['id']]['id']);
2140 jp_milcent 130 
			$docs['bazar-' . $v['id']] = array_merge($v, $res['bazar'][$v['id']]);
131 
		}
132 
 
133 
		// projet
134 
		$projet = new MoteurRecherche_PROJET();
2143 jp_milcent 135 
		foreach ($projet->get(array_filter(array_keys($res['projet']), 'intval'), $q) as $v) {
2140 jp_milcent 136 
			unset($res['projet'][$v['id']]['group_id'],
2143 jp_milcent 137 
				$res['projet'][$v['id']]['main_id'],
138 
				$res['projet'][$v['id']]['id']);
2140 jp_milcent 139 
			$docs['projet-' . $v['id']] = array_merge($v, $res['projet'][$v['id']]);
140 
		}
141 
 
142 
		// papyrus
143 
		$papyrus = new MoteurRecherche_PAPYRUS();
2143 jp_milcent 144 
		foreach ($papyrus->get(array_filter(array_keys($res['papyrus']), 'intval'), $q) as $v) {
2140 jp_milcent 145 
			unset($res['papyrus'][$v['id']]['group_id'],
2143 jp_milcent 146 
				$res['papyrus'][$v['id']]['main_id'],
147 
				$res['papyrus'][$v['id']]['id']);
2140 jp_milcent 148 
			$docs['papyrus-' . $v['id']] = array_merge($v, $res['papyrus'][$v['id']]);
149 
		}
150 
 
151 
		// coste
152 
		$coste = new MoteurRecherche_COSTE();
2143 jp_milcent 153 
		foreach ($coste->get(array_filter(array_keys($res['coste']), 'intval'), $q) as $v) {
2140 jp_milcent 154 
			unset($res['coste'][$v['id']]['group_id'],
2143 jp_milcent 155 
				$res['coste'][$v['id']]['main_id'],
156 
				$res['coste'][$v['id']]['id']);
2140 jp_milcent 157 
			$docs['coste-' . $v['id']] = array_merge($v, $res['coste'][$v['id']]);
158 
		}
159 
 
160 
		// nvjfl
161 
		$nvjfl = new MoteurRecherche_NVJFL();
2143 jp_milcent 162 
		foreach ($nvjfl->get(array_filter(array_keys($res['nvjfl']), 'intval'), $q) as $v) {
2140 jp_milcent 163 
			unset($res['nvjfl'][$v['id']]['group_id'],
2143 jp_milcent 164 
				$res['nvjfl'][$v['id']]['main_id'],
165 
				$res['nvjfl'][$v['id']]['id']);
2140 jp_milcent 166 
			$docs['nvjfl-' . $v['id']] = array_merge($v, $res['nvjfl'][$v['id']]);
167 
		}
2143 jp_milcent 168 
		//die('<pre>'.print_r($docs, true).'</pre>');
2140 jp_milcent 169 
 
170 
		// sort
171 
		$sorted = _sortArrayByArray($docs, $ids_par_poids);
172 
 
173 
		if (isset($_GET['tri']) && $_GET['tri'] == 'date') {
174 
			usort($sorted, '_actuNewerFirst');
175 
		}
2144 jp_milcent 176 
		// Transforme le poids de chaque doc en pourcentage relatif aux résultats courant de la recherche
2143 jp_milcent 177 
		array_walk($sorted, '_weight2score', $poidsMax);
2140 jp_milcent 178 
 
179 
		// var_dump($sorted);die;
180 
		$retour = $sorted;
181 
	}
182 
	return $retour;
183 
}
184 
 
2103 drzraf 185 
class MoteurRecherche_SPIP {
2105 drzraf 186 
	public function get($ids, $q = NULL) {
2103 drzraf 187 
		$content = array();
2140 jp_milcent 188 
		if (count($ids) > 0) {
189 
			$db = DB::connect($GLOBALS['_MOTEUR_RECHERCHE_']['spip'][0]['bdd_dsn']);
190 
			$requeteTpl = 'SELECT id_article AS id, titre, texte, date AS date_creation, lang as hreflang '.
191 
					'FROM spip_articles '.
192 
					'WHERE statut = "%s" '.
193 
					'AND id_article IN (%s) ';
194 
			$requete = $db->query(sprintf($requeteTpl, 'publie',implode(',', $ids)));
195 
			(DB::isError($requete)) ? die($requete->getMessage()) : '';
2103 drzraf 196 
 
2140 jp_milcent 197 
			while ($rec = $requete->fetchRow(DB_FETCHMODE_ASSOC)) {
198 
				$rec['url_simple'] = sprintf("%s/article%d.html",
199 
					 trim($GLOBALS['_MOTEUR_RECHERCHE_']['spip'][0]['url'], '/'),
200 
					 $rec['id']);
201 
				$rec['url'] = sprintf("%s?var_recherche=%s",
202 
						$rec['url_simple'],
203 
						More_Recherche::traiterMotif($q, 'url'));
204 
				$rec['description'] = More_Recherche::couperTexte($rec['texte'], MORE_RESULTAT_TAILLE_DESCRIPTION);
205 
				unset($rec['texte']);
206 
				$content[$rec['id']] = $rec;
207 
			}
2103 drzraf 208 
		}
209 
		return $content;
210 
	}
211 
}
212 
 
213 
class MoteurRecherche_BAZAR {
2105 drzraf 214 
	public function get($ids, $q = NULL) {
2140 jp_milcent 215 
		$content = array();
216 
		if (count($ids) > 0) {
217 
			$db = DB::connect($GLOBALS['_MOTEUR_RECHERCHE_']['bazar'][0]['bdd_dsn']);
218 
			$requeteTpl = 'SELECT bf_id_fiche AS id, '.
219 
					'bf_description AS texte, '.
220 
					'bf_titre AS titre, '.
221 
					'bf_date_debut_evenement AS date_creation '.
222 
					'FROM bazar_fiche '.
223 
					'WHERE bf_id_fiche IN (%s) ';
224 
			$requete = $db->query(sprintf($requeteTpl, implode(',', $ids)));
225 
			(DB::isError($requete)) ? die($requete->getMessage()) : '';
2103 drzraf 226 
 
2140 jp_milcent 227 
			while ($rec = $requete->fetchRow(DB_FETCHMODE_ASSOC)) {
228 
				$rec['url_simple'] = $rec['url'] = sprintf(trim($GLOBALS['_MOTEUR_RECHERCHE_']['bazar'][0]['url'], '/'), $rec['id']);
229 
				$rec['description'] = More_Recherche::couperTexte($rec['texte'], MORE_RESULTAT_TAILLE_DESCRIPTION);
230 
				unset($rec['texte']);
231 
				$content[$rec['id']] = $rec;
232 
			}
2103 drzraf 233 
		}
2140 jp_milcent 234 
		return $content;
2103 drzraf 235 
	}
236 
}
237 
 
238 
class MoteurRecherche_PROJET {
2105 drzraf 239 
	public function get($ids, $q = NULL) {
2140 jp_milcent 240 
		$content = array();
241 
		if (count($ids) > 0) {
242 
			$db = $GLOBALS['_MOTEUR_RECHERCHE_']['bd']['papyrus'];
243 
			$requeteTpl = 'SELECT p_id AS id, p_titre AS titre, p_description, p_date_creation AS date_creation '.
244 
					'FROM projet '.
245 
					'WHERE p_id IN (%s)';
246 
			$requete = $db->query(sprintf($requeteTpl, implode(',', $ids)));
247 
			(DB::isError($requete)) ? die($requete->getMessage()) : '';
2105 drzraf 248 
 
2140 jp_milcent 249 
			while ($rec = $requete->fetchRow(DB_FETCHMODE_ASSOC)) {
250 
				$rec['url_simple'] = $rec['url'] = sprintf("%s?id_projet=%d",
251 
					trim($GLOBALS['_MOTEUR_RECHERCHE_']['projet']['url'], '/'),
252 
					$rec['id']);
253 
				$rec['description'] = substr(strip_tags($rec['p_description']), 0, 400 + 2 * MORE_RESULTAT_TAILLE_DESCRIPTION);
254 
				unset($rec['p_description']);
255 
				$content[$rec['id']] = $rec;
256 
			}
2103 drzraf 257 
		}
2140 jp_milcent 258 
		return $content;
2103 drzraf 259 
	}
260 
}
261 
 
262 
class MoteurRecherche_PAPYRUS {
2105 drzraf 263 
	public function get($ids, $q = NULL) {
264 
		$content = array();
2140 jp_milcent 265 
		if (count($ids) > 0) {
266 
			$db = $GLOBALS['_MOTEUR_RECHERCHE_']['bd']['papyrus'];
267 
			$requeteTpl = 'SELECT mc.gmc_ce_menu AS id, '.
268 
				"	IF(gm_nom != '', gm_nom, IF(gm_titre != '', gm_titre, gm_titre_alternatif)) AS titre, ".
269 
				'	gmc_contenu AS texte, '.
270 
				'	gm_description_libre, gm_description_resume, '.
271 
				'	gm_mots_cles,gm_source, gm_auteur, gm_contributeur, gm_editeur, gm_categorie, '.
272 
				'	gm_date_creation AS date_creation '.
273 
				'FROM gen_menu AS m '.
274 
				'	LEFT JOIN gen_menu_contenu AS mc ON mc.gmc_ce_menu = m.gm_id_menu AND mc.gmc_bool_dernier = 1 '.
275 
				'WHERE mc.gmc_ce_menu IN (%s) ';
276 
			$requete = $db->query(sprintf($requeteTpl, implode(',', $ids)));
277 
			(DB::isError($requete)) ? die($requete->getMessage()) : '';
2105 drzraf 278 
 
2140 jp_milcent 279 
			while ($rec = $requete->fetchRow(DB_FETCHMODE_ASSOC)) {
280 
				// Création de l'url
281 
				// TODO : utiliser comme pour spip un fichier de config spécifique pour virer PAP_URL d'ici
282 
				$une_url = new Pap_URL(PAP_URL);
283 
				$une_url->setId($rec['id']);
284 
				$rec['url_simple'] = $une_url->getURL();
285 
				$une_url->addQueryString('var_recherche', More_Recherche::traiterMotif($q, 'url'), true);
286 
				$rec['url'] = $une_url->getURL();
287 
 
288 
				$rec['description'] =  htmlentities($rec['gm_description_resume']);
289 
				unset($rec['gm_description_resume']);
290 
				$content[$rec['id']] = $rec;
291 
			}
2105 drzraf 292 
		}
2140 jp_milcent 293 
		return $content;
2103 drzraf 294 
	}
295 
}
296 
 
2117 drzraf 297 
class MoteurRecherche_COSTE {
298 
	public function get($ids, $q = NULL) {
2140 jp_milcent 299 
		$content = array();
300 
		if (count($ids) > 0) {
301 
			// DB access is dumb, let's use this one and pray
302 
			$db = $GLOBALS['_MOTEUR_RECHERCHE_']['bd']['bota'];
303 
			$requeteTpl = 'SELECT c.flore_bdtfx_nn AS id, c.nom_sci AS titre, dsc.body AS description '.
304 
				'FROM tb_eflore.coste_v2_00 AS c '.
305 
				"	LEFT JOIN tela_prod_wikini.florecoste_pages dsc ON c.page_wiki_dsc = dsc.tag AND dsc.latest = 'Y' ".
306 
				'WHERE c.flore_bdtfx_nn IN (%s) ';
307 
			$requete = $db->query(sprintf($requeteTpl, implode(',', $ids)));
308 
			(DB::isError($requete)) ? die($requete->getMessage()) : '';
2105 drzraf 309 
 
2140 jp_milcent 310 
			while ($rec = $requete->fetchRow(DB_FETCHMODE_ASSOC)) {
311 
				$rec['url_simple'] = $rec['url'] = sprintf("http://%s/bdtfx-nn-%d", _MRS_SPHINX_BASEHOST, $rec['id']);
312 
				// TODO: interpret wikini
313 
				$rec['description'] = substr($rec['description'], 0, 400 + 2 * MORE_RESULTAT_TAILLE_DESCRIPTION);
314 
				$content[$rec['id']] = $rec;
315 
			}
2117 drzraf 316 
		}
2140 jp_milcent 317 
		return $content;
2117 drzraf 318 
	}
319 
}
320 
 
2123 drzraf 321 
class MoteurRecherche_NVJFL {
322 
	public function get($ids, $q = NULL) {
2140 jp_milcent 323 
		$content = array();
324 
		if (count($ids) > 0) {
325 
			// DB access is dumb, let's use this one and pray
326 
			$db = $GLOBALS['_MOTEUR_RECHERCHE_']['bd']['bota'];
327 
			$requeteTpl = 'SELECT b.num_nom AS id, '.
328 
			"	CONCAT(nom_sci, ' (nn: ', b.num_nom, ', nt: ', num_taxonomique, ')') AS titre, ".
329 
			'	GROUP_CONCAT(n.nom_vernaculaire) AS description '.
330 
			'FROM tb_eflore.bdtfx_v1_01 AS b '.
331 
			'	LEFT JOIN tb_eflore.nvjfl_v2007 n ON n.num_taxon = b.num_taxonomique '.
332 
			'WHERE b.num_nom IN (%s) '.
333 
			'GROUP BY n.num_taxon ';
334 
			$requete = $db->query(sprintf($requeteTpl, implode(',', $ids)));
2117 drzraf 335 
 
2140 jp_milcent 336 
			(DB::isError($requete)) ? die($requete->getMessage()) : '';
337 
 
338 
			while ($rec = $requete->fetchRow(DB_FETCHMODE_ASSOC)) {
339 
				$rec['url_simple'] = $rec['url'] = sprintf("http://%s/bdtfx-nn-%d", _MRS_SPHINX_BASEHOST, $rec['id']);
340 
				$rec['description'] = substr($rec['description'], 0, 400 + 2 * MORE_RESULTAT_TAILLE_DESCRIPTION);
341 
				$content[$rec['id']] = $rec;
342 
			}
2123 drzraf 343 
		}
2140 jp_milcent 344 
		return $content;
2123 drzraf 345 
	}
346 
}
347 
 
2103 drzraf 348 
// http://stackoverflow.com/questions/348410/sort-an-array-based-on-another-array
349 
function _sortArrayByArray($array, $orderArray) {
350 
	$ordered = array();
2140 jp_milcent 351 
	foreach ($orderArray as $key) {
352 
		if (array_key_exists($key, $array)) {
2103 drzraf 353 
			$ordered[$key] = $array[$key];
354 
			unset($array[$key]);
355 
		}
356 
	}
357 
	return $ordered + $array;
2105 drzraf 358 
}
2103 drzraf 359 
 
2140 jp_milcent 360 
function _actuNewerFirst($a,$b) {
361 
	return isset($a['date_creation']) && isset($b['date_creation']) ? strcmp($b['date_creation'], $a['date_creation']) : 0;
362 
}
363 
 
2143 jp_milcent 364 
// Transforme un score en pourcentage
2105 drzraf 365 
function _weight2score(&$item, $key, $max) {
2143 jp_milcent 366 
	$item['score'] = intval($item['poids'] / $max * 100);
2117 drzraf 367 
	$item['date_creation'] = isset($item['date_creation']) ? strftime("%d %B %Y", strtotime($item['date_creation'])) : '';
2143 jp_milcent 368 
	unset($item['poids']);
2103 drzraf 369 
}
370 
 
2140 jp_milcent 371 
?>