2103 |
drzraf |
1 |
<?php
|
|
|
2 |
/*
|
|
|
3 |
* Moteur de recherche SPHINX
|
|
|
4 |
* @author Raphaël Droz <raphael@tela-botanica.org
|
|
|
5 |
* @copyright Tela-Botanica 2013
|
|
|
6 |
*/
|
|
|
7 |
|
|
|
8 |
/*
|
|
|
9 |
sudo urpmi lib64sphinxclient-devel
|
|
|
10 |
sudo pecl install sphinx
|
|
|
11 |
|
|
|
12 |
see also: http://www.ibm.com/developerworks/library/os-sphinx/
|
|
|
13 |
see also: http://sphinxsearch.com/docs/manual-2.0.7.html#extended-syntax
|
|
|
14 |
TODO: http://sphinxsearch.com/blog/2010/08/17/how-sphinx-relevance-ranking-works/
|
2113 |
drzraf |
15 |
TODO: attention, projet,bazaar,spip et papyrus sont latin9, ainsi en est-il de l'input attendu du <form>
|
|
|
16 |
mais pour coste... c'est full utf-8
|
2103 |
drzraf |
17 |
*/
|
|
|
18 |
|
|
|
19 |
/* returned struct:
|
|
|
20 |
'poids' => 0,
|
|
|
21 |
'url' => '',
|
|
|
22 |
'titre' => '',
|
|
|
23 |
'hreflang' => '',
|
|
|
24 |
'accesskey' => '',
|
2105 |
drzraf |
25 |
'title' => '', // balise 'title'
|
2103 |
drzraf |
26 |
'date_creation' => '',
|
|
|
27 |
'description' => ''
|
|
|
28 |
*/
|
|
|
29 |
|
2113 |
drzraf |
30 |
define('SPHINX_DSN', '127.0.0.1:9306');
|
2117 |
drzraf |
31 |
define('_MRS_SPHINX_BASEHOST', $_SERVER['HTTP_HOST']);
|
2113 |
drzraf |
32 |
|
2105 |
drzraf |
33 |
// pour strftime()
|
|
|
34 |
date_default_timezone_set('Europe/Paris');
|
|
|
35 |
setlocale(LC_TIME, 'fr_FR');
|
|
|
36 |
|
2103 |
drzraf |
37 |
class MoteurRecherche_SPIP {
|
2105 |
drzraf |
38 |
public function get($ids, $q = NULL) {
|
|
|
39 |
if(!$ids) return array();
|
2103 |
drzraf |
40 |
$db = DB::connect($GLOBALS['_MOTEUR_RECHERCHE_']['spip'][0]['bdd_dsn']);
|
2105 |
drzraf |
41 |
$req = $db->query(sprintf(<<<EOF
|
|
|
42 |
SELECT id_article AS id, titre, texte, date AS date_creation, lang as hreflang
|
|
|
43 |
FROM spip_articles
|
|
|
44 |
WHERE statut = "%s"
|
|
|
45 |
AND id_article IN (%s)
|
|
|
46 |
EOF
|
|
|
47 |
,
|
2103 |
drzraf |
48 |
"publie",
|
|
|
49 |
implode(',', $ids)));
|
2105 |
drzraf |
50 |
|
|
|
51 |
(DB::isError($req)) ? die($req->getMessage()) : '';
|
2103 |
drzraf |
52 |
$content = array();
|
|
|
53 |
while($rec = $req->fetchRow(DB_FETCHMODE_ASSOC)) {
|
2105 |
drzraf |
54 |
$rec['url_simple'] = sprintf("%s/article%d.html",
|
2103 |
drzraf |
55 |
trim($GLOBALS['_MOTEUR_RECHERCHE_']['spip'][0]['url'], '/'),
|
|
|
56 |
$rec['id']);
|
2105 |
drzraf |
57 |
$rec['url'] = sprintf("%s?var_recherche=%s",
|
|
|
58 |
$rec['url_simple'],
|
2106 |
drzraf |
59 |
More_Recherche::traiterMotif($q, 'url'));
|
2105 |
drzraf |
60 |
$rec['description'] = More_Recherche::couperTexte($rec['texte'], MORE_RESULTAT_TAILLE_DESCRIPTION);
|
2103 |
drzraf |
61 |
unset($rec['texte']);
|
|
|
62 |
$content[$rec['id']] = $rec;
|
|
|
63 |
|
|
|
64 |
}
|
|
|
65 |
return $content;
|
|
|
66 |
}
|
|
|
67 |
}
|
|
|
68 |
|
|
|
69 |
class MoteurRecherche_BAZAR {
|
2105 |
drzraf |
70 |
public function get($ids, $q = NULL) {
|
|
|
71 |
if(!$ids) return array();
|
2103 |
drzraf |
72 |
$db = DB::connect($GLOBALS['_MOTEUR_RECHERCHE_']['bazar'][0]['bdd_dsn']);
|
2105 |
drzraf |
73 |
$req = $db->query(sprintf(<<<EOF
|
|
|
74 |
SELECT bf_id_fiche AS id,
|
|
|
75 |
bf_description AS texte,
|
|
|
76 |
bf_titre AS titre,
|
|
|
77 |
bf_date_debut_evenement AS date_creation
|
|
|
78 |
FROM bazar_fiche
|
|
|
79 |
WHERE bf_id_fiche IN (%s)
|
|
|
80 |
EOF
|
|
|
81 |
,
|
|
|
82 |
implode(',', $ids)));
|
2103 |
drzraf |
83 |
|
2105 |
drzraf |
84 |
(DB::isError($req)) ? die($req->getMessage()) : '';
|
2103 |
drzraf |
85 |
$content = array();
|
|
|
86 |
while($rec = $req->fetchRow(DB_FETCHMODE_ASSOC)) {
|
2105 |
drzraf |
87 |
$rec['url_simple'] = $rec['url'] = sprintf(trim($GLOBALS['_MOTEUR_RECHERCHE_']['bazar'][0]['url'], '/'), $rec['id']);
|
|
|
88 |
$rec['description'] = More_Recherche::couperTexte($rec['texte'], MORE_RESULTAT_TAILLE_DESCRIPTION);
|
2103 |
drzraf |
89 |
unset($rec['texte']);
|
|
|
90 |
$content[$rec['id']] = $rec;
|
|
|
91 |
}
|
|
|
92 |
return $content;
|
|
|
93 |
}
|
|
|
94 |
}
|
|
|
95 |
|
|
|
96 |
class MoteurRecherche_PROJET {
|
2105 |
drzraf |
97 |
public function get($ids, $q = NULL) {
|
|
|
98 |
if(!$ids) return array();
|
2103 |
drzraf |
99 |
$db = $GLOBALS['_MOTEUR_RECHERCHE_']['bd']['papyrus'];
|
2105 |
drzraf |
100 |
$req = $db->query(sprintf(<<<EOF
|
2117 |
drzraf |
101 |
SELECT p_id AS id, p_titre AS titre, p_description, p_date_creation AS date_creation
|
2105 |
drzraf |
102 |
FROM projet WHERE p_id IN (%s)
|
|
|
103 |
EOF
|
|
|
104 |
,
|
2103 |
drzraf |
105 |
implode(',', $ids)));
|
2105 |
drzraf |
106 |
|
2103 |
drzraf |
107 |
(DB::isError($req)) ? die($req->getMessage()) : '';
|
|
|
108 |
$content = array();
|
|
|
109 |
while($rec = $req->fetchRow(DB_FETCHMODE_ASSOC)) {
|
2105 |
drzraf |
110 |
$rec['url_simple'] = $rec['url'] = sprintf("%s?id_projet=%d",
|
|
|
111 |
trim($GLOBALS['_MOTEUR_RECHERCHE_']['projet'][0]['url'], '/'),
|
|
|
112 |
$rec['id']);
|
2103 |
drzraf |
113 |
$rec['description'] = substr(strip_tags($rec['p_description']), 0, 400 + 2 * MORE_RESULTAT_TAILLE_DESCRIPTION);
|
|
|
114 |
unset($rec['p_description']);
|
|
|
115 |
$content[$rec['id']] = $rec;
|
|
|
116 |
}
|
|
|
117 |
return $content;
|
|
|
118 |
}
|
|
|
119 |
}
|
|
|
120 |
|
|
|
121 |
class MoteurRecherche_PAPYRUS {
|
2105 |
drzraf |
122 |
public function get($ids, $q = NULL) {
|
|
|
123 |
if(!$ids) return array();
|
|
|
124 |
$db = $GLOBALS['_MOTEUR_RECHERCHE_']['bd']['papyrus'];
|
|
|
125 |
$req = $db->query(sprintf(<<<EOF
|
|
|
126 |
SELECT mc.gmc_ce_menu AS id,
|
|
|
127 |
IF(gm_nom != '', gm_nom, IF(gm_titre != '', gm_titre, gm_titre_alternatif)) AS titre,
|
|
|
128 |
gmc_contenu AS texte,
|
|
|
129 |
gm_description_libre, gm_description_resume,
|
|
|
130 |
gm_mots_cles,gm_source, gm_auteur, gm_contributeur, gm_editeur, gm_categorie, gm_date_creation AS date_creation
|
|
|
131 |
FROM gen_menu m
|
|
|
132 |
LEFT JOIN gen_menu_contenu mc ON mc.gmc_ce_menu = m.gm_id_menu AND mc.gmc_bool_dernier = 1
|
|
|
133 |
WHERE mc.gmc_ce_menu IN (%s)
|
|
|
134 |
EOF
|
|
|
135 |
,
|
|
|
136 |
implode(',', $ids)));
|
|
|
137 |
|
|
|
138 |
(DB::isError($req)) ? die($req->getMessage()) : '';
|
|
|
139 |
$content = array();
|
|
|
140 |
while($rec = $req->fetchRow(DB_FETCHMODE_ASSOC)) {
|
|
|
141 |
// Création de l'url
|
|
|
142 |
// TODO : utiliser comme pour spip un fichier de config spécifique pour virer PAP_URL d'ici
|
|
|
143 |
$une_url = new Pap_URL(PAP_URL);
|
|
|
144 |
$une_url->setId($rec['id']);
|
|
|
145 |
$rec['url_simple'] = $une_url->getURL();
|
2106 |
drzraf |
146 |
$une_url->addQueryString('var_recherche', More_Recherche::traiterMotif($q, 'url'), true);
|
2105 |
drzraf |
147 |
$rec['url'] = $une_url->getURL();
|
|
|
148 |
|
|
|
149 |
$rec['description'] = htmlentities($rec['gm_description_resume']);
|
|
|
150 |
unset($rec['gm_description_resume']);
|
|
|
151 |
$content[$rec['id']] = $rec;
|
|
|
152 |
}
|
|
|
153 |
return $content;
|
2103 |
drzraf |
154 |
}
|
|
|
155 |
}
|
|
|
156 |
|
2117 |
drzraf |
157 |
class MoteurRecherche_COSTE {
|
|
|
158 |
public function get($ids, $q = NULL) {
|
|
|
159 |
if(!$ids) return array();
|
|
|
160 |
// DB access is dumb, let's use this one and pray
|
|
|
161 |
$db = $GLOBALS['_MOTEUR_RECHERCHE_']['bd']['papyrus'];
|
|
|
162 |
$req = $db->query(sprintf(<<<EOF
|
|
|
163 |
SELECT c.flore_bdtfx_nn AS id, c.nom_sci AS titre, dsc.body AS description
|
|
|
164 |
FROM tb_eflore.coste_v2_00 c
|
|
|
165 |
LEFT JOIN tela_prod_wikini.florecoste_pages dsc ON c.page_wiki_dsc = dsc.tag AND dsc.latest = 'Y'
|
|
|
166 |
WHERE c.flore_bdtfx_nn IN (%s)
|
|
|
167 |
EOF
|
|
|
168 |
,
|
|
|
169 |
implode(',', $ids)));
|
2105 |
drzraf |
170 |
|
2117 |
drzraf |
171 |
(DB::isError($req)) ? die($req->getMessage()) : '';
|
|
|
172 |
$content = array();
|
|
|
173 |
while($rec = $req->fetchRow(DB_FETCHMODE_ASSOC)) {
|
|
|
174 |
$rec['url_simple'] = $rec['url'] = sprintf("http://%s/bdtfx-nn-%d", _MRS_SPHINX_BASEHOST, $rec['id']);
|
|
|
175 |
// TODO: interpret wikini
|
|
|
176 |
$rec['description'] = substr($rec['description'], 0, 400 + 2 * MORE_RESULTAT_TAILLE_DESCRIPTION);
|
|
|
177 |
$content[$rec['id']] = $rec;
|
|
|
178 |
}
|
|
|
179 |
return $content;
|
|
|
180 |
}
|
|
|
181 |
}
|
|
|
182 |
|
2123 |
drzraf |
183 |
class MoteurRecherche_NVJFL {
|
|
|
184 |
public function get($ids, $q = NULL) {
|
|
|
185 |
if(!$ids) return array();
|
|
|
186 |
// DB access is dumb, let's use this one and pray
|
|
|
187 |
$db = $GLOBALS['_MOTEUR_RECHERCHE_']['bd']['papyrus'];
|
|
|
188 |
$req = $db->query(sprintf(<<<EOF
|
|
|
189 |
SELECT b.num_nom AS id, CONCAT(nom_sci, ' (nn: ', b.num_nom, ', nt: ', num_taxonomique, ')') AS titre, GROUP_CONCAT(n.nom_vernaculaire) AS description
|
2125 |
drzraf |
190 |
FROM tb_eflore.bdtfx_v1_01 b
|
2123 |
drzraf |
191 |
LEFT JOIN tb_eflore.nvjfl_v2007 n ON n.num_taxon = b.num_taxonomique
|
|
|
192 |
WHERE b.num_nom IN (%s)
|
|
|
193 |
GROUP BY n.num_taxon
|
|
|
194 |
EOF
|
|
|
195 |
,
|
|
|
196 |
implode(',', $ids)));
|
2117 |
drzraf |
197 |
|
2123 |
drzraf |
198 |
(DB::isError($req)) ? die($req->getMessage()) : '';
|
|
|
199 |
$content = array();
|
|
|
200 |
while($rec = $req->fetchRow(DB_FETCHMODE_ASSOC)) {
|
|
|
201 |
$rec['url_simple'] = $rec['url'] = sprintf("http://%s/bdtfx-nn-%d", _MRS_SPHINX_BASEHOST, $rec['id']);
|
|
|
202 |
$rec['description'] = substr($rec['description'], 0, 400 + 2 * MORE_RESULTAT_TAILLE_DESCRIPTION);
|
|
|
203 |
$content[$rec['id']] = $rec;
|
|
|
204 |
}
|
|
|
205 |
return $content;
|
|
|
206 |
}
|
|
|
207 |
}
|
|
|
208 |
|
|
|
209 |
|
2103 |
drzraf |
210 |
// http://stackoverflow.com/questions/348410/sort-an-array-based-on-another-array
|
|
|
211 |
function _sortArrayByArray($array, $orderArray) {
|
|
|
212 |
$ordered = array();
|
|
|
213 |
foreach($orderArray as $key) {
|
|
|
214 |
if(array_key_exists($key, $array)) {
|
|
|
215 |
$ordered[$key] = $array[$key];
|
|
|
216 |
unset($array[$key]);
|
|
|
217 |
}
|
|
|
218 |
}
|
|
|
219 |
return $ordered + $array;
|
2105 |
drzraf |
220 |
}
|
2103 |
drzraf |
221 |
|
2105 |
drzraf |
222 |
// adaption aux templates existants: [score] => [weight]
|
|
|
223 |
function _weight2score(&$item, $key, $max) {
|
|
|
224 |
$item['score'] = intval($item['weight'] / $max * 100);
|
2117 |
drzraf |
225 |
$item['date_creation'] = isset($item['date_creation']) ? strftime("%d %B %Y", strtotime($item['date_creation'])) : '';
|
2105 |
drzraf |
226 |
unset($item['weight']);
|
2103 |
drzraf |
227 |
}
|
|
|
228 |
|
2105 |
drzraf |
229 |
function sphinx_search($q = NULL, $page = 1) {
|
|
|
230 |
if(!$q) return array();
|
2103 |
drzraf |
231 |
|
2113 |
drzraf |
232 |
$db = mysql_connect(SPHINX_DSN, NULL, NULL, TRUE);
|
2123 |
drzraf |
233 |
$req = mysql_query(sprintf("SELECT group_id, main_id FROM i_projet, i_spip, i_papyrus, i_bazar, i_coste, i_nvjfl WHERE MATCH('%s') LIMIT 50", $q), $db);
|
2103 |
drzraf |
234 |
|
2126 |
drzraf |
235 |
$res = array('spip' => array(), 'bazar' => array(), 'projet' => array(), 'papyrus' => array(), 'coste' => array(), 'nvjfl' => array());
|
2105 |
drzraf |
236 |
$ids_par_poids = array();
|
|
|
237 |
while($rec = mysql_fetch_array($req, MYSQL_ASSOC)) {
|
|
|
238 |
$res[$rec['group_id']][$rec['id']] = $rec;
|
|
|
239 |
$ids_par_poids[] = $rec['main_id'];
|
|
|
240 |
}
|
|
|
241 |
$docs = array();
|
2103 |
drzraf |
242 |
|
2105 |
drzraf |
243 |
// spip
|
|
|
244 |
$spip = new MoteurRecherche_SPIP();
|
|
|
245 |
foreach($spip->get(array_filter(array_keys($res['spip']), 'intval'), $q) as $v) {
|
|
|
246 |
unset($res['spip'][$v['id']]['group_id'],
|
|
|
247 |
$res['spip'][$v['id']]['main_id'],
|
|
|
248 |
$res['spip'][$v['id']]['id']);
|
|
|
249 |
// left: weight
|
|
|
250 |
$docs['spip-' . $v['id']] = array_merge($v, $res['spip'][$v['id']]);
|
|
|
251 |
}
|
2103 |
drzraf |
252 |
|
2105 |
drzraf |
253 |
// bazar
|
|
|
254 |
$bazar = new MoteurRecherche_BAZAR();
|
|
|
255 |
foreach($bazar->get(array_filter(array_keys($res['bazar']), 'intval'), $q) as $v) {
|
|
|
256 |
unset($res['bazar'][$v['id']]['group_id'],
|
|
|
257 |
$res['bazar'][$v['id']]['main_id'],
|
|
|
258 |
$res['bazar'][$v['id']]['id']);
|
|
|
259 |
// left: weight
|
|
|
260 |
$docs['bazar-' . $v['id']] = array_merge($v, $res['bazar'][$v['id']]);
|
|
|
261 |
}
|
2103 |
drzraf |
262 |
|
2105 |
drzraf |
263 |
// projet
|
|
|
264 |
$projet = new MoteurRecherche_PROJET();
|
|
|
265 |
foreach($projet->get(array_filter(array_keys($res['projet']), 'intval'), $q) as $v) {
|
|
|
266 |
unset($res['projet'][$v['id']]['group_id'],
|
|
|
267 |
$res['projet'][$v['id']]['main_id'],
|
|
|
268 |
$res['projet'][$v['id']]['id']);
|
|
|
269 |
// left: weight
|
|
|
270 |
$docs['projet-' . $v['id']] = array_merge($v, $res['projet'][$v['id']]);
|
|
|
271 |
}
|
2103 |
drzraf |
272 |
|
2105 |
drzraf |
273 |
// papyrus
|
|
|
274 |
$papyrus = new MoteurRecherche_PAPYRUS();
|
|
|
275 |
foreach($papyrus->get(array_filter(array_keys($res['papyrus']), 'intval'), $q) as $v) {
|
|
|
276 |
unset($res['papyrus'][$v['id']]['group_id'],
|
|
|
277 |
$res['papyrus'][$v['id']]['main_id'],
|
|
|
278 |
$res['papyrus'][$v['id']]['id']);
|
|
|
279 |
// left: weight
|
|
|
280 |
$docs['papyrus-' . $v['id']] = array_merge($v, $res['papyrus'][$v['id']]);
|
|
|
281 |
}
|
2103 |
drzraf |
282 |
|
2117 |
drzraf |
283 |
// coste
|
|
|
284 |
$coste = new MoteurRecherche_COSTE();
|
|
|
285 |
foreach($coste->get(array_filter(array_keys($res['coste']), 'intval'), $q) as $v) {
|
|
|
286 |
unset($res['coste'][$v['id']]['group_id'],
|
|
|
287 |
$res['coste'][$v['id']]['main_id'],
|
|
|
288 |
$res['coste'][$v['id']]['id']);
|
|
|
289 |
// left: weight
|
|
|
290 |
$docs['coste-' . $v['id']] = array_merge($v, $res['coste'][$v['id']]);
|
|
|
291 |
}
|
|
|
292 |
|
2123 |
drzraf |
293 |
// nvjfl
|
|
|
294 |
$nvjfl = new MoteurRecherche_NVJFL();
|
|
|
295 |
foreach($nvjfl->get(array_filter(array_keys($res['nvjfl']), 'intval'), $q) as $v) {
|
|
|
296 |
unset($res['nvjfl'][$v['id']]['group_id'],
|
|
|
297 |
$res['nvjfl'][$v['id']]['main_id'],
|
|
|
298 |
$res['nvjfl'][$v['id']]['id']);
|
|
|
299 |
// left: weight
|
|
|
300 |
$docs['nvjfl-' . $v['id']] = array_merge($v, $res['nvjfl'][$v['id']]);
|
|
|
301 |
}
|
|
|
302 |
|
2105 |
drzraf |
303 |
$sorted = _sortArrayByArray($docs, $ids_par_poids);
|
|
|
304 |
|
|
|
305 |
// sort
|
|
|
306 |
/*
|
|
|
307 |
// uncomment this:
|
|
|
308 |
print_r(array_keys($sorted); die;
|
|
|
309 |
|
|
|
310 |
// then:
|
|
|
311 |
ddiff
|
2123 |
drzraf |
312 |
<(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) \
|
2105 |
drzraf |
313 |
<(POST http://localhost/site:reseau<<<"more_motif=test&"|awk -F'=>' '{print $2}'|sed -e 's/ //g' -e '/^$/d')
|
|
|
314 |
|
|
|
315 |
// both should be equal.
|
|
|
316 |
// [ SELECT main_id, group_id FROM i_projet, i_spip, i_papyrus, i_bazar WHERE MATCH('test') LIMIT 50; ]
|
|
|
317 |
*/
|
|
|
318 |
|
|
|
319 |
$max = current($sorted);
|
|
|
320 |
$max = $max['weight'];
|
|
|
321 |
array_walk($sorted, '_weight2score', $max);
|
|
|
322 |
|
2117 |
drzraf |
323 |
// var_dump($sorted);die;
|
2105 |
drzraf |
324 |
return $sorted;
|
|
|
325 |
}
|
|
|
326 |
|
|
|
327 |
|
2103 |
drzraf |
328 |
/*
|
|
|
329 |
// http://www.php.net/manual/fr/sphinx.examples.php
|
|
|
330 |
$s = new SphinxClient;
|
|
|
331 |
$s->setServer("localhost", 9306);
|
|
|
332 |
$s->setMatchMode(SPH_MATCH_ANY);
|
|
|
333 |
$s->setMaxQueryTime(3);
|
|
|
334 |
var_dump($s->query("test"));
|
|
|
335 |
*/
|
2108 |
drzraf |
336 |
|
|
|
337 |
|
|
|
338 |
/*
|
|
|
339 |
Note: conversion côté client SQL:
|
|
|
340 |
mysql -h0 -P 9306 < <(iconv -f utf8 -t latin1 <<<"SELECT * FROM i_projet, i_spip, i_papyrus, i_bazar WHERE MATCH('journée');")
|
|
|
341 |
*/
|