WebSVN – Applications.papyrus – /trunk/papyrus/applettes/moteur_recherche/moteur_recherche_sphinx.php

<?php
/*
 * Moteur de recherche SPHINX
 * @author        Raphaël Droz <raphael@tela-botanica.org
 * @copyright     Tela-Botanica 2013
 */

/*
  sudo urpmi lib64sphinxclient-devel
  sudo pecl install sphinx

  see also: http://www.ibm.com/developerworks/library/os-sphinx/
  see also: http://sphinxsearch.com/docs/manual-2.0.7.html#extended-syntax
  TODO: http://sphinxsearch.com/blog/2010/08/17/how-sphinx-relevance-ranking-works/
  TODO: attention, projet,bazaar,spip et papyrus sont latin9, ainsi en est-il de l'input attendu du <form>
                mais pour coste... c'est full utf-8
*/

/* returned struct:
'poids' => 0,
'url' => '',
'titre' => '',
'hreflang' => '',
'accesskey' => '',
'title' => '', // balise 'title'
'date_creation' => '',
'description' => ''
*/

define('SPHINX_DSN', '127.0.0.1:9306');
define('_MRS_SPHINX_BASEHOST', $_SERVER['HTTP_HOST']);

// pour strftime()
date_default_timezone_set('Europe/Paris');
setlocale(LC_TIME, 'fr_FR');

class MoteurRecherche_SPIP {
        public function get($ids, $q = NULL) {
                if(!$ids) return array();
                $db = DB::connect($GLOBALS['_MOTEUR_RECHERCHE_']['spip'][0]['bdd_dsn']);
                $req = $db->query(sprintf(<<<EOF
SELECT id_article AS id, titre, texte, date AS date_creation, lang as hreflang
FROM spip_articles
WHERE statut = "%s"
AND id_article IN (%s)
EOF
                                                                  ,
                                                                  "publie",
                                                                  implode(',', $ids)));

                (DB::isError($req)) ? die($req->getMessage()) : '';
                $content = array();
                while($rec = $req->fetchRow(DB_FETCHMODE_ASSOC)) {
                        $rec['url_simple'] = sprintf("%s/article%d.html",
                                                                                 trim($GLOBALS['_MOTEUR_RECHERCHE_']['spip'][0]['url'], '/'),
                                                                                 $rec['id']);
                        $rec['url'] = sprintf("%s?var_recherche=%s",
                                                                  $rec['url_simple'],
                                                                  More_Recherche::traiterMotif($q, 'url'));
                        $rec['description'] = More_Recherche::couperTexte($rec['texte'], MORE_RESULTAT_TAILLE_DESCRIPTION);
                        unset($rec['texte']);
                        $content[$rec['id']] = $rec;

                }
                return $content;
        }
}

class MoteurRecherche_BAZAR {
        public function get($ids, $q = NULL) {
                if(!$ids) return array();
                $db = DB::connect($GLOBALS['_MOTEUR_RECHERCHE_']['bazar'][0]['bdd_dsn']);
                $req = $db->query(sprintf(<<<EOF
SELECT bf_id_fiche AS id,
           bf_description AS texte,
           bf_titre AS titre,
           bf_date_debut_evenement AS date_creation
FROM bazar_fiche
WHERE bf_id_fiche IN (%s)
EOF
                                                                  ,
                                                                  implode(',', $ids)));

                (DB::isError($req)) ? die($req->getMessage()) : '';
                $content = array();
                while($rec = $req->fetchRow(DB_FETCHMODE_ASSOC)) {
                        $rec['url_simple'] = $rec['url'] = sprintf(trim($GLOBALS['_MOTEUR_RECHERCHE_']['bazar'][0]['url'], '/'), $rec['id']);
                        $rec['description'] = More_Recherche::couperTexte($rec['texte'], MORE_RESULTAT_TAILLE_DESCRIPTION);
                        unset($rec['texte']);
                        $content[$rec['id']] = $rec;
                }
                return $content;                
        }
}

class MoteurRecherche_PROJET {
        public function get($ids, $q = NULL) {
                if(!$ids) return array();
                $db = $GLOBALS['_MOTEUR_RECHERCHE_']['bd']['papyrus'];
                $req = $db->query(sprintf(<<<EOF
SELECT p_id AS id, p_titre AS titre, p_description, p_date_creation AS date_creation
FROM projet WHERE p_id IN (%s)
EOF
                                                                  ,
                                                                  implode(',', $ids)));

                (DB::isError($req)) ? die($req->getMessage()) : '';
                $content = array();
                while($rec = $req->fetchRow(DB_FETCHMODE_ASSOC)) {
                        $rec['url_simple'] = $rec['url'] = sprintf("%s?id_projet=%d",
                                                                                                           trim($GLOBALS['_MOTEUR_RECHERCHE_']['projet']['url'], '/'),
                                                                                                           $rec['id']);
                        $rec['description'] = substr(strip_tags($rec['p_description']), 0, 400 + 2 * MORE_RESULTAT_TAILLE_DESCRIPTION);
                        unset($rec['p_description']);
                        $content[$rec['id']] = $rec;
                }
                return $content;                
        }
}

class MoteurRecherche_PAPYRUS {
        public function get($ids, $q = NULL) {
                if(!$ids) return array();
        $db = $GLOBALS['_MOTEUR_RECHERCHE_']['bd']['papyrus'];
                $req = $db->query(sprintf(<<<EOF
SELECT mc.gmc_ce_menu AS id,
           IF(gm_nom != '', gm_nom, IF(gm_titre != '', gm_titre, gm_titre_alternatif)) AS titre,
           gmc_contenu AS texte,
           gm_description_libre, gm_description_resume,
           gm_mots_cles,gm_source, gm_auteur, gm_contributeur, gm_editeur, gm_categorie, gm_date_creation AS date_creation
FROM gen_menu m
LEFT JOIN gen_menu_contenu mc ON mc.gmc_ce_menu = m.gm_id_menu AND mc.gmc_bool_dernier = 1
WHERE mc.gmc_ce_menu IN (%s)
EOF
                                                                  ,
                                                                  implode(',', $ids)));

                (DB::isError($req)) ? die($req->getMessage()) : '';
                $content = array();
                while($rec = $req->fetchRow(DB_FETCHMODE_ASSOC)) {
                        // Création de l'url
                        // TODO : utiliser comme pour spip un fichier de config spécifique pour virer PAP_URL d'ici
                        $une_url = new Pap_URL(PAP_URL);
                        $une_url->setId($rec['id']);
                        $rec['url_simple'] = $une_url->getURL();
                        $une_url->addQueryString('var_recherche', More_Recherche::traiterMotif($q, 'url'), true);
                        $rec['url'] = $une_url->getURL();

                        $rec['description'] =  htmlentities($rec['gm_description_resume']);
                        unset($rec['gm_description_resume']);
                        $content[$rec['id']] = $rec;
                }
                return $content;                
        }
}

class MoteurRecherche_COSTE {
        public function get($ids, $q = NULL) {
                if(!$ids) return array();
                // DB access is dumb, let's use this one and pray
                $db = $GLOBALS['_MOTEUR_RECHERCHE_']['bd']['papyrus'];
                $req = $db->query(sprintf(<<<EOF
SELECT c.flore_bdtfx_nn AS id, c.nom_sci AS titre, dsc.body AS description
FROM tb_eflore.coste_v2_00 c
LEFT JOIN tela_prod_wikini.florecoste_pages dsc ON c.page_wiki_dsc = dsc.tag AND dsc.latest = 'Y'
WHERE c.flore_bdtfx_nn IN (%s)
EOF
                                                                  ,
                                                                  implode(',', $ids)));

                (DB::isError($req)) ? die($req->getMessage()) : '';
                $content = array();
                while($rec = $req->fetchRow(DB_FETCHMODE_ASSOC)) {
                        $rec['url_simple'] = $rec['url'] = sprintf("http://%s/bdtfx-nn-%d", _MRS_SPHINX_BASEHOST, $rec['id']);
                        // TODO: interpret wikini
                        $rec['description'] = substr($rec['description'], 0, 400 + 2 * MORE_RESULTAT_TAILLE_DESCRIPTION);
                        $content[$rec['id']] = $rec;
                }
                return $content;                
        }
}

class MoteurRecherche_NVJFL {
        public function get($ids, $q = NULL) {
                if(!$ids) return array();
                // DB access is dumb, let's use this one and pray
                $db = $GLOBALS['_MOTEUR_RECHERCHE_']['bd']['papyrus'];
                $req = $db->query(sprintf(<<<EOF
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
FROM tb_eflore.bdtfx_v1_01 b
LEFT JOIN tb_eflore.nvjfl_v2007 n ON n.num_taxon = b.num_taxonomique
WHERE b.num_nom IN (%s)
GROUP BY n.num_taxon
EOF
                                                                  ,
                                                                  implode(',', $ids)));

                (DB::isError($req)) ? die($req->getMessage()) : '';
                $content = array();
                while($rec = $req->fetchRow(DB_FETCHMODE_ASSOC)) {
                        $rec['url_simple'] = $rec['url'] = sprintf("http://%s/bdtfx-nn-%d", _MRS_SPHINX_BASEHOST, $rec['id']);
                        $rec['description'] = substr($rec['description'], 0, 400 + 2 * MORE_RESULTAT_TAILLE_DESCRIPTION);
                        $content[$rec['id']] = $rec;
                }
                return $content;                
        }
}


// http://stackoverflow.com/questions/348410/sort-an-array-based-on-another-array
function _sortArrayByArray($array, $orderArray) {
        $ordered = array();
        foreach($orderArray as $key) {
                if(array_key_exists($key, $array)) {
                        $ordered[$key] = $array[$key];
                        unset($array[$key]);
                }
        }
        return $ordered + $array;
}

// adaption aux templates existants: [score] => [weight]
function _weight2score(&$item, $key, $max) {
        $item['score'] = intval($item['weight'] / $max * 100);
        $item['date_creation'] = isset($item['date_creation']) ? strftime("%d %B %Y", strtotime($item['date_creation'])) : '';
        unset($item['weight']);
}

function _actuNewerFirst($a,$b) {
        return isset($a['date_creation']) && isset($b['date_creation']) ? strcmp($b['date_creation'], $a['date_creation']) : 0;
}

function sphinx_search($q = NULL, $page = 1) {
        if(!$q) return array();

        // quelques aliases pour faciliter l'usage sans passer par le full sphinxQL
        // $q = preg_replace('/\<actu\>(.*)/', '\1 @group_id i_spip', $q);
        // $q = preg_replace('/\<eflore\>(.*)/', '\1 @group_id i_bazar|i_coste', $q);

        $db = mysql_connect(SPHINX_DSN, NULL, NULL, TRUE);
        // AFAICT, pas de réel risque de SQL-injection du côté de sphinx (au pire, $req = FALSE)
        // et il serait dommage de devoir limiter la puissante syntaxe offerte à l'utilisation
        $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), $db);
        if(!$req) return array();

        $res = array('spip' => array(), 'bazar' => array(), 'projet' => array(), 'papyrus' => array(), 'coste' => array(), 'nvjfl' => array());
        $ids_par_poids = array();
        while($rec = mysql_fetch_array($req, MYSQL_ASSOC)) {
                $res[$rec['group_id']][$rec['id']] = $rec;
                $ids_par_poids[] = $rec['main_id'];
        }
        $docs = array();

        // spip
        $spip = new MoteurRecherche_SPIP();
        foreach($spip->get(array_filter(array_keys($res['spip']), 'intval'), $q) as $v) {
                unset($res['spip'][$v['id']]['group_id'],
                          $res['spip'][$v['id']]['main_id'],
                          $res['spip'][$v['id']]['id']);
                // left: weight
                $docs['spip-' . $v['id']] = array_merge($v,     $res['spip'][$v['id']]);
        }

        // bazar
        $bazar = new MoteurRecherche_BAZAR();
        foreach($bazar->get(array_filter(array_keys($res['bazar']), 'intval'), $q) as $v) {
                unset($res['bazar'][$v['id']]['group_id'],
                          $res['bazar'][$v['id']]['main_id'],
                          $res['bazar'][$v['id']]['id']);
                // left: weight
                $docs['bazar-' . $v['id']] = array_merge($v, $res['bazar'][$v['id']]);
        }

        // projet
        $projet = new MoteurRecherche_PROJET();
        foreach($projet->get(array_filter(array_keys($res['projet']), 'intval'), $q) as $v) {
                unset($res['projet'][$v['id']]['group_id'],
                          $res['projet'][$v['id']]['main_id'],
                          $res['projet'][$v['id']]['id']);
                // left: weight
                $docs['projet-' . $v['id']] = array_merge($v, $res['projet'][$v['id']]);
        }

        // papyrus
        $papyrus = new MoteurRecherche_PAPYRUS();
        foreach($papyrus->get(array_filter(array_keys($res['papyrus']), 'intval'), $q) as $v) {
                unset($res['papyrus'][$v['id']]['group_id'],
                          $res['papyrus'][$v['id']]['main_id'],
                          $res['papyrus'][$v['id']]['id']);
                // left: weight
                $docs['papyrus-' . $v['id']] = array_merge($v, $res['papyrus'][$v['id']]);
        }

        // coste
        $coste = new MoteurRecherche_COSTE();
        foreach($coste->get(array_filter(array_keys($res['coste']), 'intval'), $q) as $v) {
                unset($res['coste'][$v['id']]['group_id'],
                          $res['coste'][$v['id']]['main_id'],
                          $res['coste'][$v['id']]['id']);
                // left: weight
                $docs['coste-' . $v['id']] = array_merge($v, $res['coste'][$v['id']]);
        }

        // nvjfl
        $nvjfl = new MoteurRecherche_NVJFL();
        foreach($nvjfl->get(array_filter(array_keys($res['nvjfl']), 'intval'), $q) as $v) {
                unset($res['nvjfl'][$v['id']]['group_id'],
                          $res['nvjfl'][$v['id']]['main_id'],
                          $res['nvjfl'][$v['id']]['id']);
                // left: weight
                $docs['nvjfl-' . $v['id']] = array_merge($v, $res['nvjfl'][$v['id']]);
        }

        // sort
        $sorted = _sortArrayByArray($docs, $ids_par_poids);

        $max = current($sorted);
        $max = $max['weight'];

        if(isset($_GET['tri']) && $_GET['tri'] == 'date')
                usort($sorted, '_actuNewerFirst');

        // transforme les clefs pour s'adapter aux templates existants
        array_walk($sorted, '_weight2score', $max);

        // var_dump($sorted);die;
        return $sorted;
}


/*
// http://www.php.net/manual/fr/sphinx.examples.php
$s = new SphinxClient;
$s->setServer("localhost", 9306);
$s->setMatchMode(SPH_MATCH_ANY);
$s->setMaxQueryTime(3);
var_dump($s->query("test"));
*/


/*
  Note: conversion côté client SQL:
  mysql -h0 -P 9306 < <(iconv -f utf8 -t latin1 <<<"SELECT * FROM i_projet, i_spip, i_papyrus, i_bazar WHERE MATCH('journée');")
*/

/*
  // test sorting
  // uncomment this:
  print_r(array_keys($sorted); die;

  // then:
  ddiff
        <(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) \
        <(POST http://localhost/site:reseau<<<"more_motif=test&"|awk -F'=>' '{print $2}'|sed -e 's/ //g' -e '/^$/d')
        
  // both should be equal.
  // [ SELECT main_id, group_id FROM i_projet, i_spip, i_papyrus, i_bazar WHERE MATCH('test') LIMIT 50; ]
*/
Subversion Repositories Applications.papyrus

(root)/trunk/papyrus/applettes/moteur_recherche/moteur_recherche_sphinx.php – Rev 2137
