Subversion Repositories eFlore/Projets.eflore-projets

Rev

Rev 867 | Go to most recent revision | Show entire file | Ignore whitespace | Details | Blame | Last modification | View Log | RSS feed

Rev 867 Rev 876
Line 12... Line 12...
12 
 */
 12 
 */
Line 13... Line 13...
13 
 
 13 
 
14 
// time php -d memory_limit=1024M sphinx-maj-nom-ret.php 0 > sphinx-maj.log
 14 
// time php -d memory_limit=1024M sphinx-maj-nom-ret.php 0 > sphinx-maj.log
Line -... Line 15...
- 
 
 15 
// 23 secondes
- 
 
 16 
 
- 
 
 17 
// settings
- 
 
 18 
define('USE_NVJFL', FALSE);
- 
 
 19 
define('ESCAPE_ON_SPHINX_SYNERROR', TRUE);
- 
 
 20 
 
- 
 
 21 
define('TRY_FORCE_START_LINE', TRUE);
- 
 
 22 
define('TRY_SPLIT',			   TRUE);
- 
 
 23 
define('TRY_EXACT',			   TRUE);
- 
 
 24 
define('TRY_REF',			   TRUE);
- 
 
 25 
define('TRY_SPLIT_AND_AUTEUR', FALSE);
- 
 
 26 
define('TRY_REMOVE_L',		   TRUE);
- 
 
 27 
 
- 
 
 28 
define('M_TRY_SPLIT',	0x01);
- 
 
 29 
define('M_TRY_EXACT',	0x02);
- 
 
 30 
define('M_TRY_REF',		0x04);
15 
// 23 secondes
 31 
define('M_TRY_SPLIT_AND_AUTEUR',	0x08);
16 
 
 32 
 
17 
error_reporting(E_ALL);
 33 
error_reporting(E_ALL);
18 
$db = mysql_connect('localhost', 'root', '');
 34 
$db = mysql_connect('localhost', 'root', '') or die('no mysql');
19 
mysql_select_db('tb_cel', $db);
 35 
mysql_select_db('tb_cel', $db);
Line 20... Line 36...
20 
mysql_query("SET NAMES utf8", $db);
 36 
mysql_query("SET NAMES utf8", $db) or die('no sphinx');
21 
$dbs = mysql_connect('127.0.0.1:9306', NULL, NULL, TRUE);
 37 
$dbs = mysql_connect('127.0.0.1:9306', NULL, NULL, TRUE);
22 
 
 38 
 
23 
$req = <<<EOF
 39 
$req = <<<EOF
24 
	SELECT id_observation, nom_sel, nom_sel_nn,nom_ret,nom_ret_nn,nt,famille
 40 
	SELECT id_observation, nom_sel, nom_sel_nn, nom_ret, nom_ret_nn, nt, famille, nom_referentiel
25 
        FROM `cel_obs`
 41 
        FROM `cel_obs`
26 
        WHERE nom_sel IS NOT NULL AND nom_sel != '' AND
 42 
        WHERE nom_sel IS NOT NULL AND nom_sel != '' AND
Line 51... Line 67...
51 
$stats = ['no_nom_sel' => ['count' => 0, 'data' => [] ],
 67 
$stats = ['no_nom_sel' => ['count' => 0, 'data' => [] ],
52 
		  'not found' => ['count' => 0, 'data' => [] ],
 68 
		  'not found' => ['count' => 0, 'data' => [] ],
53 
		  'too many' => ['count' => 0, 'data' => [] ],
 69 
		  'too many' => ['count' => 0, 'data' => [] ],
54 
		  'fixable' => ['count' => 0, 'data' => [] ],
 70 
		  'fixable' => ['count' => 0, 'data' => [] ],
55 
		  'sauvages' => ['count' => 0, 'data' => [] ],
 71 
		  'sauvages' => ['count' => 0, 'data' => [] ],
56 
		  'sphinx errors' => ['count' => 0, 'data' => [] ], ];
 72 
		  'sphinx errors' => ['count' => 0, 'data' => [] ],
- 
 
 73 
		  'ref pb' => ['count' => 0, 'data' => [] ], ];
Line -... Line 74...
- 
 
 74 
 
Line 57... Line 75...
57 
 
 75 
$sphinx_req = sprintf("SELECT * FROM i_bdtfx %s WHERE MATCH('%%s') LIMIT 5", USE_NVJFL ? ", i_nvjfl" : "");
58 
 
 76 
 
59 
for($current = 0; $current < intval($max/$chunk_size) + 1; $current++) {
 77 
for($current = 0; $current < intval($max/$chunk_size) + 1; $current++) {
60 
	// printf("current = %d, chunk_size = %d, max = %d (rmax = %d) [real limit: %d]\n", $current, $chunk_size, $max, intval($max/$chunk_size) + 1, $current*$chunk_size);
 78 
	// printf("current = %d, chunk_size = %d, max = %d (rmax = %d) [real limit: %d]\n", $current, $chunk_size, $max, intval($max/$chunk_size) + 1, $current*$chunk_size);
61 
	// printf(strtr($req, "\n", " ") . "\n", $start, $max, $current*$chunk_size, $chunk_size);
 79 
	// printf(strtr($req, "\n", " ") . "\n", $start, $max, $current*$chunk_size, $chunk_size);
62 
	$data = mysql_query(sprintf($req, $start, $max, $current*$chunk_size, $chunk_size), $db);
 80 
	$data = mysql_query(sprintf($req, $start, $max, $current*$chunk_size, $chunk_size), $db);
63 
	if(!$data) { var_dump(mysql_error()); die('end'); }
 81 
	if(!$data) { var_dump(mysql_error()); die('end'); }
- 
 
 82 
	while($d = mysql_fetch_assoc($data)) {
- 
 
 83 
		$n  = trim($d['nom_sel']);
64 
	while($d = mysql_fetch_assoc($data)) {
 84 
		//d: fprintf(STDERR, "$n\n");
65 
		$n  = $d['nom_sel'];
 85 
 
66 
		if(!$n) {
 86 
		if(!$n) {
67 
			$stats['no_nom_sel']['count']++;
 87 
			$stats['no_nom_sel']['count']++;
68 
			// $stats['no_nom_sel']['data'][] = [$d['id_observation'], $n];*/
 88 
			// $stats['no_nom_sel']['data'][] = [$d['id_observation'], $n];*/
Line 74... Line 94...
74 
			$stats['sauvages']['count']++;
 94 
			$stats['sauvages']['count']++;
75 
			// $stats['sauvages']['data'][] = [$d['id_observation'], $n];
 95 
			// $stats['sauvages']['data'][] = [$d['id_observation'], $n];
76 
			continue;
 96 
			continue;
77 
		}
 97 
		}
Line -... Line 98...
- 
 
 98 
 
- 
 
 99 
		$MASQUE = 0;
- 
 
 100 
 
- 
 
 101 
		if(TRY_REMOVE_L) {
- 
 
 102 
			$n = str_replace(' L.','', $n);
- 
 
 103 
		}
- 
 
 104 
 
- 
 
 105 
		$orig_n = $n;
78 
 
 106 
 
- 
 
 107 
	recherche:
- 
 
 108 
		if(TRY_FORCE_START_LINE && !_has($MASQUE, M_TRY_EXACT)) {
- 
 
 109 
			$n = '^' . $n;
- 
 
 110 
		}
79 
	recherche:
 111 
 
- 
 
 112 
		$s = mysql_query(sprintf($sphinx_req, $n), $dbs);
- 
 
 113 
 
- 
 
 114 
 
80 
		//$s = mysql_query("SELECT * FROM i_bdtfx WHERE MATCH('" . str_replace($from,$to,$n) . "') LIMIT 5", $dbs);
 115 
		if(!$s && ESCAPE_ON_SPHINX_SYNERROR) {
- 
 
 116 
			$s = mysql_query(sprintf($sphinx_req, str_replace($from,$to,$n)), $dbs);
81 
		$s = mysql_query("SELECT * FROM i_bdtfx, i_nvjfl WHERE MATCH('" . $n . "') LIMIT 5", $dbs);
 117 
		}
82 
		if(!$s) {
 118 
		if(!$s) {
83 
			$stats['sphinx errors']['count']++;
 119 
			$stats['sphinx errors']['count']++;
84 
			// $stats['sphinx errors']['data'][] = [$d['id_observation'], $n];
 120 
			// $stats['sphinx errors']['data'][] = [$d['id_observation'], $orig_n];
85 
			continue;
 121 
			continue;
Line 86... Line 122...
86 
		}
 122 
		}
- 
 
 123 
 
- 
 
 124 
		$c = mysql_num_rows($s);
87 
 
 125 
		//d: fprintf(STDERR, "\t search [nb:%d] \"%s\" (msk:%d)\n", $c, $n, $MASQUE);
- 
 
 126 
 
- 
 
 127 
		if($c == 0) {
- 
 
 128 
			if(TRY_SPLIT && !_has($MASQUE, M_TRY_SPLIT)) {
- 
 
 129 
				require_once('lib-split-auteur.php');
- 
 
 130 
				$MASQUE |= M_TRY_SPLIT;
- 
 
 131 
				// $n = RechercheInfosTaxonBeta::supprimerAuteur($orig_n);
- 
 
 132 
				// list($ret, $m) = RechercheInfosTaxonBeta::contientAuteur($orig_n);
- 
 
 133 
				$ret = RechercheInfosTaxonBeta::supprimerAuteurBis($orig_n, $m);
- 
 
 134 
				if($ret) {
- 
 
 135 
					// printf("===================== SPLIT: contientAuteur \"%s\" [@%s @%s)\n", $orig_n, $ret, $m);
- 
 
 136 
					$n = sprintf('%s @auteur %s', $ret, $m);
- 
 
 137 
					goto recherche;
- 
 
 138 
				}
- 
 
 139 
			}
- 
 
 140 
			if(TRY_SPLIT_AND_AUTEUR && !_has($MASQUE, M_TRY_SPLIT_AND_AUTEUR) && strpos($orig_n, ' ') !== FALSE) {
- 
 
 141 
				require_once('lib-split-auteur.php');
- 
 
 142 
				$MASQUE |= M_TRY_SPLIT_AND_AUTEUR;
- 
 
 143 
				$ns = RechercheInfosTaxonBeta::supprimerAuteur($orig_n);
- 
 
 144 
				if($ns) {
- 
 
 145 
					$a = trim(substr($orig_n, strlen($n)));
- 
 
 146 
					$n = sprintf("%s @auteur %s", $ns, $a);
- 
 
 147 
					// echo "===================== SPLIT N/A: $n\n";
- 
 
 148 
					goto recherche;
- 
 
 149 
				}
88 
		$c = mysql_num_rows($s);
 150 
			}
89 
		if($c == 0) {
 151 
 
90 
			$stats['not found']['count']++;
 152 
			$stats['not found']['count']++;
91 
			// $stats['not found']['data'][] = [$d['id_observation'], $n];
 153 
			// $stats['not found']['data'][] = [$d['id_observation'], $orig_n];
Line 92... Line 154...
92 
			continue;
 154 
			continue;
- 
 
 155 
		}
- 
 
 156 
 
- 
 
 157 
		if($c > 1) {
- 
 
 158 
 
- 
 
 159 
			if($c == 2) {
- 
 
 160 
				if(mysql_fetch_array($s)['group_id'] !=
- 
 
 161 
				   mysql_fetch_array($s)['group_id']) {
- 
 
 162 
					// recherche donne seulement 2 résultats dans 2 référentiels
- 
 
 163 
					// potentiellement fixable si l'on peut se référer à $d['nom_referentiel']
- 
 
 164 
					$stats['ref pb']['count']++;
- 
 
 165 
					// $stats['ref pb']['data'][] = [$d['id_observation'], $orig_n];
- 
 
 166 
					continue;
- 
 
 167 
				}
- 
 
 168 
			}
- 
 
 169 
 
- 
 
 170 
			if(TRY_EXACT && !_has($MASQUE, M_TRY_EXACT)) {
- 
 
 171 
				$MASQUE |= M_TRY_EXACT;
- 
 
 172 
				$n = '"^' . trim($orig_n) . '$"';
- 
 
 173 
				goto recherche;
- 
 
 174 
			}
- 
 
 175 
			if(TRY_REF && isset($d['nom_referentiel']) && !_has($MASQUE, M_TRY_REF)) {
- 
 
 176 
				$MASQUE |= M_TRY_REF;
- 
 
 177 
				$n = $orig_n . ' @group_id ' . $d['nom_referentiel'];
93 
		}
 178 
				goto recherche;
94 
 
 179 
			}
95 
		if($c > 1) {
 180 
 
96 
			$stats['too many']['count']++;
 181 
			$stats['too many']['count']++;
Line -... Line 182...
- 
 
 182 
			// $stats['too many']['data'][] = [$d['id_observation'], $orig_n];
- 
 
 183 
			continue;
97 
			// $stats['too many']['data'][] = [$d['id_observation'], $n];
 184 
		}
98 
			continue;
 185 
 
Line 99... Line 186...
99 
		}
 186 
 
100 
 
 187 
	ok:
Line -... Line 188...
- 
 
 188 
		$stats['fixable']['count']++;
- 
 
 189 
		// $stats['fixable']['data'][] = [$d['id_observation'], $orig_n];
- 
 
 190 
 
- 
 
 191 
	}
- 
 
 192 
}
101 
		$stats['fixable']['count']++;
 193 
 
102 
		// $stats['fixable']['data'][] = [$d['id_observation'], $n];
 194 
function _has($v, $r) {
103 
 
 195 
	return ($v & $r) == $r;