optimisations diverses (?)

ecrire/inc_charsets.php3

@@ -323,7 +323,11 @@ function unicode_to_utf_8($chaine) {
 function load_windows_1251() {
 	// extrait de la table
 	// http://www.slav.helsinki.fi/atk/codepages/win1251.html
-	return array(
+	static $table;
+	if(is_array($table))
+		return $table;
+	else
+		return $table = array(
 		chr(129)=>'Ѓ',chr(131)=>'ѓ',chr(138)=>'Љ',chr(140)=>'Њ',chr(141)=>'Ќ',
 		chr(142)=>'Ћ',chr(143)=>'Џ',chr(144)=>'ђ',chr(154)=>'љ',chr(156)=>'њ',
 		chr(157)=>'ќ',chr(158)=>'ћ',chr(159)=>'џ',chr(161)=>'Ў',chr(162)=>'ў',
@@ -377,10 +381,9 @@ function translitteration ($texte, $charset='AUTO') {
 	if ($charset == 'iso-8859-1') {
 		$texte = translit_iso8859_1($texte);
 	} else if ($charset == 'windows-1251') {
-		$texte = strtolower(translit_windows_1251($texte));
+		$texte = translit_windows_1251($texte);
 	} else if ($GLOBALS['flag_iconv']) {
-		@iconv_set_encoding("internal_encoding", "UTF-8");
-		if ($iconv = @iconv($charset, 'ASCII//TRANSLIT', $texte)) {
+		if ($iconv = @iconv(strtoupper($charset), 'ASCII//TRANSLIT', $texte) && !ereg("^?+$",$iconv)) {
 			$texte = $iconv;
 		}
 	}

ecrire/inc_index.php3

@@ -7,10 +7,7 @@ define("_ECRIRE_INC_INDEX", "1");
 
 function nettoyer_chaine_indexation($texte) {
 	include_ecrire("inc_charsets.php3");
-
-	$texte = ereg_replace("<[^<]*>", "", $texte);
 	$texte = strtolower(translitteration($texte));
-
 	return $texte;
 }
 
@@ -31,11 +28,12 @@ function spip_split($reg, $texte) {
 function indexer_chaine($texte, $val = 1, $min_long = 3) {
 	global $index, $mots;
 
-	$texte = nettoyer_chaine_indexation($texte);
+	$texte = supprimer_tags($texte);
 	$regs = separateurs_indexation();
 	$texte = strtr($texte, $regs, "                                                           ");
 	$table = spip_split(" +([^ ]{0,$min_long} +)*", ' '.$texte);
 	while (list(, $mot) = each($table)) {
+		$mot = nettoyer_chaine_indexation($mot);
 		$h = substr(md5($mot), 0, 16);
 		$index[$h] += $val;
 		$mots .= ",(0x$h,'$mot')";