public function buildDictionarySql()
{
$app = \Slim\Slim::getInstance();
foreach (['freq_threshold', 'suggest_dubug', 'length_threshold', 'levenshtein_threshold', 'top_count'] as $var) {
define(strtoupper($var), intval($app->config['sphinx'][$var]));
}
$in = fopen("php://stdin", "r");
$out = fopen("php://stdout", "w+");
$used_keywords = array();
fwrite($out, "TRUNCATE suggest;\n");
$n = 0;
$m = 0;
while ($line = fgets($in, 1024)) {
list($keyword, $freq) = explode(" ", trim($line));
$keyword = trim($keyword);
if (strlen($keyword) < 2 || $keyword === '' || $freq < FREQ_THRESHOLD || strstr($keyword, "_") !== FALSE || strstr($keyword, "'") !== FALSE || array_key_exists($keyword, $used_keywords) === TRUE) {
continue;
}
$trigrams = buildTrigrams($keyword);
$used_keywords[$keyword] = NULL;
fwrite($out, !$m ? "INSERT INTO suggest VALUES\n" : ",\n");
$n++;
fwrite($out, "( 0, '{$keyword}', '{$trigrams}', {$freq} )");
$m++;
if ($m % 10000 == 0) {
fwrite($out, ";\n");
$m = 0;
}
}
if ($m) {
fwrite($out, ";");
}
fwrite($out, "\n");
$app->stop();
}
function MakeSuggestion($keyword, $ln)
{
$trigrams = buildTrigrams($keyword);
$query = "\"{$trigrams}\"/1";
$len = strlen($keyword);
$delta = LENGTH_THRESHOLD;
$stmt = $ln->prepare("\n\t\t\tSELECT *, weight() as w, w+:delta-ABS(len-:len) as myrank\n\t\t\tFROM slimpdsuggest\n\t\t\tWHERE MATCH(:match) AND len BETWEEN :lowlen AND :highlen\n\t\t\tORDER BY myrank DESC, freq DESC\n\t\t\tLIMIT 0,:topcount OPTION ranker=wordcount");
$stmt->bindValue(':match', $query, PDO::PARAM_STR);
$stmt->bindValue(':len', $len, PDO::PARAM_INT);
$stmt->bindValue(':delta', $delta, PDO::PARAM_INT);
$stmt->bindValue(':lowlen', $len - $delta, PDO::PARAM_INT);
$stmt->bindValue(':highlen', $len + $delta, PDO::PARAM_INT);
$stmt->bindValue(':topcount', TOP_COUNT, PDO::PARAM_INT);
$stmt->execute();
if (!($rows = $stmt->fetchAll())) {
return false;
}
// further restrict trigram matches with a sane Levenshtein distance limit
foreach ($rows as $match) {
$suggested = $match["keyword"];
if (levenshtein($keyword, $suggested) <= LEVENSHTEIN_THRESHOLD) {
return $suggested;
}
}
return $keyword;
}