/**
* @param nc_search_query_expression_phrase $expression
* @return string
*/
protected function translate_phrase(nc_search_query_expression_phrase $expression)
{
$this->stack[] = "phrase";
// Given a phrase "one two", where "two" has forms of "two1" and "two2":
// - added to the main FTS condition: ((+one +two1) (+one +two2))
// (this query is used to fetch rows which contain these words using a FTS index)
// - added to the query: REGEXP condition(s) to fetch only rows where these terms
// come in sequence (REGEXP is used because there could be several
// variants of the same term in the index in the form "one two1|two2")
$term_combinations = array();
$regexps = array();
/** @var $item nc_search_query_expression_term */
foreach ($expression->get_items() as $item) {
// "for each term in the phrase"
$item_codes = $this->process_term($item);
if ($item_codes[0]) {
// not a stop word
$brackets = $this->get_brackets(count($item_codes) > 1);
$regexps[] = $brackets[0] . join("|", array_filter($item_codes)) . $brackets[1];
}
if (count($term_combinations) == 0) {
// it's a first (meaningful) term
$term_combinations = $item_codes;
} else {
// produce all possible combinations
$previous_state = $term_combinations;
$term_combinations = array();
foreach ($previous_state as $p) {
foreach ($item_codes as $c) {
$not_empty = strlen($c) > 0;
// might be a stop word
$term_combinations[] = $p . ($not_empty ? " {$c}" : "");
}
}
}
}
// there can be a space in the beginning if the first term is stop word
if ($term_combinations && $term_combinations[0][0] == " ") {
$term_combinations = array_map('trim', $term_combinations);
}
// all terms in the phrase are stop words?!
if (!$term_combinations || !$term_combinations[0]) {
return "";
}
// make FTS query to pre-filter results before running regexp match
$fts_query = "";
if (!$expression->is_excluded()) {
$brackets = $this->get_brackets(count($term_combinations) > 1);
$fts_query = "{$brackets['0']}(+" . join(")(+", $term_combinations) . "){$brackets['1']}";
$fts_query = strtr($fts_query, array(" " => " +", ")(" => ") ("));
}
// add exact sequence condition
$table_names = $this->get_table_names($expression);
// [.vertical-line.] is not redundant: it prevents from matching a partial code
$term_boundary = "([[.vertical-line.]][^[.space.][.slash.]]+)?[[.space.]]";
$skip_regexps = false;
$num_regexps = count($regexps);
if ($expression->get_distance()) {
// it's a "proximity search"
if ($num_regexps < 2 || $num_regexps > (int) nc_search::get_setting("DatabaseIndex_MaxProximityTerms")) {
// the proximity phrase is too long (or too short)!
$skip_regexps = true;
} else {
// get all possible permutations
$skipped_term = "[^[.space.][.slash.]]+[[.space.]]";
$distance = min((int) nc_search::get_setting("DatabaseIndex_MaxProximityDistance"), $expression->get_distance());
$gaps = $distance == 1 ? $term_boundary : "{$term_boundary}({$skipped_term}){0,{$distance}}";
$regexps = $this->permute($regexps, $gaps);
}
} else {
// ordinary phrase, distance=0
if ($num_regexps > 1) {
$regexps = join($term_boundary, $regexps);
} else {
$skip_regexps = true;
}
}
if (!$skip_regexps) {
$this->query_builder->add_field_regexps($table_names, $regexps, $expression->is_excluded());
} elseif ($expression->get_field()) {
$this->add_field_matches($expression, $fts_query);
}
// remove "phrase" from the stack
array_pop($this->stack);
// done
return $fts_query;
}
/**
* @param nc_search_query_expression_phrase $expression
* @return void
*/
protected function translate_phrase(nc_search_query_expression_phrase $expression)
{
$terms = array();
foreach ($expression->get_items() as $term) {
$terms[] = $term->get_value();
}
$string = 'PHRASE "' . join(" ", $terms) . '"' . $this->get_modifiers($expression);
$this->print_line($string);
}
