Question

I am using SphinxQL to query Sphinxsearch engine. I want to simulate the SPH_MATCH_ANY which is implemented in the php API like this :

$cl->SetMatchMode(SPH_MATCH_ANY);
$cl->Query("test query", "index");

=> search for docs matching with “test” OR “query”

So, I have written a function (php) to replace spaces and other special chars with pipes (|) in order to use it with SphinxQL :

function formatQuery($str) {
   return trim(preg_replace('/[^-_\'a-z0-9]+/', '|', $str), ' |');
}

$str = "test query";
$sql = "SELECT * FROM index WHERE MATCH('" . addslashes(formatQuery($str)) . "')";

=> SELECT * FROM index WHERE MATCH(‘test|query’);

The problem is, for some characters like – (minus), it can break the query, example :

$str = "i-phone is great";
$sql = "SELECT * FROM index WHERE MATCH('" . addslashes(formatQuery($str)) . "')";

=> SELECT * FROM index WHERE MATCH(‘i-phone|is|great’)

=> ok

$str = "i - phone is great";
$sql = "SELECT * FROM index WHERE MATCH('" . addslashes(formatQuery($str)) . "')";

=> SELECT * FROM index WHERE MATCH(‘i|-|phone|is|great’)

=> broken query because of “|-|”

Do you know a better way to make SphinxQL queries work in SPH_MATCH_ANY mode? or a better regexp to make it works for all cases?

I know I could use a more restrictive regexp like this:

preg_replace('/[^a-z0-9]+/', '|', $str)

but it would split strings like “i-phone is great” in ‘i|phone|is|great’ and I don’t want that…

Thank you,
Nico

Answer 1

One way might be to use quorom

$sql = "SELECT * FROM index WHERE MATCH('\"" . addslashes($str) . "\"/1')";

you will need to add – to your charset_table tho, so it becomes part of a word.