[RESOLVED] Any regular expression gurus out there? - REG_BADRPT

[greenie2600]

I'm trying to use eregi_replace() to replace the column list in a SQL SELECT query with the string "COUNT()". In other words, the query "SELECT username, email, password FROM users" would become "SELECT COUNT() FROM users".

I've devised a regex that works beautifully as long as the query only contains one SELECT statement:

$count_query = eregi_replace( "
    ^[ \t\r\n]*     # zero or more whitespace characters
    SELECT
    [ \t\r\n]+      # one or more whitespace characters
    .+              # one or more characters
    [ \t\r\n]+      # one or more whitespace characters
    FROM
    [ \t\r\n]+      # one or more whitespace characters
    (.+)            # one or more characters - load into backreference \\1
    $/sx", "SELECT COUNT(*) FROM \\1", $original_query );

However, I'm having problems with queries that contain SQL subselects. For example, the query "SELECT foo FROM bar WHERE baz IN ( SELECT bop FROM yyz )" produces "SELECT COUNT() FROM yyz )", which is not what I want. (It should be "SELECT COUNT() FROM bar WHERE baz IN ( SELECT bop FROM yyz )".)

I know this has to do with greediness, but that's one aspect of regular expressions that I don't know very well. At first, I thought all I had to do was change the line that reads ".+" to ".+?" (thus making that match ungreedy), but that produces a REG_BADRPT warning.

I putzed around with a few other ideas, but nothing is working. Can some kindly soul show me the error of my ways?

[NogDog]

I'm more comfortable with the preg_ functions, so I came up with this:

$regex = '/^\s*SELECT\s.+\sFROM/isU';
$replace = 'SELECT COUNT(*) FROM';
echo preg_replace($regex, $replace, $query);

Note: the "i" modifier makes it case-insensitive, the "s" makes "." include newlines, and the "U" makes it ungreedy.

[greenie2600]

Wow...it works beautifully, and it's much more concise than my version. That \s metacharacter certainly makes life easier. Thanks!