Although I do not trust long regular expressions…

@nerik:

I had the same problem. I solved it this way:

Replace the preg_match_all(….); stuff with

preg_match_all(’#^\s*((\['.$s.'\])|((”?)’.$s.’\\5\s*=\s{0}(”?)(.*?)\\7))\s*(;[^\n]*?)?$#ms’, @file_get_contents($iIniFile), $aMatches, PREG_SET_ORDER);

Could be done better, I’m new to regular expressions - it looked like chinese to me at first.

Many thanks.

first of all nice work!

Just one issue.
With an INI file like:

[first_section]
foo=
bar=fast

i expected an array like:
Array ( [first_section] => Array ( [foo] => (–empty/null–) [bar] => fast))

But when I run the script this is my output:
Array ( [first_section] => Array ( [foo] => bar=fast))

It seems there’s a problem if one of my key values is blank.

Can you help me ?

I made some changes to suit my needs, here’s what’s come out of it… it’s slightly verbose, I thought of leaving it as is for anyone to better inspect… it handles multiline the smarty way (with triple quotes), as the code you provided would not work with conditions like the following:

var = “blabla
This cose is gonna “break”
because of the quote on line end”

My version goes single line with both single and double quotes, and implements the multiline like this:

var = “”"
This is a smarty-like multiline variable
“”"

I didn’t implement delimiter escaping, but at this point it would be trivial.

Here it goes:

function parseDataFile($pFile) {
$aResult =
$aMatches = array();

$a = &$aResult;
$s = ‘\s*([[:alnum:]_\- \*]+?)\s*’;

$RX = ”;

$RX .= ‘[\s\n]*’; // Any space and empty lines, and discard them
$RX .= ‘(?:’;
$RX .= ‘(?:\[\s*(?P[^\n]+?)\s*\])’; // if it’s a section, match it as “section”
$RX .= ‘|’; // else…
$RX .= ‘(?:’;
$RX .= ‘(?P”?)(?P[^\n]+?)(?P=vardelim)’; // get the var name into “var” and it’s delim (should there be one) into “vardelim”
$RX .= ‘\s*=\s*’; // skip the equal sign
$RX .= ‘(?:’; // two possible layouts:
$RX .= ‘(?s:”"”\n(?P.*?)\n”"”\n)’; // multiline, delimited by “”", as in PHP ini files
$RX .= ‘|’; // or…
$RX .= ‘((?P["\']?)(?P.*?)(?P=delim))’; // single line, catching the delimiter, which could be ” or ‘
$RX .= ‘)’;
$RX .= ‘)’;
$RX .= ‘)’;
$RX .= ‘\s*(?P;[^\n]*?)?$’; // Any comment afterwards gets fetched as “comment”, just in case

preg_match_all(’#’ . $RX . ‘#m’, @file_get_contents($pFile), $aMatches, PREG_SET_ORDER);

foreach ($aMatches as $aMatch) {
if (empty($aMatch['section'])) $a[ $aMatch['var'] ] = ( !isset($aMatch['val']) ? $aMatch['multiline'] : $aMatch['val'] );
else $a = &$aResult[ $aMatch['section'] ];
}

return $aResult;
}

Any run against test-cases is very welcome, would save me debug time

Cya!

I needed it to handle Smarty-style multi-line values, like so:

longtext=”"”
this is a bunch of text
“”"

I thought it would be easy to add an additional processing block in the middle. This is what I added:

|((”?)’.$s.’\\5\s*=\s*(”"”)(.*?)”"”)

between the two existing blocks in the (|) section. I figured it would capture the multi-line values while leaving alone the standard single-quote values. Sadly, I am but a young padawan to your Jedi master regex skills, and my change killed normal single-quote handling. Sigh.

The PHP function “empty” considers that the string “0″ is empty.

The line “if (!empty($aMatch[8]))”
has to be changed to : “if (empty($aMatch[2]))”
for safer Section detection.

The code above has been updated

Thanks