Fastest way to achieve this (mine is slow)

[asmith]

Hello,

 

I'm rewriting my boards names. The original is like SMF: (first number is board id, second is board page)

example.com/index.php?board=1.0

to

example.com/name-of-board/

 

if page number is 0 (board=1.0), then page won't be in url:

example.com/name-of-board/

if page number is not (board=1.5):

example.com/name-of-board-5/

 

I have an array, with board IDs as keys and values as board names:

 

$new_names = array(
    1 => 'new-discussion',
    2 => 'feedback'
);

 

Here's the first batch I tried. It is very slow, since I have more than 200 boards:

 

<?php

$url_input = array();
$url_output = array();

foreach ($new_names as $id_board => $name)
{
    // When url has other variables. The output put ? at the end to not break other GET variables.
    $url_input[] = "'example.com/index.php?board=" . $id_board . "\.([1-9][0-9]*)[;&]'";
    $url_output[] = 'example.com/' . $name . '-$1/?';

    // When url has nothing at the end.
    $url_input[] = "'example.com/index.php?board=" . $id_board . "\.([1-9][0-9]*)'";
    $url_output[] = 'example.com/' . $name . '-$1/';

    // Should I go another round for when page number is 0? too nasty!
    $url_input[] = "'example.com/index.php?board=" . $id_board . "\.0[;&]'";
    $url_output[] = 'example.com/' . $name . '/?';

    // omg 
    $url_input[] = "'example.com/index.php?board=" . $id_board . "\.0'";
    $url_output[] = 'example.com/' . $name . '/';   
}

$content = preg_replace($url_input, $url_output, $content);
?>

 

That's 4 rule per board (x 200). 800 rules, it is taking 0.2s to 0.5s!

Any advise is greatly appreciated.

[trq]

Is there a reason your doing this in php and not mod_rewrite? It's pretty difficult to see how this is actually being used.

[asmith]

isn't mod_rewrite an Apache/webserver module for resolving friendly urls to their actual address? I'm trying to do the other way around. Make my boards friendly then use nginx and use its mod_rewrite to resolve these back.

[DavidAM]

Use the force, Luke.

 

Let the system do the work:

 

// The list of boards keyed by the Board's ID
$boards = array(1 => 'Board-A', 2 => 'Board-B', 3 => 'Board-J', 42 => 'Grok All');

// The original URL we are trying to convert
$content = 'example.com/index.php?order=date&board=42.4&search=mike';

$urlParts = parse_url($content);	// Breakup the URL

$qsParts = array();		// Breakup the Query String (if any)
if (isset($urlParts['query'])) parse_str($urlParts['query'], $qsParts);

// Convert the BoardID.Page to BoardName-Page
if (isset($qsParts['board'])) {
$boardParts = explode('.', $qsParts['board']);
$boardName = $boards[$boardParts[0]];
if ( (count($boardParts) > 1) and (!empty($boardParts[1])) ) $boardName .= '-' . $boardParts[1];
unset($qsParts['board']);
} else {
// No Board ID -- use the default
$boardName = $boards[1];
}

// Build the new URL
$newUrl = 'example.com/' . $boardName;
if (! empty($qsParts)) $newUrl .= '?' . http_build_query($qsParts);

print($content . ' => ' . $newUrl . PHP_EOL);

 

No repeating code! It handles all boards and all extra parameters in one fell swoop.

 

If any of it is not clear, feel free to ask.

 

[asmith]

Thanks for the reply.

 

It is all clear.

The $content is my example is the whole html output. My code scans the whole output and replace urls with new ones. Your code is dealing only with one url. So I assume you mean, I get all the urls in the output myself in an array, then run each of them by this function.

[Christian F.]

asmith: mod_rewrite is used to do all kinds of address manipulation and redirection from the server-side, including sending HTTP errors. It's not limited just to making SEO-friendly (fake) URLs.

[DavidAM]

The $content is my example is the whole html output. My code scans the whole output and replace urls with new ones. Your code is dealing only with one url. So I assume you mean, I get all the urls in the output myself in an array, then run each of them by this function.

 

Oops, I kind of forgot about that preg_replace. Here's another stab at it. I've turned the conversion into a function. See if you can follow this:

 

// The original Content we are trying to convert
$content = '<BODY>
<UL>
<LI><A href="example.com/index.php?board=1.0">First Board</A></LI>
<LI><A href="example.com/index.php?order=date&board=42.4&search=mike">Another</A></LI>
</UL>
<P>Try <A href="google.com">Google</A> (not replaced)</P>
</BODY>';

$findReplace = array();	// Collect old and new URLs

$matches = array();	// Array of what we find with preg_match_all
if (preg_match_all('~href=([\'"])(example\.com/[^\1]+?)\1~i', $content, $matches, PREG_PATTERN_ORDER)) {
# print_r($matches[2]);		// Testing to see what we found

foreach ($matches[2] as $oldUrl) {	// [2] is an array of the URLs (inside the quotes) found
	if (!isset($findReplace[$oldUrl])) {	// If we have not seen this one already
		$findReplace[$oldUrl] = changeUrl($oldUrl);	// Convert it to the new style
	}
}

// Now replace what we found with the new style
// str_replace is faster and we don't have any regexp's in there now, anyway
$content = str_replace(array_keys($findReplace), $findReplace, $content);
print $content;
}

exit;

function changeUrl($oldUrl) {
// The list of boards keyed by the Board's ID
static $boards = array(1 => 'Board-A', 2 => 'Board-B', 3 => 'Board-J', 42 => 'Grok All');

$urlParts = parse_url($oldUrl);	// Breakup the URL

$qsParts = array();		// Breakup the Query String (if any)
if (isset($urlParts['query'])) parse_str($urlParts['query'], $qsParts);

// Convert the BoardID.Page to BoardName-Page
if (isset($qsParts['board'])) {
	$boardParts = explode('.', $qsParts['board']);
	$boardName = $boards[$boardParts[0]];
	if ( (count($boardParts) > 1) and (!empty($boardParts[1])) ) $boardName .= '-' . $boardParts[1];
	unset($qsParts['board']);
} else {
	// No Board ID -- use the default
	$boardName = $boards[1];
}

// Build the new URL
$newUrl = 'example.com/' . $boardName;
if (! empty($qsParts)) $newUrl .= '?' . http_build_query($qsParts);

return $newUrl;
}

 

By the way, in your original post, you were building up regular expressions for preg_replace. When you do that, you need to remember to escape the regexp special characters. For instance, here is your first line and the correction below it:

 

$url_input[] = "'example.com/index.php?board=" . $id_board . "\.([1-9][0-9]*)[;&]'";
$url_input[] = "'example\.com/index\.php\?board=" . $id_board . "\.([1-9][0-9]*)[;&]'";

 

Looks like you were using the single-quote as the delimiter. I usually use tilde ("~") because it is highly unlikely that I need to include it in the regexp.

 

 

The regexp explained: ~href=([\'"])(example\.com/[^\1]+?)\1~i

 

~		# A delimiter to indicate the beginning of the pattern
href=		# A literal string "href=" to find since it usually introduces a URL in a link
(		# Start a capture group - # 1
[\'"]		# A Character class - find either a single-quote or double-quote - the single-quote is escaped because I used single-quotes for the string itself
)		# End of capture group - # 1
(		# Start a capture group - # 2
example\.com/	# A literal string "example.com/" - we had to escape the full-stop (".") because it is special to regexp
[^\1]		# A character class - ^ means NOT when used in the first position - match any character that is NOT the character found in Capture Group # 1
+?		# Repeat the preceeding match one or more times but don't be greedy about it
)		# End of capture group - # 2
\1		# Match the character found in Capture Group # 1
~		# The Delimiter marking the end of the pattern
i		# A modifier to make the matches case-INsensitive

[asmith]

Thanks DavidAM. Very detailed. I'll try this and see how much time it will take.

 

Thanks for going into trouble for me

[asmith]

Ok here's an update.

 

str_replace() was kinda messing the string, because if 2 urls had same start and if str_replace() were doing the shorter one first, the second was getting half-way done and it produced broken urls. So, I sorted urls by length first:

 

function bylength($a, $b)
{
$length1 = strlen($a);
$length2 = strlen($b);
if ($length1 == $length2)
	return 0;
return $length2 > $length1 ? 1 : -1;
}

uasort($matches[0], 'bylength');

 

Everything's so far so good and fast enough