Debugging a recursive while loop (php) - php

Question

I'm writing a function to take the tags from a mustache template and generate a hash (the reason for this is to be able to take any given template and quickly show a developer what the expected variables are).
I extract the tags into a flat array (easy enough), but the next step is tricky - I need to turn the flat array into a multi-dimensional array to indicate nested variable.
Here's my sample flat array:
$arr = array(
'one',
'#two',
'sub1',
'sub2',
'/two',
'three'
);
And the expected output:
$newArray = array(
'one'=>'',
'two'=>array(
'sub1'=>'',
'sub2'=>''
),
'three'=>''
);
I have been getting close, but am not quite there yet. I thought a recursive function would be the way to go (though I am open to a different solution). Here is what I have so far:
function recurse($array, $i = 0) {
$nested = array();
while ($i < count($array)):
$tag = $array[$i];
if (preg_match('/\//',$tag)) {
return $nested;
} elseif (preg_match('/^#/',$tag)) {
$tag = str_replace('#','',$tag);
$nested[$tag] = recurse($array, $i+1);
$i+= count($nested[$tag])+1;
} else {
$nested[$tag] = '';
$i++;
}
endwhile;
return $nested;
}
I think the bug may be that it hits the first 'if' and returns all the way out of the function, but I'm not certain, nor am I sure how to fix it.

Answer 1

Just for fun I decided to make you one without recursion and using references instead (more efficient that recursion, storing array element aliases on a stack). Works with nested subsets too.
$arr = array(
'one',
'#two','sub1',
'#twotwo','sub1','sub2','/twotwo',
'sub2','/two',
'three'
);
$out = array();
$stack = array();
$sp = 0;
$stack[$sp] = &$out;
foreach ($arr as $item) {
$cur =& $stack[$sp];
if ($item[0] == '#') {
$item = substr($item, 1);
$cur[$item] = array();
$stack[++$sp] = &$cur[$item];
}
elseif ($item[0] == '/') {
$sp--;
}
else {
$cur[] = $item;
}
}
var_dump($out);
Output:
array
0 => string 'one' (length=3)
'two' => &
array
0 => string 'sub1' (length=4)
'twotwo' => &
array
0 => string 'sub1' (length=4)
1 => string 'sub2' (length=4)
1 => string 'sub2' (length=4)
1 => string 'three' (length=5)
You can ignore the fact in the output you see & array in places instead of simply array. This signifies that in the symbol table the reference count for that particular element is > 1.
The reason for this is that $stack is still maintaining a reference. If you do an unset($stack); before returning the output, the additional references are removed and the &s in the output will disappear.

Answer 2

I modified your function a bit to match your needs, see if it works for you:
$arr = array(
'one',
'#two',
'sub1',
'#sub2',
'subsub1',
'subsub2',
'subsub3',
'subsub4',
'/sub2',
'sub3',
'/two',
'three'
);
function recurse($array, &$i, $current_tag = "")
{
$nested = array();
while ($i < count($array)):
$tag = $array[$i];
if ($tag == '/'.$current_tag)
{
$i++;
return $nested;
}
elseif (preg_match('/^#/',$tag))
{
$tag = str_replace('#','',$tag);
$i++;
$nested[$tag] = recurse($array, $i, $tag);
} else
{
$nested[$tag] = '';
$i++;
}
endwhile;
return $nested;
}
$i = 0;
$a = recurse($arr, $i);
echo '<pre>'.print_r($a, true).'</pre>';
You had some issues with that $i... I gave it as reference so that it will automatically update with the function system, and used another parameter to match exactly the next closing tag..., so that it will validate.

Answer 3

Yes, recurse function is the way. Some advices :
Do not include "count" functions in loops when you have not to do (your "$array" is not updated, so his size still the same from the begening to the end)
Do not use preg_match when you have simple comparison to do.
Use references, else you should quickly get a memory error with huge arrays used in recurse functions.
Here an other way to do what you want to :
<?php
function recurse(&$array, &$return = array(), &$i = 0, $limit = NULL)
{
if(!isset($limit)){
$limit = count($array) ;
}
for(;$i < $limit;$i++){
if($array[$i]{0} == '#'){
//opening
$key = substr($array[$i++], 1) ;
$return[$key] = array();
recurse($array, $return[$key], $i, $limit) ;
}elseif($array[$i]{0} == '/'){
return ;
}else{
//same level
$return[$array[$i]] = '';
}
}
}
$arr = array(
'one',
'#two',
'sub1',
'#t2',
'sub1.1',
'sub1.2',
'/t2',
'sub2',
'/two',
'three'
);
$nested = array();
recurse($arr, $nested);
var_dump($nested);
?>

Answer 4

This might be more of what you are looking for (and a little more closer to true recursion), but I didn't test it because I don't have a PHP instance to work off of at the moment
Usage:
$input = array(
'one',
'#two',
'sub1',
'sub2',
'/two',
'three'
);
$result = array();
recurse($input, $result, '', 0);
Steps:
If the position is greater than the array count, we are done.
If we need to go back up to root, remove tag and call again
If we need to go into a tag, add tag and call again
If we are in root, add the key and blank entry
If we are in a tag, add the key to the tag with a blank entry
Code:
function recurse($input, &$result, $tag, $position)
{
if($position >= count($input))
{
return;
}
if(preg_match('#\/#',$input[$position]))
{
recurse($input, $result, '', $position + 1);
}
else if (preg_match('#^##',$input[$position]))
{
$result[substr($input[$position], 1)] = array();
recurse($input, $result, substr($input[$position], 1), $position + 1);
}
else if($tag == '')
{
$result[$input[$position]] = '';
recurse($input, $result, $tag, $position + 1);
}
else
{
$result[$tag][$input[$position]] = '';
recurse($input, $result, $tag, $position + 1);
}
}

Answer 5

Off by one error
$tag = str_replace('#','',$tag);
$nested[$tag] = recurse($array, $i+1);
$i+= count($nested[$tag])+1;
When you return the nested array, you have to skip over the closing tag, so it should be $i += count($nested[$tag]) + 2;.