How can I loop through an array, split it into two arrays and run a function for each possible combination? Order does not matter.
// Original array
$array = array('a','b','c','d','e');
// Result 1
array('a');
array('b','c','d','e');
// Result 2
array('a', 'b');
array('c','d','e');
// Result 3
array('a', 'c');
array('b','d','e');
And so on...
Here's my take at this:
<?php
$ar = ['a','b','c','d','e'];
function permuteThrough($ar, $callback, $allowMirroredResults = true, $mode = 'entry', $desiredLeftCount = null, $left = [], $right = [])
{
switch($mode)
{
case 'entry':
// Logic:
// Determine how many elements we're gonna put into left array
$len = $allowMirroredResults ? count($ar) : floor(count($ar)/2);
for($i=0; $i <= $len; $i++)
{
call_user_func(__FUNCTION__, $ar, $callback, $allowMirroredResults, 'permute',$i);
}
break;
case 'permute':
// We have nothing left to sort, let's tell our callback
if( count($ar) === 0 )
{
$callback($left,$right);
break;
}
if( count($left) < $desiredLeftCount )
{
// Note: PHP assigns arrays as clones (unlike objects)
$ar1 = $ar;
$left1 = $left;
$left1[] = current(array_splice($ar1,0,1));
call_user_func(__FUNCTION__, $ar1, $callback, $allowMirroredResults, 'permute', $desiredLeftCount, $left1, $right);
}
// This check is needed so we don't generate permutations which don't fulfill the desired left count
$originalLength = count($ar) + count($left)+count($right);
if( count($right) < $originalLength - $desiredLeftCount )
{
$ar2 = $ar;
$right1 = $right;
$right1[] = current(array_splice($ar2,0,1));
call_user_func(__FUNCTION__, $ar2, $callback, $allowMirroredResults, 'permute', $desiredLeftCount, $left, $right1);
}
break;
}
}
function printArrays($a,$b)
{
echo '['.implode(',',$a).'],['.implode(',',$b)."]\n";
}
permuteThrough($ar, 'printArrays', true); // allows mirrored results
/*
[],[a,b,c,d,e]
[a],[b,c,d,e]
[b],[a,c,d,e]
[c],[a,b,d,e]
[d],[a,b,c,e]
[e],[a,b,c,d]
[a,b],[c,d,e]
[a,c],[b,d,e]
[a,d],[b,c,e]
[a,e],[b,c,d]
[b,c],[a,d,e]
[b,d],[a,c,e]
[b,e],[a,c,d]
[c,d],[a,b,e]
[c,e],[a,b,d]
[d,e],[a,b,c]
[a,b,c],[d,e]
[a,b,d],[c,e]
[a,b,e],[c,d]
[a,c,d],[b,e]
[a,c,e],[b,d]
[a,d,e],[b,c]
[b,c,d],[a,e]
[b,c,e],[a,d]
[b,d,e],[a,c]
[c,d,e],[a,b]
[a,b,c,d],[e]
[a,b,c,e],[d]
[a,b,d,e],[c]
[a,c,d,e],[b]
[b,c,d,e],[a]
[a,b,c,d,e],[]
*/
echo "==============\n"; // output separator
permuteThrough($ar, 'printArrays', false); // doesn't allow mirrored results
/*
[],[a,b,c,d,e]
[a],[b,c,d,e]
[b],[a,c,d,e]
[c],[a,b,d,e]
[d],[a,b,c,e]
[e],[a,b,c,d]
[a,b],[c,d,e]
[a,c],[b,d,e]
[a,d],[b,c,e]
[a,e],[b,c,d]
[b,c],[a,d,e]
[b,d],[a,c,e]
[b,e],[a,c,d]
[c,d],[a,b,e]
[c,e],[a,b,d]
[d,e],[a,b,c]
*/
My permuteThrough function takes three arguments.
The array, the callback, and an optional boolean indicating whether or not you want to allow mirrored results.
The logic is pretty straight forward:
First decide how many elements we want to put into our left array.
Then Recursively call the function like so:
Our working array with the remaining elements to sort.
Shift an element off and put it into the left array. The result gets sent to another layer of recursion.
Shift an element off and put it into the right array. The result gets sent to another layer of recursion.
If there's no elements left to shift off, call our callback with the resulting left and right arrays.
Finally, make sure we're not going over the desired left array element size determined by the for loop at the beginning and make sure the right size doesn't become so big that it makes the desired left size impossible to satisfy. Normally this would be done by two separate functions. One to decide how many elements should go into the left array. And one to be used for recursion. But instead I tossed in another argument to the recursion function to eliminate the need for a separate function so it can all be handled by the same recursive function.
This is the best I can do:
class Combos
{
/**
* getPossible then getDivide
*
* #param array $input
* #return array
*/
public function getPossibleAndDivided( array $input )
{
return $this->getMultiShiftAndDivided( $this->getPossible( $input ) );
}
/**
* return all possible combinations of input
*
* #param array $input
* #return array
*/
public function getPossible( array $inputs )
{
$result = [];
if ( count( $inputs ) <= 1 ) {
$result = $inputs;
} else {
$result = array();
foreach($inputs as $input){
//make it an array
$first = [ $input ];
//get all inputs not in first
$remaining = array_diff( $inputs, $first );
//send the remaining stuff but to ourself
$combos = $this->getPossible( $remaining );//recursive
//iterate over the above results (from ourself)
foreach( $combos as $combo ) {
$last = $combo;
//convert to array if it's not
if( !is_array( $last ) ) $last = [ $last ];
//merge them
$result[] = array_merge( $first, $last );
}
}
}
return $result;
}
/**
* shift and divide a multi level array
*
* #param array $array
* #return array
*/
public function getMultiShiftAndDivided( array $mArray )
{
$divided = [];
foreach ( $mArray as $array ) {
$divided = array_merge($divided, $this->getShiftAndDivided( $array ));
}
return $divided;
}
/**
* shift and divide a single level array
*
* #param array $array
* #return array
*/
public function getShiftAndDivided( array $array )
{
$divided = [];
$array1 = [];
while( count( $array ) ){
$array1[] = array_shift( $array );
$divided[] = [ $array, $array1 ];
}
return $divided;
}
}
How it works
Top level, I don't want to get into all the details. This is basically a 2 step process or at least it was easier to figure it out that way. I built it in a class to keep everything neat. It also allows better unit testing and re-usability.
This requires 2 operations, or at least it was easier for me to do it in 2 instead of as 1. They are combined in this method
public function getPossibleAndDivided( array $input )
{
return $this->getMultiShiftAndDivided( $this->getPossible( $input ) );
}
This is the main reason I made it a class, to keep everything packaged nicely together. I would call this a wrapper method.
Step One
$Combos->getPossible(array $input)
if only one item remains, return $inputs.
this is all the combinations it can ever have (its just a single element after all).
else It iterates thought $inputs with foreach as $input
Wraps $input as an array named $first
this is a single element from $inputs
Gets the remaining elements in $inputs in a non-destructive way as $remaining
using array_diff we need both elements as arrays (see aabove)
recursive call to $this->getPossible($remaining) (see #1) and returns as $combos
Iterates over $combos foreach as $combo
$combo is assigned to $last and turned into an array, if its not
sometimes combo is an array with several elements
sometimes is a single element. It depends on the recursive call
We add to our result set the merger of $first and $last
we need both as arrays so we can merge, without causing nesting
End Anything in $result is returned.
This basically rotates though all the combinations of the array and returns them in an array like this:
Array
(
[0] => Array
(
[0] => a
[1] => b
[2] => c
[3] => d
[4] => e
)
[1] => Array
(
[0] => a
[1] => b
[2] => c
[3] => e
[4] => d
)
[2] => Array
(
[0] => a
[1] => b
[2] => d
[3] => c
[4] => e
)
...
[117] => Array
(
[0] => e
[1] => d
[2] => b
[3] => c
[4] => a
)
[118] => Array
(
[0] => e
[1] => d
[2] => c
[3] => a
[4] => b
)
[119] => Array
(
[0] => e
[1] => d
[2] => c
[3] => b
[4] => a
)
)
Yes it returns 119 results, no I will not include them all.
Step Two
Don't forget the above output is a multi-dimensional array (this is important below).
$Combos->getMultiShiftAndDivided(array $mArray)
This method is intended to be used with multi-dimensional arrays (hence its name). We get this from "Step 1". Its basically a wrapper for $Combos->getShiftAndDivided($array)
Iterates over $mArray foreach as $array
It calls $this->getShiftAndDivided($array) returns and merges with $divided.
there was no need to store the results, so I didn't waste a variable on it.
Output example:
$input = array(array('a','b','c','d','e'));
print_r($combos->getMultiShiftAndDivided($input));
Array
(
[0] => Array
(
[0] => Array
(
[0] => b
[1] => c
[2] => d
[3] => e
)
[1] => Array
(
[0] => a
)
)
....
[4] => Array
(
[0] => Array
(
)
[1] => Array
(
[0] => a
[1] => b
[2] => c
[3] => d
[4] => e
)
)
)
$Combos->getShiftAndDivided(array $array)
This method is intended to be used single level arrays.
Loops as long as the count of $array is more then 0, while loop
$array1 gets the first element from $array added and that element is removed from $array (destructively)
we store both $array and $array1 in our results $divided
this records there current "state" at that moment
when there are no more items in $array we return our results
Output example:
$input = array('a','b','c','d','e');
print_r($combos->getShiftAndDivided($input));
Array
(
[0] => Array
(
[0] => Array
(
[0] => b
[1] => c
[2] => d
[3] => e
)
[1] => Array
(
[0] => a
)
)
....
[4] => Array
(
[0] => Array
(
)
[1] => Array
(
[0] => a
[1] => b
[2] => c
[3] => d
[4] => e
)
)
)
Basically this shifts the elements of a single array to two result arrays and records their state on each shift. I made it 2 functions so that it could be tested easier and be re-used easier.
Another issue is its kind of hard to check for multi-dimensional arrays. I know how to do it, but I didn't feel like it because it's kind of ugly and there is a better way. I say that because its possible to use a one level array in what is getMultiShiftAndDivided and it wouldn't give you what you would expect. Probably you would get an error like this:
//I say probably, but I actually test it ... lol
Warning: array_shift() expects parameter 1 to be array
Which could be confusing, one could think the code is buggered. So by having the second method call with a type set into it getShiftAndDivided( array $array ). When the wrapping method tries to call this with a string its going to blow up, but in a better way:
Catchable fatal error: Argument 1 passed to Combos::getShiftAndDivided() must be of the type array, string given
Hopefully that makes sense, it's something I always try to do in cases like this. It just makes life easier in the long run. Both function return the data in the same format, which is convenient (your welcome).
Summary
So the sum of what this does, is find all the combinations of our input, then it takes those and breaks each one into shifted and divided up array. There for it stands to reason that we will have all the possible combinations divided up into 2 arrays anyway possible. Because that is pretty much exactly what I said.
Now I am not 100% it does that, you can check them if you want, it returns like 599 elements at the end. So good luck on that, I would suggest testing just the results of $combos->getPossible($input). If that has all the combinations like it should, then this will have all that it needs. I am not sure if it returns duplicates, I don't think that was specified. But I didn't really check it.
You can call the main method like this:
$input = array('a','b','c','d','e');
print_r((new Combos)->getPossibleAndDivided($input));
Test It!
P.S. I spell breaks as brakes, but I can write code this, go figure...
This question already has answers here:
How can I remove duplicates in an object array in PHP?
(2 answers)
Closed 8 years ago.
When I print $online_performers variable I want to get a unique value for id 2. Do I need to convert them in standard array first or is that possible without it? (remove all duplicates).Please check my new code for this.
Array
(
[0] => stdClass Object
(
[id] => 1
[username] => Sample1
)
[1] => stdClass Object
(
[id] => 2
[username] => Sample1
)
[2] => stdClass Object
(
[id] => 2
[username] => Sample1
)
[3] => stdClass Object
(
[id] => 4
[username] => Sample4
)
)
to
Array
(
[0] => stdClass Object
(
[id] => 1
[username] => Sample1
)
[1] => stdClass Object
(
[id] => 4
[username] => Sample4
)
)
PHP has a function called array_filter() for that purpose:
$filtered = array_filter($array, function($item) {
static $counts = array();
if(isset($counts[$item->id])) {
return false;
}
$counts[$item->id] = true;
return true;
});
Note the usage of the static keyword. If used inside a function, it means that a variable will get initialized just once when the function is called for the first time. This gives the possibility to preserve the lookup table $counts across multiple function calls.
In comments you told, that you also search for a way to remove all items with id X if X appears more than once. You could use the following algorithm, which is using a lookup table $ids to detect elements which's id occur more than ones and removes them (all):
$array = array("put your stdClass objects here");
$ids = array();
$result = array();
foreach($array as $item) {
if(!isset($ids[$item->id])) {
$result[$item->id]= $item;
$ids[$item->id] = true;
} else {
if(isset($result[$item->id])) {
unset($result[$item->id]);
}
}
}
$result = array_values($result);
var_dump($result);
If you don't care about changing your keys you could do this with a simple loop:
$aUniq = array ();
foreach($array as $obj) {
$aUniq[$obj->id] = $obj;
}
print_r($aUniq);
Let's say we have:
$array = [
//items 1,2,3 are same
(object)['id'=>1, 'username'=>'foo'],
(object)['id'=>2, 'username'=>'bar'],
(object)['id'=>2, 'username'=>'baz'],
(object)['id'=>2, 'username'=>'bar']
];
Then duplication depends of what do you mean. For instance, if that's about: two items with same id are treated as duplicates, then:
$field = 'id';
$result = array_values(
array_reduce($array, function($c, $x) use ($field)
{
$c[$x->$field] = $x;
return $c;
}, [])
);
However, if that's about all fields, which should match, then it's a different thing:
$array = [
//1 and 3 are same, 2 and 3 are not:
(object)['id'=>1, 'username'=>'foo'],
(object)['id'=>2, 'username'=>'bar'],
(object)['id'=>2, 'username'=>'baz'],
(object)['id'=>2, 'username'=>'bar']
];
You'll need to identify somehow your value row. Easiest way is to do serialize()
$result = array_values(
array_reduce($array, function($c, $x)
{
$c[serialize($x)] = $x;
return $c;
}, [])
);
But that may be slow since you'll serialize entire object structure (so you'll not see performance impact on small objects, but for complex structures and large amount of them it's sounds badly)
Also, if you don't care about keys in resulting array, you may omit array_values() call, since it serves only purpose of making keys numeric consecutive.
I have an array whose values are all arrays of a specific format that looks like this:
Array
(
[0] => Array
(
[0] => '8227'
[1] => ' 8138'
)
[1] => Array
(
[0] => '8227'
[1] => ' 8138'
[2] => ' 7785'
)
)
and I would like to have this:
Array
(
[0] => 8227
[1] => 8138
[2] => 7785
)
How can I do this ?
$result = array();
foreach ($input as $sub) { // Loop outer array
foreach ($sub as $val) { // Loop inner arrays
$val = trim($val);
if (!in_array($val, $result)) { // Check for duplicates
$result[] = $val; // Add to result array
}
}
}
$result = array();
foreach($array as $arr){
$result = array_merge($result, $arr);
}
$result = array_unique($result);
array_merge_recursive() can be used to flatten the array. Then, array_unique() to get the unique values, with array_values() to "reindex" the resultant array.
$flat = call_user_func_array('array_merge_recursive', $subject);
$uniq = array_values(array_unique($flat));
<?php
$array = array(
0 => array(
0 => 8227,
1 => 8138
),
1 => array(
0 => 8227,
1 => 8138,
2 => 7785
)
);
$newArray = array();
array_walk_recursive($array, function($item, $key) use(&$newArray) {
if(!in_array($item, $newArray)) {
$newArray[] = $item;
}
});
print_r($newArray);
?>
I don't like the idea of iterated calls of in_array() since it can cause a bit of drag on big arrays.
Now, my methods to follow are probably not going to set any speed records (I didn't bother to benchmark), but I thought I would post a couple of unorthodox approaches in case they may inspire future readers.
Method #1:
convert the multi-dimensional array to a json string
split the string on all non-digital substrings (this also trims the values)
eliminate duplicates using array_flip()
re-index the resultant array using array_keys() (output values are integer-type)
Method #2:
convert the multi-dimensional array to a json string
extract the words (which in this case include numbers and there aren't any letters to worry about)
eliminate duplicates using array_flip()
reindex the resultant array using array_keys() (output values are integer-type)
Code: (Demo)
$array = [['8227', '8138'], [' 8227', ' 8138', ' 7785']];
echo "regex method: ";
var_export(
array_keys(
array_flip(
preg_split(
'/\D+/',
json_encode($array),
0,
PREG_SPLIT_NO_EMPTY
)
)
)
);
echo "\n\nnon-regex method: ";
var_export(
array_keys(
array_flip(
str_word_count(
json_encode($array),
1,
'0..9'
)
)
)
);
Output:
regex method: array (
0 => 8227,
1 => 8138,
2 => 7785,
)
non-regex method: array (
0 => 8227,
1 => 8138,
2 => 7785,
)
If either of these methods perform well, it will be because they don't make iterated function calls.
p.s. Okay, because I was curious, I just did a very small / unofficial speed test on my two methods and DaveRandom's method (only 1000 iterations using the OP's data - I didn't want to abuse 3v4l.org) and...
Method #1 ran about as fast as Method #2
Both Method #1 and Method #2 ran faster than DaveRandom's method.
Again, I'll state that fabricated tests for micro-optimization may be pointless and real tests should be done on your REAL data IF it is actually important. If you merely prefer the logical flow of another answer on this page, I totally respect that.