PHP MySQL hierarchy data optimization - php

Question

I have a table employees, which contains columns : employee_id, name, employee_manager_id.
employee_manager_id references to employee_id. It's a hierarchal data.
I have this output using PHP but couldn't achieve it using only one mySQL query. As of now, I need to process the data in PHP using recursive function so i can achieve this kind output.
Sample Array Output
0 => (
employee_id => 2,
name => Jerald,
employee_manager_id => 1,
depth => 1
),
1 => (
employee_id => 3,
name => Mark,
employee_manager_id => 2,
depth => 2
),
2 => (
employee_id => 6,
name => Cyrus,
employee_manager_id => 3,
depth => 3
),
3 => (
employee_id => 4,
name => Gerby,
employee_manager_id => 2,
depth => 2
)
As of now, this is my recursive function in PHP to achieve the output above.
function get_employees_by_hierarchy( $_employee_id = 0, $_depth = 0, $_org_array = array() ) {
if ( $this->org_depth < $_depth ) {
$this->org_depth = $_depth;
}
$_depth++;
$_query = "SELECT * FROM employees WHERE ";
if ( !$_employee_id ) {
$_query .= "employee_manager_id IS NULL OR employee_manager_id = 0";
}
else {
$_query .= "employee_manager_id = " . $this->dbh->quoteSmart( $_employee_id );
}
$_result = $this->query( $_query );
while ( $_row = $_result->fetchRow() ) {
$_row['depth'] = $_depth;
array_push( $_org_array, $_row );
$_org_array = $this->get_employees_by_hierarchy(
$_row['employee_id'],
$_depth,
$_org_array
);
}
return $_org_array;
}
My question is, is there anyway so I can achieve the array output i want using just one mysql query?
If not possible in mysql query, is there anymore to optimize in my current code?
Any help would greatly be appreciated.
Thanks

Answer 1

You can try a nested set a.k.a. celko tree but insert and delete is very expensive. There is also closures and path enumeration (materialized path) but I'm not an expert. MySql doesn't support recursive queries.

Answer 2

I don't think you can get the depth with your current model without doing any processing on the results, but you don't need to make multiple queries.
Assuming $employees is the list of employees indexed by employee_id, you could do something like this:
function set_employee_depth(&$employees, $id) {
if (!isset($employees[$id]['depth'])) {
$employee_manager_id = (int) $employees[$id]['employee_manager_id'];
if (!$employee_manager_id) {
$employees[$id]['depth'] = 0;
} elseif ($employee_manager_id !== $id) {
$employees[$id]['depth'] = 1 + set_employee_depth($employees, $employee_manager_id);
} else {
throw new \Exception('Employee cannot be its own manager!');
}
}
return $employees[$id]['depth'];
}
foreach ($employees as $id => $employee) {
set_employee_depth($employees, $id);
}

Answer 3

So, your table is composed of 3 columns (employee_id, name, employee_manager_id). employee_manager_id is a self reference to employee_id. You want to construct an array with all records, adding an extra field called depth which represents the distance of said employee to the "big boss", with only one query to the database. Is that correct? I'm also assuming that the database structure can't be changed.
If these assumptions are correct, this is a basic HIERARCHICAL/TREE data structure and thus, you have a couple of ways you can tackle this problem.
First Script
The first script runs the results array sequentially, finding the main node / trunk (the big boss) first and then adding it's children, then it's grandchildren and so on. Each time a node is "sorted", it will be removed from the cycle until no nodes are left. It assumes that:
There are no Orphan records (employees with invalid managers_ids)
There are no Circular References, either simple (A is manager of B and B is manager of A) or complex (*A manager of B, B manager of C and C manager of A)
Each path (from the main node to last node) can have an infinite number of nodes
$results are produced by running a simple query SELECT * FROM employees ORDER BY employee_manager_id
Code:
$finalArray = array();
$limit = count($results);
while (count($results) > 0) {
$results[0]['cnt'] = isset($results[0]['cnt']) ? $results[0]['cnt']++ : 0; // set num of times each element was already visited
if ($results[0]['cnt'] === $limit) { //prevent an infinite cycle
break;
}
$manId = $results[0]['manager_id'];
if ($manId === null) {
$results[0]['depth'] = 0;
} else if ( ($key = searchForId($manId, $finalArray)) !== null ) {
$results[0]['depth'] = $finalArray[$key]['depth'] + 1; //use the depth of parent to calculate its own
} else {
$results[] = $results[0]; //parent was not visited yet so we add it to the end of array
array_shift($results);
continue;
}
unset($results[0]['cnt']);
$finalArray[] = array_shift($results);
}
function searchForId($id, $array) {
foreach ($array as $key => $val) {
if ($val['id'] === $id) {
return $key;
}
}
return null;
}
This script is pretty straightforward. It only runs one query to the DB. In best case scenario, it will only traverse the array once. In worse case scenario, it will visit each element count(array) - 1, which can be slow with big arrays. However, since the results are pre-sorted, best case scenario will probably be more common.
Second Script
The second script builds an actual tree of elements. It's a bit more complex but achieves similar results. Also, the Depth is calculated dynamically.
class Employee {
public $id;
public $name;
public $manager;
public function __construct($id, $name, Employee $manager = null) {
$this->id = $id;
$this->name = $name;
$this->manager = $manager;
}
public function setManager(Employee $manager) {
$this->manager = $manager;
}
public function getDepth() {
if ($this->manager === null) {
return 0;
} else {
return $this->manager->getDepth() + 1;
}
}
}
$finalArray = array();
$paths = array();
foreach ($results as $r) {
$finalArray[(int) $r['id']] = new Employee((int)$r['id'], $r['name']);
if ($r['manager_id'] !== null) {
$paths[(int) $r['id']] = (int) $r['manager_id'];
}
}
foreach ($paths as $k => $v) {
if (isset($finalArray[$k]) && isset($finalArray[$v])) {
$finalArray[$k]->setManager($finalArray[$v]);
}
}

Answer 4

Here is a link for answer
Here is full code for making tree structure for hierarchy management using php and mysql.