使用连接从单个查询中获取嵌套数组

Question

I have the common situation, but I can't find a common solution...

Imagine I have two tables like authors and corresponding books:

Authors

+----+-----------+
| id | name      |
+----+-----------+
|  1 | O.Connor  |
|  2 | F.Myler   |
+----+-----------+

Books

+----+-----------+-----------+----------------+
| id | author_id | title     | number_of_pages|
+----+-----------+-----------+----------------+
|  1 | 1         |Book 1     |315             |
|  2 | 1         |Book 2     |265             |
+----+-----------+-----------+----------------+

I want to get the data from database in one sql query (I use mySQL but it doesn't matter, I would like to have a database independent solution) with JOIN like:

SELECT A.*, B.title as book_title, B.number_of_pages as book_pages FROM authors A LEFT JOIN books B ON B.author_id=A.id

I would like to avoid multiple queries in a loop because of productivity issues on big databases. And then I need to nest the data from the second table as a subarray to avoid having multiple copy of the same data. In other words I need to transform a flat result from database to a nested multidimensional array. It should give something like:

[
    {
        "id": 3,
        "name": "O.Connor",
        "birth_date": "05.06.1985",
        "from": "England",
        "mother": "",
        "father": "",
        "email": "email@example.com",
        "books": [
            {
                "title": "Some Title",
                "pages": 235
            },
            {
                "title": "Some Title",
                "pages": 267
            },
            {
                "title": "Some Title",
                "pages": 317
            },
            {
                "title": "Some Title",
                "pages": 235
            },
            {
                "title": "Some Title",
                "pages": 298
            }
        ]
    },
    {
        "id": 28,
        "name": "O.Henri",
        "birth_date": "05.06.1300",
        "from": "England",
        "mother": "",
        "father": "",
        "email": "email@example.com",
        "books": [
            {
                "title": "Some Title",
                "pages": 235
            },
            {
                "title": "Some Title",
                "pages": 267
            },
            {
                "title": "Some Title",
                "pages": 317
            },
            {
                "title": "Some Title",
                "pages": 235
            },
            {
                "title": "Some Title",
                "pages": 298
            }
        ]
    }
]

The only related resources I've found so far are:

SQL query with join to get nested array of objects

https://phpdelusions.net/pdo_examples/nested_array

The first one uses either multiple queries or some SQL Server functionality that is not available in MySQL. The second one proposes to fetch each row from the result separately and construct the resulting array on-the-go. Maybe it's not a bad solution, but maybe there are better ones, more productive.

There are some fetch modes in PDO, like FETCH_GROUP etc. But nothing is enough for my task.

I wrote my own function with an idea to create further some DB class and insert this function as a method (so it should be really universal for most cases).

function group_by($array, $criterias=NULL, $group_name='group') {
    if ($criterias == NULL) return $array; // add also a verification of type
    // transform input parameters for array_diff_key()
    $criterias = array_flip($criterias);
    $group = array($group_name=>0);

    $result_array = [];
    $push_index = 0;
    foreach ($array as $row) {
        $sup_array = array_diff_key($row, $criterias);
        $sub_array = array_intersect_key($row, $criterias);
        // add verification of not NULL elements in future
        $isInArray = false;

        foreach ($result_array as $index => $grouped_array) {
            // if there is no difference between arrays then this sub-array is already present in a result array.
            $array_without_groups = array_diff_key($grouped_array, $group);
            if ( empty(array_diff_assoc($array_without_groups, $sup_array)) && (count($array_without_groups)==count($sup_array)) ) {
                $isInArray = true;
                $push_index = $index;
                break;
            }
        }
        if ($isInArray) {
            array_push($result_array[$push_index][$group_name], $sub_array);
        } else {
            $sup_array[$group_name] = array();
            array_push($sup_array[$group_name], $sub_array);
            array_push($result_array, $sup_array);
        }
    }

    return $result_array;
}

And then you call this on the fetchAll result as:

$result = group_by($result, array('book_title', 'book_pages'), 'books');

I don't think it's the best solution. I am a noobie to PHP, so I'm not even sure if array_diff and array_intersect that I use are the efficient way to extract one array from another. I would be thankful for any advice.

Answer 1

Here's a solution that works in MySQL 5.7 or later.

SELECT A.id, A.name, A.birth_date, A.`from`, A.mother, A.father, A.email, 
  JSON_ARRAYAGG(JSON_OBJECT('title', B.title, 'pages', B.number_of_pages)) AS books
FROM authors A LEFT JOIN books B ON B.author_id=A.id
GROUP BY A.id;