We have users, questions and unlimited levels of categories. The users can get some points from questions. Questions can have multiple categories.
What I want to do is to calculate the top users per category: It's simply total points taken from the questions under that category AND it's sub-categories too.
So, I have these tables:
questions
--------------
id
title
question
categories
--------------
id
parent_id
category
lft
rgt
question_categories
--------------
question_id
category_id
users
--------------
id
username
user_points
--------------
id
user_id
question_id
point_type
points
user_category
--------------
user_id
category_id
points
What I want to do is to calculate user_category.points value.
Summing up the points for each category is easy but including the sub-categories is getting complicated.
What might be the best way to do this?
Example calculation:
Let's say the categories are:
Programming
PHP
Zend Framework
Symfony
Java
Ruby on Rails
Assume that the user got 3 points from Zend Framework, 2 points from PHP, 5 points from java and 1 point from Rails. The points for this user per categories will be:
Programming 11 (5+5+1)
PHP 5 (2+3)
Zend Framework 3
Symfony
Java 5
Ruby on Rails 1
Perhaps it would be best to use tags instead of a hierarchy. For instance, anything with a "Zend Framework" will also have "PHP" and "Programming" tags. This also helps when some categories can appear in multiple places. For instance, I can use ajax in jQuery and also Javascript. Then, add 1 to each tag listed in the category for the user.
I would create a user_categories table in which I would store 3 values: user_id, category_id and user_score. It's easy to maintain (need only to INSERT or UPDATE) and it's also easy to query for top-users of every category.
If you're only going to sum per top-level category, then you should add a field to your categories table called root_id (holding the id of the transitive parent of the category).
Then your sum would be calculated as:
select up.user_id, ctg.root_id, sum(up.points)
from user_points up
join question_categories qc on up.question_id = qc.question_id
join categories ctg on qc.category_id = ctg.id
group by up.user_id, ctg.root_id
This php and SQL should get you the top 3 users for each category including sub categories:
$query = "SELECT id, parent_id FROM categories";
$parent = array();
...fetch mysql data loop depending on what connection you use, mysqli or pdo...
{
$parent[$result['id']] = $result['parent_id'];
}
$childs = array();
foreach($parent as $id => $parrent_id)
{
$childs[$parrent_id][$id] = $id;
$next_parrent_id = $parrent_id;
while($next_parrent_id = $parent[$next_parrent_id])
{
$childs[$next_parrent_id][$id] = $id;
}
}
foreach($parent as $id => $parrent_id)
{
$current_categories = array($id => $id) + $childs[$id];
$query = "SELECT user_id, username, SUM(points) AS total_points
FROM user_points
LEFT JOIN users ON (user_id = users.id)
LEFT JOIN question_categories USING (question_id)
WHERE category_id IN (" . implode(', ', $current_categories). ")
ORDER BY total_points DESC
LIMIT 3";
...fetch mysql data loop...
}
Related
Okay so I have a simple category table and a separate posts table easy right but when the user posts a post I wast think should I store both the sub and parent cat in the posts table but would that not be a lot of data duplication so I instead just store the sub_cat then I use a few PHP functions to query the database for the primary cat and its name.
categories table
ID | cat_name | main_cat
1 | Dinner | 0
2 | Chicken | 1
posts table
ID | title | sub_cat | fields that are not related to Q
1 | test | 2 |
Get parent(main) category
$sub_cat = is from a selection query that gets posts and their sub_cats
function main_cat($sub_cat){
require("conn_posts.php");
$stmt = $conn_posts->prepare("SELECT `main_cat` FROM `cats` WHERE `ID` = ?");
$stmt->bind_param("s", $sub_cat);
$stmt->execute();
$stmt_results = $stmt->get_result(); // get result
while($row_get = $stmt_results->fetch_assoc()){
if($row_get['main_cat'] == 0){
return $sub_cat;
}elseif($row_get['main_cat'] !== ""){
return $row_get['main_cat'];
}
}
}
This function gets any category name as long as the id is valid
function cat_name($cat_number){
require("conn_posts.php");
$stmt = $conn_posts->prepare("SELECT `cat_name` FROM `cats` WHERE `ID` = ?");
$stmt->bind_param("s", $cat_number);
$stmt->execute();
$stmt_results = $stmt->get_result(); // get result
$row_get = $stmt_results->fetch_assoc();
if($stmt_results->num_rows <= 0){
return 0;
}elseif($stmt_results->num_rows == 1){
return $row_get['cat_name'];
}
}
My question is is this a good way to process my posts sub-category and parent category are there better ways of doing what I am currently doing? eg. is my database schema good(by good I mean is it better to just include the parent cat id in the posts table than to do the PHP server-side processing)?
Your database schema is good: it doesn't include any replication, I wouldn't change it. The way you're handling fetching the categories in PHP isn't really optimal though: you should almost always aim to minimize the number of queries as it (in general) will affect performance more than the complexity of a query.
If you're running MySQL 8+, a great way to do this is with a recursive CTE; it will allow you to fetch all parents with one query:
WITH RECURSIVE cte AS (
SELECT id, cat_name, main_cat, 0 as depth FROM categories WHERE ID=3
UNION ALL
SELECT categories.id, categories.cat_name, categories.main_cat, cte.depth+1 as depth
FROM cte inner join categories
ON cte.main_cat = categories.id
)
SELECT cat_name FROM cte order by depth ASC
The number '3' in that query can be replaced by the category you're trying to retrieve. You can check this DB fiddle for a live example. If I see your code, incorporating it into your PHP should be fairly trivial. If not, leave a comment and I'll try to expand.
I have a categories table:
id | name | parent_id
1 | Camaro | 0
2 | Chevelle | 0
3 | Sale - Camaro Parts | 1
4 | Bestselling Parts | 1
My first request looks like:
'SELECT
*
FROM
`categories`
WHERE
parent_id = :parent_id';
And after I'm fetching result set I make sub query to check if row has child elements:
foreach($result as $r) {
$r->hasChild = count(ORM::forTable('categories')->where('parent_id', $r->id)->findArray());
$data[] = $r;
}
Is any way to avoid multiple connection to DB in foreach loop and get data in first query?
Thanks!
This isn't awful to do, so long as you only want the count of children below the selected rows. If you want the entire hierarchy, you'll need to use a better RDMS.
The main part of the solution here is self joining the same table. Then we can use the count() aggregate function to see how many children are attached to each item.
select
categories.id
, categories.name
, categories.parent_id
, count(chld.id)
from
categories
left join categories chld
on categories.id = chld.parent_id
where
parent_id = :parent_id
group by
categories.id
, categories.name
, categories.parent_id
You can do a self join to the table on parent_id and id. Based on whether you want categories with child or not you can do a left join or inner join. Kind of a similar question is mentioned here -
Mysql Self Join to find a parent child relationship in the same table
I have a table that looks like this
id | itemID | catID | Title
----------------------------------------------------------------------------
0 3 4 Hello
1 3 6 Hello
2 4 4 Yo
3 4 8 Yo
4 5 2 Hi
5 1 3 What
I want to do a MySQL PHP Select that only gets one occurrence of the itemID. As you can see they are the same item, just in different categories.
This is what I tried
SELECT * FROM Table GROUP BY itemID
That didn't seem to work, it still just shows duplicates.
Is this what you are looking for? http://www.sqlfiddle.com/#!2/5ba87/1
select itemID, Title from test group by itemID;
As far as MySQL is concerned, the data is all unique, since you want all of the columns. You have to be more specific.
Do you just want the itemID (or other column)? Then say so:
select [column] from Table GROUP BY itemID
Do you want the last entry of a particular item ID? Then say that:
select * from Table where itemID = 1 ORDER BY id DESC
Or the first one?
select * from Table where itemID = 1 ORDER BY id
If none of these are what you want, then you probably need to restructure your tables. It looks like you want different categories for your items. If so, then you'll want to split them out into a new join table, because you have a many-to-many relationship between Items and Categories. I recommend reading up on database normalization, so you're not duplicating data (such as you are with the titles).
If you want everything for the distinct itemIDs, you could certainly take a long route by doing one selection of all of the distinct itemIDs, then doing a series of selections based on the first query's results.
select distinct(`itemID`) from Table
Then in your PHP code, do something like this:
while ($row = mysql_fetch_array($result, MYSQL_ASSOC))
{
$itemID = $row['itemID'];
$sql2 ="SELECT * FROM Table WHERE 1 and `itemID`=\"$itemID\" limit 1";
$result2 = #mysql_query($sql2, $connection);
while ($row2 = mysql_fetch_array($result2))
{
$id = $row2['id'];
$itemID = $row2['itemID'];
$catID = $row2['catID'];
$Title = $row2['Title'];
}
}
I have a comics website which currently allows users to choose which comics they view by category: Charts, Office, Life, Misc, etc.
I'd like to implement a tagging system, similar to what we have here on StackOverflow, which will describe more of the content of each comic rather than its category. Ex: In Charts category, I have several business related...
My simple solution would be to handle it just how I've handled my categorization-
Create a "Tags" table with tagid, tagname, tagdescription
Add a tagid_ForeignKey field in comics table, and add a tag to each post.
When a user clicks on a tag, it will show only those posts with that tag... or if there is also a category specified, it will show that specific category with that specific tag.
This approach, however, seems to limit me to one tag per category. What if I have a comic that is business and relationships related... so It'd need both of those tags.
How would I be able to attach multiple tags per comic?
EDIT:
A few more questions:
1) What do I insert into my new relational table... anything?
2) Also, for while ($row = $tag->fetch_assoc()) {, how can I loop through a table if there is a join? Isn't that an associative array?
3) The issue is that I am echoing out the tag choices as such, so once a user clicks on a link, how would you be able to allow them to then click on another link to assign a 2nd tag?
function getTags() {
include 'dbconnect.php';
global $site;
$tag = $mysqli->query("SELECT tagid, tagname FROM tags");
//$tag = $mysqli->query("SELECT * FROM comics c INNER JOIN comictags ct ON (c.comicID = ct.comicID) WHERE ct.tag_id IN (1, 2, 3) GROUP BY c.comic_id");
mysqli_close($mysqli);
while ($row = $tag->fetch_assoc()) {
echo "<a href='?action=homepage&site=" . $site . "&tag=" . $row['tagid'] . "&tagname=" . $row['tagname'] . "'/>" . $row['tagname'] . "</a><br />";
}
}
Just add another table. Then you have three: One for Tags, one for Comics, and one for the relationship. You have to have this indirection table to properly store a many-to-many relationship. This allows each comic to have zero or more tags (and vice versa).
You can accomplish this with a many-to-many relationship. A many-to-many relationship uses a relational join table that would look like this:
+---------------+---------------+
| comic_id | tag_id |
+---------------+---------------+
| 1 | 2 |
+---------------+---------------+
| 1 | 3 |
+---------------+---------------+
| 1 | 4 |
+---------------+---------------+
Now, in your query:
SELECT * FROM comics c INNER JOIN comic_tags ct ON (c.comic_id = ct.comic_id) WHERE ct.tag_id IN (1, 2, 3) GROUP BY c.comic_id
Where 1, 2, 3 are the tags the user selected that they would like to see.
I want to pull from my database records corresponding to parent_id, like this:
function getChildren($id, $parent_id) {
$q = $this->db->select('id, name, slug, plat');
$q = $this->db->from('games');
$q = $this->db->where('parent_id',$id);
$q = $this->db->or_where('id',$parent_id);
$q = $this->db->get();
return $q->result_array();
}
It - if it's a children game - get parent_id and search for a game with such id and for other games that has parent_id same as this one. If it's the parent game, it only looks for games with parent_id same as it's id.
The problem is... it's not always working. I have four games in db:
id | parent_id | title
15 | 0 | Abe
19 | 15 | Abe
20 | 0 | RE2
21 | 20 | RE2 DS
First two works, last two - only children (id = 21) shows parent.
You likely could not do that with a relation database. RDBMS are not intended to manage any form of trees.
You can in some simple case, like one-level hierarchy, but as soon as it becomes more complex, it's getting messier and messier.
Keeping your structure, you have to make ONE JOIN per LEVEL, and that means knowing the depth in advance.
A solution to store trees in database is called Nested Tree, it basically stores interval for in each rows, but it is a bit complex to implement by yourself.
Take a look at Wikipedia explanation. There are however library which allows you to programmatically abstract such operations.
Use this query
select
*
from games as g
where parent_id = 0
union all
select
l.*
from games r
left join (select * from games) as l on r.id = t.id
where
r.parent_id != 0
Just after posting my problem, I tried another solution. Funny, I sit with this for more than an hour and after wasting your time, I came with the solution ;)
Anyway, here it is:
function getChildren($id, $parent_id) {
$q = $this->db->select('id, name, slug, plat');
$q = $this->db->from('games');
$q = $this->db->where('parent_id',$id);
$q = $this->db->or_where('id',$parent_id);
$q = $this->db->or_where('parent_id',$parent_id);
$q = $this->db->where('id !=', $id);
$q = $this->db->get();
return $q->result_array();
}