I'm really hoping someone can help me with this. I have a number of product attribute types that users can select from to refine the products that are returned to them on screen. What I'm trying to do is, for each product attribute type, I want to list all attributes that relate to either the selected category or search term, then once they've made their selections, I still want to display each of the attributes that relate to the category or search term, but only display a clickable link if the product count for that particular attribute is greater than 1 and for those that have a product count of zero, I want to list them, but make them unclickable. An example of what I'm trying to achieve can be found on the ASOS website, in the left hand menu
http://www.asos.com/Women/Dresses/Cat/pgecategory.aspx?cid=8799#state=Rf961%3D3340%2C3341%40Rf-200%3D20&parentID=Rf-300&pge=0&pgeSize=20&sort=-1
Initially I tried using just joins to achieve this, but I wasn't able to do it, successfully. So I decided to create a temporary table for each attribute type which held a list of all the attributes that related to the main query and then created a refined query, with a left join. Here's my code:
CREATE TEMPORARY TABLE temp_table
SELECT su_types.id, type AS item FROM su_types
INNER JOIN su_typerefs ON su_types.id=su_typerefs.id
INNER JOIN su_pref ON su_typerefs.mykey = su_pref.mykey
WHERE wp_category_id =40 GROUP BY su_typerefs.id
$sudb->query($query);
if ($sudb->affected_rows > 0) {
SELECT temp_table.id,item,COUNT(su_typerefs.mykey) AS product_count FROM temp_table
LEFT JOIN su_typerefs ON temp_table.id=su_typerefs.id
LEFT JOIN su_pref ON su_typerefs.mykey = su_pref.mykey
LEFT JOIN su_stylerefs ON su_pref.mykey = su_stylerefs.mykey
LEFT JOIN su_productrefs ON su_pref.mykey = su_productrefs.mykey
WHERE wp_category_id =40 AND su_stylerefs.id in (91) AND su_productrefs.id in (54) AND su_typerefs.id in (159) GROUP BY su_typerefs.id
if ($itemresults = $sudb->query($query)) {
while($itemresult = $itemresults->fetch_array(MYSQLI_ASSOC)) {
$id=$itemresult['id'];
$item=$itemresult['item'];
$product_count=$itemresult['product_count'];
build_link($list_type, $item, $product_count, $id);
}
}
In the above example the first query selects all the product types that relate to a particular category, say dresses. And the second query is based on the refinements the user has made on the category, in this example this is product, product type and style. A user can also refine their search by colour, fit, fabric and design.
There are a couple of issues with this:
1) The number of results returned in the second query do not match the results of the first. Using the above as an example, I wish to list all products that relate to the chosen category, then using the second query return the product count for each of these products as I described above. So if the temporary table returns, trousers, jeans and skirts. I expected these three items to be displayed on screen based on the conditions applied in the second query, however my results may only show trousers and jeans, if there is not a match for skirts in the second query. I thought that using a left join would mean that all the results of the temporary table would be displayed.
2)Also I wonder if I'm doing this the most efficient way. I have a total of 8 attribute groups, and therefore need to do the above 8 times. If the user choses to refine the results using all 8 attribute groups then in addition to the temp table join, there will be a total of 9 joins for each type. It's taking a while to execute, is there a better way to do this? There are approximately 1/2 million products in the table, and this will probably be 5 times this, once my site goes live.
I really hope all that I have written makes sense and I'd really appreciate the stackoverflow community's help with this, if anyone can help. I apologise for the essay ;). Thanks in advance
To answer your first question; yes, a LEFT JOIN will indeed keep all data from the initial table. That, however, isn't the problem.
The reason why you lose empty categories, is most likely (I say this because I don't fully know your db structure) because of the where condition filtering out all results based on the data in the joined tables.
If for a category all items get filtered out (possibly including the NULL joined values), you will not get this category back from that query anymore. Also the GROUP BY is done on a joined column, that might also effectively wipe out your other categories.
As for the second question, you already state it's taking long; so it's probably not the way to go if you want things to work fast ;) (okay, obvious answer, low hanging fruit, etc). What you might want to do, is get a collection of keys from the filterable categories first, and use that data to select items.
This prevents that you have to join up your entire products table in a temp table (at least, that's what I think you're doing), which of course will take long with the given number of entries. Selecting a list of matching IDs from the given attributes also gives you the advance of using your indexes (more), which a temp-table probably won't have. If this is possible and feasible mainly depends on your schema's structure; but I hope it might lead you to the direction you want to go :)
Related
I am using codeigniter and MySQL to build an ecommerce web application.
This one required three level of categories. So I have created 3 tables. These are-
category
category_id, category_name
subcategory
subcategory_id,subcategory_name,subcategory_category_id
subsubcategory
subsubcategory_id,subsubcategory_name,subsubcategory_subcategory_id
Here they are linked as parent of one another. Finally I have the product table
product
product_id, product_name, product_subsubcategory_id
Now, I need a sql query on this to fetch all product of any specific category.
Something like
$this->Mdl_data->select_products_by_category($category_id);
Please help me on this. I have tried PHP programming to solve this. But it was too slow with lot's of nested loops.
If you need to select all products, that match some specific category, try this request:
SELECT p.product_id, p.product_name, p.subsubcategory_id FROM category c
JOIN subcategory sc ON sc.subcategory_category_id = c.category_id
JOIN subsubcategory ssc ON ssc.subsubcategory_subcategory_id = sc.subcategory_id
JOIN product p ON p.subsubcategory_id = ssc.subsubcategory_id
WHERE c.category_id = 1;
But you should think about changing your database structure to make your requests faster and simpler.
Edit: Answering the comment about how to improve DB.
Current design of database looks correct, according to actual data relations. 1-many for cat-subcat and 1-many for subcat-subsubcat. But this leads to complicated (and possibly slow) queries while usage.
One way I see is to implement many-many relation with additional restriction. You can create additional table cat-subcat, just as you would do if you needed many-many. But in that table you can set unique limitation to subcat_id, so every subcat could belong only to 1 cat and it becomes in fact 1-many relation. But in this case you can move both up- and downwards the hierarchy. This approach will reduce number of JOINs in your query only by 1, but the whole logic of the query would be easier to understand.
Another way. As I understand this is the query for web-store filter. So, new products will be inserted much more seldom, than viewed by category. You can just add subcat_id and cat_id fields to your product, which is not good idea from the point of data structure, but for this particular situation this might be good solution. Every time new product is inserted to DB, you should control the correctness of those 2 fields by PHP or whatever you use on server. But when products are searched by category you will have simple request without JOINs at all.
Both approaches are based on the idea to sacrifice some space for speeding up and simplifying the queries, that are frequently used. Maybe there is even better solution, but I can't find it right now.
I was wondering if mysql has a way to look at a column and only retrieve the results when it finds a unique column once. For example
if the table looks like this:
id name category
1 test Health
2 carl Health
3 bob Oscar
4 joe Technology
As you can see their are two rows that could have the same category. Is their a way to retrieve the result where the array will one only return the category once?
What I am trying to do is get all the categories in the database so I can loop through them later in the code and use them. For example if I wanted to created a menu, I would want the menu to list all the categories in the menu.
I know I can run
SELECT categories FROM dbname
but this returns duplicate rows where I only need the cateogry to return once. Is there a way to do this on the mysql side?
I assume I can just use php's array_unique();
but I feel like this adds more overhead, is this not something MYSQL can do on the backend?
group by worked perfectly #Fred-ii- please submit this as answer so I can get that approved for you. – DEVPROCB
As requested by the OP:
You can use GROUP BY col_of_choice in order to avoid duplicates be shown in the queried results.
Reference:
https://dev.mysql.com/doc/refman/5.5/en/group-by-handling.html
By using database normalization, you would create another table with an unique id and the category name and by that link those two together, like
select * from mytable1
on mytable1.cat = mytable2.id
group by mytable1.cat
You can ofcourse also use group by without multiple tables, but for the structure, I recommend doing it.
You can use select distinct:
SELECT DISTINCT categories
FROM dbname ;
For various reasons, it is a good idea to have a separate reference table with one row per category. This helps in many ways:
Ensures that the category names are consistent ("Technology" versus "tech" for instance).
Gives a nice list of categories that are available.
Ensures that a category sticks around, even if no names currently reference it.
Allows for additional information about categories, such as the first time it appears, or a longer description.
This is recommended. However, if you still want to leave the category in place as it is, I would recommend an index on dbname(categories). The query should take advantage of the index.
SELECT id, name from dbname GROUP BY categoryname
Hope this will help.
You can even use distinct category.
I want to store reviews in a flexible system of categories and subcategories, and am currently in the process of designing the database structure for that. I have an idea how to do that, but I'm not entirely sure if it couldn't be done more elegant and/or efficient. These are my thoughts - if anybody can comment on if/how this can be improved I'd be really grateful.
(To keep this post concise, I only list the important field for the tables)
1.) The reviews are stored in the table "reviews". It has the following fields:
id: uniquite ID, auto-incrementing.
title: the title that will show up in <head><title>, etc.
stub: a version of the title without spaces, special chars, etc. so it can be part of the URL/URI
text: the actual content
2.) All categories are in the same table "categories"
id: unique ID, auto-incrementing.
title: the full title/name of the categorie how it will be output on the website
stub: version of the title that will be shown in the URL/URI.
parent_id: if this is a subcategory, here is the categories.id of the parent category. Else this is 0.
order_number: simple number to order the categories by (for display in the navigation menu)
3.) Now I need an indicator which reviews are in what categories. The can be in multiple. My first idea was to add a "review_list" field to the categories and have it contain all reviews.id's that should be in this category. However I think that adding and removing reviews from categories would be a hassle and "unelegant". So my current idea is to have a table "review_in_category" and have an entry for every review-category relation. The structure is:
id: Unique ID, auto-increment.
review_id: the reviews.id
category_id: the categories.id
So if a review is in 3 different categories it would result in 3 entries in the "review_in_category" table.
The idea is, that when a user opens www.mydomain.de/animation/sci-fi/ the wrapper script will break up the URL into its parts. If it finds more than one category with category.stub = "sci-fi", it will check which of those has a parent category with the stub "animation". Once the correct category is identified (most the time the stubs are unique anyway so this check can be skipped) I want to SELECT all review_id's from "review_in_category" where the category_id matches the the one determined by the wrapper script. All the review_id's are put into an array. A loop will iterate through this array and compose the SELECT statement for listing all review titles (and create links to them using the stub values) by "SELECT title, stub FROM reviews WHERE id=review_list[$counter]" and then add "OR id=review_list[$counter]" until the array is completely travelled.
SO my questions are:
- Is the method my creating a single SELECT statement with potentially a large number of "OR id=" parts an "elegent" and/or efficient way to handle this situation or are there better variants?
- Does using a "taxonomy"-style table (review_in_category) make sense or would it be better to store the "membership"/"relation" directly in the reviews or category tables?
- Any other thoughts... I just started to learn this stuff and appreciate any feedback.
Thank you
Your design looks sound.
To retrieve all reviews in a category, you should use a join:
SELECT reviews.title, reviews.stub FROM reviews, review_in_category WHERE reviews.id = review_in_category.review_id AND category_id = $category
Really hope someone can help me or at least point me in the right direction. I working on a product site, that allows visitors to filter their results using a menu on side of the product results, similar to what www.asos.com have done. Basically, they can chose product type, style,price,fit ... etc. Each of these fiter types, have their own table and a reference table which links each value to the products table. EG There is a
Products table with productkey and all other product information
Colours table, with the following fields, Colour_ID, Name
Colourref table with productkey, Colour_ID
I'm using MySQL and PHP. I know how to query the database and display the data if the visitor makes one selection from each filter type and then to display the counts for each attribute, but I'd really like to be able to allow them to make multiple selections and then calculate the counts for each attribute, based on what's been selected. I've looked into how this should be done, and I've seen that subqueries are an option, but I'm concerned about how many subqueries I would need to create as I have 9 filter groups, that can have a large number of values. There are currently, 1/2 million products in the database, and this grow over time. I'm capturing the values that need to be queried via the url, so as an example
page=1&view=20&product=125,137,147&type=1,3,5&colour=3,9,5&material=187,345
As you can see from the example, I can have multiple values for each time. I tried writing a query using AND. Example, product = 125 AND product = 137, but that doesn't work.
Does anyone have any advice on the best way to go about doing this even if it's just a point in the right direction?
Any help will be greatly appreciated
Thank you in Advance
Vivien
Basically you answered your own question already:
SELECT ...
FROM ...
WHERE (product = 125 OR product = 137) AND
(colour = 3 OR colour = 8 OR colour = 5) ...
You need to use OR instead of AND if you want to select several products, colours and so on.
If you want both a product and colour then you need to combine those using AND in between. There is no need for subqueries here.
It's easier to use IN though:
SELECT ...
FROM ...
WHERE product IN (125, 137, 147) AND colour IN (3, 5, 8)
A more complete example of this SQL code:
SELECT p.*
FROM Products p
LEFT JOIN Colourref cr
ON cr.productkey = p.productkey
LEFT JOIN Colours c
ON c.Colour_ID = cr.Colour_ID
WHERE
p.productkey IN (1, 2, 4)
AND c.Colour_ID IN (1, 2)
This will select all products that have the ID 1, 2 or 4 which have the colours 1 or 2.
It left joins the required tables on the IDs and then filters the desired values.
So I am picking up a project that was quit halfway through by the last guy so that I could get some more practice with PHP and databases. I have run into a problem, and I am sure it is common enough that there is a standard solution, but I am unable to find one.
The db I am working with has 4,600, so reorganizing is out of the question. It is a db of liquers for a wholesaler. Here is what the results page looks like currently:
What I am trying to set it up so the results are returned in list form, with only one title and dropdown menus for the different sizes/prices of products that looks like this:
The problem is that there are multiple entries in the db for each product. In this example there are 3, while some have 1, and some have 2.
I am really not sure how to go about this, and any help would be greatly appreciated. Thank you.
I'm not sure about the PHP syntax, but pseudocode here's what you could do:
allProductsReturnedFromMySQL = QueryYourDatabaseForAllProducts()
Hashtable[productId, List[productSizes]] dropDownsByProduct;
Hashtable[productId, commonProductInformation] uniqueProducts;
foreach (product in allProductsReturnedFromMySQL) {
if product.productId not in uniqueProducts
then add it with the product information that does not vary
if product.productId not in dropDownsByProduct
then add it with an empty list
append the size of this product to the corresponding list in dropDownsByProduct
}
After that little bit of logic you'll have all your unique products with the common properties for each one, and a way to fetch the corresponding sizes drop down. If you wanted to do this purely in SQL to minimize the data that's transferred, you could do something like this:
-- this would get you your products
select distinct id, property1, property2 from product
-- this would get you your drop downs by product
select id, size from product order by id
You can then build the same drop down hashtable by iterating through the second result set.
I'm not sure if this is the best way, but I've always approached this by altering the query so that it is sorted by product name. Then as you iterate through the rows, check to see if the product name matches the one you just processed. If it's the same, then this row is a different size of the same project.