SQLite GROUP BY lowest price missing results - sql

Question

I've got a SQLite DB that houses basically a list of products and prices.
CREATE TABLE brand_list (
id INTEGER NOT NULL,
brand_name TEXT NOT NULL,
PRIMARY KEY (id)
);
CREATE TABLE product_list (
id INTEGER NOT NULL,
prod_generic VARCHAR NOT NULL,
prod_pretty VARCHAR NOT NULL,
PRIMARY KEY (id),
UNIQUE (prod_generic)
);
CREATE TABLE qty_list (
id INTEGER NOT NULL,
qty_initial VARCHAR NOT NULL,
PRIMARY KEY (id)
);
CREATE TABLE vendor_list (
id INTEGER NOT NULL,
vendor_name TEXT NOT NULL,
vendor_url VARCHAR NOT NULL, url_pretty varchar not null default 1, datacollect varchar not null default 'Y',
PRIMARY KEY (id)
);
CREATE TABLE listing (
id INTEGER NOT NULL,
brand_id INTEGER,
qty_id INTEGER,
vendor_id INTEGER,
prod_id INTEGER,
qty_actual INTEGER,
price INTEGER,
unit_price FLOAT,
timestamp VARCHAR(255),
sale INTEGER DEFAULT 0 NOT NULL,
PRIMARY KEY (id),
FOREIGN KEY(brand_id) REFERENCES brand_list (id),
FOREIGN KEY(prod_id) REFERENCES product_list (id),
FOREIGN KEY(qty_id) REFERENCES qty_list (id),
FOREIGN KEY(vendor_id) REFERENCES vendor_list (id)
I pull a query looking at the listing table with the goal of grabbing the lowest price and listing all of the relevant information for end users (ie brands, products, qty, price, when it was pulled, where it was pulled from).
select qty_initial, qty_actual, price, MIN(unit_price), brand_name, prod_pretty, sale, timestamp, vendor_name from listing left join brand_list on listing.brand_id = brand_list.id left join product_list on listing.prod_id = product_list.id left join vendor_list on listing.vendor_id = vendor_list.id left join qty_list on listing.qty_id = qty_list.id group by prod_pretty order by brand_name, prod_pretty
What I'm seeing is that most of the data is showing up but not all of the data is showing up. And it doesn't seem consistent - but one brand might show seven out of eight entries, another brand might have everything.
The interesting part is that if I add a WHERE clause just prior to the group by and specify a brand that I know is missing entries in the original output, I'll get all of the expected output for that brand.
This isn't my world - I'm just hacking something together for my own fun. I'm certain I'm missing something fundamental here and it's likely around the GROUP BY. Any direction would be great - even a way to rethink how I'm doing this to get to where I want to be in a simpler manner.

Answer 1

Your query is malformed -- although SQLite does accept it. The GROUP BY columns are inconsistent with the SELECT list.
Instead, use window functions:
select qty_initial, qty_actual, price, unit_price, brand_name, prod_pretty, sale, timestamp, vendor_name
from (select qty_initial, qty_actual, price, unit_price, brand_name, prod_pretty, sale, timestamp, vendor_name,
row_number() over (partition by prod_pretty order by unit_price) as seqnum
from listing left join
brand_list
on listing.brand_id = brand_list.id left join
product_list
on listing.prod_id = product_list.id left join
vendor_list
on listing.vendor_id = vendor_list.id left join
qty_list
on listing.qty_id = qty_list.id
group by prod_pretty
) p
where seqnum = 1
order by brand_name, prod_pretty